|
1. |
Contents pages |
|
Chemical Society Reviews,
Volume 12,
Issue 4,
1983,
Page 007-008
Preview
|
PDF (90KB)
|
|
摘要:
Chemical Society Reviews Vol 12 No 4 1983 Page The Glass Transition: Salient Facts and Models By R. Parthasarathy, K. J. Rao, and C. N. R. Rao 361 Electron Spin Resonance of Haemoglobin and Myoglobin By L. Charles Dickinson and Martyn C. R. Symons 387 Hydrido Complexes of the Transition Metals By D. S. Moore and S. D. Robinson 415 High Resolution Laser Spectroscopy By G. Duxbury 453 1983 Indexes 505 The Royal Society of ChemistryLondon Chemical Society Reviews EDITORIAL BOARD Professor K. W. Bagnall, B.Sc., Ph.D., D.Sc., C.Chem., F.R.S.C. Professor K. R. Jennings, M.A., D.Phil., C.Chem., F.R.S.C. Professor G. W. Kirby, M.A., Ph.D., Sc.D.,-F.R.S.E., C.Chem., F.R.S.C. Professor G. Pattenden, Ph.D., C.Chem., F.R.S.C. Professor B.L. Shaw, B.Sc., Ph.D., F.R.S. Professor P. A. H. Wyatt, B.Sc., Ph.D., C.Chem., F.R.S.C. (Chairman) Editor: K. J. Wilkinson, B.Sc., M.Phi1. Chemical Society Reviews appears quarterly and comprises approximately 20 articles (ca. 500 pp) per annum. It is intended that each review article shall be of interest to chemists in general, and not merely to those with a specialist interest in the subject under review. The articles range over the whole of chemistry and its interfaces with other disciplines. Although the majority of articles are intended to be specially commissioned, the Society is always prepared to consider offers of articles for publication. In such cases a short synopsis, rather than the completed article, should be submit- ted to the Managing Editor, Books and Reviews Section, The Royal Society of Chemistry, Burlington House, Piccadilly, London, W 1V OBN.Members of the Royal Society of Chemistry may subscribe to Chemical Society Reviews at 514.00 per annum (beginning 1984, Z15.50); they should place their orders on the Annual Subscription renewal forms in the usual way. All other orders accompanied with payment should be sent directly to The Royal Society of Chemistry, The Distribution Centre, Blackhorse Road, Letch- worth, Herts. SG6 1HN, England. 1983 annual subscription rate U.K. f39.50, Rest of World f42.00, U.S.A. $85.00 (beginning 1984, f43.50,&45.50, $87.00 respectively). Air freight and mailing in the U.S.A. by Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. U.S.A. Postmaster: Send address changes to Chemical Society Reviews, Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. Second class postage paid at Jamaica, New York 11431. All other despatches outside the U.K. by Bulk Airmail within Europe. Accelerated Surface Post outside Europe. @ Copyright reserved by The Royal Society of Chemistry 1984 ISSN 0306-0012 Published by The Royal Society of Chemistry, Burlington House, London, W1V OBN Printed in England by Eyre & Spottiswoode Ltd, Thanet Press, Margate.
ISSN:0306-0012
DOI:10.1039/CS98312FP007
出版商:RSC
年代:1983
数据来源: RSC
|
2. |
Front cover |
|
Chemical Society Reviews,
Volume 12,
Issue 4,
1983,
Page 013-014
Preview
|
PDF (56KB)
|
|
ISSN:0306-0012
DOI:10.1039/CS98312FX013
出版商:RSC
年代:1983
数据来源: RSC
|
3. |
Back cover |
|
Chemical Society Reviews,
Volume 12,
Issue 4,
1983,
Page 015-016
Preview
|
PDF (33KB)
|
|
摘要:
Chemical Society Reviews Vol 12 No 4 1983 Page The Glass Transition: Salient Facts and Models By R. Parthasarathy, K. J. Rao, and C. N. R. Rao 361 Electron Spin Resonance of Haemoglobin and Myoglobin By L. Charles Dickinson and Martyn C. R. Symons 387 Hydrido Complexes of the Transition Metals By D. S. Moore and S. D. Robinson 415 High Resolution Laser Spectroscopy By G. Duxbury 453 1983 Indexes 505 The Royal Society of ChemistryLondon
ISSN:0306-0012
DOI:10.1039/CS98312BX015
出版商:RSC
年代:1983
数据来源: RSC
|
4. |
The glass transition: salient facts and models |
|
Chemical Society Reviews,
Volume 12,
Issue 4,
1983,
Page 361-385
R. Parthasarathy,
Preview
|
PDF (1684KB)
|
|
摘要:
The Glass Transition: Salient Facts and Models* By R. Parthasarathy, K. J. Rao, and C. N. R. Raoy SOLID STATE AND STRUCTURAL CHEMISTRY UNIT, INDIAN INSTITUTE OF SCIENCE, BANGALORE-56001 2, INDIA 1 Introduction Glasses constitute a class of amorphous solids prepared by melt-quenching and are distinguished by the unique transition, the so-called glass transition, that they exhibit. The glass transition is marked by more or less discontinuous changes from solid-like to liquid-like values in the second derivatives of the Gibbs free energy (e.g. the specific heat, C,) around a temperature, Tg,called the glass transition temperature; the first derivatives (e.g. volume, entropy) change continuously through Tg.Many models have been proposed to explain the glass transition but a clear picture of the nature of the transition has not yet emerged.An early review on the phenomenon of the glass transition by Kauzmann’ appeared in 1948. Since then, a large body of valuable experimental data has become available in the literature in respect of this transition. We have been investigating the glass transi- tion by employing various spectroscopic and other techniques for some time. In this article, we shall discuss the important experimental results on the glass tran- sition including the results from computer simulation studies, an aspect reviewed by Cohen and Grest2 in relation to a modified version of the free volume model. We shall then critically examine the different theoretical models for the glass transition. It would be particularly instructive if the phenomenological similarity between various disordered solids could be correlated with the existence of a structural motif common to these solids.We shall attempt to show in this article how the concept of clusters could provide such a unifying element. Disordered solids may be visualized as conglomerates of islands of more ordered material (density fluctuations) or clusters dispersed in a less ordered matrix or ‘connective tiss~e’.~ In the most general case, clusters may be differentiated from the tissue by a higher degree of correlation in position, orientation, or some other appropriate degree of freedom. In the context of glasses, the cluster-tissue material theme has had its genesis in the microcrystallite model, earlier workers having used terms such as ‘vitron~’~ to identify the ordered regions.It seems that the use of or ‘amorphon~’~ *Contribution No. 21 1 from the Solid State and Structural Chemistry Unit. tTo whom all correspondence should be addressed. * W. Kauzmann, Chem. Rev., 1948, 3, 219. * M. H. Cohen and G. S. Grest, Adv. Chem. Phys., 1981,48, 455. M. R. Hoare and J. Baker, in ‘The Structure of Non-Crystalline Materials’, ed. P. H. Gaskell, Taylor and Francis, London, 1976; M. R. Hoare, Ann. N. Y. Acud. Sci., 1976, 279, 186. L. W. Tilton, J. Res. Nut. Bur. Stand., U.S.A., 1957, 59, 139. R. Grigorovici, J. Non-Cryst. Solids, 1969, 1, 371. 36 1 The Glass Transition: Salient Facts and Models the cluster-tissue formalism may be useful in interpreting experimental data related to the glass transition.Recent studies have demonstrated that phenomena such as the glass transition, p-relaxations and so on, ascribed to solids with long-range positional disorder, are also found in solids that are characterized by disorder in other degrees of free- d~m.~,~This implies that any model of the glass transition as it occurs in the usual glasses, should be applicable to other disordered solids as well. A logical corollary is that such a model should not be restricted by the assumption of long-range positional disorder. We shall briefly discuss orientational glasses (or glassy crys- tals) obtained by quenching orientationally disordered (plastic) crystals towards the end of the article alongwith the so-called dipolar glasses.2 The Glass Transition The glass transition, unlike other phase transitions,8 exhibits unusual changes in the derivatives of the Gibbs free energy, G, around the transition temperature. In Figure l(a), we show the variation of entropy, S, or volume, I/, of a substance with temperature; a plot for the second derivatives of G (the heat capacity, Cp,expan-sivity, a, or compressibility, fi) is given in Figure l(b). It is evident that the supercooled liquid departs from the equilibrium line (broken) at a ‘fictive’tem-perature, T,, which depends upon the cooling rate. T, is usually estimated as the temperature at which the specific heat increases steeply, and is generally quite close to Tf. The response of a system to a perturbation from equilibrium is governed by a spectrum of relaxation times, zi,which define the times at which the equilibrium response to the perturbation will have decayed to l/e of its initial value.9 At low temperatures or at high frequencies of the applied field, the system cannot respond to the field within the time scale of the experiment. The shortest relaxation time of the system then exceeds the experimental time scale and the parameter associ- ated with that relaxation time remains constant during the experiment.lo The liquid, at this point (T, or Tf),is said to have fallen out of equilibrium or to display ‘solid-like behaviour’; zi itself shows an exponential temperature dependence.Considering that a glass can be described as a snapshot picture of a liquid, the glass transition can be defined as that event wherein the translational degrees of freedom (capable of inducing liquid-like configurational rearrangements) are activated. However, it is evident that the glass transition temperature observed in a given experiment is a function of the time scale employed‘ --the shorter the time scale, the higher is the Tg;it is therefore not obvious that the glass transition is a thermodynamic phenomenon. Empirically, T, is generally around 0.67 TF,where TFis the temperature of fusion.12 H. Suga and S. Seki, Faraday Discuss. Chem. Soc., 1980, 69; J. Non-Cryst. Solids, 1974, 16, 171. G. P. Johari, Phil.Mag., B, 1980, 41, 41; Ann.N. Y. Acad. Sci., 1980, 279, 117. C. N. R. Rao and K. J. Rao, ‘Phase Transitions in Solids’, McGraw Hill, London, 1978. R. 0. Davies and G. 0.Jones, Adv. Phys., 1953, 2, 370. lo C. A. Angell and W. Sichina, Ann. N. Y. Acad. Sci., 1976, 279, 53. C. A. Angell, Ann. N. Y. Acad. Sci., 1980, 282, 123. l2 S. Sakka and J. D. Mackenzie, J. Non-Cryst. Solids, 19’71, 6, 145. Parthasarathy, Rao, and Rao 2)I I I 11 Tf, Tf2 Tf T T Figure 1 Schematic variation with temperature of (a) entropy, S, volume, V, and (b) heut capacity, C,, expansivity, oc,and compressibility, P,fbr a typical glass-forming liquid. The Tf represent jctive temperatures with the lower one representing the slower cooling rate The decrease in heat capacity as the liquid vitrifies is interesting.If the transition at Tgwere purely relaxational, would a sufficiently slow cooling rate obviate the transition? Figure 2(a) shows the temperature dependence of the normalized en- 'I AS (a) CP=f Figure 2 (a) Schematic dependence on temperature of the normalized difference in entropji between glass and crystal for a typical glass-former. S, = entropy of fusion. (h) A schematic Kauzmann plot of C versus In T-for a glass f 0)and crystal (a).The hatched area is equal to the entropy of,fu:ion tropy difference between the solid and liquid.' Should no transition intervene, the entropy of the liquid would equal that of the solid well above absolute zero. This is unphysical in that it implies that a supercooled liquid has an entropy equal to that of an ordered solid.This problem, first studied by Kauzmann,' has been circumvented by postulating the existence of a transition at a temperature Towhich limits the decrease in entropy -the so-called 'thermodynamic' glass transi- 3-1ti~n.~, To is obtained using the construction' shown in Figure 2(b). The l3 M. H. Cohen and D. Turnbull, J. Chem. Phys., 1959, 31, 1164. l4 J. H. Gibbs and E. A. DiMarzio, J. Chem. Phys., 1958, 28, 373. l5 M. H. Cohen and G. S. Grest, Phys. Rev. B, 1980, 20, 1077. l6 C. A. Angell, J. Chem. Educ., 1970, 47, 583. 363 The Glass Transition: Salient Facts and Models hatched area represents melt entropy, which at To must equal the entropy of fusion. The line joining the points represents an experimental trace.The self-conflicting situation that would arise for T< To,in the absence of a transition at To,is often called the Kauzmann paradox. There has been extensive debate about the existence of a transition at To.The justification given for extend- ing the free energy curve on both sides of a second-order transition has been questioned.’ ’Kauzmann’ himself resolved the paradox by stating that the glass would crystallize at a temperature greater than To.According to him, this would be possible if the barrier to homogeneous nucleation decreases with temperature while that impeding molecular motion (and hence the maintenance of equilibrium) increases. A cross-over temperature, TK,where the two are equal may be expected and Kauzmann suggests that TK> To.Assuming the existence of a Kauzmann limit, Angelll* has developed the concept of an ‘ideal’ glass, wherein a glass is reckoned to be the more ideal, the closer Tgis to To.Ideality of a glass is thus a measure of the entropy frozen in at Tg(this should be zero if Tgequals To)and present indications are that ionic glasses may be amongst the most ideal.lg The temperature dependence of dynamic properties such as viscosity, q, fluidity,4, and ionic conductivity, A, is described by the empirical Vogel-Tammann-Fulcherzo (VTF) equation, * = $0 exp[B/(T -To11 (1) where $o, B, and Toare constants ($o is weakly temperature dependent) evaluated by a least-squares fit to the data. Setting Toequal to zero reduces the VTF equation to the Arrhenius equation.It is significant that Toso determined is often equal to the To obtained from calorimetric data.21 This finding has been taken to support the notion that Tomay mark the occurrence of a transition. The value of q at To is formally infinite so that Tois said to be the lower limit of the liquid regime. Tg is also defined as that temperature at which 7 is equal to 10l3 P, the ‘isoviscous’ definition of the glass transition. 3 Experimental Studies of the Glass Transition Viscosities of many glass-forming melts, both organic and inorganic, have been measured as functions of temperature.2~z2 For T > Tg,q shows an Arrhenius dependence while at temperatures close to Tg,the dependence is often non-Arrhenius. In some case^,^^'^^ reversion to Arrhenius behaviour is noted close to l7 L.V. Woodcock, in ‘The Structure of Non-Crystalline Materials’, ed. P. H. Gaskell, Taylor and Francis, London, 1976. C. A. Angell, J. Am. Ceram. SOC.,1968, 51, 117. l9 K. J. Rao in ‘Preparation and Characterization of Materials’, ed. J. M. Honig and C. N. R. Rao, Academic, New York, 1981. zo H. Vogel, Physik. Z., 1921,22,645; G. Tammann and W. Hesse, 2.Anorg. Allg. Chem., 1976,56,245; G. S. Fulcher, J. Am. Ceram. Sor., 1925, 77, 3701. 21 A. J. Easteal and C. A. Angell, J. Chem. Phys., 1972, 56, 4231; H. Tweer, H. Laberge. and P. B. Macedo, J. Am. Ceram. SOC.,1971, 54, 121; C. T. Moynihan, L. R. Smalley, C. A. Angell, and E. J. Sare, J. Phys. Chem., 1969, 73, 2287. 22 A.Bondi, ‘Physical Properties of Molecular Crystals and Glasses’, Wiley, New York, 1968. 23 R. Weiler, S. Blaser, and P. B. Macedo, J. Phys. Chem., 1969, 73, 4147; P. B. Macedo and A. Napolitano, J. Chem. Phys., 1968, 49, 1887. z4 A. C. Ling and J. E. Willard, J. Phys. Chem., 1968, 72, 1918. Parthasarathy, Rao, and Rao Tg. Heat capacity measurements indicate that Tg is greater than the Debye tem- perature since the value of Cp at Tg is nearly always of the order of the Dulong-Petit value of 3NR, where N is the number of independently vibrating particles. * An interesting correlation has been found between the temperature dependence of q and of AC, of some glasses; strong network glasses such as SiO, and GeO, show a small change in C, at Tgand their viscosities show an Arrhenius behaviour over large ranges of q.l0 Ionic glasses such as acetate,1° nitrate,1° and ~ulphate~~glasses show pronounced changes in C, at T, and marked non- Arrhenius variation of viscosity (for the first two) around T,.Glassy ZnCI,, which is of intermediate character, shows only a moderate change in Cp and a small deviation from Arrhenius behaviour. A. Relaxation Methods.-If the glass transition is indeed relaxational, relaxation spectroscopy would be particularly useful for studying the phenomenon. 26 Gold-stein and co-w~rkers~~ have found sub-T, dielectric loss peaks (8-relaxations) in several glasses. Since 8-relaxations are present both in glasses containing rigid molecules as well as flexible ones, they may be regarded as a consequence of amorphous packing.Recent work2*- 30 has shown that 8-relaxations exist in ‘glasses’ formed by supercooling liquid crystals and plastic crystals. The presence of these relaxations therefore seems to be less dependent on the nature of long- range disorder than supposed earlier. In covalent network inorganic glasses, dis- persions in dielectric constant and dielectric loss are observed near Ts where large scale reorganization of the glassy matrix OCCU~S.~ Mechanical relaxations under low frequency to 1 rad s-’) alternating stress fields have been studied in multi-component commercial glasses.32 A loss peak appears at T > Tgin all the systems so far studied.32 The systems studied to date are thermorheologically simple and the relaxation function shifts along the time axis with temperature, but shows no change in In silicate glasses,34 a single mechanism seems to be responsible for both flow and relaxation at T > Tg.Perez et aI.35have, however, found the existence of a double relaxation for T < T,; the a-and 8-relaxations appear to correspond to viscous flow and delayed elasticity respectively. Ultrasonic relaxation phenomena of glass-forming melts can be divided into LS K. J. Rao, Bull. Mat. Sci.,1980, 2, 357. 26 J. Wong and C. A. Angell, ‘Glass: Structure by Spectroscopy’, Marcel Dekker, New York,1976. 27 G. P. Johari and M. Goldstein, J. Chern. Phys., 1971, 55,4245; L. Hayler and M. Goldstein, J. Chern. Phys., 1977, 66, 4736; J.Haddad and M. Goldstein, J. Non-Cryst. Solids, 1978, 30,1. G. P. Johari and J. W. Goodby, J. Chem. Phys., 1982,77, 5:65; G. P. Johari, J. Chem. Phys., 1982, 77, 4619. 29 H. R. Zeller, Phys. Rev. Lett., 1982, 44, 583. 30 R. Parthasarathy, K. J. Rao, and C. N. R. Rao, J. Phys. Chem., 1983, in press. 31 J. M. Stevels, ‘Handbook der Physik’, Vol. 20, Springer, Berlin, i957. 32 A. Zdaniewski, G. E. Rindone, and D. E. Day, J. Mater. Sci., 1979, 14, 763. 33 S. M. Rekhson, J. Non-Cryst. Solids, 1980, 38, 457 and 1980,39, 457; C. R. Kurkjion, Phys. Chem. Glasses, 1963, 4, 128. 34 J. J. Mills, J. Non-Cryst. Solids, 1974, 14, 255. 35 J. Perez, D. Duperray, and D. Lefevre, J. Non-Crysr. Solids,1981, 44, 113. The Glass Transition: Salient Facts and Models three categorie~.~~ In organic liquids, the real, K', and the imaginary, K", com-ponents of the bulk modulus, K, do not in general, fit single relaxation time formulations and the relaxation time distribution function is not a symmetric function of In z.In molten oxides, the single relaxation time behaviour at T > Tg broadens into a distribution symmetric in lnz as the temperature is lowered through Tg.In molten salts, however, the departure from single relaxation time behaviour occurs well above Tg.36The relation between the dielectric relaxation time, zD,and the shear relaxation time, zs, has been explored.37 The ratio of (z,}, the conductivity relaxation time (which is more appropriate for conducting mate- rials than zD), to (zs) in ionic salts varies from near unity well above Tgto about 10-15 just above Tg.37In covalent materials the ratio is 104-10s at low tem- peratures and could be as high as 1O'O in fast ion conducting glasses3* The divergence of the ratio through Tgindicates that fundamentally different processes are likely to be involved in the two cases in the glassy state.B. Spectroscopic Methods.-Resonance spectroscopy yields valuable information about structure as well as particle dynamics.26 The latter feature has, however, found little application in glasses since the static broadening due to site distribu- tions, for example, often obscures the more subtle dynamical effects. Molecular shapes of glass formers are also often asymmetric, rendering model-based interpre- tation difficult.It is recognized that resonance spectroscopy is ideally suited to probe changes in local site symmetry and hence valuable in investigating the glass transition. A Raman of a Ca(NO,),-KNO, glass has shown that band- widths change through Tg, with the depolarized bandwidth increasing more rapidly than that of the polarized line at T > Tg.Raman bandshape analyses of glassy methyl salicylate carried out in this laboratory show that the vibrational correlation time, zv, is practically constant through Tg,while the reorientational correlation time, zR,decreases sharply in this region. It is possible that the glass transition is associated with the activation of roto-diffusional modes. An early variable temperature n.m.r.study on o-terphenyl has shown40 that there is a marked increase in the relaxation and correlation times as the super- cooled liquid vitrifies. The times do not reach values characteristic of a rigid lattice until T > Tg.This may be due to the arrest at Tgof the long range diffusion and rotation characteristic of the supercooled liquid; local relaxation in statistically distributed regions of lower density may, however, continue to lower tempera- ture~.~~A recent n.m.r. hole-burning experiment with glycerol has shown that relaxation around Tgoccurs through large angle (-45 ") steps and not through continuous diff~sion.~' llB n.m.r. studies of glassy B203 and B2S3has indicated 36 C. A. Angell and L. M. Torell, J.Chem. Phys., 1983, 78, 937. 37 F. S. Howell, R. A. Bose, P. B. Macedo, and C. T. Moynihan, J. Phys. Chem., 1974, 68, 739; C. T. Moynihan, N. Balitactac, L. Boone, and T. A. Litovitz, J. Chem. Phys., 1970, 55, 3013. 38 S. I. Smedley and C. A. Angell, Solid State Commun., 1978, 27, 1; Mat. Res. Bull., 1980, 15, 421. 39 C. A. Angell 'Vibrational Spectroscopy of Molecular Liquids and Solids' ed. S. Bratos and R. M. Pick, Plenum, New York, 1980; J. H. R. Clarke and S. Miller, Chem. Phys. Left.. 1972, 13, 97. 40 D. W. McCall, J. Chem. Phys., 1973, 47, 530. 41 P. L. Kuhns and M. S. Conrad], J. Chem. Phys., 1982, 77, 1771. Parthasarathy, Rao, und Rao that a narrow line emerges for T > Ts in B,O,, corresponding to the motionally narrowed powder pattern.42 In B,S,, this line overlaps the powder pattern, a feature that the authors have explained using a cluster model of glass.Due to the low Tg(-380K), the B2S3molecular units in the less dense tissue material are expected to rotate readily giving rise to the motionally narrowed line.42 A rise in temperature would increase the fraction of such units and, hence, the intensity of the spectral line. Another possible explanation for the simultaneous presence of both patterns would be in terms of an equilibrium between three- and four-co- ordinated boron atoms.43 E.s.r. studies have shown that there is motional narrow- ing and change in lineshape as the glass transforms into a The available relaxation theories are rather difficult to apply to glasses so that the few studies reported in this area have employed lineshape simulation to explain the experi- mental observations.In their e.s.r. study of glycerol-water mixture using the ‘spin-probe’ peroxy- lamine disulphonate (PADS) Antsiferova et reported that the relation be- tween the correlation time T~,and q/T deviated from linearity in the region of slow tumbling. The activation barrier to tumbling was found to be cu. 50 kJ mol-’. In a study of o-terphenyl glass using vanadyl(1v) chelates as probes, Mat~unaga~~ found a sudden decrease in the average value of the tumbling angle for T > Tg. Molecular motion appears to be anisotropic and the results seem to substantiate the presence of interlocked clusters in the melt (‘the cogwheel effect’) around Tg.A spin-probe e.s.r. study on some glass-forming liquids carried out in this laboratory47 showed a marked decrease in the spin correlation time, zc, at a temperature, Tk,Tk 2Tg,as the glass was warmed (Figure 3). This seems to indicate that the spin probe is immobilized within clusters in the glass; these clusters would dissolve at Tk.The difference between Tgand Tk depends on the liquid and seems to be related to the intermolecular forces present. An analogous study by Spielberg and Gelerinter4* shows similar features and these authors suggest that translational, but not rotational diffusion is inhibited as the tempera- ture is lowered through Tg. E.s.r. studies of MnZ + and Fe3+ in silicate glasses show that resonance intensity decreases with tem~erature.~~ There seems to be no change in the linewidth even for T> Tg.The rate of the decrease appears to be larger than what might be expected from the Boltzmann term alone.It has been that the observed decrease in the intensity of the Fe3+(g = 4.3) resonance may be due to the disap- pearance of the orthorhombic distortion associated with this signal; the intensity of the isotropic g = 2.0 resonance remains nearly constant. We have carried out 42 M. Rubinstein, Phys. Rev. B, 1976, 14, 2778. 43 J. R. Hendrickson and S. G. Bishop, Solid State Commun., 1975, 17, 301. 44 ‘Electron Spin Relaxation in Liquids’, ed. L. D. Muus and P. W. Atkins, Plenum, New York, 1972. 4s L. I. Antsiferova, N. N. Korst, V.B. Strynkov, A. N. Ivanova, N. S. Nazemets, and N. V. Rabin’kina, Mol. Phys., 1978, 25, 909. 46 Y. Matsunaga, Bull. Chem. Soc., Jpn., 1977, 51, 422. 47 R. Parthasarathy, K. J. Rao, and C. N. R. Rao, J. Phys. Chem., 1981, 85, 3085. 48 J. I. Spielberg and E. Gelerinter, J. Chem. Phys., 1982, 77, 2159. 49 F. Momo, G. A. Ranieri, and A. Sotgui, J. Non-Cryst. Solids, 1981, 46, 115; E. Baiocchi, A. 367 The Glass Transition: Salient Fucts and Models F I F--8 r E -8 3 , A J-= -9 -1 --9 -eu (r 0 4 -10 --10" -1 d , I I I 0.6 0.9 1.2 1.5 TgIT Figure 3 Plot of log 5, against TJTfor (A) glycerol; (B) o-toluidine; (C) methyl sulicylute: (D) propylene carbonate; (E) dimethyl phthalate and ( F) p-anisaldehyde.The broken line indicates Tg (After R. Parthasarathy, K. J. Rao, and C. N. R. Rao, 1981, Ref. 47) e.s.r. spectroscopic studies of ionic and covalent glasses containing Fe3 + and Mn2+ through the glass transition temperature. We find that the anomalous decrease in signal intensity at T > Tg can be related to the marked decrease in configurational entropy expected at Tg.Changes in the dielectric constant may also affect e.s.r. resonance intensity, particularly at the glass transition. Mossbauer spectroscopic studies (using 7Fe nuclei) have been conducted as a function of temperature in some glasses, both organi~~~~~~ Theand inorgani~.~~~~ main experimental finding from these studies, particularly those on organic glasses, is that the Lamb Mossbauer factor (the recoil-free fraction) decreases sharply (along with a change in the quadrupole splitting) near Tg.Diffusion appears to be significant only at temperatures greater than Tg.Fe Mossbauer studies of inorganic glasses carried out in this laboratoryss show that a decrease in the R. Parthasarathy, K. J. Rao, and C. N. R. Rao, Chem. Phys., 1982, 68,393. 51 F. J. Owens, C. P. Poole, and H. A. Farach, 'Magnetic Resonance in Phase Transitions', Academic, New York, 1979. s2 P. A. Flinn, B. J. Zabransky, and S. L. Ruby, J. Phys. (Paris), C6, 1976, 37, 739; S. L. Ruby, B. J. Zabransky, and P. A. Flinn, J. Phys. (Paris), C6, 1976,37,745; A. Vasquez and P. A. Flinn, J. Chem. Phys., 1980,72, 1958, J. A. Eliott, H. E. Hall, and D.St. P. Bunbury, Proc. Phys. SOC.,1966,89, 595. 53 D. C. Champeney, Rep. Prog. Phys., 1980, 42, 1017. 54 J. P. Gosselin, U. Shimony, L. Grodzins, and A. R. Cooper. Phys. Chem. Glasses, 1968, 8, 56. s5 S. Bharati, R. Parthasarathy, K. J. Rao, and C. N. R. Rao, Solid State Commun., 1983, 46,457. 368 Parthasarathy, Rao, and Rao 0.8 0.6 I I I I 400 500 600 700 T(K) Figure 4 Variation of the normalized area under resonance with temperuture in a 57Fe Mosshauer study of (a) borate (h) borosilicate and (c) vanadute glasses. Inset: A similar plot for glassy glycerol FeCl (data from Elliott et al., Ref. 52). Points are experimental data while the lines are theoretical3ts obtained using the cluster model of the glass transition (Section 5)The arrows denote T (After Seeta Bharatiget al., 1983, Ref.55) recoil-free fraction is generally found around Tg(Figure 4).Such a behaviour may be due to the onset of librational motion or of diffusional motion and can be explained, using the cluster model of the glass transition in terms of mode softening at Tg.It may be noted that the concept of soft modes has been most useful in understanding phase transitions in crystalline substances.* Extended X-ray Absorption Fine Structures6 (EXAFS) has been useds7 to study s6 R. Parthasarathy, P. R. Sarode, K. J. Rao, and C. N. R. Rao, Proc. Indian Nat. Sci.Acad., Sect. A, 1982, 48, 119. 51 J. Wong and F. W. Lytle, J. Non-Cryst.Solidv., 1980, 37,273. The Glass Transition: Salient Facts and Models the glass transition in GeO, and ZnC1, in order to observe the possible evolution of disorder.An increase in the vibrational disorder of ZnC1, was noticed at the Tg. ZnCI, is known to be a weak structural analogue of GeO, and hence shows more disorder, even at lower temperatures. The experimental results presented hitherto suggest that the glass transition can be associated with the emergence of degrees of freedom in the supercooled liquid that are inactive in the glass. The manner in which these motional modes are activated is not very clear but a probable mechanism may be described in terms of a cluster model. There is some evidence that local motion in the connective tissue is frozen out only at T < Tg;upon reheating the glass, these are the modes which would be excited first, leading, in suitable cases, to the simultaneous presence of both ‘cluster’ and ‘tissue’ contributions to the spectrum.Glasses are known to possess anomalous properties at low temperatures. Some of these are:58 (a) an excess specific heat proportional to T, (b) thermal conduc- tivity that increases as T2,and (c) ultrasonic velocity proportional to In T. The model used to explain these phenomena suggests that these are due to the presence of some structural entity of the glass in a double-well potential in which tunnelling takes place.59 The number of such particles appears60 to be proportional to (Tg)--’ or (Tf)-’.There are as yet no data available on the dependence of such states upon the thermal history of the sample,61 and the identity of the tunnelling particles has not been established.Nevertheless, the dependence of Tg on the number of such particles is taken to reflect the structure frozen in at Tg.58,60.61 4 Ideality of Glasses An issue central to the study of the glassy state has been that of defining the thermodynamic state of a glass -in particular, the so-called amorphous ground state. We noted earlier that an ideal glass can be described in Kauzmann’s terms as a glass whose Tg is equal to To, i.e., one which contains no frozen-in configurational entropy.’ Such a glass may also be said to be in the ‘amorphous ground state’ and one may then ask if a distinction can be drawn between such an amorphous ground state and the crystalline state.It has been suggested by Kauz- mann that TKis usually greater than To so that there cannot be a distinct amor- phous ground state. The alternative is to consider TK < Towhich can lead to such a state. The work of Donnella and Ange1P2 hints at the possibility of such a situation which one may also expect to encounter in atactic polymers. The problem essentially reduces to one of deciding which alternative is physically more plausi- ble: a distinct amorphous ground state (TK< To)or a liquid-to-crystal transition at TK,TK G TF,which is implied by the condition, TK > To. It is interesting to regard the existence of a unique temperature, To,as a conse- quence as much of geometric, as of energetic considerations.’* In that case, To 58 M.H. Cohen and G. S. Crest, Solid State Commun., 1981, 39, 145. 59 P. W. Anderson, B. I. Halperin, and C. M. Varma, Philos. Mug., 1972, 25, I; W. A. Phillips, J. Low Temp. Phys., 1972, 7, 351. 6o A. K. Raychaudhuri and R. 0. Pohl, Solid State Commun., 19x1, 37, 105. 61 C. L. Reynolds, J. Non-Cryst.Solids, 1980, 37, 125. 62 J. Donnella and C. A. Angel], J. Chem. Phys., 1977, 67, 4560. Parthasarathy, Rao, and Roo would be that temperature at which the close-packing limit is reached. This high density limit was first described by BernaP3 in his experiments with mechanical assemblies, but computer simulation experiments have recently been shown to produce amorphous packings of marginally higher den~ity.~~T,would then be that temperature at which the excess volume disappears, but volume does not appear to be a significant parameter in determining the occurrence of the glass transition (see Section 7).An answer to the question of the possible existence and nature of the amorphous ground state may be expected from computer simulation studies. These may also answer the more general question of whether a glass can exist in a definite thermo- dynamic state, or whether it is merely trapped, metastably, in a local potential minimum. The former would imply the possibility of phase transitions from one glassy state to another at well-defined temperatures, but there is no evidence to date for such phenomena. In a later section we shall examine the implications of the cluster model in this regard.5 Computer Simulation Studies Simulation of local structures of liquids has had its genesis in the classic work of BernaP3 who used mechanical assemblies to this end. Studies by Tilt~n,~ Grigoro-vi~i,~and Hoare3 have shown the importance of non-space filling symmetries such as the pentagonal and the icosahedral to the ‘ordered’ aggregates in the amorphous state (Figure 5). Such aggregates can indeed be far larger than originally consi- dered.3 The glass transition from this viewpoint is regarded as the interlocking or ‘congelation’ of such clusters followed by the gradual freezing out of the tissue materiaL3 The role that dynamic simulation can play in describing the glass transition has been recognized re~ently.~~.~~ The use of high quenching rates and (often) of small sample sizes limit the applicability of the results.Nonetheless, these studies yield useful information on the prototype glass transition in simple liquids, particularly because the transition at Tgis known to be a non-equilibrium transition that is closely related to the cooling rates employed. The diffuse nature of the simulated glass transition requires the definition of ‘upper’ and ‘lower’ glass transition tem- peratures, but more often, Tg in these studies is quite simply defined as the intersection point of glassy and liquid density-temperature plots.65 The high fictive temperatures of the simulated glasses, however, imply high diffusion rates. Both hard and soft spheres can be compacted into amorphous assemblies but the typical discontinuity in Cpis absent in these cases.65 The C, jump is apparent with the Lennard-Jones (LJ) potential and the attractive component of the potential appears to be responsible for this feat~re.~’ This also leads one to expect that the 63 J.D. Bernal, Nature, 1960, 185, 68, Proc. R. Soc. London, Ser. A, 1964, 280, 299. 64 L. V. Woodcock and C. A. Angel], Phys. Rev. Lett., 1980, 47, 1129. 65 C. A. Angel], J. H. R. Clarke. and L. V. Woodcock, Adv. Chem. Phys.. 1981, 48, 397. O6 D. Frenkel and J. P. McTague, Ann. Rev. Phys. Chem., 1980, 31, 49. 67 J. H. R. Clarke, J. Chem. Soc., Furaduy Truns. 2, 1979, 75, 1371. 371 The Glass Transition: Salient Facts and Models Figure 5 Cluster obtained spontaneously on cooling a simulated Lennard-Jones liquid drop of 129 atoms in a molecular dynamics study.The heavy lines indicate elements of pentagonal symmetry(After M. R. Hoare and J. Barker, 1976, Ref. 3) Cpanomaly would be difficult to observe under pressure as is, in fact, the case.68 In a system of soft spheres, diffusivity, D,is proportional to a reduced volume at higher temperatures, consistent with the variation in the heat capacity.65 At lower temperatures though, a recent study using a hard sphere ensemble shows that D does not vary linearly with free volume.64 Experimentally, the variation of D with volume is exponentially dependent upon the free volume, or fluidity as described by the Doolittle equation,69 4 = A exp[ -B/(V -c)] (2) where and A are constants with the dimensions of volume.Woodcock and Ange1P4 have found that is the Bernal dense random packing of hard spheres limit. These workers were able to produce liquids of higher density and the significance of the Bernal limit is not entirely clear. Radial or pair distribution functions have also been used to characterise Tg,at which point the ratio of the intensity of the first peak to the first trough is found to show a change in slope.7o This is probably due to the change in density (which determines the height of the first peak) and to the commencement of diffusion (to T. Atake and C. A. Angell, J. Phys. Chem., 1977, 81, 232. 69 A. K. Doolittle, J. Appl. Phys., 1951, 22, 1471.'O H. R. Wendt and F. F. Abraham, Phys. Rev. Lett., 1978, 41, 1244. 372 Parthasarathy, Rao, and Rao which the depth of the first trough is related). The well known splitting of the second peak in the radial distribution function appears as a liquid is compacted, though it has not been observed with soft interaction potentials. The distant component of the split peak is more intense, probably due to the effect of diffusion. An intriguing result is the discovery of bond orientational fluctuations above Tg suggesting a broken icosahedral symmetry in a molecular dynamics study of a Lennard-Jones fluid by Steinhardt et al.7 Inspite of such simulation studies, some important questions yet remain un- answered. First, it is not clear whether an underlying phase transition does exist.Secondly, properties of the system at temperatures close to Tg have not been explored in detail. Finally, the heating run of a glass remains to be simulated -the existence of the C, overshoot, the hallmark of the laboratory glass transition, has not yet been demonstrated in simulated glasses.65 6 Models for the Glass Transition A. Free Volume Model.-This model was first developed for liquids and was thereafter used to describe the glass transition in polymers by Fox and F10ry.~~ Turnbull and Cohen13*73 proposed a generalization of this model to glass transi- tions in other materials. Essentially, the model rests on the following assumptions: (a) it is possible to associate a free volume v with each particle; (b) when v exceeds some value vc, the excess may be regarded as free; (c) transport occurs only when voids of volume greater than some v* (v* -vm) form by redistribution of free volume and (d) no local free energy is required for the redistribution of free volume.The model concludes that a liquid can be vitrified only if it can be sufficiently undercooled without crystallization. It relates dynamic quantities, e.g.diffusivity, to the free volume, vf, so that these vanish at To where vf is zero. Since vf cc (T -To)in this approach, the VTF equation (equation 1) is easily derived. Cohen and Grest2*15,58 have refined this model further in order to evaluate thermodynamic properties. In this refinement, individual cells are described as being either liquid-like or solid-like, depending on whether or not their volume exceeds a certain volume v,.Only liquid-like cells have free volume,' and the change in free energy associated with redistribution of free volume arises from the change in entropy due to such redistribution. It is crucial to note that exchange of free volume is possible only between liquid-like cells. A further restraint on such exchange is that a given liquid-like cell must have at least a certain number of near neighbour liquid-like cells so that the volume of neighbouring solid-like cells is not constrained to change simultaneously. This defines the underlying percolation problem. The fraction of liquid-like cells is, P = P(V)dV (3) VCi P. J. Steinhardt, D.R. Nelson, and M. Ronchetti, Phys. Rev. Lett., 1981, 47, 1297. l2 T. G. Fox and P. J. Flory, J. Polym. Sci., 1954, 14, 315. 73 D. Turnbull and M. H. Cohen, J. Chem. Phys., 1961, 34,1120; J. Chem. Phys., 1970, 52, 3038. 373 The Glass Transition: Salient Facts and Models For p # 0 there are clusters of liquid-like cells each one of which has at least z liquid-like neighbours. For p larger than a value pcz,an infinite cluster exists. With this model, Cohen and Grest2 view the thermodynamic glass transition as a first order transition at which p changes discontinuously from p1> pcz to py < pcz. Nevertheless, molecular dynamics studies have shown that the percolation transi- tion need not be first order, but this finding could be an artefact of the non- equilibrium ensembles, inevitably found in simulated glasses.74 The equilibrium transition is associated with the decrease in p with temperature. Such a decrease demands an activated exchange of volume between solid-like and liquid-li ke cells.This is progressively frozen out as T -P Tgat a value that depends on the cooling rate, so that below the glass transition, p does not attain its equilibrium value but remains at pfroz.The presence of liquid-like cells in the glass even for T < Tymakes it possible to discuss sub-glassy relaxations in terms of this model. B. Configurational Entropy Model.-This model was first developed by Gibbs and DiMarzi0~~9~~in the context of glass transitions in polymers. These authors relate the configurational entropy, S,, to the number of configurations available to the system. In polymers, S, can be calculated by considering the entropy due to the mixing of holes and links on a model lattice.The major conclusion of this approach is that a solution of the Kauzmann Paradox requires a transition at a temperature To where the configurational entropy vanishes. First, a region of z particles is rigorously defined as that region that can undergo a transition to a new configuration without requiring a simultaneous change on or outside its boundary. Transport properties can then be calculated, assuming the existence of these small co-operatively rearranging regions; the dependence of z on temperature yields the temperature dependence of the property itself.76 At To, z is of the order of a macroscopic portion of the sample itself, and the dearth of configurations at this temperature is responsible for the observed sluggishness of mass transport.Dy- namic properties can be shown to be related to the configurational entropy as76 $ (73 = ’4 exPP/TS,I (4) where A and B are constants. It is evident that the size of the co-operatively rearranging regions is determined by configurational restrictions related to melt entropy. C.Cluster Model.-The notion that glass may consist of ordered aggregates has been prevalent since Tammann.77 Recent studies by Hoare and co-workers3 sug- gest the possibility that these clusters may be embedded in a more ordered, less dense, ‘tissue material’.The cluster model of the glass transition proposed from this laboratory7* suggests that the size of the clusters decreases with increasing 74 Y. Hiwatari, J. Chem. Phys., 1982, 76,5502. l5 J. H. Gibbs, ‘Modern Aspects of the Vitreous State’, ed. J. D. Mackenzie, Butterworths, London, 1963. l6 G. Adams and J. H. Gibbs, J. Chem. Phys., 1965,43, 139. ’’ G. Tammann, ‘Der Glaszustand’, Leopold Voss, Berlin, 1933. K. J. Rao and C. N. R. Rao, Muter. Res. Bull., 1982, 13, 1337. 374 Parthasarathy, Rao, and Rao temperature due to the surfacial melting of the clusters. The relative size, r/r, of the clusters is used as an order parameter, (; the cluster model is the first model to describe the glass transition in terms of an order parameter.Cluster melting would increase the tissue fraction and, at Tg,the clusters would be about the size of critical nuclei (or ‘flickering’ clusters) which would spontaneously dissolve. The variation in the configurational properties arises from particle motion in the tissue material which is supposed to consist of extremely anharmonic potential wells as in Figure 6.With this qualitative picture in mind, we proceed to analyse the cluster model in some detail. Figure 6 Schematic diagram of 3-connected potential wells. Broken lines indicate the an- harmonic pseudo-wells. V is the potential energy The total volume of the glass is given by, ’g = ‘cluster + ‘tissue (5) Now, the particles on the cluster surfaces are regarded as vibrating in highly anharmonic potential wells where the well depth, V, is given by V = 4 kX2 -j ckx3 (6) where k is the vibrational force constant, x is the displacement co-ordinate and c is the anharmonicity constant. With increase in temperature, a particle in a given well is excited to a higher vibrational state.Owing to the anharmonicity of the potential, the wells get connected. A modified pseudo-well, caused by the coalescing of two anharmonic wells, is described by V’= + k’(x + x’)’ -f C’ k’(x + x’)~;C’ 6 c (7) At still higher temperatures, motion in the tissue material is even more an- harmonic, but now the surfaces of the clusters begin to ‘melt’. The dislodged The Glass Transition: Salient Facts and Models particles fall into the ground vibrational state.It is assumed for simplicity that the ground state population does not vary with temperature till the clusters dissolve completely. In other words, if N, is the total number of particles and fois the fraction in the ground state, This is the key assumption in our simple mathematical treatment (and not of the model). A canonical partition function, z, may now be written for the ensemble of particles: n z = exp(-AEi/kT) (9) i= 1 so that the population in the ith level is given by, f; =exp (-AE,/kT) (10) We use a heuristic scheme for calculating the Ei, AE, =AE,xl,fi I where AE, is the separation between the ground state and the first excited state; it is apparent that the separation between levels progressively decreases with increasing i.The normalized variation with temperature of the order parameter is given by, where r is the radius of the cluster, ro is its initial radius and a is the initial ratio of Ve,the total volume, to V, (a = 1 for an ideal glass in this model). The variation of the order parameter with respect to kT/AE, is shown in Figure 7. Clearly, the temperature at which the rate of descent is steepest is a function of both a and i, the number of levels used in the calculation. It is also clear that r,, the critical cluster size, tends to zero in first order fashion. It is clear from the above discussion that particles in the higher vibrational states execute large amplitude oscillations. If there is a simultaneous change in the co-ordinates of the environment, memory of the initial position of the particle is lost and the particle will essentially have undergone diffusive motion. Because of such a close correspondence between the vibrational state of a particle and its dynamics, it is possible to calculate configurational properties based on the popu- lations in the different states.Writing the total enthalpy involved in the excitation, AH,, as, AH, = V,/V,1Ef; (1 3) I We differentiate AH, with respect to Tto obtain Cp.Subtracting the specific heat Parthasarathy, Rao, and Rao Figure 7 The variation of the normalized cluster radius, (, with temperature. The upper value in each rectangle is a, and the lower, the number of levels used in the calculation (After K.J. Rao and C. N. R. Rao, 1982, Re$ 78) due to excitation to the first excited state, 6C,, from C,, we obtain the configurational heat capacity, CpCOnf = c, -6C, (14) A plot of CpConfversus temperature is shown in Figure 8 for various values of the input parameters. Configurational entropy, S,, is given by n sc = <K/KJ c RA1n.h (15) i= 1 so that using equation (4), dynamic parameters may be determined. The dimensionless parameter, t, describes the radius of an ordered aggregate and may thus be used as an order parameter. It also defines a region in which there is positional correlation, permitting phonon propagation. It is thus analogous to a correlation length and can be used in Landau’s expansion of the free energy to show that the cluster model belongs to the class of mean free models.The salient aspects of this model are (a) the order parameter, t, shows the expected behaviour, tending to zero as T 4Tg.(b) When the cluster radius reaches a critical value, V,, the clusters dissolve spontaneously or t, = (rJr) -+ 0 in first order fashion, a feature that has been considered to be quite realistic.2’1s*s8 (4 A distribution of cluster sizes will not vitiate the conclusions given here which were drawn assuming a uniform cluster size. A distinctive type of mode softening is indicated in the cluster model, described as follows. It is known that the vibrational frequency miis related to the force The Glass Transition: Salient Facts and Models g 4.0 20 0“ 2.0 10-1 Log T Figure 8 Plot ojC ‘Onf versus log temperature.The parameters used in the calculation are a, dE,(kJ mol-l), i =” 1.8, 4200, 10(A); 1.8, 4200, 4(B); 1.8, 4200, 8(C); 1.8, 2100, 6(D); 1.8, 2100, 4(E) constant, ki, of the bond as wi cc Jk,. Considering the two-well case shown in Figure 6, f k2(X + x’)’-3k24x2= 2k,.~’= kx2 -f ck.x3 (16) Then, For the n-well case, This implies that the soft mode does not steeply decrease at the glass transition, but that it decreases in discrete steps. This is an important difference between the glassy ‘soft mode’ and the soft mode in crystals. Another difference between the two is the absence, for the glass transition, of a restoring mode in the transformed phase, the liquid. D. Other Approaches to the Glass Transition.-The ‘bond-lattice’ model of Angell and R~o’~ was one of the first models to study the glass transition as a solid state phase transition.These authors consider a glass as a lattice of bonds each of which may be ‘on’ or ‘off. Configurational heat capacities were calculated using this model and it was found that experimental C, versus T plots could be matched 7g C. A. Angell and K. J. Rao, J. Chem. Phys., 1972,52,470; K. J. Rao and C. A. Angell, ‘Amorphous Materials’, ed. R. W. Douglas and B. E. Ellis, Wiley, New York, 1971. Parthasarathy, Rao, and Rao theoretically without requiring the existence of a transition at To.The significance of Toin this treatment is of an operationally defined quantity which is related to a simple temperature dependence of excess thermodynamic quantities.In the ‘potential barrier’ model, Goldsteinso regards glass as a system trapped kinetically in a potential energy well of depth, U, accessible to the equilibrium liquid at Tg. The number of such wells, D (U), increases with U so that the transition from deeper to shallower wells as Tg is approached is favoured by entropic considerations. The softening of vibrational modes and the concomitant gain in the amplitude of the P-relaxations also favour the transition between wells. The model is qualitative so that comparisons with experimental results are not easily made. It is worth noting that by assuming that the energy well has a ‘corrugated’ bottom, the potential barrier model can account for the existence of P-relaxations.Eyring and co-workersS1 have approached the glass transition through a theory of viscosity. In their ‘significant structures’ theory, the partition function of a glass is regarded as a product of solid-like and gas-like contributions. Only the former is assumed to contribute to viscosity which, for a system of hard spheres is given by’ 7 =AT1’2j(V-V,)exp(E#/kT) (19) where A is a constant related to the distance between nearest neighbours, particle mass, and a constant transmission coefficient, and V and V,are total and solid-like volumes respectively; E ’is the activation barrier to particle motion. Ts in this model, is evaluated using the isoviscous criterion. The glass transition has been the focus of much interest recently. Kannog2 regards configurational entropy as a sum of communal and positional entropy.Communal entropy is defined, after Kirkwo~d,~~ as the entropy due to the sharing of the available volume in the liquid by all the particles. Since positional entropy is always greater than zero, Kanno concludes that the glass transition is due to the vanishing of communal entropy. Thus the uniqueness of To as a limit of Tg disappears. Phillipsg4 uses the cluster formalism to describe the temperature de- pendence of viscosity which can also be used to understand the glass transition. Other workers have regarded glass as a crystal with a high density of dislocations and have proceeded to discuss its properties from this point of view.85 An analogy with spin glasses has also been described using a model of entangled rod-like Most of these notions are as yet in preliminary states of development and their validity remains to be proven.M. Goldstein, J. Chem. Phys., 1969, 51, 3728; J. Chem. Phys., 1976, 64,4767; J. Chem. Phys., 1977, 67,2246. 81 L. Faerber, S. W. Kim, and H. Eyring, J. Ph.ys. Chem., 1970, 74, 3510. 8L H. Kanno, J. Non-Cryst.Solids, 1980, 37, 203. 83 J. G. Kirkwood, J. Chem. Phy.~., 1950, 18, 380. a4 J. C. Phillips, J. Non-Crysl.Solids, 1981, 44,17. S. F. Edwards and M. Warner, Philos. Mug. A, 1971, 40, 257. 86 S. F. Edwards and K. E. Evans, J. Chem., SOC..Furuduy Trans.2, 1982, 78, 113. The Glass Transition: Salient Facts and Models 7 Pressure as a Variable in the Glass Transition Studies of the glass transition as a function of applied pressure are rather small in number and the data available are scanty.These studies are nonetheless essential if one is to determine the merits of the various models of the glass transition. Simulation studies have shown that an amorphous packing may be produced by the application of pressure alone but these need to be further explored.65 If the thermodynamic excess quantities, volume, V, (V,iquid -Vglass/ and entropy, S,, are functions of a single parameter, Z(P, T), we mean that P and T may simultaneously be changed so that 2remains constant.87 If V, = V,(2)and, S, = S, (2)and Tgis determined by a single ordering parameter, it can be shown thats7 rs)ve= AB/Aa, d V, = r$)se 0 or = TgVgAcljAC,, dS, = 0 where Vgis the volume of the glass and da and dB are changes in expansivity and compressibility, respectively, at Tg.Should dS, and d V, vanish simultaneously at Tg,9387 where 17 is the Prigogine-Defay ratio.88 To date, it appears that 17 is always greater than unity and equation (21) rather than equation (20) is found to be experimentally valid. However, equation (21) also fails for ZnC1, glass where dT,/dP is less than TgV,LI~/~C,.~~If Tgis taken to be that temperature at which a relaxation time, z,is constant,89 or (aTgIap) = (aseiap),(Tg/ACp)+ Vg TgAa/ACp (24) In glassy ZnCl,, (i?S/dP),appears to be negative indicating a possible route to the ideal glass.89 From the standpoint of the cluster model, increase of pressure should increase the cluster fraction which is just what the notion of ideality entails.In any event, the validity of equation (21), noted in many experiments, is strange since it suggests that zis a function of S, alone, while z = z(TS,) in the configurational entropy model. In summary, a single ordering parameter appears to be insufficient to describe the glass tran~ition.~'~~ The configurational entropy model seems to describe the M. Goldstein, J. Chem. Phys., 1963, 39, 3369. I. Prigogine and R. Defay, 'Chemical Thermodynamics', Longmans Green, London, 1954. 89 C. A. Angel], E. Williams, K. J. Rao, and J. C. Tucker, J. Phys. Chem., 1977, 81, 238. Parthasarathy, Rao, and Rao variation of Tg with pressure more accurately than the free volume theory.87 Further constraints on the ordering parameters have also been extensively dis- cussed.O 8 Concluding Remarks It is clear that the experimental transition at Tgoccurs when a relaxation time, T, attains a constant value on the time scale of the e~periment.~.~~ A thermodynamic transition has been proposed at a temperature To( T,) but this transition has not so far been (can never be!) observed, since relaxation times are too long below T,. The configurational entropy model is unable to describe the change in viscosity with temperature ac~urately.~~ Its description of the variation of Tgwith pressure is in error in the case of ZnC1,.89 Further, its extension to non-polymeric glasses is not obvious.Most importantly, perhaps, the entropy model requires the value of S, as input data to yield the values of other parameters. The free volume model has no means of accounting for transitions with negative expansi~ity.~~ In the Its prediction of dT,/dP is usually seriously in error.87*89,93 Cohen-Grest modification of the original theory, the distinction between solid-like and liquid-like cells is rather arbitrary and interconversion of one into the other at sub-T, temperatures is ignored. In this form, though, the theory can account for an entropy catastrophe. However, both the free volume and the configurational entropy model are able to account for the observed identity of Tovalues obtained from various experiments. With respect to the order of the transition at To,the free volume theory predicts that it is of first order,2 but so far there has been no evidence for such a tran~ition.~~ The bond lattice model does not require the existence of a transition at Toin order to match the variation of C,; in this respect this model is unique.The cluster model predicts that the transition at Tgitself is of second order with a first order com- ponent where the clusters dissolve spontaneously. Experimentally, relaxation effects are likely to preclude verification of this prediction. The cluster model, in particular, does not distinguish between an ideal glass and the corresponding crystal, at least in a restricted class of glass formers; even though proof to the contrary is scarce,62 this appears to be a singular conclusion.As with Goldstein’s potential barrier approach,8O change in lattice frequencies, increasing an-harmonicity, and an increase in the number of particles participating in secondary relaxations (the last two are closely linked) are the factors governing the approach to Tgin the cluster model. The free volume model does not explicitly consider configurational entropy, and the configurational entropy model does not consider communal entropy. The 90 P. K. Gupta and C.T. Moynihan, J. Non-Cryst.Solid.y, 1978,29. 143; J. Kovacs, J. Phys. Chem., 1981, 85, 2060. 91 W. T. Laughlin and D. R. Uhlmann, J. Phy.\. Chem., 1972, 76, 2317; R. J. Greet and D. Turnbull, J. Chem.Phjs., 1967, 46, 243. 92 E. Williams and C. A. Angell, J. Polym. Sci., Polym. Lett., 1973, 11, 382. 93 E. Williams and C. A. Angell, J. Phys. Chem., 1977, 81, 232. 94 L. Boehm, M. D. Ingram, and C. A. Angell, J. Non-Crysf.Solids, 1981,44, 305. 381 The Glass Transition: Salient Facts and Models distinction between the two due to Kanno, is however, likely to be only of heuristic interest, since the apportioning of this quantity between .these two components is rather arbitrary. Further, it has also been stated that Kanno’s approach neglects the observed dependence of Tgon cooling/heating rates, and is therefore unreal- istic.l9 To the extent that Tgis not determined by a single excess parameter, (Z7> 1 for most of the systems studied) there may be little to choose between the free volume and configurational entropy models.It is worth noting here that the free volume model in its current form resembles the configurational entropy model rather closely. The free volume model appears to describe the ‘cells’ of the Gibbs-DiMarzio approach in physical terms that are applicable to glasses other than polymeric ones. The cluster model is successful in many respects. Primarily, there is visual proof that clusters exist. High resolution electron microscopy (HREM) work has re- vealed the existence of positionally correlated regions -50 8, across in a number of glasses.95 Interestingly, if Y, is taken to be 5-108, (in the steeply decreasing region in Figure 7), then the cluster radius in the glass may be estimated to be around 25-50 8,.Simulation studies have shown that these are of pentagonal or related symmetry so that they cannot be space filling.3 The observation of an anomalous temperature dependence of viscosity (equation 1) at temperatures just greater than Tg can be explained by the formation of clusters (and therefore, increased co-operativity) at T > Tg.96The observation of ,!I-relaxations in glasses consisting of rigid molecules suggests that these relaxations are features character- istic of amorphous packing.22 It has also been shown that decreasing the fictive temperature reduces the amplitude of these relaxations but not their fre-q~ency,~~,~*so that it is appealing to consider that p-relaxations originate in the tissue material.Mechanical relaxation studies have also used the cluster model to identify the a-and p-relaxations with viscous flow and delayed elasticity re- ~pectively.~N.m.r. studies have detected sub-glassy relaxations and these have been attributed to particle motion in regions of lower density.41 In certain cases, it is indeed possible to study both cluster and tissue contributions to the spectrum simultaneously.43 Spin probe e.s.r. studies4’. 48 indicate that cluster dissolution takes place only at Tk 2 Tg.Current indications are that the low-temperature anomalies in glasses arise in the tissue material. Current theoretical work has tended to study the relaxation phenomena charac- teristic of the glass transition using concepts developed for other solid state phase tran~itions.~’It is well-known that the non-Arrhenius dependence of dynamic variables near Tgis indeed due to co-operative processes.Primarily, there appears to be a rapid increase in the correlation length in this temperature regime.98 This 9s P. H. Gaskell, D. J. Smith, C. J. D. Catto, and J. R. A. Cleaver, Nature. 1979,881,465; L. A. Bursill, J. M. Thomas, and K. J. Rao, Nafure, 1981, 289, 157. 96 E. McLaughlin and A. R. Ubbelohde, 7rans. Faraday SOC.,1958, 54, 804; A. J. Matheson, J. H. Magill, and A. R. Ubbelohde, Trans. Faraday Sw., 1958, 54, 181 1. 97 P. W. Anderson, ‘TI1 Condensed Matter’, ed. R. Balian, R. Maynard, and R. Toulouse, North-Holland, Amsterdam 1979. 98 B. I. Halperin and P.C. Hohenberg, Rev.Mod. Phys., 1977, 49, 435. Parthasarathy, Rao, and Rao is very easily rationalized using the cluster model, in which an increase in the correlation length would imply cluster growth. Thus, Tono longer marks merely the termination of the liquid state, i.e. a point at which viscosity is infinite. It is, theoretically, a temperature at which the correlation length would be infinite in In glasses, it represents a temperature at which the correlation length is large enough to suppress relaxations even on the longest practicable time scales.99 The major advantage in using the cluster model to explain glass transition phenomena lies in the close relation that it builds between the structure of a glass and its behaviour at Tg.The term ‘structure’ need not connote only translational disorder -the cluster model assumes the existence merely of correlationally differentiated regions.This flexibility makes it possible for one to discuss glass-like transitions in plastic crystals, using the cluster model, for which conventional glass transition models that rely on positional disorder are not easily employed. Further- more, the cluster model regards the glass transition as an aspect of solid state phenomenology and not primarily as an event that marks the end of the liquid regime. This enables one to explain the glass transition in terms of well-understood concepts developed for solid state phase transitions (e.g.soft modes). The possible existence of soft modes in glasses has been indicated by spec- troscopic ~t~die~.~~,~~,~~*~~~ These can be accounted for by the cluster model, where excitation of particles to higher states naturally leads to the weakening of vibrational force constants.In contrast to soft modes in crystals, however, the ‘soft’ modes in glasses are likely to soften continuously. Orientational and Dipolar Glasses.-We have seen earlier that disorder in other degrees of freedom may be frozen-in to yield glass-like solids which are akin to bona jide (positionally disordered) glasses. In particular, it has been shown that orientational disorder found in plastic crystals may be quenched in to yield ‘glassy crystals’. Calorimetric studies6 have demonstrated the existence of glass-like tran- sitions in these solids at a temperature, TA.Dispersions in dielectric loss and dielectric constant are seen30*101 both at and below Ti (Figure 9). E.s.r. spin probe spectroscopic studies indicate that there is a marked decrease in the correlation time around TA as the glassy crystal is warmed.30 Raman and i.r. band shape studies on glassy crystalline cyclohexanol carried out in this laboratory also show the onset of molecular motion around Ti. Dielectric studies on glasses formed by supercooling liquid crystals yield results very similar to these. Disorder in dipole interactions may be frozen-in to yield dipole glasses, sub- stances which have features similar to those of spin glasses. A typical case is that of KBr doped with CN-in which relaxation has been discerned by dielectric studie~’~~~’~~(Figure lo).Yet another kind of dipole glass is exemplified by 99 M. Shablakh, R. M. Hill, and L. A. Dissado, J. Chem. Soc., Faraday Trans. 2, 1982, 78, 625. loo G. J. Exarhos, P.J. Miller, and W. M. Risen, J. Chem. Phys., 1974, 60,4145. lol K. Adachi, H. Suga, S. Seki, S. Kubota, S. Yamaguchi, D. Yano. and Y. Wada, Mol. Cryst. Li4. Cryst., 1972, 18, 345. lo2 S. Bhattacharya, S. R. Nagel, L. Fleishman, and S. Susman. Phys. Rev. Lett., 1982, 46, 1267. Io3 A. Loidl, R. Feile, and K. Knorr, Phvs. Rev. Lerf., 1982, 48, 1263. The Glass Transition: Salient Facts and Models A 10 c)51 X rp c51 I I I 150 2 00 250 T (K) Figure 9 A semi-log plot of tan 6 against temperature for camphor at the following frequencies in kHZ: (A) 1; (B) 10; (C)50; (D) 100.Inset: Plots of upagainst T-' .for the low temperature, j-.and high temperature, ti-, relaxations (After R. Parthasarathy, K. J. Rao, and C. N. R. Rao, 1983, Re$ 30) Li+-doped KTaO, studied by Hochli et al.lo4 The off-centre impurity seems to stabilize locally polarized clusters in the matrix and the distributions of relaxation times at various dopant concentrations appear to reflect cluster interactions. We have reported dielectric studies on dilute solutions of ferroelectrics in anti- ferroelectrics in which we found that tan 6 goes through a minimum and dielectric constant through a maximum around 5% solute concentration.' O5 This anoma- lous variation of dielectric parameters could reflect the presence of locally polar- ized clusters in these transitional phases.Thus the glassy state seems to include long range disorder of many types while the glass transition, in very general terms, manifests itself when relaxational and experimental time scales intersect. Nevertheless, the existence of an ideal glass transition temperature in glassy crystals as well as in glasses hints at the possibility lo4 U. T. Hochli, H. E. Weibel, and L. A. Boatner, J. Phys. C;Solid Sute Phys., 1979, 12, L563. lo5 R. Parthasarathy, K. J. Rao, and C. N. R. Rao, J. Muter. Sci., 1981, 16, 1424. Parthasarathy, Rao, and Rao 1.12 1 I 1 I 20Ht -- 1.10 100 HZ 750HZ -.- 2500HZ - 1.08 25000H~... . & 1.06 -1 €7 1.04 1.0ot’ , , 4I 0 40 80 120 160 200 T (K) i 1 I 0 20 40 60 80 Temperature Figure 10 (a) The normalized dielectric constant versus temperature at vurious ,frequencies, of (KBr),. (KCN),,. Inset: 75KHz data up to 200K. (h) Dielectric loss versus temperature for severaifreyuencres (After Bhattacharya et al., 1982, Ref: 102) that the glassy state is a thermodynamic entity and not a kinetic accident. Further- more, the evolution of excess entropy with temperature in these materials is known to parallel that in glasses.’ It is clear, therefore, that a theory proposed for glasses and the glass transition should be extensible even to those glasses that are not positionally disordered.
ISSN:0306-0012
DOI:10.1039/CS9831200361
出版商:RSC
年代:1983
数据来源: RSC
|
5. |
Electron spin resonance of haemoglobin and myoglobin |
|
Chemical Society Reviews,
Volume 12,
Issue 4,
1983,
Page 387-414
L. Charles Dickinson,
Preview
|
PDF (1803KB)
|
|
摘要:
Electron Spin Resonance of Haemoglobin and Myoglobin By L. Charles Dickinson DEPARTMENT OF CHEMISTRY, UNIVERSITY OF MASSACHUSETTS, AMHERST, MA, USA 01003 M. C. R. Symons DEPARTMENT OF CHEMISTRY, UNIVERSITY OF LEICESTER, LEICESTER, LEI 7RH 1 Introduction The purpose of this review is to give sufficient detail of the past electron spin resonance (e.s.r.) work on haemoglobin and myoglobin to bring the general reader up to date (early 1983), and to present a critical appraisal of some of the unresolved problem areas which are areas of active research. It is assumed that the reader has a foundation in transition metal e.s.r. spectroscopy from experience with other systems or from any of the several introductory works on the technique. See, for example, references 1-3 or specific works on haemoglobin and my~globin.~, A.E.s.r. Parameters of Haems: An Overview. -The native form of deoxyhaemoglobin [Fe"; d6]is diamagnetic and hence can only be studied by e.s.r. spectroscopy if it has a paramagnetic ligand [e.g. NO or O,-). Also, oxyhaemoglobin gives no e.s.r. signals. However, good spectra can be obtained from the oxidized [Fell'; d5]forms, and these have been widely studied by e.s.r. spectroscopy. For many purposes the porphyrin moiety may be regarded as an axially symmetric tetradentrate ligand for iron (Figure 1). In FeMb'(H,O) the five electrons are unpaired, as has been demonstrated by magnetic susceptibility studies, and thus form a 6Sstate with manifolds of S, = & i, k :,k where S, Abhrrviations: FeHb or Hb, ferrous haemoglobin; FeMb or Mb, ferrous myoglobin; FeHb+, ferric (met) haemoglobin; FeMb+, ferric (met) myoglobin; "Hb, cobaltous haemoglobin; 'OHbO,, oxycobalt haemoglobin; ""Mb, cobaltous myoglobin; '"MbO,, oxycobaltomyoglobin; HbKansas haemoglobin Kansas; Mb(Im), myoglobin with imidazole sixth ligand.H. M. Swartz, J. Bolton, and A. Borg, 'Biological Applications of Electron Spin Resonance', Academic Press, New York, 1972. A. Carrington and A. D. MacLachlan, 'Introduction to Magnetic Resonance', Harper and Row, New York, 1967. (a)M. C. R. Symons. 'Chemical and Biochemical Aspects of Electronic Spin Resonance Spectroscopy', Van Nostrand, New York, 1978. (b) P. Atkins and M. C. R. Symons, 'Structure of Inorganic Radicals', Elsevier, New York, 1968.M. Weissbluth, 'Haemoglobin', Springer Verlag, New York, 1974. E. Antonini and M. Brunori, 'Haemoglobin and Myoglobin in Their Reactions with Ligands', Elsevier, New York, 1971. Electron Spin Resonance of Haemoglobin and Myoglobin H,C =CH '\! 2 Figure 1 Diagram of haem moiety showing axes chosen, angle qf tilt of A-B ligund, und numbering of porphyrin nitrogen system is the component of magnetic spin in the direction of'the magnetic field, z. This is the 'high spin' case. The apparent g-values which result are g z 2 and g, % 6. These are limiting positions for one of the many allowed trarkitions. This large anisotropy has provided very accurate determination of the orientation of the haem groups in haemoglobin single crystals.There is with Fe3+ the possibility of an intermediate spin, S = case, but this has been observed only in model haems.6 In Fe3+ with strongly bound ligands (e.g., CN-, N,) the electronic structure changes drastically to 2Swith one unpaired electron. The g-values become 2.82, 2.20, 1.70 in Hb'N,. This is the 'low-spin' case. Hyperfine splitting is not directly seen in the e.s.r. spectra of ferric haems although some is seen in ENDOR studies that we describe below. This is because of the fairly large linewidths and quite small splitting parameters of the low natural abundance (2.2%) 57Fe, and low unpaired electron density on porphyrin and protein nuclei. For this reason much interest has been directed towards metal- replaced cases such as '"Hb which give an indication of electron density on the metal through the large hyperfine splitting by 59C0.In order to detect superhyperfine splitting from ligand nuclei, it is necessary again to examine non- native species such as 'OHbO,, HbNO, or HbO; in order to gain some insight into the electronic structures and densities within the ligand-metal-porphyrin moiety (Figure 2). C.A. Reed, T. Mashiko, S, P. Bentley, M. E. Kastner, W. R. Scheidt, K. Spartalian, and G.Lang. J. Am. Chem. SOC.,1979, 101, 2948. Dickinson and Symons L /lI- ---Ff --- /I 6 d5 HS d5 LS d HS E.s.r.+ ve E.ac + ve Es.r.-ve Mb+/ Hb' Mb+/ H b+ Mb /Hb L =H20, FY HCO;,OCN- t L =N3-,CN', OH',OCN- "deox y " OH0 I/---Ff---4 7 5 2-d I--(NO) d6(02) d (0,)or d (4 )+ Esf.+ ve Esx-ve Es.r.+ ve.-, MbNO/H bNO Mb02/HbOz MbO,-/H b02-\\ .nitrosyl " *oxy " "perox y " /---?---4 d7 LS d7(02)or d6(0,-) Es.r.t ve E.sr.+ ve CoMb/CoHb , coMb02/'OH b02 "cobalt-deoxy * *cobalt -oxy '' Fi7ure 2 Structures and abbreviations of species discussed. L are sixth ligunds which prefer 'Fe" high-spin (HS)and L are Iigandsfavouring low-spin Fe"' (LS) B. E.s.r. Measurements.-E.s.r. spectral measurements are most commonly made on randomly orientated samples, either in solution or as polycrystalline materials. Such samples which contain all possible orientations of the paramagnetic species, yield spectra which are the sum of all angle-dependent spectra.[See the appendix of reference (3b) for a detailed discussion.] In principle, this spectrum contains all the information about the e.s.r. parameters and indeed there are powerful computer programs for extracting such information from the 'envelope spectra.' However, in cases such as '"MbO, where the g-tensor and s9C~hyperfine tensor do not share principal axes, the envelope spectrum can give misleading parameters. Thus, there is an increase of information and accuracy in e.s.r. crystallographic results which also allow direct comparison of orientation of e.s.r. parameter tensors with atomic co-ordinates obtained from X-ray crystallography. Because some spectra are complex and ambiguous it is often advantageous to 389 Electron Spin Resonance of Haemoglobin and Myoglobin obtain spectra at several frequencies, usually 9 GHz (X-band) and 35 GHz (Q-band).Good hyperfine parameters can be defined since the separation between g features is proportional to field strength and hyperfine features are invariant with field strength; this greatly aids spectral interpretation. C. Haem-Globin Complexes.-In this work haemoglobin refers to the tetrameric haem-globin complex which is the primary oxygen carrier from lung to tissue in most higher animals, but is also found in some molluscs and plants. Haemoglobin has four haems, each in a separate protein chain, all of which closely resemble myoglobin. In haemoglsbin A, of adult humans, there are two each of two distinct protein sequences, labelled a- and P-chains.Each haem is firmly packed in a protein matrix of hydrophobic amino-acids and there is a covalent bond between the iron and the F8 imidazoyl E-nitrogen. (F8 refers to the eighth residue of the F helix and is histidine at the 87th residue in a-chains, and at the 92nd in P-chains. The haem is ensconced in a pocket: on one side is the F8 histidine, on the other an open cavity where ligation can occur (Figure 1). Nearby, but not bonded to the iron, is the distal histidine (E7) which may influence ligand bonding in some cases. One of the intriguing questions in haemoglobin research has been the mechanism of change in oxygen affinity of haemoglobin depending on the oxygen pressure. That is, at low oxygen tension the haemoglobin has a low affinity for oxygen; with increasing addition of oxygen molecules to the haemoglobin tetramer the affinity for oxygen increases greatly.This phenomenon of allosterism (co-operativity, or ‘haem-haem’ interaction) requires that there be some linkage between the subunits so that bonding of some oxygen molecules changes the molecular geometry of the remaining haem moieties. The low affinity conformation corresponding to deoxyhaemoglobin has been labelled ‘7“ for tense; the high affinity or oxygenated form ‘R,’for relaxed. The terms tense and relaxed can be related to the Fe-porphyrin distances as seen in X-ray crystallography of haemoglobin and model compounds. In the Tform the Fe nucleus is as much as 0.75A out of the porphyrin ring towards the F8 histidine; in the R form the Fe atom is virtually in the ring.For further details on the physical properties of haemoglobin the reader is referred to an excellent monograph by Antonini and Br~nori.~ The most commonly accepted structural interpretation of R and Teffects is that of Perutz, details of which can be found in his original paper.’ A number of abnormal haemoglobins have been found in humans. These are caused by genetic defects such that there is a substitution of an amino-acid in the ususal peptide chain. For example, HbKansas has asparagine G4( 102)in the P-chain replaced by threonine and is designated Hb,a,,a,[AsnG4( 102)P +Thr]. This is of interest, not because of a direct influence on the substitution of the haem environment, but because of the indirect influence caused by a change in T conformation.’ M. F. Perutz, Nulure, 1970, 228, 726. Dickinson and Syrnons N .". s h QU L,m -ra .e i! 0: E ... D P 30 1 Electron Spin Resonance of Huemoglobin and M.yoglobin 2 Monomeric Species: Ferric Forms Myoglobin offers the simplest case for study of haem-globin complexes because it has no co-operative haem-haem interaction, and in the single crystals (type A)* there are only two inequivalent haem orientations. Hence it has been extensively studied and was one of the first haemproteins investigated by e.s.r. in the classic work from Ingram's lab~ratory.~ We present myoglobin results first so that the reader may appreciate 'basic' haem-globin interactions before being introduced to the complexities which arise in the tetrameric haemoglobins.A. High-spin Derivatives Initial observations on aquo-metmyoglobins employed the large anisotropy of the gap,-tensor (g,,= 2.0, g, = 6.0) to determine the orientation of the normals of the haem plane with respect to the crystallographic unit-cell axes (Figure 3). This was done prior to the solving of the detailed atomic structure by X-ray crystallography and illustrates the power of e.s.r. crystallography. At that time the nature of the sixth ligand was not known but was later determined to be a single water molecule. In this case the axial symmetry of the porphyrin ligand field generates an axial g,,,;tensor, as seen at 77 K.The gap,-value, once the principal values are known, is given by g2(0)= gf,coszU + g: sin20 where 0 is the angle between the magnetic field direction and the symmetry axis of the paramagnetic system. The distortion of the ligand field from axial symmetry by the fifth ligand, an imidazoyl nitrogen. is not seen at 77K, although it is observable in some haem proteins if the temperature of observation is lowered to near 1 K.l0 Although a small rhombicity is seen in crystal studies on aquo-metmyoglobin,'l, l2 frozen solution studies show only a reduction of linewidth from 75 to 36 G* as the temperature is lowered from 77 to 1.6 K.l0 The direction of the minimum gap,-value in the haem plane is found to correspond to the projected proximal imidazole plane.The rhombicity of the gap,-values arises because the E term is non-zero. The theory for such cases has been treated by Kotani eta1.13 and by Dowsing and GibsonI4 and others. Since variation of the gap,-value on magnetic field strength is small, there is a large error in the estimates of the values of D and E. The best estimates of D and E come from the high microwave frequency single crystal study of Slade and Ingram' where D = 8.8 f0.8 cm-and E = 0.2 cm-l. *G = Tesla. J. C. Kendrew and R. G. Parrish, Proc. R. Soc. London. Ser. A, 1957, 283, 305. J. E. Bennett, J. F. Gibson, and D. J. E. Ingram, Proc. Chem. Soc. (London). 1957, ,4240, 67. lo J. Peisach, W. E. Blumberg, S. Ogawa. E. A. Rachmilewitz, and R.Oltzik, J. Bid. Chem., 1971, 246, 3342. J. F. Gibson, D. J. E. Ingram, and D. Schonland, Discuss. Furduy Soc., 1958, 28, 72. '* G. A. Helcke, D. J. E. Ingram, and E. F. Slade. Pro(,. R. Soc. Lmzclon, Ser. B., 1968, 169, 275. l3 M. Kotani, and H. Watari, in 'Magnetic Resonance in Biological Systems'ed. C. Franconi, New York. Gordon and Breach, 1971. p. 75. l4 R. D. Dowsing, J. F. Gibson, M. Goodgame, and P. J. Hayward, J. C'hrrn. Soc. (A), 1969, 187. Is E. F. Slade and D. J. E. Ingram, Proc. R. Soc. London, Ser. A, 1969, 312, 85. Dickinson and Symons Table 1 E.s.r. parameters of aquo-metmyoglobin Ref: 2.0 7, 10, 16 6.0 7 5.92 10 5.98, 5.86 0.01 D 1Ocm-' 12 4.4 13 E 0.02 13 N haem NAx 20.5 1.5MHz 4 equivalent 30.0 2.5 14 "4, 7.6 0.02 57Fe Ax, -27.05 & 0.01 14 A zz -27.77 & 0.2 Nu A; 11.46MHz 14'Elis HA 6.02 0.08 15H2* haem meso 'H 0.79 & 0.008 15 Because the directions of the anisotropy in the haem plane do not correspond to those for low-spin derivatives, it was concluded that a factor beyond the influence of fifth and six ligands, such as puckering of the porphyrin ring, was responsible for the anisotropy.There seems to be no final consensus on the size or molecular cause of the rhombic distortion of the gap,-tensor. Even when linewidths of e.s.r. spectra are very much larger than any hyperfine splitting, it is possible to determine, in favourable cases, the hyperfine splitting parameters by strongly irradiating n.m.r.transitions and observing changes in the e.s.r. intensity. This is ENDOR or electron nuclear double resonance spectroscopy. By ENDOR it has been possible to determine the z or parallel components of the porphyrin nitrogen, and the proximal histidine nitrogen, hyperfine coupling.' The small 14N quadrupolar splitting constants have also been obtained from ENDOR line spacings. Proton ENDOR reveals hyperfine coupling from haem- bound water protons and from the four rneso-protons of the haems. ' These small hyperfine couplings, as measured by ENDOR, agree accurately with the dipolar calculations based on the known co-ordinates in Mb', and thus are not a consequence of electron delocalization. The complete hyperfine tensor for 'Fe has been determined.An intriguing effect, seen in single crystals of myoglobin and its derivatives, is the dramatic change in linewidth with change in angle to the magnetic field. For example, as the magnetic field direction is rotated in the haem plane, the linewidth varies from 40 to 130G. This was studied in great detail by early workers in the field' 2, who concluded that in aquo-metmyoglobin there was a slight random misorientation of molecular axes within the crystal, and because of the large g-anisotropy, a standard deviation of only 1.6O was sufficient to explain their data. l6 G. Feher, R. A. Isaacson, C. P. Scholes, and R. Nagel, Ann. N.Y. Ac~.%i., 1973, 222, 86. " C. F. Mulks. C. P. Scholes, L. C. Dickinson, and A. Lapidot, J. Am. C'Iien7. Soc., 1979, 101, 1645.P. Eisenberger and P. S. Pershan, J. C'lzrm. Phys., 1967, 41,3321. Electron Spin Resonance of Haemoglobin and Myoglobin More recently. Calvo and BemskiI9 have offered a detailed model that purports to explain the variation in linewidth in the haem plane based on a distribution of positions of fifth and sixth ligands. Probably both of these factors contribute. Fluo~.o-metmyoglobin.The water molecule co-ordinated to the haem iron can be replaced by a fluoride ion, although high concentrations are rcquired for complete conversion (pK, = 1.9). In the single crystal study the 19Fnucleus (S = 4 j causes a splitting of 43 G on gIIand slightly anisotropic splitting of 23.5 and 21.5 G in the haem plane.'O This change of sixth ligand to a charged electronegative species reduces the zero field splitting parameter, D,to 5 cm- I.The ENDOR spectrum of fluoro-metmyoglobin is very similar to that of aquo-metmyoglobin except that the hyperfine splitting from the histidine nitrogen is reduced in the former case by 0.5 MHz. Curiously, the meso-protons of the porphyrin show a slight increase in hyperfine splitting. The Mb' F-complex has been shown in careful single crystal studies to have its haem axis of symmetry titled 5 ' from that of the Mb'(H,Oj.Z1 Mb'(HC0; j derivatives have been shown, in a single crystal study, to exist in two forms having distinguishable ga,,-tensors, each with distinct orientation.2 The observation of two species may be related to the observation of two species for MbNO and '"MbO, derivatives.Mb+(OCN-) has been studied in single crystals. The relatively large rhornbic splitting gave in-plane g-values of 5.82 and 6.1 1 .22 The large E term is thought to arise from tilt of the OCN- with respect to the haem plane, but it is not clear why there are two forms. This species is in thermal equilibrium with a low-spin form (vide irzfia). OCN- is isoelectronic with N, which only forms low-spin form and is discussed below. B. Ferric Forms: Low-spin Derivatives.--Addition of strong ligands such as CN , N3, imidazole or OCN causes the energy levels of the iron atom to split to such ~ a degree that the five electrons are confined to the three fZg orbitals. Thus a single electron remains unpaired. The measured g-values are now true g-values, which are a function of the total ligand field.Most theoretical interpretations originate from that of Griffith,23-ZS who divided the ligand field into rhombic and tetragonal components. Hence the data have been divided into five types depending on the ratio of rhombic to tetragonal components as shown in Figure 4.2s Mb'CN-has been studied in single crystal forms by several workers who determined principal g-values of (0.93, 1.89, 3.45) with g, making an angle of approximately 30 with the projection of the proximal imidazoyl group onto the haem plane, g, lying within 10' of the haem normal of Mbf(H20) and g,, making l9 R. Calvo and G. Bemski, J. Clrern. Phys., 1976, 64,2264. 2o H. Morimoto and M. Kotani. Biochrm.Biophys. Actu. 1966, 126, 176. z1 E. F. Slade and R. H. Farrow, Binchum. Biophys. Actu. 1972, 278, 450. 22 H. Hori, Biochem. Biophys. Ai,tu, 1971, 251, 227. 23 (a)J. S. Griffith, Nafure (London), 1957, 180. 30. (h)J. S. Griffith, 'The Theory of Transition Metal Ions', Cambridge University Press, Cambridge. 1961. (c) J. F. Gibson, D. J. E. Ingram, and D. S. Schonland, Furuduy Soc. Dismw., 1958, 24, 72. 24 G. M. H. Loew, Biophys. J. 1970, 10. 196. 2s W. E. Blumberg and J. Peisach, 'Probes of Enzymes and Haemproteins', ed. T. Yonetani and A. Mildvan, Academic Press, New York, 1976, p. 215. 394 Dickinson and Sywions .9t ,,I I Pure rhombi c .5i1 /. I ng 0 Haemoglobin A 1-Pure axial 1 I I I I 1 I Tetragonal field, [A/h] Figure 4 Grouping diagram for muny lowspin harm compound.s.Thc.,five c1us.sr.s ure dividcd according to relative vulurs of V und A, wmputed,frorn obscwwd lowspin K-vu1ue.v an angle of 9" above the haem plane (Figure 5). Scholes has made a detailed ENDOR study of Mb'(13CN-) and deuteriated Mb'('*CN -)I7 determining IA,,(l3CC>l= 28.64 0.08 MHz with evidence that the hyperfinc splitting is nearly isotropic. This most probably means that spin polarization of the o-electrons exceeds the expected dipolar coupling due to x-delocalization. Nitrogen ENDOR of Mb+(CN-) has proven somewhat ambiguous in that the four lines detected can be interpreted as a single four-line multiplet from one type of nitrogen, or consid- ering some weaker lines in the spectrum, there may be two inequivalent types of haem nitrogen.In Mb+(C' 5N-)additional resonances are seen indicating weak I'c(l)-e--u21--I=* ---\N/ I IC(11 Ii Figure 5 Orientations of' various lowspin huem derivative gvulues with respect to the haeni plane 395 Electron Spin Resonunre qf' Haemoglobin and Myoglobin coupling of the 15N to the iron electrons. No 57Fe hyperfine splitting has been reported in Mb+(CN-). All 14Nsplittings seen by ENDOR are attributed to haem nitrogen atoms. Mb+N, is perhaps the most thoroughly studied low-spin derivative of myo- globin.'*,23a, 26 These single crystal studies substantially agree on (gx,gy,g,) values of (1.72, 2.22, 2.80) but the reported angles from direction cosines, relating the directions of the principal values to the crystallographic (a, b, c*) axes, differ by as much as 20 *.There is agreement that the direction of g, in Mb'N, deviates about 7 * from the direction of the haem normal. The g-tensors of CN-, N;, and Im complexes have very nearly identical orientations of the x, y, z axes. The linewidth variation in the spectra of Mb'N, with orientation has attracted a great deal of attention because of the possibility of observing variation of the strong ligand field.12, IR The minimal linewidth of 250G is larger than that for Mb+(H,O), indicating that such an influence is present. Eisenberger and PershanI8 include terms for the variation of the rhombic distortion of the crystal field and form misalignment.They ignore distortions of the tetragonal field be- cause these are expected to be much smaller. A +3.5% variation in the rhombic distortion parameter and 2 'in haem orientation satisfactorily explain the vari- ation of linewidth in Mb'N;. No hyperfine splitting was seen directly in the e.s.r. spectrum of Mb'N,, but ENDOR signals on Mb+N, with 5N-enriched haem have been studied in detail'' at the g, and gypositions in the frozen glass. The nitrogen coupling constants were approximately equal along x and y. No azide I4N splitting was seen, but two types of haem nitrogen were seen with 14AZz= 5.64 and 6.14MHz. The inequivalence could arise either from direct ligand electronic influence (either N, or proximal imidazole nitrogen) or from porphyrin ring puckering. It is puzzling that this haem nitrogen inequivalence was observable in Mb+N, and not in Mb+CN--.The azide derivative is known from X-ray crystallography to contain a linear ion bound to the haem at an angle of 11 *. This bending probably arises primarily from steric reasons as is discussed below for MbNO and '"MbO,. This asymmetric influence of the azide ligand on the electronic structure of the porphyrin appears to be a source of inequivalence. One study attributes half of the g-anisotropy to electronic influences of N, and the other to a dynamic Jahn-Teller effect.27 The latter seems unlikely in the absence of any remaining degeneracy. Single crystal studies of Mbt(OH-) have been reported. Mb+(H,O) and Mb+(HO-) exist in equilibrium with a pK of 8.9.Two forms of Mb+(OH-) (low-spin) are seen which interconvert at about the same PH.~~' The principal g-values of Mb'OH- (1.85, 2.17, 2.55),2sshow a lower anisotropy than the CN- or N; derivatives, but theoretical considerations show that both rhombic and tetragonal fields are largest for this ligand. The interconversion of H,O and OH -forms has been studied in detail by other techniques but will not be treated here. No Mb+(OH-) ENDOR data are available. 26 J. F. Gibson and D. J. E. Ingram, NLiturP (London), 1957. 180. 29 27 S. Mizumashi, J. Pliys. So(,.Jpri.. 1969, 26, 468. 396 Dickinson and Syrnons Mb'(1m) e.s.r. results have been reported for crystals of Mbf(H,O) into which imidazole was diffused.22 The spectra show that both high- and low-spin haems are present.The high-spin form is identical to Mb+(H20). The observed g-values for the second species in the crystal are (1.53, 2.26, 2.91) and the alignment of the tensor is nearly identical to that of MbfN;. Mb+(OCN-), as mentioned above, has the ligand co-ordinated with the haem iron in such a way that there are high-spin and low-spin forms. In this case the high-spin form is quite different from Mb+(H20) as discussed above. The low-spin form shows a rather large g-anisotropy with g = (1.08, 2.02, 3.32) and an orien- tation of g, in the haem plane intermediate between that of the high spin form and the N; forms (see Figure 5). The z axis for Mbf(H20) makes angles of 39 O, 123", and 0" with the gx,gy, and g, directions respectively in MbfOCN-.22 Haernichrornes.The protein conformation of myoglobin can be distorted from its native average structure by various chemical and mechanical agents. In the native ferrous case, this can result in an increase of oxidation to the inactive ferric state and attachment of an endogenous amino-acid side-chain as a sixth ligand to the iron. This has been extensively investigated for haemoglobins but not myoglobins. Some low-spin forms are seen in aged or poorly handled samples of myoglobin, so this possibility should be kept in mind.*O Sulphrnyoglobin. A pathological physiological condition in which a single sulphur atom is added to the porphyrin ring of haemoglobin has been studied by preparing the simpler sulphmyogloblin.The ferric sulphmyoglobin gives g-values for the OH-, N3, and CN- derivatives which differ from the corresponding native deriv- ative.28 3 Monomeric Species: Ferrous Forms The liganded ferrous forms of myoglobin are of keen interest because they more closely resemble native oxygenated myoglobin than do the ferric forms. The nitro- syl derivatives and the superoxide derivative that is formed by electron addition to MbO, at low temperature, however, are the only species to have an unpaired electron, which comes from the ligand. To what extent these are ferrous derivatives is discussed below. A. MbNO Most studies of nitrosyl haemproteins have been on frozen solution samples that yield spectra which are, in general, complicated by overlap and not unambiguous resolution. However, two types of spectra are seen for MbNO as discussed by two recent papers on the temperature dependence of the c.s.r.spectra. A frozen solu-tion study by Morse and Chan29 resolved the e.s.r. features into two components, one with a rhombic (type I) tensor (2.080, 1.998, 1.979) and the other with an axial (type 11) g tensor (2.041, 1.983). A recent single crystal study by Hori et~l.~O 28 J. A. Berzofsky. J. Peisach. and E. W. Blumberg, J. Rid. Clirm., 1971, 246. 3367 29 R. H. Morse and S. I. Chan. J. Siol. Chem., 1980, 255, 7876. 3o H. Hori, M. Ikeda-Saito. T. Yometani. J. Bid. Clicm., 1981, 256. 7849. 397 Electron Spin Resonance qf Haemoglobin and Myoglobin resolved similar rhombic g-tensors at low temperature (2.076,2.002, 1.979) as well as the complete hyperfine tensor (15, 17, 19G) for one species. A second species, observed in MbNO could not be fully described.Room temperature spectra gave not only a different g-tensor (2.050, 2.022, 1.993) and non-coincident A(I4N)-tensor (10.8, 14.6, 18.6) but the principal g-values (but not the A-values), had radically changed their direction! 14N hyperfine splitting was seen along the direction of gminin all cases. This has very important implications for changes of haem environment upon sample freezing. We consider herein only the room temperature data. This effect was not considered in earlier single crystal work on MbNO at 77KZ9ENDOR spectra on MbNO have been obtained31 that yielded, in addition to the above 14N splittings, 'H splittings assignable to meso-protons of the porphyrin, N,, of the distal histidine, and the methyl protons of valine El 1 on the distal side of the haem pocket.Replacement of the native iron with S7Fe (I = t)gave a splitting of 5.5G observed only along gmin.32For completeness, it is mentioned that a third type of nitrosyl-haem e.s.r. spectrum has been and only recently assigned as a six co-ordinated complex with variant Fe-N distances and bond angles.34 Among the many studies there have been warnings as to the use of e.s.r. spectra in distinguishing clearly between five and six co-ordinate nitrosyl haems and one cannot be too cautious about the ease of generation of denatured species.35 In solution at 26 "C MbNO shows an apparently axial e.s.r. spectrum with g,, g, = 2.026, and g3 = 1.998.The spectrum is not isotropic because of the slow tumbling time of the protein molecule. The spectrum is unchanged over a pH range of at least 6.3-9.4.36 Because of the convenience of describing directional results from single crystal studies by stereographic projections, we insert here a description of this depiction (Figure 6) so that the reader may easily grasp their content. The directional results from single crystal work on MbN037*38 are shown in Figure 7. The data of reference 38 have been transformed to a co-ordinate system constructed from vectors connecting nitrogen atoms I and IT, and connecting atoms I and 111.The cross product of these vectors comes out of the plane of the paper and represents the haem normal on the distal side of the haem pocket. The co-ordinates used are Kendrew'~.~~No great stress is to be laid on the exact angles but we wish to discuss the general location of components. It must be realized that for all myoglobin type A crystals, which are P2, monoclinic space group, there are two haems per unit cell. In the case of met- myoglobin where the g-tensor was axially symmetric, it was easy to assign the e.s.r. 31 M. Hohn and J. Hiittermann, personal communication. 32 L. C. Dickinson and J. C. W. Chien, Biochem. Biophys. Res. Commun., 1973, 51, 587. 33 T. Yonetani, H. Yamamoto, J. E. Erman, J. S. Leigh, jun., and G.H.Reed, J. Biol. Chem., 1972, 247, 2447. 34 T. Yoshimura, T. Ozak, Y. Shintani, and H. Watanabe, Arch. Biochem. Biophys., 1979. 193, 301. 3s D. M. Scholler, M.-Y. R. Wang, and B. M. Hoffman, J. Biol.Chem., 1979, 254, 4072. 36 E. Trittlevitz, H. Sick, and K. Gersonde, Eur. J. Biochem., 1972, 31, 578. 37 L. C. Dickinson and J. C. W. Chien, J. Am. Chem. Soc., 1971, 95, 5036. 38 H. Hori, M. Ikeda-Shito, and T. Yonetani, J. Biol. Chem., 1981, 256, 7849. 39 J. E. Kendrew and H. C. Weston, personal communication. Dirkinson and Symons A I n Figure 6 Stereographic projection of' principal directions qf'tensors ure the hest rvuy to displayall ofthe directional infbrmution,from e.s.r. crystallography at u glance. The projection can be simply envisioned by beginning with u sphere of unit vector rudius about an origin.We take the hem plune as the plane onto which we project the unit vectors of the principul directions. ii is the normal to the haem plane. The projection point in the harm plane is theii,formetl by 'sighting along' the line containing the tip of the vector, Cx, to be projected und the point at the -ri (or i-i? if Ci is below the plune). The point, X,at which the line strikes through the huem plune then represents the vector V, in projection. If V, is above the hucm plane, MV represent X by u solid symbol, if below, by an unfilled symbol. Stereographic projwtion distorts by spreading udjuccwt vectors near n and adjacent vectors near the haem plune, but is very useful,fiw cleurly combining relationships of vectors in a single diagram.The linear radial scale ,from the centre qf the projection plune has value.sfrom 0 to 1.0 and are converted easily to ungle vulues b.v the table below Lineur value 0.0 Angle to n 0 0.1 5 0.2 12 0.3 18 0.4 27 0.5 36 0.6 47 0.7 57 0.8 67 0.9 78 1.0 90 g = 2 direction to a given haem normal from X-ray crystallographic results. How-ever, this simplicity is lost once the g-tensor becomes rhombic and the rationale for assignment of one of thc two sets of e.s.r. directions to a given haem becomes indirect. In the NO case at room temperature, Hori etcrl.38assign the direction of the Fe-"(NO) bond to the minimum g-value, both because it is near frec-spin (g = 2.0) and because it is close to a haem normal direction based on other e.s.r.and X-ray results. The assumption that the haem does not move significantly seems soundly based on a number of results on several myoglobin derivatives. However, based on our assignment of the electronic structure discussed below and variations expected in g-values with angle, the assignment ofgminas the Fe- N direction does Electron Spin Resonance of Haemoglobin and Myoglobin Figure 7 Stereographic prqjection of'principal directions fixthe g-tensor and A ( ' 5N)-tensor of MbNO onto haem co-ordinates as described in text. Both projections arc shown us there arc two sites observed and there is no a priori way of selecting which one belongs to which haem., A-tensor component direction; A,g-tensor component dircction; A, = 15.6, A, = 21.4; AZ = 26.7; g, = 2.050, g, = 2.022, g, = 1.993 not seem on secure ground. Additionally, it should be pointed out that as g,,, (=g,) lies along the b (dyad) axis of the crystals, it remains the same in both haem co-ordinate systems of the crystal, but depending upon which set of e.s.r. tensors is assigned to which haem co-ordinate system, the directions of g,, and g, are virtually reversed. While there are arguments for assigning a given e.s.r. tensor to one haem in the crystal, as presented in the cited papers, we wish to emphasize that these assignments are by no means final. In the MbNO case, as mentioned above, there is no ambiguity in the assignment of gmaxto (8,) in the haem co-ordinate system, because of a coincidence of the direction of the x vector in that system.However, the assignment of 2 and is open to question. These are the near free-spin g-values. As we argue below, the negative deviations from g = 2.0023 can arise from several terms so there can be no predic- tion as to whether gminwill occur perpendicular to the haem normal or perpendic- ular to the Fe-NO plane. The nitrogen hyperfine splitting shows a rather small anisotropy and cannot be resolved into axial dipolar tensors in a convincing way so that it does not contribute unambiguously to the assignment of a given tensor to a set of haem co-ordinates. Thus with MbNO the e.s.r. data in single crystals, except for the g,,, assignment, can only be ambiguously related to the haem co-ordinates. As a further caveat on the interpretation of single crystal e.s.r. data it must be realized that standard methods of diagonalization may be misleading in cases where the point group symmetry drops below orthorhombic. This situation has been recently reviewed40 and it seems that off-diagonal elements in the g-tensor are insignificant, but that those for the A-tensor can be quite significant resulting in 40 J.Piibrow and M. R.Lowrey, Rep. frog. PIiys., 1980, 43, 432. 400 Dickinson and Symons a rotation of the apparent principal axis of the metal hyperfine tensor with respect to those of the g-tensor. The electronic structural meaning of these off-diagonal elements in A cannot be simply interpreted.For cases where no axes of g and A are coincident, such as we observe here, the theory has not been derived. Thus one may expect that future analysis may greatly modify the directional data currently derived from e.s.r. for these haem proteins. B. MbO, This species is of recent discovery. When oxymyoglobin was y-irradiated in an aqueous glass at 77 K, electron capture occurred at the FeO, This species is isoelectronic with '"MbO,, discussed below. Principal values of the g-tensor of the species formed (2.20, 2.11, 1.97) show smaller anisotropy than has been ob- served for normal low-spin Fe3 + forms, as is shown in Figure 8. A second signal, resolved in glycol-water glasses when the sample was annealed to temperatures above 77 K, had a larger anisotropy with g(2.295, 2.164, 1.942). The bonding in this complex requires the unpaired electron to be in a molecular orbital which includes the iron prominently and yet has sufficient density on the peroxy-ligand to explain the major splittings of 65 and 52 G for the '0 atoms.The orientation of the g-tensor with respect to the haem axes (Figure 9) has been determined41 and the results have been used to rationalize t_he bonding scheme (see below). The MbO, case offers a complicated electronic structure in that the spin-orbit coupling on the iron atom increases the g-anisotropy. The direction of g,,, is expected to be near the normal of the haem plane. We have revised interpretation of the single crystal results given by Symons and Peter~on,~' and xis now believed to be about 30" off the haem normal.If our predictions are correct and gminis along the 'hinge orbital' (n*)direction then x is roughly the direction of the 0-0 tilt. The results can be compared with the recent X-ray diffraction study of Phillips, who found the dioxygen tilt in oxymyoglobin to be ca. 65" from the haem normal.4 C. Electronic Structure For the normal ferric derivatives, high- or low-spin, there are no serious problems regarding electronic structure. In particular, e.s.r. studies clearly establish that the unpaired electrons are strongly confined to the metal atom, with relatively minor delocalization onto the porphin and histidine ligands. However, for the NO and 0, derivatives there are several reasonable alternative structures, some with exten- sive delocalization onto these ligands.The structural problems are far from being resolved, but it may help if we recall some results for electron addition to the nitroprusside ion, [Fe(CN),N0I2 -. Elec-tron addition at 77 K gave a species in which the SOMO is largely confined to one of the n* orbitals of NO. The relatively small value for g, (along the NO bond) of 51 M. C. R. Symons and R. L. Peterson, Biochem. Biophys. Actu, 1978, 535, 241; M. C. R. Symons and R. L. Peterson. Proc. R. Soc. London, Ser. B, 1978, 201, 285-300. 42 S. E. V. Phillips. J. Mol. Bid., 1980, 142, 531. 40 1 Electron Spin Resonance of Haemoglobin and Myoglobin 0i A/I/ I I 0 I I 3.0 2.5 2.0 1.5 9 Values Figure 8 Display of g-values jbr an arbitrarily selected range of low-spin ,firric haem com-plexes ( Ag gin,;Vgmin).The trend in g,,, va1ue.v has been plotted linearlv, and dutu,for the various ?%Fof(FeO,)-derivutives (CI,[j, y) discussed in the text have been included.This shows the (Fe0,)-derivatives, with extensive clelocalizution qf the unpaired electron onto oxygen I .928 shows that the n-degeneracy is strongly lifted, there must have been a major bending distortion on electron addition.43 On warming, this species was lost irreversibly, being replaced by a species having a 3dj2 structure with major G* interaction with nitrogen. This was established by the form of the g-tensor com- ponents (g, $g,, z free-spin) and of the 14N hyperfine coupling (A,,> A, >> 0).It seems clear that the axial cyanide ligand moves away as the electron moves onto the metal. In this case, there is less incentive for bending, and deviations from linearity are small. E.s.r. results for MbNO and HbNO complexes can be used to distinguish between these alternatives. 43 M. C. R. Symons, J. G. Wilkinson, and D. X. West, Rudiuf. Phys. Ch~m..1982, 19, 309. 402 Dickinson and Sytnons Figure 9 Projection of the principal directions of the g-tensor (4 MbO, onto the haem co-ordinate system x = g,,,; 7 = g,,,,; z = g,,,. Both possible projections of the e..s.r. crj~.staldata are shown FeNO Derivatives. At least four different electronic structures have been proposed, attention being focused on the orbital of the unpaired electron (SOMO).We now outline these suggestions in their simplest form, and endeavour to select the most satisfactory in terms of the e,s.r. parameters. Structure (I). The o* model. This places the electron into the 3dz2iron orbital, with slight delocalization onto the N(o) orbital of N, and greater delocalization onto the NO(a, TC)orbital.45 *Yc-Structure (2). The n* (dyz)model. In this case, TC: is placed below the two ojn levels, and for this structure, the major contributor is thought to be dyz(Fe). Structure (3). The TC*(NO) model. This is the same TC;level as in (2), but the contribution from the TC*(NO) orbital is envisaged as being greater than that from the dyz(Fe) ~rbital.~~,~’ Electron Spin Resonance of Haemoglobin and Myoglobin Structure (4).Doetschman’s model. In a detailed attempt to accommodate the g-and A-values for various nitrosyl derivatives, Doets~hman~~ has proposed a level which involves an admixture of almost all possible levels, namely n: (0.18), n,* (0.35), d,, (0.24), dz2(0.23). In order to account for the large I4N (NO) hyperfine coupling he invokes admixture of the a-(NO) level in addition to all these. We assume that this implies that n,* is the (o/n)level in our scheme. In that case, this orbital somewhat resembles an admixture of our (o/.n),and n: levels. A preferred-level scheme. The reason for this proliferation of models is surely the fact that the e.s.r.results are remarkably complex, especially when directions for the various tensor components are taken into account. However, there are certain aspects of the results common to all examples which, in our view, point firmly towards structure (l), the a*-structure, as the correct one. The case for (1) is as follows:-(a) The g-shifts are small. Almost all complexes with &-type structures exhibit large g-shifts, including the (FeOJ species, which have two electrons more than the (FeNO) species. (b) The large value of Aiso(I4N) is typical of a a*-stru~ture,~~ and cannot, in our view, be accommodated by a n*-structure. Indeed, the limit of a n*-structure is NO itself, which has Aiso z 7G--a factor of three less than the values for (FeNO) complexes, which have a far lower spin density on NO.(c) The very small magnitude of the 14Nanisotropic hyperfine coupling for N, also firmly establishes a o-interaction rather than a n-interaction. The small mag- nitude of Aiso (ca. 6G) shows that delocalization onto this ligand is small, which probably indicates weak bonding. This, in turn, may explain why this bond is relatively readily broken in the R%T change. Partial displacement of this ligand accords nicely with results for Fe(CN),N03 -discussed above. (d) The form of the S7Fe coupling also accords best with a di2 configuration on iron. The maximum coupling of ca. 6G lies close to the haem normal, as required for di2.Its magnitude (ca. 6 G) requires considerable population of this level, comparison with results for CoMb indicating at least 50% spin-density therein.(e) The negative g-shift. The fact that there is a small negative shift for one of the g-values has been taken as clear evidence against the a*-str~cture.~~ However, our scheme suggests that the D*-and n,*-levels are relatively close together, in which case a negative g-shift stems naturally from coupling between them. Thus the a*-model requires that there be one negatively shifted g-value, as observed. Against structure (2) for the Fe(N0) derivatives (x*:dlz)is the small magnitude of dg. Relatively large positive g-shifts are expected for this structure. The only way to explain the small shifts observed would be to postulate very low spin- density on iron: this leads us to structure (3) (n*:NO).If the unpaired electron is largely in a pure n*-level on NO, the hyperfine 44 D. C. Doetschman, S. A. Schwartz, and S. G. Utterback, Chc~m.Phys., 1980, 49, I. 404 Dickinson and Sqwions coupling to I4N should tend towards that for NO itself, and the results should resemble those for the x*-structure of the nitroprusside ion. In particular, small isotropic and a relatively large anisotropic coupling constants to 14N are required, quite different from those observed. Again, this model would require negligible spin-density of N,. None of these expectations is fulfilled, so we neglect this model. We conclude that the simple a*-structure (I) is quite satisfactory for the nitrosyl derivatives. In particular, it nicely explains why there is some tendency for the proximal histidine ligand to move away from iron for some of these complexes.Fe0,-Derivatives. Two possible structures can again be envisaged. The oxy- derivatives contain one electron more than the NO derivatives, and it is therefore tempting to suggest that the extra electron be placed in the dzLorbital to give the diamagnetic complex. However, this neglects the extra nuclear charge on the ligand, and does not explain the large degree of bending observed for HbO, and MbO,. It seems to us possible that the dz2a*-orbital remains unoccupied for the oxy-derivatives, the extra two electrons being accommodated in one of the n*-orbitals on oxygen, with some antibonding interaction with the iron orbitals.Addition of a third electron once again presents us with a problem, since this could go into the dz2a*-orbital or into the other n*-orbital (the 'hinge' orbital). The e.s.r. results strongly support the latter concept. Thus the form of the g-tensor components is remarkably similar to that for low-spin Fe"' complexes, except that the g-shifts are reduced. Also, the form of the 7O hyperfine coupling is in accord with this n*-structure, and rules out the a*-structure. The fact that when these unstable (Hb0,)- complexes are warmed to room temperature there is loss of peroxide, leaving an Fe'" derivative, also strongly supports this formulation. The (Mb0,)- derivative discussed herein is isoelectronic with oxycobalt myo- globin, which has been extensively studied by e.s.r.spectroscopy, as outlined below. The results show that the SOMO is now largely confined to oxygen. Replacement of iron by cobalt must lower the dzzorbital so that there is now a high probability that the dz2a*-orbital will be doubly occupied. Indeed, the structure can be simply understood in terms of binding 0, t6 the deoxy-cobalt derivative, which has its unpaired electron largely confined to the dzzorbital. On binding, one of the n*-electrons of oxygen pairs with this electron, leaving the other in the 'hinge' n*-orbital largely confined to oxygen. In fact, there are several forms for these oxy-cobalt complexes, including a remarkable change in structure on cooling. It is just possible that there is an electronic switch in structure between the (nl*)2(n2*)1structure envisaged for the (Mb0,)-complexes and the (dz2),(x,*)l structure normally envisaged for the cobalt analogues.4 Haemoglobin Derivatives: Ferric Forms A. High Spin The haemoglobin tetramer has received a great deal of attention using e.s.r. techniques because the e.s.r. parameters are sensitive to subtle electronic structural changes at the 0, bonding site. Thus, for example, e.s.r. spectroscopy can be used Electron Spin Resonance of Haemoglobin and Myoglobin to detect conformation shifts between 'R' and 'T'states. One must appreciate the difficulties in partitioning the results between differences in R and T states, and differences between a-and P-~hains.~~ -47 Hbf(H20).As with myoglobin the initial classical work resulted in the deter- mination of the orientation of the haem planes relative to the crystallographic axes,9 and relative to the then crudely pictured protein chain.48 In the monoclinic horse methaemoglobin crystals there is by symmetry effectively one tetramer per unit cell, so that only four e.s.r.signals are observed. No differences can be observed between a-and P-chains in the single crystals. Early frozen-solution work with signals from isolated a- and p-chains in the ferric state led to observations of differences, later related to denat~ration.~~ An observation has also been made that the ferric a-chain in frozen solution shows a splitting of the g = 6 feature which is removed upon binding of the a-chains to P-chain~.~~ However, later studies show no such anisotropy.An extended theoretical analysis has been made to understand the anisotropy of the fundamental linewidth/relaxation times of the Hb' e.s.r. signal. It was concluded that a major contribution to TI aniso-tropy arises from nitrogen spin-state mixings2 of the electronic wave function at 4 K. No change was seen in the fundamental linewidth in Hb+ single crystals upon replacement with D20, although a reduction in the minimal linewidths or second moment was expected. A 4" haem misalignment in both a-and P-haems can explain the angular dependence of the linewidths. 53 The linewidths of haemoglobin solutions have been found not to vary significantly with temperat~re.~~ The effects of pH on the intensity and form of the e.s.r.signals of ferric haemoglobins have been followed showing alkaline (pH I I), acid (pH 4-5), and heat-denatured (pH 34) forms, as well as the conversion from the aquo into the hydroxy form. ENDOR studies of ferric haemoglobin gave similar results to those for myo- globin.I6 Hb, Hyde Park (P92His 'Tyr) does not give a nitrogen histidine ENDOR signal. In HbMilwaukee (P67Val +Glu) where the P-chains are ferric, a shift in ENDOR signals is seen with oxygenation of the a-chain,16 however, hybrid haemoglobin A(a2P2+)did not show the same shift. Other e.s.r. work with hybridss6 gave evidence for a-11 interaction in the u,'[j2 hybrid only, although the method of [1-chain preparation has been called into q~estion.~' The hyperfine 45 S.K. Mun, J. C. Chang, and T. P. Das, Proc. Notl. Accirl. Sci. USA, 1979, 76, 4842. 46 J. C. W. Chien, J. Chem. Phys., 1969. 51, 4220. 47 E. Trittlevitz. K. Gersonde, and K. H. Winterhalter. Eur. J. Biochem.. 1975, 51. 33. 48 D. J. E. Ingram, J. F. Gibson, and M. F. Perutz, Nufure (London), 1956, 178, 906. 49 Y. Henry and R. Banerjee, J. Mol. Biol., 1970, 50, 99. J. Peisach, W. E. Blumberg, B. Wittenberg, J. Wittenberg, and L. Kampa, Proc. Null. Acad. Sci. USA, 1969, 63, 934. 51 J. Peisach, W. E. Blumberg, S. Ogawa. E. A. Rachmilewitz, and R. Oltzik, J. Bid. C'hrw., 1971,246, 3342. 52 A. S. Brill, C.-I. Shyr, and T. C. Walker, Mol. Piiys., 1975, 29, 437. 53 A. S. Brill and D. A. Hampton, Biophy.s.J., 1979. 25,313. 54 T. Asakura, G. Reed, and J. S. Leigh, jun., Biorhnistry, 1972, 11, 334. 55 T. C. Hollocher and T. M. Buckley, J. Biol. C'hem., 1966, 241, 2976. 56 R. Bannerjee, F. Stekowski, and Y. Henry, J. Mol. Biol.. 1973, 73, 455. 57 M. F. Perutz, E. J. Heidner, J. E. Ladner, J. G. Beethstone, C. Ho, and E. F. Slade, Biorhunistry, 1974, 132, 187. 406 Dickinson and Sjimons splitting parameters for 57Fe are about 2% smaller than in ferric my~globin.~~ Hb'F-. Hb'F-gives an histidine ENDOR splitting about 0.5 MHz lower than in Hb+(H,O) indicating that there is a trans effect of fluoride through the haem plane.' The zero-field splitting in Hb' F-- is also substantially reduced compared to Hb+H20.lS We conclude that no clear cut inequivalence of Q-and a-chain structures is seen in high-spin ferric haemoglobin.B. Low Spin The OH-, CN-, and N; derivatives of haemoglobin have been studied to some degree but none show significant interpretable differences from the corresponding myoglobin derivatives discussed above and none show any a-fi inequivalence. Attention has been addressed to low-spin forms generated by protcin dena- turation.sl,5s ENDOR spectra for Hbi(13CN-) show a 1.5 MHz (12%) smaller z component of the A(13C) in haemoglobin as compared with myoglobin. How- ever, the splitting from rneso-protons is l .19 MHz in myoglobin cyanide and l .34 in haemoglobin cyanide. The ENDOR of haemoglobin azide is also somewhat different from that of myoglobin azide with an unclear but possible indication of alpha-beta inequivalence in the former.' 5 Haemoglobin Derivatives: Ferrous Forms A.Nitrosylhaemog1obin.-In many ways, for the e.s.r. spectroscopist this is the most important haemoglobin derivative since the spectra of HbaNO and HbgNO are quite distinct and the e.s.r. spectra are remarkably sensitive to the R -+ T shift in conformation of the tetramer. This haemoglobin derivative was curiously neglected until about 10 years ago following a paper on horse single-crystal HbN046 and model compoundss9 though there were earlier cursory reports on this derivative.60 -62 Rein et aL60 noted that the tetramer, human haemoglobin A, gave two distinct types of e.s.r. spectra (Figure 10) which were interconverted by addition of inositol hexaphosphate, IHP, an effector molecule known to promote the R -+Ttransition.In the absence of IHP the rhombic spectrum has a feature at g = 2.06 and a number of less well resolved shoulders in the g = 2.0 region. The addition of IHP gives rise to a complicating change in the g = 2.06 region and a very characteristic three line hyperfine pattern centred at g = 2.00 split by 16.6G. Work on separated subunits6' showed that the HbNO Q-and P-chains gave distinct e.s.r. spectra, the fi-subunit spectrum being more symmetric than that of the a. At pH7.4 the 58 C. P. Scholes, R. A. Isaacson, T. Tohetani, and G. Feher, Biochem. Biophys. kta, 1973, 322, 457. 59 ((I) H. Kon. J. Biol. Chem., 1968, 243.4350. (h)H. Kon and N. Katoaka, Biochemistr!, 1969,8, 4757.60 H. Rein, 0. Ristau, and W. Scheler, FEBS Lott., 1972. 24, 24. b' T. Shiga. K.-J. Hwang, and I. Tyuma, J. Biol. Chem., 1968. 243. 203. 62 (a)D. J. E. Ingrain and J. E. Bennett, Discuss. Fururluy Soc., 1955, 19, 140. (h)W. Gordy and N. Rexroad, in 'Free Radicals in Biological Systems,' ed. M. S. Blois ef uI.,Academic Press, New York, 1961, p.263. (c) K. M. Sancier, G. Freeman. and J. S. Mills, Scicwc., 1962, 137, 752. (d)A. Ehrenberg, Avkiii. Kewi. 1962, 19, 119. 407 Electron Spin Resonance of Haernoglobin and Myoglobin Figure 10 E.s.r. spectra of Hb,NO ut pH 6.3 with and without IHP spectrum of the combined tetramer was that of the arithmetic sum of the spectra of the separated chains. The e.s.r.spectra of a number of nitrosyl haemoprotein complexes have been grouped into two types based on g, < or > 1.96,63 and more recently a third type of haem-NO e.s.r. spectrum has been characterized, as discussed in the section on MbNO above. This work has been extended by Henry and Banerjee64 to a study of HbNO chains and tetrameric hybrids of the type cr,(X)P,(NO) or a,(NO)P,(X) where X means deoxy, oxy, carboxy, or ferric (H,O, F-, N3, or CN-). The e.s.r. spectrum of a,(NO)P,(X) was found to depend upon the spin state carried by the haem in the P-chain. For low-spin X the e.s.r. spectrum of a,(NO)[j,(X) was identical to that of the free chain, but for high-spin X the e.s.r. spectrum was modified. The ct2(X)f12(NO)spectra were found to be invariant with X.Further work by Gersonde and co-~orkers~~ on HbNO extended careful mea- surement with both 14N0 and l5NO to several mutant haemoglobins. Although 63 T. Yonetani, H. Yamamoto, J. E. Erman, J. S. Leigh, jun., and G. H. Reed, J. Bid. Ckrm., 1972,247. 2447. O4 Y. Henry and R. Banerjee, J. Mol. Biol.. 1973, 73, 469. 65 E. Trittlevitz. K. Gersonde, and K. Winterhalter, Eur. J. Biochern., 1975, 51, 33. Dickinson and Symons structural inequivalence is unambiguously demonstrated by the inequivalence of e.s.r. spectra of a-and fl-nitrosyl derivatives, the two subunits have been shown by rapid-freeze e.s.r. techniques to react at the same rate with In contrast, a long time-scale NO titration technique showed that fl-haems react in deoxy- haemoglobin four-fold faster than a-haems.68 Trout blood contains four distinct haemoglobin tetramers, which show a range of oxygen affinities and dependence of P,, on pH and temperature, thus offering a variety of examples of R-T shift with conditions.Component 4,with a strongly pH-dependent oxygen affinity, also shows a pH dependence of the e.s.r. spectrum of Hb,,NO. Component 1, with pH-independent oxygen affinity, shows no change in the e.s.r. spectrum of Hb,NO with PH.~’ A frozen-solution work on Hb,,,,,,NO suggested the presence of four spectral component^.^^ Subsequent single-crystal work on Hb,,,,,,NO, which is locked in the T state, showed two types of haem-NO moieties. The fl-haems show a splitting from the 14NE histidine atom, and the a-haems show no 14NE histidine splitting.70 This gave e.s.r.spectral evidence for five and six co-ordinated haems. It is inter-esting to note that infra-red spectroscopic evidence7 had been previously used to infer such a change in structure. This structural difference was first deduced from frozen-solution e.s.r. by Szabo and Perut~.’~ In marked contrast, a study of carp nitrosyl haemoglobin in the R and T states reveals only slight differences for these two forms in the e.s.r. parameters in comparison with human nitrosyl hacmoglobin. The authors stress the ease of generation of denatured HbNO species35 and the fact that this has a unique e.s.r. spectrum assigned to the ligand-off species. A very recent single crystal e.s.r.study of Hb,NO by Doetschman and Utterba~k’~reported the photolytic replacement of NO by 0, and re-attachment kinetics of NO in subunits at 9T finding that the a-subunits exchange with 0, faster than /?-subunits. The crystal e.s.r. spectra of the tetramer should display hyperfine coupling to 14N0 and [14N]histidine ligands for both a- and P-chains. B. Peroxyhaemog1obin.-As was described above for y-irradiated oxymyoglobin, electron-addition centres were first discovered in o~yhaemoglobin.~~ Because these centres are isoelectronic with oxycobalt haemoglobin they offer an exciting struc- tural pair for comparison. Additionally, separate centres were found that are associated with a- and with fl-chains. Further centres are formed upon annealing to higher temperature and upon using a different matrix, i.e., glycol versus water- b u ffe r on 1y .66 R. Hille, G. Palmer, and J. S. Olson J. Biol. Chcw.. 1977, 252, 403. 67 P. Reisberg, J. S. Olson, and G. Palmer, J. Riol. Cliem.. 1976, 251, 4379. 68 M. Brunori, G. Falcioni, and G. Rotilio. Proc. Ncit. Auid. Sci. USA, 1974, 71, 2470. 69 H. Twifler and K. Gersonde, Z. Naturforsck., Ted. C‘, 1976. 31. 664. ’O J. C. W. Chien and L. C. Dickinson, J. Biol. Chmi.. 1977, 252, 1331. 71 J. C. Maxwell and W. S. Caughey, Biociiemi.ytr.v, 1976, 15, 3XX. 72 A. Szabo and M. Perutz, Biocliemistrj,. 1976, 15, 4427. 73 D. Doetschman and S. G. Utterback, J. Am. Ckm. Soc., 19x1. 103, 2847. 409 Electron Spin Resonance qf Huemoglobin and Myoglobin 6 Metal-replaced Derivatives The replacement of iron by other metal ions can yield species which are benefically studied by e.s.r.As with myoglobin, oxy- and deoxy-cobalt haemoglobins are both paramagnetic. ''Hb reversibly and co-operatively binds 0, at a reduced affinity compared to Hb.74* 75 A. Cobalt Derivatives "Hb. The first successful metal-replacement technique for cobalt was accom- plished by H~ffman.'~ Subsequent work in other laboratories led to a single crystal of horse "Hb which demonstrated that the haem normals differ in direc- tion from those in methaemoglobin crystals by as much as 30 ". Resolution in one plane was poor so that the complete tensor and orientation could not be deter- mined. Frozen solution spectra give a large A ll(Co)= 76 G, with I4N splitting of 17G. Proto-, meso-, and deutero-cobalt porphyrins were substituted into hae-moglobin by Yonetani et al.with slight changes resulting in the e.s.r. parameters. A table of e.s.r. properties for these species can be found in reference 75 for the oxy- and deoxy-derivatives. These workers observed a small but distinct narrowing of the e.s.r. lines when D20 replaced H20 in solutions of "HbO,. This is taken as evidence that a water molecule is involved in the influence of the distal histidine upon oxygen bonding. This effect was not seen in Glycera haemoglobin which lacks a distal histidine." Subsequent of 1702-c0Hb showed that there are two non-equivalent oxygen atoms which eliminated the Griffith n-bonding model for the Co-0--0 bond.A very productive aspect of metal replacement has been the preparation of hybrid haemoglobins, that is, tetramers such as a,(Co)P,(Fe) or a,(Fe)p,(C~).~~The iron-containing subunit is e.s.r. silent and thus the differences between Co in the a-and P-subunits can be observed by e.s.r. in the tetrameric state. The e.s.r. spectra of each oxygenated hybrid differ, but each is the same as the spectrum of the corresponding separated subunit.80 The tetrameric '"Hb gives an e.s.r. spectrum which is the arithmetic sum of the spectra of the separated subunits. The main difference between the e.s.r. spectra of '"Hb,(O,) and '"Hb,(O,) is that the former has sharper lines. The deoxy-separated chains give indistinguishable e.s.r.spectra. Thus the Co-Fe hybrids are not useful indicators of co-operative effects between subunits. '"Mb. As with the ferric forms, cobalt myoglobin offers a simpler case more amenable to single-crystal study than cobalt haemoglobin. 'OM b gives e.s.r. param- eters shown in Table 2. The orientation of the g-tensorsl was found to be identical 74 B. M. Hoffman and D. H. Petering, /'roc,. Noti. Aud. Sci. USA. 1970. 67. 637. 75 T. Yonetani, H. Yamamoto, and T. Itzuka, J. Biol. Chem., 1974, 249, 2168. l6 L. C. Dickinson and J. C. W. Chien, Biochem. Bioph>.s. Rcs. Commun.. 1973, 51, 587. " M. Ikeda-Saito, T. Itzuki, H. Yamamoto. F. J. Kayne, and T. Yonelani, J. Biol. Chem.. 1977, 252, 4882. l8 R. J. Gupta, A. S. Mildvan, T. Yonetani, and T.S. Srivastava, Biochrm. Biophys. Re.\. Commun., 1975. 67, 1005. 79 M. Ikeda-Saito. H. Yamamoto, and T. Yonetani, J. Biol. Chcm.. 1977, 252. 8639. 8o M. Ikeda-Saito, H. Yamamoto. K. Imai. F. J. Kayme, and T. Yonetani, J. Bid. Chem., 1977.252,620. 8' L. C. Dickinson and J. C. W. Chien, Proc. hicrtl. Acad. Sci. USA, 1972, 69. 2783. Dickinson und Symons Table 2 E.s.r.parumcters for cohult haemoglohin und niyoglohin g1,1 K1.2 R11 A1('O AIIC'O (Ih '"Mb 2.33 2.32 2.02 6 79 17.5 7 "MbO, 1.989 2.006 2.083 16.7, 5.95 9.3 -7 '"Hb 2.31 2.037 6.2 76 17.5 6, 12 'OMbO, 2.00 2.008 -16 2 to that in F'Mb+(H,O) with respect to the crystallographic axes. A very small anisotropy is observed in the haem plane g-value. Because "Mb is an S = t species there are no zero-field complications as in the case of iron species. Thus, this g-anisotropy must arise from ligand influence.'"MbO,. The single crystal '"MbO, study showed that the g-tensor and the cobalt hyperfine tensor do not share the same principal axes.8 In fact, it was assumed that the cobalt hyperfine tensor had a principal plane in the porphyrin plane and that the g-tensor was orientated with the 0-0 axis along the major g-principal value direction. The latter assumption was proven correct in single crystal '"M b' 702 studies, but some corrections to the first assumption causcd a revision of the estimated Co-0-0 angle from 90" to 120'.82 The complete "0 hyperfine tensor was resolved for each inequivalent oxygen atom allowing spin-density calculations for both n-orbitals of the 0, moiety.The e.s.r. data for CoMb0274. show a markedly reduced 59C0hyperfine797 interaction, typical of a superoxide derivative rather than a Cot]species. Thus, in terms of the n* structure postulated for the isoelectronic Mb(0,) species discussed above, the spin density has shifted markedly from the metal to the di-oxygen ligand. This is expected on electronegativety grounds for the antibonding electron, and indeed, it establishes the antibonding character of the orbital. The 59C0 coupling can be interpreted in terms of a maximum of about 10% spin population of the cobalt dyzorbital, but it may be less than this, since spin-polarization is expected to make a considerable contribution also.Studies of 70,-enriched samples82 confirm that the spin is largely localized on the two oxygen atoms. The data also confirm the ny*character of the orbital in good accord with our simple model. One important aspect of this works' was the discovery of two forms of '"MbO, at 77 K which differ markedly in their g- and A-tensor orientation with respect to the haem plane. It now seems that at room temperature there exists only one form, which clearly differs from both low-temperature forms.83* 84 This result ties in well with the behaviour of MbNO outlined above. It is of great importance in that it implies a significant change in protein conformation upon freezing. Since many such studies are carried out at 77 K or below, the possibility of such changes must always be borne in mind.As in all these studies, difficulties are cxperienced when analysing tensor direc- 82 L. C. Dickinson and J. C. W. Chien, Proc. Natl. Acuci. Sci. USA. 1980. 77, 1235 83 H. Hori, M. Ikeda-Saito, and T. Yonetani, J. Biol. Chem., 1982, 257, 3636. 84 H. Hori, M. Ikeda-Saito, and T. Yonetani. Nature (London), 1980, 288, 501. 41 1 Electron Spin Resonance of Haemoglobin and Myoglobin tions. Ideally this should be done in the light of single-crystal X-ray results at the same temperature. The near room temperature e.s.r. results have indeed been compared with the -3OT X-ray data, but the placement of the near free spin g-value for ''MbO, along the 0-0 axis leads to serious contradiction with all previous sense of the theory of bound n-ligands.Thus some error must be sus-pected. Our computation of the g and A principal directions in the porphyrin co-ordinate frame from the direction cosines in the references 83 and 84 do not lead to the directions reported in those references. Further, as discussed above there is a two-fold ambiguity in assigning a given tensor orientation to either of the two haems per unit cell. The two possibilities are depicted in Figure 1 1. The correspon- Figure 11 Stereographic projections .for principal g and A directions of '"MbO, at room temperature.for our assignment as discussed in the text which disagrees with that of the original authors.8'-82A,A component direction, 0,g-component direclion. A, = 186, A, = 11, AZ = 1.3; g, = 2.056, g, = 2.01 1, g, = 2.003.Both possible projections are shown, although we favour the projwtion on the left dence of the 0-0 hirection with g = 2.00 presents us with severe difficulties. Our expectations for an 0; ligand is that g,,, should lie along the 0-0 bond direc- tion, as with 0; itself. Indeed this argument was originally used to decide upon the Co-0-0 bond angles and direction for the 77 K species8 However, it is inferred for the room temperature data that gmin(2.003) lies close to this direction! We reject this assignment on the grounds that 0, is a n-ligand and must have its largest g-value along, or nearly along, the 0-0 axis. Our own calculations based on the angles given by Hori et al. yield similar results for the most reasonable assignment of the '"A tensor.That is, z is near the porphyrin normal, and x and JJare near the porphyrin plane. Our calculations of the g-tensor orientation do not agree with those of Hori et al. in that gmax(gx) is found to lie near the X-ray crystallographic O(1)-0(2) direction, g = 2.03 is within about 18 of the porphyrin normal, and g = 1.98 is 43 O from the porphyrin plane. This seems to remove a major objection to our general bonding scheme. The observation that the '0 principal directions coincide with the g-tensor direction Dickinson and Symons also supports our scheme. The major splitting is along the hinge direction, or J direction in Figure 1 la, thus supporting n*-occupancy for '"MbO,. When oxygenated '"MbO, containing mesoporphyrin is photolysed at low tem- perature and observed at high microwave power, broad features appear in the e.s.r.spectrum at g = 3.87 and 1.9,attributable to an intermediate containing high-spin cobalt (S = 5). This has been observed in '"Mb of several species.75. 77* 833 85 y-Radiolysis of ColI1myoglobin at 77 K has been shown to generate in high yield a six co-ordinate Co" species which is not observable in room temperature CO" solutions. This means that CoilL myoglobin has an endogenous strong sixth ligand, probably the distal histidine nitrogen.86 B. Manganese Derivatives Manganese porphyrins substituted into haemoglobin and myoglobin give e.s.r. spectra which are somewhat sensitive to the environment of the prosthetic group and are thus of value as probes. Mn" porphyrin complexes, like the Fellr ones, have five unpaired electrons, so that observed g-factors are dependent upon zero-field splitting parameters. Unfortunately, M"Hb and M"Mb oxidise very rapidly to Mn1I1 and the oxygenated state cannot be observed.Frozen-solution spectra of M"Hb at X-band show a prominent seven-line feature at g = 5.9 and a weaker sextet centred at g = 2.0. The hyperfine splitting is approximately A, = 0.0073 cm-', AII = 0.011 cm-1.87 The hyperfine lines on the perpendicular region serve as an effective vernier on the g-rhombicity and this is used to estimate zero-field splitting. For M"Hb, D = 0.5, E = 0.0056 cm-'; for M"Mb the respective parameters are 0.56 and <0.0028crn-l.No single crystal e.s.r. work on M"Hb or M"Mb has been published, although MnHb+ has been shown to crystallize isomorphously with the native Hb'.88 C. Zinc Derivatives '"Hb and '"Mb have been prepared and give upon photolysis an excited triplet state. The e.s.r. parameters of these species are sensitive to the environment of the zinc porphyrin and to the presence or absence of the vinyl side-chains on the porphyrin. D. Other Derivatives Metalloporphyrins of copper, silver, nickel, chromium. ruthenium, and rhenium have been added to apohaemoglobin and apomyoglobin. The copper and silver(r1) derivatives show hyperfine splitting in the e.s.r. spectrum. Nickel(i1) surprisingly gives an e.s.r. spectrum indicating high symmetry and the chromium(rrr) complex 85 M.Ikeda-Saito, M. Brunori, and T. Yonetani. Biorhem. Biophys. Actu, 1978, 533, 173. L. C. Dickinson and M. C. R. Symons, J. P/IJ.F.Chem., 1982, 86, 917. B. M. Hoffman, Q. H. Gibson, C. Bull, R. H. Crepeau, S. J. Edelstein, R. G. Fisher, and J. J. McDonald, Ann. N.Y. Acud. Sci.USA, 1974, 224, 1975. K. Moffat, R. S. Loe, and B. M. Hoffman, J. Am. Chem. Soc., 1974, 96, 5259. 8g B. M. Hoffman, J. Am. C'hem. Soc.., 1975, 97, 1688. 413 Electron Spin Resonance of Haemoglobin and Myoglobin gives spectra very different from those for model compounds, showing significant interaction with the protein ligand~.~~ 7 Leghaemoglobin This review has been concerned with mammalian haemoglobin. We conclude by making brief reference to some e.s.r.studies of a species of haemoglobin which has been extracted from the roots of soy bean and from cow pea plants. It differs from mammalian haemoglobin in its amino-acid sequence, but is very similar in the region of the haem group. It is normally described as leghaemoglobin. E.s.r. studies of the ferric form co-ordinated with F-, OH-, N,, CN -,and MeCO; have been reported, the results being similar to those for the corresponding mammalian haemoglobin derivative^.^^' 92 The nitrosyl derivative has also been as have the oxy- and deoxy-forms of the cobalt deri~ative.~~ 90 T. S. Srivastava and T. Yonetani, Fed. Proc., 1974, 33, 1449. 91 C. A. Appleby, W. E. Blumberg, J. Peisach, and B. J. Wittenberg, J. Bid. C’hern., 1976, 251, 6090. 92 S. Maskall, J. F. Gibson, and P. J. Dart, Biochem. J., 1977, 167, 435. 93 M. Cristahl, A. Roap, and K. Gersonde, Biophys. Strucl. M~ch.,1981, 7, 171. 414
ISSN:0306-0012
DOI:10.1039/CS9831200387
出版商:RSC
年代:1983
数据来源: RSC
|
6. |
Hydrido complexes of the transition metals |
|
Chemical Society Reviews,
Volume 12,
Issue 4,
1983,
Page 415-452
D. S. Moore,
Preview
|
PDF (2558KB)
|
|
摘要:
Hydrido Complexes of the Transition Metals By D. S. Moore DEPARTMENT OF CHEMISTRY, DOVER COLLEGE, DOVER, KENT S. D. Robinson DEPARTMENT OF CHEMISTRY, KING’S COLLEGE, STRAND, LONDON WCZR 2LS 1 Introduction The ability of molecular hydrogen to interact with transition metals was first recognized in the early decades of the 19th century. In 1823 Dobereiner developed a spongy form of platinum capable of catalysing the combination of hydrogen and atmospheric oxygen at room temperature.’ In the same year Dulong and Thenard achieved a similar effect with finely divided palladium and iridium.2 The first well established binary hydride, CuH, was prepared by Wurtz in 1844: and the capacity of palladium metal to occlude large volumes of hydrogen gas was first explored by Graham in 1866.4 From these and similar early observations rich areas of solid state chemistry, surface chemistry, and heterogeneous catalysis have developed.Binary hydride phases are now known for the early transition metals (Ti. Zr, Hf, V, Nb, Ta, and Cr), palladium, the lanthanides and most of the actinide^.^ The ability of transition metal surfaces to chcmisorb and thus activate gaseous hydrogen has been intensively studied over many years and has found important applications in heterogeneous catalysis. Finally, current interest in molecular hydrogen as a fuel has led to much recent work on the use of d and f’block metal alloys as agents for its safe storage and transport in high concentrations.6 In marked contrast the discovery and characterization of molecular complexes containing hydride ligands co-ordinated to transition metals occurred relatively recently.Although earlier claims exist, the first fully authenticated complex hy-drides appear to be the carbonyl hydrides FeH,(CO), and CoH(CO),, reported by Hieber in 193 1 and 1934 re~pectively.~ Following Hieber’s initial discovery, almost 25 years elapsed before further examples were identified and characterized. During this period several new hydrido complexes were undoubtedly synthesized but, in J. W. Dobereiner, Ann. Chini. Phys., 1823, 24, 91. P. L. Dulong and L. G. Thenard, Ann. Chim. Phys., 1823, 23, 440. 4. Wurtz, Compt. Rend., 1844, 18, 702. T. Graham, Philos. Trans. R. SOL‘.London, 1866, 156, 415.G. G. Libowitz, ‘The Solid State Chemistry of Binary Metal Hydrides’, Benjamin, New York, 1965; K. M. MacKay, ‘Hydrogen Compounds of the Metallic Elements’, E. and F. N. Spon Ltd.. London, 1966: G. C. Bond, ‘Catalysis by Metals’. Academic Press, New York, 1962. J. J. Reilly and G. D. Sandrock, Scientific Americcm, 1980, 242, 98.’ W. Hieber and F. Leutert, Nciturir.i,ssensc,Aafien, 1931, 19. 360; Z. Anorg. AIIg. Chern., 1932, 204. 145: W. Hieber, Z. Elektrochem., 1934, 40, 158. 415 Hvdrido Complexes of the Transition Metals the absence of modern spectroscopic techniques, were incorrectly characterized and thus passed into the literature unrecognized.8 Others were claimed but could not subsequently be ~onfirmed.~ Three seminal papers published in the mid 1950's changed this situation dramatically and initiated a period of rapid growth in the field which continues to this day.In 1955 Wilkinson and Birmingham prepared ReH(C,H,), and were able to detect the hydride ligand using the newly emergent technique of n.m.r. spectroscopy.1° The same year Fischer, Hafner, and Stahl reported the chromium and molybdenum hydrides MH(C,H,)(CO),.' I Two years later Chatt, Duncanson, and Shaw described the remarkably stable platinum(r1) hydride, tvans-PtHCI(PEt,),, and recognized the ability of phosphorus (and arsenic) donor ligands to 'stabilize' transition-metal hydrogen bonds.' Rapid progress followed these discoveries and within ten years literally hundreds of transition metal hydrido complexes had been characterized. Twenty-five years later the field is still expanding rapidly.Structural characterization of the hydride ligand by diffraction methods proved difficult because of the small X-ray scattering cross-section of the H-atom and the close proximity of the heavy transition metal atom. Early diffraction studiesI3* l4 revealed 'vacant' co-ordination sites that were presumed to contain the hydride ligands. However, in 1963 LaPlaca and IbersIS succeeded in locating the hydride ligand in RhH(CO)(PPh,), by X-ray diffraction methods and found a Rh-H distance of 1.72( 15) A. The following year Ginsberg et al. * reported a neutron diffraction study of K,ReH, that unequivocally located all nine hydrogen atoms bound to rhenium and finally settled the long-running controversy concerning the stoicheiometry of this salt.These structure determinations also established that the hydride ligand does exert a stereochemical influence, has a 'normal' metal-ligand distance and is not, as was previously suggested, buried in the metal orbitals. In 1957 Hieber, Behrens, and co-workers' ' prepared the first polynuclear metal hydrido complexes. Subsequent diffraction studies on these and related poly- nuclear species have revealed a multitude of complex structures containing p,-and p3-bridging hydride ligands. The first report of an interstitial hydride ligand encap- sulated in a non-metallic solid appeared in 1967 when Simon1s published the structure of polymeric (Nb,HI, The H-atom was not located directly but its presence at the centre of the Nb, octahedron could be inferred.More recent G. E. F. Lundell and H. B. Knowles, J. Rrs. Nat. Burrciu SI~IIZ~II~~J.1937, 18. 624; F. P. Dwyer and R. S. Nyholm, J. Proc. R. Soc. Nrw South Wrrlos, 1941,75. 122, 127: 1942.76, 133, and 1944,77, 116. '' W. Hieber and H. Stallmann, Z. E/ck/roc~licr~t..1943, 49. 2x8: W. Hieber and H. Lagally, Z. ,4norg. Allg. C'hm?.,1940, 245, 321; 1943, 251. 96. lo G. Wilkinson and J. M. Birmingham, J. An?. C'hcwi. Soc., 1955, 77, 3421. II E. 0. Fischer, W. Hafner. and H. 0. Stahl. Z.Anorg. Alig. C'hrw?., 1955. 282, 47. IZ J. Chatt, L. A. Duncanson. and B. L. Shaw. Proc CIirwt. So(,.,1957, 343. P. C. Owston. J. M. Partridge, and J. M. Rowe. Actri Cr~..s/dlogr.,1960, 13.246. l4 P. L. Orioli and L. Vaska, Proc. Clrcwi. Soc., 1962. 333. Is S. J. LaPlaca and J. A. Jbers. J. Am. Cltrm. Soc.. 1963, 85. 3501. lo S.C. Abrahams. A. P. Ginsberg. and K. Knox, Inorg. C'iicwi.. 1964. 3, 558. W. Hieber and G. Brendel. Z. Anorg. Allg. Climi., 1957, 289. 324; H. Behrens and W. Klek, Z.Anorg. Al@. C'Iwni.. 1957. 292, 151. A. Simon, Z. Aiiorg. Allg. C'lici??., 1967, 355,31 1. 416 Moore and Robinson neutron diffraction studies on M, and M molecular hydride clusters have permit- ted direct location of H atoms within the interstitial cavities of M, octahedra.lQ Although interstitial binary hydrides of the,flblock elements have been known for many years5 complex molecular hydrides of these metals have only rccently been synthesized.The first actinide hydrido complexes were reported in 197820 and the first lanthanide hydrido complexes were characterized in 1982? Finally, the ability of certain transition metal salts to activate molecular hydrogen in homoge- neous solution was first reported by Calvin in 193822 and the involvement of hydrido complexes in many catalytic proccsses, notably olefin hydrogenation, has served to further stimulate interest in their chemistry. 2 Terminal Hydride Ligands These were the first to be observed and are still the most numerous, examples are known for virtually all the d-block transition elements. The majority are stabilized by the presence of ancillary phosphine, arsine, or carbonyl ligands. A.Preparative Methods.-These are both numerous and and a selection of the more important routes is given below: (1) Hydrogenation of Metal Comp1e.w.~.Examples are given in equations 1--3; in those instances where the complex is formed in situ (equations 2 and 3) the method constitutes a direct synthesis from the metal. ~i.~-PtCI,(PEt,),+ Hz /rufz.\-PtHC1(PEt,)2+ HCI (1) 2C0 + XCO + H, -----+ 2CoH(CO), (2) Fe + 2 o-C,H,(PEt2), + H, --+ frun.s-FeH2 ((Et,P)2C,H,-o)2 (3) (ii) Hjdrolj9sis yf'Comp1o.u Sults. This method has been widely employed to prc- pare hydrido complexes containing CO or PF, ligands (equations 4 and 5). Na,[Fe(CO),] -Hs FeH,(CO), (4) Na[Rh(PF,),] -H=RhH(PF,), (iii) From Metul Salts und Reducing Agents. The 'reduction' of metal halide, carboxylate, or acetylacetonate complexes with a variety of reducing agents no- tably LiAIH,, NaBH,, N2H,, and EtOH-base has been widely employed in hydride synthesis (equations 6--9 respectively).mc~r-IrCl,(AsEt,Ph), + LiAIH, TC'F,fuc'-IrH3(AsEt,Ph), (6) '' R. W. Broach, L. F. Dahl. G. Longoni, P. Chini, A. J. Schultz, and J. M. Williams, Adv. C/7cn7.Ser.. 1978. 167, 93. 2" J. M. Manriquez. P. J. Fagan. and T. J. Marks. J. Am. C/iom.So(,..1978. 100. 3939. l' W. J. Evans. J. H. Meadows. A. L. Wayda. W. E. Hunter, and J. L. Atwood, J. Ani. C'hmi. Soc.. 1982. 104, 2008, 201s. zz M. Calvin, Trms. Firrcdrij. So(.., 1938. 34. 1181; J. Am. C/70~7.Soc.. 1939. 61, 2230. 2J For a fuller discussion of preparative methods see H. D.Kaesz and R. B. Saillant, Clicw. Rri,.. 1972. 72. 23 I; M. L. H. Green and D. J. Jones. At/i,. Itzorg. ~hiW7.R(/dk)dlW7.,1965. 7. 1 15. 417 Hydrido Complexes of' the Transition Metals CoCI,.6H,O + PEtPh, + NaBH, t'oH*CoH,(PEtPh,)3 (7) pi.r-PtCl,( PPh,), + I\i,H,.H,O E'o"*~ran.s-PtHCI(PPh,), IrCI,(CO)(PEt,Ph), + EtOH + KOH -IrHC1,(CO)(PEt,Ph)2 (8) (9) Other reducing agents employed include HCO,H, H,PO,, H,S,O,, and dialkyl aluminium compounds R,AIX (X = H or OR). The ethanol-base and formic acid reactions involve [I-elimination (equation 10) and carbon dioxide extrusion (equation I 1) steps respectively. ~-M -0~,--M--HM-CI otl, ,H + MeCI10 (10) ?'(+Me (iv) Protonation bj> Strong Acid.*. This reaction was first noted in 1955 by Wilkinson etaI.'O who found that ReH(C,H,), displays a base strcngth similar to that of ammonia and is protonated by HCI to yield the cation [ReH,(C,H,),]+. A variety of strong protonic acids including HSO,F and BF,.H,O have been used.(v) Oxidative Addition oj Weak Acids. Weak acids, HX, where the conjugate base X-is a halogen, oxygen, sulphur, nitrogen. phosphorus, carbon, silicon, germanium, or tin donor frequently generate hydrides by oxidative addition across unsaturated metal centres. These reactions are particularly common for complexes containing d8metal ions (notably RuO, OsO, Rh', Irl). A few illustrative examples are given below: RhCI(PPh,), + HSiCI, --RhHCI(SiCI,)(PPh,), + PPh, (12) Pt(PPh,), + HCN -Pt€4(CN)(PPh,)2+ PPh, (13) [Ir(dppe),]CI + PH %-c~r.~-[IrH(PH,)(dppe),]Cl (14) Carbon acids usually participate in these reactions only if the C-H bonds are activated by adjacent electron-withdrawing groups (i.e.MeNO,, HC=CR),, or, in the case of cyciometallation rca~tions,~~ are held in close proximity to the metal as part of a co-ordinated ligand (equation 15). IrCI(PPh,), -----+ IrCHCI(F77C,H,)Ph2)(PPh,), (15) However, recent work involving very electron-rich complexes has yielded a few *Since the hydride ligand and the proton are formally assigned oxidation states of -1 and -tI respectively. hydride formation by protonation involves an increase of two units in the formal oxidation state of the metal concerned. 74 D. M. Roundhill. Ah. Orgmom~/.C'/i0/77..IY75. 13. 273 and references therein. 2s C. W. Parshall. Aw. C/wm. RPS.,1970. 3. 139 and references therein. 418 Moore and Robinson examples involving cleavage of C-H bonds in simple aromatic26 (equation 16) and aliphatic27 hydrocarbon substrates (equation 17). W(C,H,),(CO) + C6H6 +WH(C6H,)(C,H,), + CO'I' Ir(C,H,)(CO), + CH,CMe, --IrH(CH,CMe,)(C,H,)(CO) + CO (17) B. Characterization.-For obvious reasons the classical technique of elemental analysis by combustion does not afford a practicable method of detecting or estimating hydride ligands in metal complexes. Chemical reactions with halogens, mineral acids, and halogenated hydrocarbons (see Section 2F) have been employed for this purpose but cannot be relied upon to give a quantitative answer.Thus IrH,(PEt,Ph), was originally formulated as IrH,(PEt,Ph), on the basis of the volume of hydrogen liberated by treatment with hydrochloric acid.28 In contrast, the hydride ligands in Cu,H,(PPh,), were estimated by acidolysis after diffraction and spectroscopic methods failed to detect their presence.2Q Molecular stereo- chemistry in metal hydride complexes can often be deduced from dipole moment measurements (the M-H bond has a small dipole moment)30 or chemical lability studies (hydride ligands process a very strong trans effcct). However, the over- whelming majority of hydride ligands are now detected and characterized by infrared and n.m.r. spectroscopy. The application of these and other spectroscopic techniques is discussed in the following section. C.Spectroscopic Properties.-These are of paramount importance in transition- metal hydride chemistry and the rapid development of the field in the past 25 years owes much to the parallel emergence of modern spectroscopic techniques, partic- ularly nuclear magnetic resonance. Spectroscopic data have been collected and tabulated in several major review article^.^' (i) Vibrational Spectroscopy. The infrared spectra of most hydrido complexes show M-H stretching and bending modes in the ranges cu. 2300-1600 and ccc. 900-600 cm -respectively. The stretching mode is usually the easiest to ob- serve and affords a valuable diagnostic test for hydride ligands, particularly in complexes unsuited to n.m.r. spcctroscopy because of paramagnetism or low solubility. In those instances where the M-H bond is of low polarity a weak or undetectable band in the infrared spectrum is often matched by a strong Raman absorption. Synthesis and examination of the corresponding deuteride offers a means of distinguishing v(M-H) from other infrared active modes-notably v(CO), v(CN), and v(NN)-that occur in the same region of the spectrum.Accord- K. I,. Tang Wong, J. L. Thomas, and H. H. Rrintzinger. J. Am. Clirvn. Soc., 1974, 96. 3694. 27 J. K. Hoyano and W. A. G. Graham, J. Am. Chcni. Soc., 1982, 104, 3723. 28 B. E. Mann, C. Masters, and €3. L. Shaw, J. C'li~~tn. Cotnmun., 1970, 703.Sot,., CIIP~. l') S. A. Bezman, M. R. Churchill. J. A. Osborn. and J. Wormald, J. Am. Chew. Soc,., 1972, 93, 2063.30 J. Chatt and G. J. Leigh, Angot*. C'I1~ni..1/11, Ed. EngI.. 197X, 17, 400. J1 A. P. Ginsberg, Trmsition Met. Clicni.. 1965. I, 11 I; H. D. Kaesr and R. B. Saillant, Clzwi. Rev.. 1972, 72. 231: G. L. Geoffroy and J. R. Lehman, Adr,. Inorg. Ciiml. Rrrtiiochcw., 1977, 20, 189; D. M. Roundhill. Adr. Orgcmoniri. Chmi., 1975, 13, 273; M. L. H. Green and D. J. Jones, Ah. Itiiirg. Cliwi, Rdioihetv., 1965. 7, 115. 419 Hydrido Comp[exes of the Transition Metals ing to the diatomic oscillator approximation the ratio v(M-H)/v(M-D) should equal 42;observed shifts are roughly in agreement with this prediction. For complexes containing carbonyl trans to hydride the Fermi resonance interaction between the two ligands leads to shifts in v(C0) and anomalous values of V(M-H)/V(M--D).~~ In those instances where v(M-D) is obscured by other absorptions the assignment must be based on the absence or reduced intensity of the v(M-H) absorption for the deuteriated sample.The value of v(M-H) reflects the nature of the ligand in the trans position, and is therefore a useful adjunct in the determination of stereochemistry; ligands with high trans influence weaken the M-H bond and produce a concomitant reduction in the M-H stretching fre- quency (Table l).33.34 The effect is particularly marked in trans-dihydrides where V(M--H),,~, can be as low as ca. 1600 cm -I. The nature of the trans-ligand also affects the solvent dependency of v(M-H). Table I Infrurcd und proton n.tn.r.dutu" fiw the mmplese s trans- PtHX( PEt,) X I Br NO, C1 SCN NO, CN v(Pt-H)/cm-' 2156 2178 2242 2183 2112 2150 2041 t(Pt-H)/p.p.m. 22.65 25.55 33.6 26.8 22.95 29.4 17.6 IJ(Pt-H)/Hz 1369 1346 1322 1275 1233 1003 778 K(Pt-H)/C.U.* 204.1 200.7 197.1 190.1 183.8 149.5 116.0 " Data froin rcf. 33 (1.r ) and ref. 34 (n.1n.r.). 'C.U. = coupling units, see ref. 41 Thus, changing from chloroform to hexane (a less polar solvent) causes a shift of ca. 30cm-l to lower frequencies for hydride trans to halogen but no shift for hydride trans to P or As donor ligand~.~~ Changes in the nature of the cis-ligands can also produce surprisingly large shifts in v(M-H). Thus, in trans-RuHCl(chelate),, v(Ru-H) ranges from 1804 to 1978cm-for a series of closely related P and As donor chelate~.~~ The bonding mode d(M-H) is frequently masked by ligand vibrations and has therefore attracted much less attention.(ii) Nuclear Magnetic Resonance Spectroscopy. The most reliable spectroscopic evidence for the presence of hydride ligands in transition metal complexes is provided by the resonances observed to high field of TMS in the proton n.m.r. spectrum. These are readily detected by modern instruments except in a small minority of complexes where poor solubility or broadening of the signal due to inter- or intra-molecular exchange, nuclear quadruple interactions (i.e. 59C0), or the presence of paramagnetic centres may cause problems. For terminal hydride ligands the high-field resonance is typically in the region z 15-35 but can range 32 P.S. Braterman. R. W. Harrill, and H. D. Kaesz, J. An?. Chmn~.Sor,., 1967, 89, 2x51 P. W. Atkins, J. C. Green, and M. L. H. Green, J. CIiiw. So(,.(A), 1968, 2275. 34 J. Powell and B. L. Shaw, J. C'hrnt. Sou., 1965, 3879. 35 D. M. Adams. Proc. Clii~nr.Soc.., 1961, 43 I. 3o J. Chatt and R. G. Hayter, J. Choni. Soc., 1961, 2605. 420 Moore and Robinson from ca. z5 to cu. 260 in special circumstance^.^^^ 38 These high-field shifts, which are almost exclusive to hydride ligands in transition metal complexes, arise prin- cipally from two effects, a paramagnetic shielding term attributed to the mixing of excited electronic states into the ground state, and a diamagnetic shielding term.The latter becomes increasingly important for shorter M-H distances.39'40 The chemical shift z(M-H) is sensitive to the nature of the ligand trans to the hydride; low values are associated with hydride ligands truns to groups having high trans influence (see Table 1). The highest recorded values are found for hydride ligands trans to the vacant co-ordination sites in tetragonal pyramidal complexes, e.g. ca. 260 for IrHCl,(PBu'~), (R = Me, Et, or The multiplicity of the hydride resonance can be particularly informative. For hydride ligands bound to mag- netically active transition-metal nuclides-notably 03Rh (100% abundance, I II = :), lS3W (14%, I = $) 1890s(16.10/~,= i) and 195Pt(33.70/~, = :)-the spin-spin coupling lJ(M-H) provides strong evidence for the presence of a metal-hydrogen bond.The values obtained, which can range from > 1000 Hz (195Pt-H) to < 15Hz (Io3Rh--H), reflect the magnetogyric ratio of the metal concerned, the co-ordination number of the metal and the nature of the trans-ligand, and can yield information concerning the oxidation state and stereo- chemistry of the complex. Since the magnetogyric ratios, y, differ from nucleus to nucleus it is desirable in comparative studies to eliminate them from 'J(M--H). To do this Pople and Santry41 have defined reduced coupling constants K(M-H) = (2n/hyMyH)J(M-H) which are expressed in coupling units (Table 1). Since the majority of terminal hydride complexes contain phosphorus donor ligands the coupling ,J(PH) between phosphorus (31P loo%, I = i) and hydride ligands is particularly useful in rigid planar or octahedral complexes, 2J(PH),,,,, is usually larger than ,J(PH),,,; typical values for octahedral iridium(1lr) complexes are 100-180 Hz and 10-30 Hz re~pectively.~~ However, in a few complexes, usually involving first row transition metals, where ligand geometry leads to significant distortion of the co-ordination sphere the magnitudes of these couplings are The majority of hydride resonance patterns are first order; analysis of second order spectra where they are found reveals that 2J(PH)I,,,, and ,J(PH),,, are of opposite sign.44 Many polyhydrido complexes of general form MH,(PR,), display stereochemical non-rigidity in solution: all hydride ligands appear equiv- alent and couple equally to all the phosphorus nuclei present.45 In these systems the number of phosphorus donor atoms and hydride ligands can be determined directly from the multiplicities of the 'H and 31P (organic 'H narrow band decoupled) n.m.r.patterns respectively. The coupling 3C-1 H) has been 37 B. D. James, R. K. Nanda, and M. G. H. Wallbridge, Inorg. CIwm.. 1967, 6. 1979. 3x H. D. Empsall, E. M. Hyde, E. Mentzer, B. L. Shaw, and M. F. Uttley, J. Chem. Soc.. Dtrllon Truns.. 1976. 2069. 3Q A. D. Buckinghain and P. J. Stephens, J. C'h~m.So(..,1964, 2747 and 4583. 4o L. L. Lohr and W. N. Lipscomb. Inorg. C/ipm..1964. 3, 22 and references therein. J1 J. A. Pople and D. P. Santry. Mol. P/iy.r.. 1964, 8, 1.'* H. D. Kaesz and R. B. Saillant, CAm. Rev.. 1972, 72. 231. 4.3 F. N. Tebbe, P. Meakin, J. P. Jesson, and E. L. Muetterties, J. Am. C/ltv??.Sot,., 1970, 92. 1068. 44 K. C. Dewhirst, W. Keim, and C. A. Reilly, Inorg. C/iwi.,1968, 7, 546. 45 J. P. Jesson in 'Transition Metal Hydrides', ed. E. L. Muetterties, Dekker. New York, 1971. 42 1 Hydrido Complexes of the Transition Metals observed for a range of 13C enriched ~arbonyl~~,~~ and cyanide46 hydrides. For rigid multihydrido complexes containing chemically non-equivalent hydride li-gands, small couplings [,J(HH') = ca. 5-10Hzl are usually observed. Broad line n.m.r. data on MnH(C0),,48 FeH,(C0),,49 and COH(CO),~O have been used to determine M--H bond distances; the very small values originally obtained (ca.I. 1-1.4 A) were subsequently revisedS * to take account of quadru- pole effects and are now in accord with results from other sources. Electronic Spectroscopy. Analysis of the visible and ultra-violet spectra recorded for the species RuX,(Me,PCH,CH,PMe,), (X = C1, Br, I, or CN) and RuRCl(Me,PCH,CH,PMe,), (R = H, Me, Ph, or p-tolyl) placed hydride along- side alkyl, aryl, and cyanide groups as a high-field ligand.s2 This conclusion is supported by spectra data for [CoH(CN)J3 -(rqf53) and [ReH,], (ref.16) and is in keeping with the high trans-influence of the hydride ligand, the Mossbauer spectra of iron(I1) hydridesS4 and the pale colours displayed by most mononuclear hydrido complexes. An examination of the electronic spectra of the rhodium(1rr) amine complexes [RhX,(en),]+ and [RhXCl(en),] + (X = H, C1, Br, I, N,, or NO,) has, however, indicated a somewhat lower position for hydrogen in thc spec- trochemical series, somewhere between ammonia and water ligands.s s In an at- tempt to rationalize this apparent anomaly it has been suggested that the hydride ligand, because of its unique nature, is highly polarizable and has a variable ligand field strength which is unusually dependent upon the co-ordination environment in which it is situated.56 Miscellaneous Spectroscopic and Physical Techniques. Mass spectroscopy has been extensively employed in the study of volatile hydride complexes containing PF, 57 and/or CO 58 ligands. Parent molecular ions are usually detected and competitive loss of hydride and carbonyl ligands is Polyhydrido complexes tend to lose hydride ligands in pairs.The scarcity of authentic paramagnetic hydrido complexes has severely restricted the application of e.s.r. spectroscopy in this field. However, the high sensitivity of the technique has permitted the detection and characterization of paramagnetic hydrides present in solutions at low concentrations.59, 6o Hyperfine 46 G. M. Whitesides and G. Maglio. J. Am. Clit,m. Soc., 1969, 91. 4980. 47 J. W. Faller. A. S. Anderson, and C.-C. Chen, J. Clim. Soc., Cliem. Cornmuti., 1969, 719. 4x T. C. Farrar. W. Ryan, A. Davison. and J. W. Faller, J. Am. Chmi. Soc., 1966. 88. 184. 40 E. 0.Bishop, J. L. Down, P. R. Emtage, R. E.Richards. and G. Wilkinson, J. Chrtn. Soc.., 1959,2484. T. C. Farrar, F. E. Brinckman, T. D. Coyle, A. Davisun, and J. W. Faller, [tmrg. Clirm., 1967, 6, 161. st G. M. Sheldrick, J. Cliorn. Soc..,Clim. Conitnun., 1967, 751. 52 J. Chatt and R. G. Hayter, J. Cliem. Soc., 1961. 772. s3 W. P. Griffith and G. Wilkinson, J. Cliem. Soc., 1959, 2757. 54 G. M. Bancroft. M. J. Mays. and R. E. Prater. J. Chrm. Soc. (A), 1970, 956. s5 J. A. Osborn, R. D. Gillard, and G. Wilkinson, J. C'hrrn. Soc., 1964, 3168. s6 K. M. MacKay, 'Hydrogen Compounds of the Metallic Elements', E. and F. N. Spon Ltd., London. 1966, p. 141. s7 F. E. Saalfeld. M. V. McDowell, S. K. Gondal. and A. G. MacDiarmid, J. Am. Cltun. Soc., 1968,90. 3684. jX B. F. G. Johnson, J. Lewis, and P.W. Robinson, J. C'1im.r. Soc. (A], 1970, 16x4. 5g V. V. Saraev, F. K. Shmidt. N. M. Ryutina, V. A. Makarov, and V. A. Gruznykh, Koord. Khim., 1977, 3, 1364; Sov. J. Coord. Chem. (Engl. TransLj, 1977, 3, 1064. 6o I. H. Elson, J. K. Kochi, U. Klabunde, L. J. Manzer, G. W. Parshall. and R. N. Tebbe, -1.Am. Clirw. Soc., 1974, 96, 7374. 422 Moore and Robinson coupling constants have been measured and spin density distributions calculated. A fuller discussion is given in Section 2G. X-Ray photoelectron spectroscopy has been used to investigate charge distribu- tion within metal hydrido complexes, data for MnH(CO),, FeH,(CO),, and CoH(CO), indicated H-atom charges of -0.8, -0.3, and -0.75 re~pectively.~~ Similar measurements on polyhydrido complexes indicate that the hydride ligands carry little negative charge.62 Mossbauer studies on iron(rr) hydrides place hydride near or above cyanide in the spectrochemical series.54 D.Diffraction Studies.-Work on this area commenced in 1939 with a report by Ewens and Lister6, on the molecular structures of CoH(CO), and FeH,(CO), as determined by gas phase electron diffraction methods. The tetrahedral arrange- ment of carbonyl ligands about the metal centre was established but the hydride ligands could not be located. Similar difficulties were encountered in the first X-ray diffraction studies on metal hydrido complexes reported in the early 1960's. Thus the positions of the hydride ligands in trans-PtHBr(PEt,),,' MnH(C0),,64 and OsHBr(CO)(PPh,), were inferred from the presence of 'vacant' co-ordination sites. The first location of a hydride ligand by X-ray diffraction methods was reported in 1963 by LaPlaca and Ibers.' In a classic study they reasoned that since the X-ray scattering from hydrogen is largely confined to low angle data it should be possible to enhance hydrogen peaks in the difference-Fourier maps by removing the high angle reflections from the calculation.Using this approach a Rh-H distance of 1.72( 15) 8, was determined for RhH(CO)(PPh,),. A neutron diffraction study completed in the following year' finally established the true stoicheiometry of K,ReH, and revealed a Re-H bond length of 1.68A.These results are consis- tent with the presence of normal covalent M-H bonds and did much to resolve the long-running controversy concerning the nature of the metal hydrogen linkage in complex metal hydrides.Location of hydride ligands by X-ray diffraction methods still poses problems except in the most favourable circumstances when sufficient low angle data are available. Consequently neutron diffraction, though requiring larger crystals and access to a suitable neutron source, remains the preferred method for accurate structure determinati~n.~~ Neutron diffraction offers important advantages over X-ray diffraction for the determination of hydride structures. Since neutrons, unlike X-rays, are diffracted by most elements with very similar efficiencies, the hydride ligand is not masked by the adjacent heavy metal atom and can therefore be more easily located.However, this equality of diffracting power precludes the use of heavy atom methods and most neutron diffraction studies are preceded by an X-ray structure determination to locate all atoms save hydrogen. It should also be noted that since neutron diffraction locates true nuclear positions whereas X-ray 61 H.-W. Chen, W. L. Jolly, J. Kopf, and T. H. Lee, J. Am. C/wn. Soc., 1979, 101, 2607. 62 G. J. Leigh, Inorg. Chim. Actu, 1975, 14, L35. 63 R. V. G. Ewens and M. W. Lister, Trms. F-cirudq Soc., 1939, 35. 681. 64 S. J. LaPlaca, W. C. Hamilton, and J. A. Ibers, Inorg. Chem., 1964, 3, 1491. b5 R. Bau and T. F. Koetzle, Pure Appl. CIietn., 1978, 50,55. 423 Hydrido Complexes of the Transition Metals diffraction methods measure electron density distribution, M-H bond lengths determined by the former technique are often significantly shorter than those measured by the latter.66 Thus values for the Zn-H bond in (ZnH(MeNCH,CH,NMe,)}, as determined by X-ray and neutron diffraction methods are 1.70 and 1.60A re~pectively.~’ E.Nature of the Terminal M-H Bond.-(i) Historical Perspectives. Stereo-chemicalproperties. Many of the early physical measurements on the simple carbo- nyl hydrides gave results consistent with the conclusion that the stereochemical influence of the hydride ligand was minimal. Thus the electron diffraction studies on the cobalt and iron carbonyl hydrides CoH(CO), and FeH,(CO), established tetrahedral arrangements for the carbonyl ligands about the metal atoms.63 Like- wise the physical properties of MnH(CO),, including the infrared spectrum which was initially interpreted in terms of a molecule of C, or C,” rather than C,v symmetry,68*69 appeared consistent with a trigonal bipyramidal structure similar to that found for Fe(CO),. To account for these observations, structures in which the hydrogen atom was buried in the electron cloud of the metal atom70 or attached to a carbonyl ligand (linear C-0-H group)63 were proposed.The second of these theories received no support from infrared evidence7 and fell from favour but the first, originally proposed by Hieber, enjoyed wide acceptance until the late 1950’s. According to Hieber the groups ‘CoH’ and ‘FeH,’ functioned as pseudo-nickel atoms and were tetrahedrally co-ordinated by carbonyl ligands in the same manner as nickel in Ni(C0)4.70 This idea was compatible with broad line n.m.r.data for FeH,(CO), which indicated a H . . * . . H distance of I .88 & 0.05 8, and an Fe-H bond length of 1.1 ,&slightly less than the covalent radius of iron.49 It also appeared to explain, in terms of shielding by ‘&-electrons, the high-field proton resonances observed for hydride ligands in the early solution n.m.r. studies. This situation changed dramatically when X-ray and neutron diffraction studies located the hydride ligands and revealed ‘normal’ covalent bond lengths, essen- tially equal to the sum of the metal and hydrogen covalent radii.Around the same period B~ckingham,~ were able to show that the high-field proton and Lips~omb~~ n.m.r. shifts associated with metal hydrides could be explained without recourse to very short M-H distances. These results finally disproved the ‘buried proton’ model and led to recognition of the metal-hydrogen linkage as a normal covalent bond. Although the hydride ligand occupies a normal co-ordination site its small volume (-7A3) relative to more bulky carbonyl and phosphine ligands (CO z 45 A3; PPh, z 370 w3) leads to sterically induced distortion within the b6 For a discussion of this problem see J. L. Petersen and J. M. Williams, Inorg. Chem.. 1978, 17, 1308. 67 P. T. Moseley, H. M. M. Shearer, C. B. Spencer, Ac,tn Crystallogr., 1969, ,425, S169.6M W. E. Wilson, Z. Natu~f~rsch.,Td B, 1958, 13. 349. 69 F. A. Cotton, J. L. Down, and G.Wilkinson, J. Chem. Soc., 1959. 833. 70 W. Hieber, Die Chemie, 1942, 55, 24; W. Hieber, K. Kramer, and H. Schulten, AngeKp. Chem., 1936, 49, 463. W. F. Edgell, C. Magee, and G. Gallup, J. Am. Chem. Soc., 1956, 78, 4185. 424 Moore and Robinson co-ordination spheres of many hydrido com~lexes.~~ Distortions are present in four, six, seven, and eight co-ordinate structures but are most readily quantified in five co-ordinate trigonal bipyramidal complexes MHL, where the equatorial li- gands are displaced toward the small hydride ligand to form a pseudo-tetrahedral ML, skeleton.73 Whereas most hydrido complexes are stereochemically rigid the five co-ordinate species MHL,, some six and many seven, eight, or nine co-ordinate polyhydrides MH,L, (L =P donor ligand) display stereochemical non-rigidity in solution.Variable temperature n.m.r. is particularly suited to the study of this phenomenon and has yielded much information concerning the kinetics and mechanisms of the rapid polytopal rearrangements occurring within these systems4 Under condi- tions of fast rearrangemcnt all the hydride ligands become magnetically equivalent and couple equally to all 31P nuclei present in the molecule. On several occasions failure to recognise this phenomenon has led to incorrect interpretation of n.m.r. data and wrong assignment of structure. For the tetraphosphine complexes MHJPR,), (n = I, 2, 3, or 4) the rearrangements are thought to involve rapid intra-molecular migration of the hydride ligands between trigonal faces of the M(PR,), tetrahedron.All distinct permutational isomerization reactions of these complexes have been determined and their importance for the interpretation of temperature dependent line shape behaviour has been Retention of internuclear couplings 'J(MH) and 2J(PH) eliminates dissociative mechanisms. Activation energics are usually less than 60 kJ mol -.7 (ii) Stability of M---H Bonds. The stability of transition metal hydrido complexes ranges from extremely high [i.e. PtHCl(PEt,),, distils at 130 "C, 0.01 mm] to notoriously low [i.e.CoH(CO),, decomposes -20 "C]. The high stability displayed by many phosphine-containing hydrido complexes and by the binary hydrido anion [ReH,I2- can in part be ascribed to kinetic factor^.^^'^^ According to this theory a large energy gap LIE between the highest occupied molecular orbital and the lowest unoccupied anti-bonding molecular orbital prevents promotion of elec-trons to the latter and thus stabilizes the complex.The observation that hydrides of highest stability usually involve the heavier members of a given metal triad and ligands of high-field strength lends support to this idea. Data on the thermodynamic stability of M-H bonds are sparse even today. However, those now available indicate that the strengths of M-H bonds are significantly greater than those of M-CH, bonds and do not fall far short of the values recorded for M-CI bonds in comparable complexes78 (see Table 2).(iii) Trans-influence and trans-efect of Hydride. In keeping with its high position 72 B. A. Frenz and J. A. Ibers. in 'Transition Metal Hydrides'. ed. E. L. Muetterties, Dekker, New York, 1971. 73 R. W. Baker and P. Pauling, J. Chem. Soc., Chem. Commun., 1969, 1495. 74 W. G. Klemperer, Inorg. Chem., 1972, 11, 2668. 75 P. Meakin, L. J. Guggenberger, W. G. Peet, E. L. Muetterties, and J. P. Jesson, J. Am. Chem. SOC., 1973, 95, 1467. 76 J. Chatt, Proc. Chem. SOC.,1962, 318. "A. P. Ginsberg, Transition Me[. Chem., 1965, 1, 1 I I. J. A. Connor, Top. Curr. Chem.. 1977, 71, 71. (‘5 Hydrido Complexes of the Transition Metals Table 2 Enthalpy contributions D(M-X) for M-X bonds (X = H, Me, or 0)in some transition metal complexes -Complex D( M-X) kJ mol-’ X=H Me c1 ReJ Me 5 12 242 a MoX2(C5H5)2 258.7 b 25 I .4 149.5 304.2 c WX,(C 5 H 5 12 286.8 b 305.2 197.8 346.0 C MnX(CO), ca.300 ca. 130 I:a. 300 d [FeX(CO>41-ca. 310 d cox(co)4 cu. 290 d IrX2 Cl(CO)( PR,), ca. 240-255 e ”J. E. Bercaw, R. H. Marvich, L. G. Bell, and H. H. Brintzinger, J. Am. Chem. Soc., 1972, 94, 1219. V. I. Tel’noi, I. B. Rabinovich, K. V. Kir’yanov, and A. S. Smirnov, Dokl. Acud. Nuuk, Ser. Khim., 1976, 231, 903. ‘J. C. G. Calado, A. R. Dias, J. A. Martinho-Simoes, and M. A. V. Ribeiro Da Silva, J. Organonlet. Chem., 1979, 174, 77. J. A. Connor, Top. Curr. Chem.. 1977, 71, 77, L. Vaska, Trans. NCW York Acutl. Sci., 1971, 33, 70.in the spectrochemical series (see Section 2C) the terminal hydride ligand exerts a strong trans-influence and trans-effect. The former is demonstrated by long M-L bond distances79’ 8o and low v(M-L) valuesso for ligands (L) trans to hydride; the latter by the facile exchange of phosphine or halide ligands trans to hydride in many Group VIII metal complexes.81, The high trans-influence of the hydride iigand is attributed to its inductive effect in drawing a large amount of p-orbital character into the M-H a-bond and thus weakening the a-bond to the trans- Iigand.83 The high trans-effect is thought to arise largely from ground-state de- stabilization and thus follows from the high trans-infl~ence.~~ (iv) Acidity of Metal Hydrides.It has long been known that many transition metal hydrides display Brransted acid character in polar media; acid strengths range from very strong to immeasurably weak.8 Thus FeH(SiCl,),(V-C,H J(C0) is reported to be a very strong acid in MeCN solution, comparable with HC104,86 whereas IrHCl,(CO)(PMe,Ph), (H trans to CO) does not lose HCI even in boiling sodium metho~ide.~~On the basis of the modest number of quantitative data available it appears that hydride acidity increases on crossing the transition metal block from l9 R. Mason, R. McWeeny, and A. D. C. Towl, Discuss. Faruday Soc., 1969, 47, 20. 8o T. G. Appleton, H. C. Clark, and L. E. Manzer, Coord. Chem. Rev., 1973, 10, 335 and references therein. B1 P. G. Douglas and B. L. Shaw, J.Chem. Soc. (A), 1970, 1556. 82 J. Chatt and B. L. Shaw, J. Chem. Soc., 1962, 5075. H3 C. H. Langford and H. B. Gray, ‘Ligand Substitution Processes’, Benjamin, New York, 1965, Chapter 2. 84 A. Pidcock, R. E. Richards, and L. M. Venanzi, J. Chem. Soe. (A), 1966, 1707. 85 R. G. Pearson and P. C. Ford, Comments Inorg. Chem.. 1982, I, 279. 86 W. Jetz and W. A. G. Graham, Inorg. Chem., 1971, 10, 1159. 87 A. J. Deeming and B. L. Shaw, J. Chem. Soc. (A), 1968, 1887. Moore and Robinson left to right and decreases on moving down a metal triad.85 Replacement of strong n-acceptor ligands with weaker ones also produces a marked decrease in acidity thus pK, values for the complexes CoH(CO),L (L = CO, P(OPh),, and PPh,) are <2, 1.13 x and 1.09 x lo-’ Finally there is evidence to suggest that for some iridium(Ii1) complexes, acid strength is proportional to Ir-H bond strength; complexes with high v(1r-H) values tend to release protons most readily.87 F.Chemical Reactions.-In this section some of the more important chemical reactions displayed by terminal hydride ligands are discussed and representative examples given. Since the reactivities of individual hydrido complexes vary very widely it should not be assumed that any particular reaction type discussed is general to all, or even most, transition metal hydrido complexes. (i) Zsornerization. Some rigid octahedral hydrido complexes of the platinum group metals have been shown to undergo geometrical isomerization on photolysis.A dissociative mechanism involving rearrangement of five-co-ordinated inter-mediates, formed by photo-induced loss of a phosphine ligand trans to the hydride, has been (ii) Hydrogen-deuterium Exchange. H ydrido complexes that are susceptible to protonation, oxidative addition, and ‘insertion’ reactions undergo H-D exchange with Df-D,O (equation 18),90 D, (equation 19),91*92 and C,D, (equation 20)92 respectively. In combination with a reversible cyclometallation reaction (equation 21) these processes can affect H-D exchange at selected sites in the organic moieties of co-ordinated ligands (i.e. the ortho sites in triarylphosphine and phosphite ligand~).~, (Et,P),PtHCl + DCl-(Et,P),PtHDCl,-(Et,P),PtDCI + HCI (18) (Ph,P),(CO)IrH + D, -(Ph,P),(CO)IrHD, + PPh, -(Ph,P),(CO)IrD + HD (19) (dppe),IrH + C,D, -(dppe),IrCD,CD,H-(dppe),IrD + C,HD, (20) {(PhO),P),IrH -{(PhO),P)~,I~H,{P(O~,H,)(OPh),)+ P(OPh), (21) (iii) Reactions with Acids. A wide variety of acids ranging from strong mineral acids to very weak carbon or silicon acids (i.e.HC-CR, HSiR,) can react with hydrido complexes. Strong acids with weakly co-ordinating conjugate bases usu-ally form salts (equation 22)94 whereas weak acids with strongly co-ordinating conjugate bases generally afford oxidative addition products (equation 23).9 8n W. Hieber and W. Hubel, 2.Eleklrochem., 1953,57,235;W. Hieberand E. Lindner, Chem. Ber., 1961, 94, 1417. 89 P. R. Brookes, C. Masters, and B. L. Shaw, J. Chem. Soc. (A), 1971, 3756. 90 C.D. Fdik and J. Halpern, J. Am. Chem. Soc., 1965, 87, 3523. L. Vaska, lnorg. Nud. Chem. Lelt., 1965, 1, 89. 92 R. A. Schunn, Inorg. Chem., 1970,9, 2567. 93 E. K. Barefield and G. W. Parshall, lnorg. Chem., 1972, 11, 964. y4 L. Malatesta, G. Caglio, and M. Angolettd, J. Chem. Soc., 1965, 6974. 95 F. Glockling and J. G. Irwin, lnorg. Chirn. Acta, 1972, 6,355. Hydrido Complexes of the Transition Metals These reactions are frequently accompanied by reductive elimination of dihydro-gen and afford a clean, convenient route for the introduction of the conjugate bases (i.e. C=CR, SIR,, acac, SPh, 0,CR) into the co-ordination sphere of the metal (equation 24).96 IrH(CO)(PPh,), + HCIO, -[IrH,(CO)(PPh,),]ClO, (22) IrH(CO)(PPh,), + HSnMe, -IrH,(SnMe,)(CO)(PPh,), + PPh, (23) MHL, + HX-MH,XL,-MXL, + H, (24) Treatment with trityl fluoroborate can lead to hydride abstraction and salt formation (equation 25).97 RuH,(PPh,), + (Ph,C)BF, -[RuH(PPh,),]BF, + Ph,CH (iv) Reactions with Bases.Many of the more acidic hydrides react with KOH or, failing this, with stronger bases such as LiR (R = alkyl or ar~l)~~ or KH99 to form salts (equations 26 and 27). ReH(C,H,), + LiBu-Li[Re(C,H,),] + BuH (26) CoH{P(OMe),}, + KH-K[Co(P(OMe),),] + H, (27) These reactions afford useful routes to salts of complex anions. (v) Reactions with Halogens and Organic Halides. Most transition metal hydrides react readily with halogens and halogenated hydrocarbons to yield the correspond- ing metal halide complexes.The reactivity of the hydrocarbon increases with the degree of halogenation CH,CI < CH,Cl, < CHCI, < CCl,. Treatment of a com- plex hydride with CCI, and subsequent detection of CHCI, is a useful but not totally reliable method for establishing the presence of hydride ligands. Conversely, caution should be exercised in the use of halogenated hydrocarbon solvents when the presence of labile hydride ligands is suspected. (vi) ‘Insertion’ (Hydrogen Migration) Reactions. This large and diverse group of reactions, many of which participate in important catalytic cycles, are of the general form: L,MH + X-L,M-X-H The reverse reaction is termed extrusion. A selection of the more important exam- ples is given in Table 3; the classification is self-explanatory.A number of the reactions listed merit special mention. The ‘insertion’ of carbon monoxide into M-H bonds, though much sought because of its probable participation in Fischer-Tropsch chemistry,’ O0 has only recently been detected and the example given in the table was the first to be reported. Insertion of carbonyl ligands into q6 K. R. Laing, S. D. Robinson, and M. F. Uttley, J. Chem. Soc., Dalton Trans., 1974, 1205; A. Dobson, S. D. Robinson, and M. F. Uttley, J. Chem. Soc., Dalton Truns., 1975, 370. q’ J. R. Sanders, J. Chem. Soc., Dalton Truns., 1973, 743. D. Baudry and M. Ephritikhine, J. Chem. Soc., Ciwm. Commun., 1979, 895. 99 E. L. Muetterties and F. J. Hirsekorn, J. Chem. Soc., Chem. Commun., 1973, 683. loo C. Masters, Ah.Organomet. Chem., 1979, 17, 61. Moore and Robinson Table 3 H-ydride migration (insertion) reactions Reaction ReJ 1f 1 Insertions (0ctaethylporphyrin)RhH + CO + (octaethylporphyrin)Rh(CHO) a (dmpe),TaH( ECCMe,)I -+ (dmpe),Ta(=CH-CMe,)I b -+(q-CsH,),WH(CH,PMe,Ph) (v-C H s)2W(CH3)(PMe2 Ph) C (?-C,H,)(CO),MoH + CH,N, -+ (q-C,H,)(CO),MoMe + N, d (tl-C,H ,)(CO), MoH + SnR, -+ (V-c H s)(CO)3M0(SnR2H) e (R = CH,SiMe,) 112 Insertions (Me,P)3Co(C2H,)H -+ (Me,P),CoC,H, f (Ph,P),CI,RhH + CZH, -+ (Ph,P),CI,Rh(CH =CH,) g (rl-C,Hs),MoH, + Et0,C.N =N.CO,Et +(~-C,H,),MoH(EtO,C~N-NH~CO,Et) h +trans-[Rh(en),H(OH)] + 0, + trans-[Rh(en),(OH)(OOH)]+ 1 (Ph,P),CoH 3 + co, -+ (Ph3P),Co(0,CH) + H, i (Ph,P),(CO)CIRuH + cs, -+ (Ph,P),(CO)ClRu(S,CH) + PPh, k 113 Insertions (tl-C,H,)(CO),MoH + N,CHSiMe, -+ (q-C,H,)(CO),Mo(N,CH,SiMe,) + CO I 114 Insertions (v-C,H ,)(CO), FeH + CH, =CH-CH =CH, -+ (q-C,H,)(CO),Fe(CH,CH:CHCH,) m 'B.B. Wayland and B. A. Woods, J. Chem. Soc., Chem. Commun.. 1981,700. M. R. Churchill, H. J. Wasserman, H. W. Turner, and R. R. Schrock, J. Am. Chem. Soc.. 1982,104, 1710. 'N. J. Cooper and M. L. H. Green, J. Chem. Soc., Chem. Commun.. 1974, 761. 'T. S. Piper and G. Wilkinson, J. Inorg. Nucf. Chem.. 1956, 3, 104. 'I. D. Cotton, P. I. Davison, D. E. Goldberg, M. F. Lappert, and K. M. Thomas, J. Chem. Sor., Chem. Commun.. 1974, 893. IH-F. Klein, R. Hammer, J. Gross, and U. Schubert, Angew. Chem.,hi.Ed. Engl., 1980,19,809. @ M. C. Baird, I. T. Mague, J.A. Osborn, and G. Wilkinson, J. Chem. Soc. (A), 1967, 1347. hS. Otsuka, A. Nakamura, and H. Minamida, J. Chem. Soc., Chem. Commun., 1969, 1148. 'R. D. Gillard, B. T. Heaton, and D. H. Vaughan. J. Chem. Soc. (A), 1970.3126 and references therein. 'L. S. Pu, A. Yamamoto, and S. Ikeda, J. Am. Chem. Soc., 1968,90,3896. 'S. D. Robinson and A. Sahajpal, Inorg. Chem., 1977, 16,2718. 'M. F. Lappert and J. S. Poland, J. Chem. SOC..Chem. Commun., 1969, 1061. "M. L. H. Green and P. L. 1. Nagy, J. Chem. Soc., 1963, 189. M-H bonds has also been cited as a possible explanation for the lability of certain carbonyl hydrides to ligand substitution. O1 Hydrogen migration reactions involving carbene ligands have been postulated as intermediate steps in alternative mechanisms for the stereospecific poly- merization of olefins by Ziegler-Natta catalysis.O2 Hydrogen migration onto co-ordinated olefin (or acetylene) ligands is a key step in hydrogenation reactions' O3 and in one established mechanism for olefin isomerization. *04 The formation of hydro-peroxides is crucial in many transition metal mediated reac- tions for the oxidation of organic substrates. O5 Finally, the 'insertion' of carbon lol R. G. Pearson, H. W. Walker, H. Mauermann, and P. C. Ford, Inorg. Chem., 1981, 20, 2741. Io2 K. J. Ivin, J. J. Rooney, C. D. Stewart, M. L. H. Green, and R. Mahtab, J. Chem. Soc., Chem. Commun., 1978, 604. Io3 B. R. James, Ah. Organornet. Chem., 1979, 17, 319; 'Homogeneous Hydrogenation', Wiley- Interscience, New York, 1973.Io4 D. Evans, J. A. Osborn, and G. Wilkinson, J. Chem. Soc. (A), 1968, 3133. Io5 M. T. Atlay, M. Preece, G. Strukul, and B. R. James, J. Chem. Soc., Chem. Commun.. 1982, 406. Hydrido Complexes of the Transition Metals dioxide into M-H bonds to yield formates is a first step in one possible mechanism for the conversion of this abundant source of carbon into useful C, organic compounds. O6 (vii) Reductive Elirninution Reactions. These lead to the elimination of dihydrogen, hydrocarbons RH, or acids HX and require that the metal involved has a stable oxidation state two units below that existing in the parent complex (unless metal-metal bond formation occurs). Reductive elimination of dihydrogen from polyhydrides can readily be achieved by thermal or photolytic rnethods,lo7 the presence of a donor ligand capable of reacting with the product often promotes the reaction.The species produced are electron deficient and are frequently highly reactive. Some are capable of taking up dinitrogen (equation 28)Io8 or cleaving C-H bonds to form organo-hydrido complexes (equation 29).' O9 'I"MoH,(dppe), +2N, + Mo(N,),(dppe), +2H, The instability of CoH(CO), can be attributed in part to the formation of Co,(CO), by facile intermolecular elimination of dihydrogen. Reductive elimi- nation of alkanes occurs readily, particularly if the participating hydride and alkyl ligands are mutually cis, and is a key step in the catalytic hydrogenation of olefins. The process can proceed by inter- or intra-molecular pathways. Thus, elimination of methane from cis-OsH(Me)(CO), in the absence of added ligand affords (OC), HOsOs(Me)(CO), whereas in the presence of a ligand L the mononuclear species Os(CO),L is produced.O The ease with which alkane elimination occurs in many systems accounts for the scarcity of organo-hydrido complexes. G. Complexes of Particular Interest.-(i) Binary Hydride Species. There are several reports in the early literature purporting to describe simple binary hydride species. In I926 Weichselfelder' ' reported on the hydrogenation of PhMgBr/first row transition metal halide mixtures in ethereal suspensions and described black oily solid products which were formulated as the simple hydrides CrH,, FeH,, FeH,, CoH,, and NiH,.Later workers confirmed the dihydrogen uptake but characterized the black solids as mixtures probably containing organometallic hydrides. However, a recent X-ray diffraction study on one of these products FeH,Mg,Br3,,C1,,,(THF), has revealed a FeHb4- anion surrounded by a tet- rahedral arrangement of magnesium ions (Figure I).' An extensive series of solid state ternary hydrides, prepared by heating together powdered alkali metal hydrides and transition metals under hydrogen at ca. 600 -ln(' R. Eisenberg and D. E. Hendriksen, Adi~.C(i~d.,1979, 28. 79. lo' G. L. Geoffroy, Prog. Inorg. Chem., 1980, 27, 123. I"" G. L. Geoffroy. unpublished results cited in rr.f. 107. loo C. Giannotti and M. L. H. Green, J. c'licwi. Soc..Clicwi. Conrniuti.. 1972. I114. 'Io J. Halpern, Arc. Cliem. Rrs.. 1982, 15, 332. I I T. Weichselfelder and B. Thiede. Lihg's Ann. C/rct77..1926, 447.64. 'IzS. G. Gibbins, Inorg. Clicnr.. 1977, 16. 2571 and references therein; R. Bau. D. M. Ho. and S. G. Gibbins, J. Am. Clic~m.Soc., 1981. 103, 4960. 430 Moore and Robinson Figure I Vieti,ofthe [FeHJ4 -anion surrounded by u tetruhedrulurruy of mugnesium c'utionx 800 "C appear to contain covalent transition metal -hydrogen linkages. * Diffraction studies on 'Sr,RuD,' and 'Li,RhH,' reveal the presence of RuD, octahedral (Ru-D = 1.69 A) and RhH, square planar units respectively.' l4 However, further work on these interesting products is required before their true (c) (d) Figure 2 Struc.ture.s udopted by pold~ydrridc.~(a) ReHi-, (b) ReH,(PR,),, (c) 1rH5(PR3)?.(d) MoH,(PR,), (P = phosphine) ' J.D. Farr. J. Itiorg. Nucl. C'hom.. 1960, 14,202;A. F. Graefe and R. K. Robeson, J. Inorg. Nucl. C'lrcm.. 1967. 29, 2917; R. 0.Moyer, R. Lindsay, and D. N. Marks, Adv. C'/imm. Su., 1978, 167, 366. L. B. Lundberg. D. T. Cromer, and C. B. Magee, Inorg. C'hm., 1972. 11, 400; R. 0. Moyer, C. Stanitski. J. Tanaka. M. 1. Kay. and R. Kleinberg, J. Solid Siutr C'hm., 1971, 3, 541. 43 1 Hydrido Complexes of the Transition Metals nature becomes fully apparent. In contrast the celebrated rhenium complex K,ReH, and its technetium analogue have been very thoroughly characterized. Their structure (Figure 2a), fluxional character, stability, and chemical properties have been noted elsewhere in this review. The existence of isoelectronic species such as [WHgI3- and [OsH,]-has been speculated upons6 but to date none has been isolated.(ii) Polyhydrides. The small size of the hydride ligand favours high co-ordination numbers and although few binary polyhydrides have been reported a wide range of other polyhydrido complexes is known. Two particularly extensive series con- taining tertiary phosphine and cyclopentadienyl ligands respectively merit special mention. Phosphine-stabilized polyhydrido complexes, MHJPR,), (see Table 4a), are generally prepared by 'reduction' of metal halide complexes with NaBH, or Table 4 Pufyhydrido Complexes (a) MH,L, (n 3 3; L = PR,; m = 1-4)" FeH,L, CoH,L, NbH,L, MoH,L, RuH,L, RhHL, RuH,L, RhH,L, TaH,L, WH,L, ReH ,L, OsH, L, IrHL, [WHsL4lt WH6L3 ReH,L, [ReH4L4It OsH,L, [OsH,L,I+ IrH,L, ReH, L, OsH6L, IrH,L, [ReH L] -(b) MHJcp), (n = 2,3; cp = q-C,H, or q-C,Me,)' " Also included (in italics) where known, are the mono and dihydrido complexes which complete the various series of compounds. LiAIH,, often in the presence of dihydrogen.They are all white or pale yellow solids of moderate stability. All obey the 18-electron rule but most undergo photolytic or thermolytic loss of dihydrogen (see Section 2F). The variable tem- perature n.m.r. spectra and fluxional character of phosphine-stabilized poly- hydrides in solution has been discussed above (Section 2C). Diffraction studies confirm that the solid-state structures are dominated by the bulky P donor ligands Moore and Robinson (Figures 2b-d).The chemistry of these compounds has recently been reviewed.* Cyclopen tadienyl-stabilized polyhydrides (Table 4b) have been prepared by numerous methods including hydrogenation of bis(cyclopentadieny1) complexes, 'reduction' of cyclopentadienyl metal halides, and treatment of anhydrous metal halides with NaC,H, and LiAIH,; the cationic species are usually obtained by protonation with strong acids. The simple mononuclear complexes MH, (C,H,)Z (n = 1, 2, or 3) have an 'oblique' structure (Figure 3) with three chemically active Figure 3 'Oblique' slructurc adopted by bis(c~c.lopentadienyl)metulhydridcs M H,( q-C H 5)2 (n = 1, 2, or 3) orbitals in the equatorial plane.These orbitals can each accommodate a hydride ligand or a lone pair of electrons. The presence of a lone pair bestows base character and permits protonation or formation of adducts with Lewis acids (i.e. BF,). Electron deficient species often adopt more complex hydrogen bridged binuclear structures. Some members of this group of polyhydrides are para- magnetic (see Sub-section V) others lose dihydrogen on photolysis to yicld highly reactive unsaturated intermediates (see Section 2F). They are usually non-fluxional, thus the trihydrides MH,(C,H 5)2 display AB, patterns in their high field proton n.m.r. spectra. Cyclopentadienyl metal hydrides have recently been reviewed."(j (iii) Cyanide Hydrides.Although the cyanide anion is a strong field ligand and ought to 'stabilize' M-H bonds, remarkably few cyanide hydride complexes have been reported. The best known examples are the cobalt(Ir1) anion [CoH(CN)J3 and its rhodium(rI1) and iridium(II1) analogues. The ability of cobalt(I1) cyanide solutions to absorb dihydrogen and catalyse the hydrogenation of activated olefins has been known since 1942.'17 However detection and characterization of the [CoH(CN),I3-anion in these solutions was first accomplished in 1958 using n.m.r. 'I A. P. Borisov. V. D. Makhaev. and K. N. Semenenko, Koortl. Khini.. 1980, 6. 1139; SOL'.J. Coor.d. Clicwi. (Etigl. Trrinsl.), 1980. 6, 549. 'Io G. L. Soloveichik and B. M. Bulychev. Usp. Khitn., 1982.51, 597; Russ. C'iimi. RPV..1982. 51. 386.' M. lguchi, J. Clicm. Soc. Jpn., 1942, 63, 634, 1752. Hydrido Complexes of the Transition Metals methods.Il* All three anions [MH(CN)J3 -have now been isolated as stable alkali metal salts.119 (iv) Hydrides Stabilized by Weak Field Ligands. The majority of hydrido com- plexes owe their stability to the presence of high field ligands. A few hydrido complexes are now known, however, in which the ancillary ligands are nitrogen, oxygen, sulphur, or halogen donors. A representative selection of these complexes, some of which display moderate stability is given in Table 5. (v) Paramagnetic Hydrides. Since most hydride complexes obey the 16-or 18-electron rule, paramagnetic hydrides are comparatively rare.However, in the past decade a few fully authenticated examples have been isolated and others, including some very unstable intermediates, have been detected and characterized in solution by e.s.r. spectroscopy. Hyperfine couplings to the hydride ligands and, in appropriate cases, to the central metal nuclei have been detected and electron spin density distributions have been calculated. A representative selection of com-plexes, together with magnetic data and attendant references, is given in Table 6. 3 M-H. ...X and M -* -* H-X Interactions Numerous structures are now known in which terminal hydride ligands bound to transition metals interact with a second (non-transition) element. Conversely, there Table 5 Hydrido complexes stabilized by Mvak ,field ligands Comments Ref.transient intermediate generated photochemically in a solution. Similar species reported for Fe3+, Ti4+, and Ag2+ product from heterolytic cleavage of b H, by [RuC1,l3 -[OsH ,(en),l[ZnC1,1 C' [RhH (N H 3) J2 examples from substantial range, + ci.s/tran.s-[RhHX(en),] stability due in part to kinetic d,0+ (X = H or halide) inertness of Rhlll I rHCl,(Me,SO) hydrogen transfer catalyst .f " D. A. Ryan and J. H. Espenson. Inorg. Cliem., 198I, 20,4401. P. J. Brothers, Prog. Inorg. C'hrm., 1981. 28, 1 and references therein. J. Malin and H. Taube, Inorg. C'licm., 1971, 10, 2403. * R. D. Gillard, and G. Wilkinson, J. C%em.Soc., 1963, 3594. J. A. Osborn, R. D. Gillard, and G. Wilkinson, J. Chem.Soc., 1964, 3168. Y. M. Y. Haddad, H. B. Henbest, and J. Trocha-Grimshaw, J. Chem. Soc., Perkin Trans. I, 1974, 592. 11* W. P. Criffith, L. Pratt. and G. Wilkinson. Nature, 1958, 182, 466. II9 1. B. Baranovskii, Zli. Nrorg. KIiir~.~1978. 23. 2579; Russ. J. Inorg. Chem., 1978, 23, 1429 and references therein. Moore and Robinson Table 6 Puramugnetic hydride.s Ref. -U -h -h 2.99 c 1.80 d 2. I6 d 2.19 e 2.15 .f' J. E. Bercaw and H. H. Brintzinger, J. Am. Chcni. Soc., 1969. 91, 7301. I. H. Elson, J. K. Kochi. U. Klabunde, L. E. Manzer, G. W. Parshall. and F. N. Tebbe, J. Am. Chcm. Soc... 1974,96, 7374. M. Frcni. P. Romiti, and D. Gusto. J. //iorg. Nud. C%cm..1970. 32, 145. M. Gargano. P. Giannoccaro, M.Rossi. G. Vasapollo. and A. Sacco, J. Ch~w.Soc.. Drrlron Trcrns.. 1975, 9. '' M. Nakajima. H. Moriyama. A. Kobayashi. T. Saito. and Y. Sasaki, J. C'licwi. SOI,..CIicwi.C'ommun.. 1975, 80. J. R. Sanders, J. C'hcm. Soc., Ddioti Trrms., 1973, 748. are many complexes in which a H atom in a co-ordinated ligand approaches to within bonding distance of the central metal atom. Only a very brief account of these structures can be given here; for a fuller discussion readers are referred to a recent excellent review article.' 2o A. M-H-B Linkages.-This type of bridge is encountered in borohydrido com- plexes where the tetrahedral BH,- anion can be bound to a transition metal in mono-, bi-, or tri-dentate manner. Examples of these three bonding modes have been confirmed by diffraction methods in Cu(HBH,)(PPh,Me), (ref: I21), Hf(H,BH,)(q-C,H,Me), (ref: 122), and Zr(H,BH), (wf; 123) respectively.In each case the bridging hydride ligands have been located by neutron or electron diffraction. Bidentate co-ordination of the BH,- anion is most common, mono- and tri-dentate co-ordination are relatively rare. The chemical and structural properties of these systems have recently been reviewed.' 24 Transition metal com- plexes containing larger borane (i.e. B,H,, B,H,) or carbaborane (i..e. C,B,H, ,) ligands co-ordinated through M-H-B bridges are also known. B. M-H-AI Linkages. -These are still rather rare. A neutron diffraction study on (~-CsHs)Ti(~-H)(~~-H,AlEt2)(~-C~H,-C,H,)Ti(~-C5H5)confirmed the pres- lZo R.G. Teller and R. Bau, Strwi. Bonding (Brdin), 1981. 44,I. I2l C. Kutal, P. Grutsch, J. L. Atwood, and R. D. Rogers, Inorg. Chem.. 1978, 17, 3558; F. Takusagawa, A. Fumagalli, T. F. Koetzle, S. G. Shore, T. Schmitkons, A. V. Fratini, K. W. Morse, C. Y. Wei, and R. Bau, J. Am. Chem. Soc, 1981, 103, 5165. 122 P. L. Johnson. S. A. Cohen, T. J. Marks. and J. M. Williams J. Am. Chmi. Soc,., 1978, 100, 2709. lz3 P. H. Bird and M. R. Churchill, J. Clicvn. Soc., Chrm. Comm1/7., 1967. 403; V. Plato and K. Hedberg Inorg. Chcm., 1971, 10, 590. T. J. Marks and J. R. Kolb. Chmi. RcI'., 1977, 77. 263. 435 Hydrido Complexes of the Transitibn Metals ence of a H,AlEt, bridge; molecular parameters within the Ti-H-A1 linkages are Ti-H, 1.69 A; AI-H, 1.70A; Ti.. . . Al, 3.13 A.125Similar M-H-A1 bridges have been postulated for a number of early transition metal complexes.116' 120 C. M-H-C Linkages. -These are of considerable importance because of their possible implications in homogeneous catalysis. Their presence is usually detected by diffraction studies which reveal ligands arranged to bring one or more C-H groups within close proximity of the central metal atom. Early examples in which X-ray diffraction studies revealed short M . * + H-C interactions between the metal atom and an ortho-hydrogen of a phenyl phosphine ligand include RuCI,(PPh,), (Ru . . H, 2.59 A)126trans-PdI,(PMe,Ph), (red form Pd *. . * H. 2.84, 2.85 A; yellow form, Pd-.. H, 3.28 A),127and RhCl(PPh,), (red form, Rh .. . *. H, 2.77 A; orange form Rh . .. * H 2.84 A).,,, More recently, structures involving much shorter M . . . . H distances have been reported including [Fe(~3-C,H,,)(P(OMe),},][BF,] in which the Fe . . . H distance at 1.874(3) 8, is only 0.1 A longer than the normal terminal Fe-H bond, and the C-H bond involved is longer than usual [1.164(3) A].129Other examples are found in metal clusters [i.e. Fe4(q2-CH)H(CO),,I1 30 and in alkylidene complexes [i.e. Ta (=CHBu')(rne~ityl)(PMe,),].~~ Finally, severe distortion or the alkyl ligands in the complexes TiRCl,(dmpe) (R = Me or Et) has been attributed to the presence of strong intramolecular bonding interactions between the titanium atom and an 2-H of the methyl group [Ti....H, 2.03(4) A] or a P-H of the ethyl group (Ti... . H, 2.29 D. M-H*.*Si Linkages. -Although several complexes have been claimed to display M-H-Si bridges these claims have rarely been fully substantiated. The best established M-H-Si linkage is found in W, (p-HSiPh,),(CO), where W-Si distances of 2.586 and 2.703 8, are attributed to W-Si and W * . * H . . . . Si bonds respectively.133 Structures of this type have been discussed in a recent review.',O 4 Edge Bridging (p,) Hydrides A. A Note on Polynuclear Clusters. -Unlike terminal hydrides which are confined largely, though not exclusively, to mononuclear complexes, ,u2, p, and encap- sulated hydride ligands are, of necessity, found only in polynuclear clusters where they occupy edge, face, and interstitial sites respectively.Therefore, before embark- L. J. Guggenberger and F. N. Tebbe, J. Am. C'hmi. Soc~.,1973, 95. 7870. Iz6 S. J. LaPlaca and J. A. Ibers, Itzorg. C'lwni., 196.5, 4, 778. 12' N. A. Bailey and R. Mason, J. Chrrn. Soc. (A), 1968, 2.594. 128 M. J. Bennett and P. B. Donaldson, Inorg. Chem., 1977, 16, 655. J. M. Williams, R. K. Brown, A. J. Schultz, G. D. Stucky, and S. D. Itlel, J. hi. Clirni. Soi., 1978. 100, 7407 and with R. L. Harlow. 1980. 102, 981. 13" M. A. Reno. J. M. Williams, M. Tachikawa, and E. L. Muetterties, J. Am. Clioin. Soc., 1980, 102.4542. 13' M. R. Churchill and W. J. Youngs. Inorg. Chrni.. 1979. 18, 1930. 132 Z. Dawoodi. M. L. H. Green, V. S. B. Mtetwa, and K. Prout, J. ('h~niSoc.. C%Cm.C'oniniun., 1982. 802 and 1410.13) M. J. Bennett and K. A. Simpson. J. Am. C'h~wi.Snr., 1971. 93, 71.56. Moore and Robinson ing upon a discussion of these types of hydride ligand some reference to the properties of metal clusters seems appropriate. The structure and stoicheiometry of the smaller (2, 3, and 4 membered) clusters can usually be rationalized in terms of the E.A.N. (18-electron) rule. Larger clusters require more sophisticated treat- ments and these have been provided by Chini, Hoffmann, Mingos, Wade, and others.134A full account of this work is given in a recent review by Johnson and Lewis. Turning to the more immediate question of stoicheiometry and structure in polynuclear metal hydrides we note the formal analogy between these systems and polynuclear boron hydride clusters.This relationship has recently been recog- nized by the extension of Lipscomb's 'styx' formulae for boranes13'j to cover polynuclear transition metal hydrido cluster^.^ 37 The mechanistic features of dy- namic rearrangements in metal cluster species including polynuclear complex hy-drides have recently been reviewed.I3* \ 1 I Figure 4 Tj>picalstructures involving bridging (p2)irjjdride ligunds 134 P. Chini, Inorg-. Chim.Actcr Rev.. 1968,2. 31; P. Chini and B. T. Heaton, Top. Curr. Chem.. 1977,71, I: P. Chini, G. Longoni, and V. G. Albano, Arh Orgmomet. Chmi., 1976, 14, 285; M. Elian and R. Hoffmann, Inorg. CIiem., 1975, 14, 1058; D. M. P. Mingos, Nciiurc (London) Phj.s. Sci., 1972, 236, 99; J. Chem.Soc., Dtilron Trcrns., 1974, 133; K. Wade, Chem. Br. 1975, 11, 177; Adv. Inorg. Chem. Rtidiochetn., 1976. 18, I . B. F. G. Johnson and J. Lewis, Ah. Inor'y:. Chcm. Rdiochun., 1981, 24, 225; 'Transition Metal Clusters'. ed. B. F. G. Johnson, Wiley, New York, 1980. W. N. Lipscomb, 'Boron Hydrides', Benjamin, New York, 1963. 13' J. C. Green. D. M. P. Mingos, and E. A. Seddon, J. Orgmomot. C'hcm., 1980, 185, C20; D. M. P. Mingos, Purr. Appl. Ch~ni.,1980, 52, 705. E. Band and E. L. Muetterties, Chm. Kt.11.. 1978, 78. 639. Hydrido Complexes of the Transition Metals B. Edge-bridging (pJ Hydrides. -Whereas the ability of hydrogen to bridge between boron atoms has been known since the 194O’s, the formation of similar H-bridges between transition metal atoms was discovered relatively recently.How- ever, following Hieber’s synthesis’ 39 of the first polynuclear metal carbonyl hy- drides, by treatment of metal carbonyls with base, a very extensive chemistry of p2-hydride bridged specks has developed. Structures in which pairs of metal atoms are linked by mono-, di-, tri-, or tetra-hydride bridges, or by mixed bridges in which one or more hydrogens are replaced by other bridging groups (i.e. halide, OR, SR, PR,, or CO) have now been established. A representative selection of examples is given in Figure 4. Edge bridging (p,)hydride ligands are also frequently encountered in larger metal cluster polyhedra. C. Preparative Methods. -Many of the early pu,-hydrido complexes were the products of serendipitous reactions.However, as reactivity patterns in hydride chemistry have emerged more systematic approaches to synthesis have been devel- oped. Thus, recognition of a formal scheme whereby mononuclear hydrido com-plexes act as donor ‘ligands’ binding through one or more hydride atoms to a co-ordinatively unsaturated acceptor complex has permitted the synthesis by de- sign of many binuclear H-bridged species. 40 Examples of these reactions, some of which involve subsequent elimination of dihydrogen from the adduct, are given in equations 30-32. Rh-H-Ir -PEt3 Ph3P’‘H’ ‘PEt3 I-’’ W. Hieber and Ci. Brendel, Z.Anorg. Allg. Cltcm., 1957, 289. 324; W. Hieber and R. Werner. Clrtwi Bcr., 1957. 90. 286. I4O L. M. Venanzi. Goorti.c‘hcwi. Rev., 1982, 43. 251. Moore and Robinson More complex polynuclear species have been synthesized from metal carbonyl cluster precursors by a diverse variety of highly specific reactions. These include protonation, oxidative addition of dihydrogen, activated hydrocarbons or func- tional organic molecules, cyclometallations, reactions with water, and base or borohydride 'reduction'. Some examples are given in equations 33-36, a fuller account may be found in a recent review.I4I Os,(CO),, + H,-Os,H2(CO),, + 2CO (34) D. Characterization. -Bridging (p2) hydride ligands are usually more firmly bound than their terminal counterparts, consequently their detection and esti- mation by chemical means (reactions with halogenated hydrocarbons or halogen acids) is not usually feasible.Most bridging hydride ligands are characterized by one or more of the spectroscopic methods discussed below -notably n.m.r. and mass spectroscopy. E. Spectroscopic Properties. -(i) Vibrational Spcctro.rc-opj*. Bands due to v(M-H-M) modes are frequently difficult to detect; they occur in a rather congested region of the spectrum (1700-700 cm-I) and are usually broad (dvr -100cm-I) and weak.142 However, in some instances, cooling of the sample to liquid nitrogen temperatures has been shown to effect a marked increase in intensity; a result which has been interpreted in terms of a double potential minimum for the M-H-M intera~tion.'~~The frequencies of v(M-H-M) modes have been shown to correlate with the corresponding M-H-M an-145gle~;'~~.conversely the v(Mo-H-Mo) frequencies of [Mo,HC1,I3 -have been'used to calculate L Mo-H-Mo and hence, given the Mo .. . . Mo distance, the Mo-H bond The dark colours and sensitivity to laser radiation associated with many cluster compounds have hindered Raman studies. Where work has been reported bands are described as broad, weak,144 and nurnero~s.'~~ The literature on vibrational spectra of hydride bridged species has recently been surveyed.142 ''I A. P. Humphries and W.D. Kaesz. Prqq. //ior,q. <'/ll't?7.. 1979. 25. 145. 142 C. B. Cooper. D. F. Shriver. and S. Onaka. Ath. Chi. Sw.. 1978. No. 167. p. 232. 143 D. C. Harris and H. B. Gray. J. Am. c/llw7.Sor,.. 1975, 97, 3073. H. D. Kaesz and S.W. Kirtley. Ahs/r.trc/.v l63r.d Am. C/iom..Tor,. Mrc~/i/igBo.cio/i Mo.ss..1972, Inorg. 131."' M. W. Howard. U. A. Jayasooriya. S. F. A. Kettle. D. B. Powell. and N. Sheppard. J. C'hwi. Soc... Chcv?l.C'on?n?lm.,1979, 18. V. Katovik and R. E. MoCarley /mwg. C/iwi.,1978, 17. 1268. 439 14' Hydrido Complexes of the Transition Metals (ii) Nzrclear Magnetic Resonance Spectroscopy. The application of high field n.m.r. spectroscopy to hydride bridged species is sometimes prevented by low solubility or by very long (or short) relaxation times.147 Where spectra have been obtained resonances due to p,-bridging hydrides usually occur at rather higher field (ca. 25---35z) than those of their terminal counterparts. Temperature-dependent n.m.r.spectroscopy has been employed to study dynamic equilibria involving rapid intramolecular exchange of p,-hydride ligands (see Section 4H). Preliminary re- sults from solid state studies on several polynuclear carbonyl hydrides have been described, newer multiple-pulse techniques suppress the proton- proton dipolar interaction and allow data on the smaller interactions to be obtained.'48 Nematic phase n.m.r. studies coupled with structural information concerning heavy metal atoms have permitted the precise location of the hydride ligands in Ru,H,(CM~)(CO),.'~~ (iii) Miscellaneous Spectroscopic and Physical Techniques. Since many hydrido clusters contain numerous carbonyl ligands and are moderately volatile they are ideal candidates for study by mass spectrometry. Whereas terminal hydride ligands are frequently lost upon ionization, bridging hydride ligands are tenaciously re- tained -right down to the bare hydrido metal cluster in some instances.147 Mass spectroscopy therefore offers a means of distinguishing between terminal and bridging hydride ligands. U.V.photoelectron spectra of Re,H,(CO), 2, Os,H,(CO),,, and Os,H,(CO),, show broad ionization bands at ca.1 1.9 -12.7eV that have been interpreted in terms of localized three centre, two electron (3c, 20) bonds centred largely on the hydrogen atom.'37 A combination of Mossbauer and diffraction methods was used to determine the structure of Fe,H(SPr')(CO),. ' Mossbauer spectroscopy has also been used to demonstrate the presence of two different iron environments in Fe,H(C=NMe,)(CO), o.F. Diffraction Studies. --Because bridging hydrides are situated in close proximity to two or more heavy metal atoms the problems encountered in their location by X-ray diffraction methods are even more acute than those experienced with termi-nal hydrides. The situation can be improved by use of data treatment techniques, notably the elimination of high angle data and the application of Fourier averaging methods. However, even with these aids results are unpredictable. Therefore, in spite of the technical problems involved, neutron diffraction is the preferred method for location of bridging hydride ligands. A hybrid method, recently devel- IJi M. J. Mays and R. N.F. Simpson, J. C'/wt?i.Soc (A), 1968, 1444. '4x ?..T. Nicol and R. W. Vaughan. Ath. C'/icwi. SP~.,1978. No. 167. p. 248. IJ9 G. M. Sheldrick and J. P. Yesinowski. J. Cht~Soc,..Lkrltoti Trotis.. 1975. 873 and references therein. 'jo R. F. G. Johnson, J. Lewis. and P. W. Robinson. J. C%otn. So(,..Dcrltoti Trotis.. 1970. 1684. Is' R. B~u.R. Don, R. Greatrex. R. J. Haines. R. A. Love, and R. D. Wilson, Itiorg. Ciimi.. 1975. 14. 302 I. Is2 R. Greatrex. N. N. Greenwood. I. Rhee. M. Ryang. and S. Tsutsumi, J. C'hori. Soc.. C'hmi. Cot?m~un.. 1970. 1193. Moore und Robinson oped by Sheldrick et a/. 53 involves simultaneous least squares refinement of complementary X-ray and neutron diffraction data and thus reduces the number of neutron data required. In those instances where hydride ligandv have not been located directly by diffraction methods, their positions can often be deduced from indirect evidence including M-M bond distances (see Section 4G), electron counts, and ligand orientations.A combination of X-ray diffraction studies, to establish the heavy atom framework, and potential energy calculations to optimise the position of hydride ligands within that framework has been used to predict hydride positions in several cluster structures. 54 G. Nature of M-H-M Bridges. - - (i) Monohjdridc Bi-idqrd Species M-H-hi. Early X-ray diffraction studies on unsupported single bridged systems, notably [Cr,H(CO),,]-gave electron density projections which were taken to imply the presence of a linear M-H-M linkage.However, a later neutron diffraction study established that the bridging H position in [Cr,H(CO),,] is two-fold disordered and that the angle Cr-H-Cr is in fxt 158.9".' '' Numerous systcms of this type, including several hetero-bimetallic complexes, have all been shown to contain bent M-H-M linkages ( LM-H-M cu. 160' to 123 ). Neutron diffraction studies on several molecules with 'unsupported' M-H-M bonds including W,H(CO),(NO)' 56and W,H(CO),(NO)(P(OMe),~ 57 rcvealed that the bridging hydridc ligands are located at 'off axis' positions (Figure 5). This arrangcmcnt is consistent with the presence of 'closed' three centre, two electron (3c, 2~)bonds (Figure 6) analogous to thosc found in boranc chemistry and thus implies a certain degree of metal-metal bonding.Consequently thc concept of bond order is difficult to quantify in these systems. Although most M-H-M systems are symmetrical, the presence of non-equivalent ligand sets can produce markcd asymmetry. For example, the complex salt [NEt,][Mo,(p-H)(CO),(PPh,)] has Mo-H bond lengths of 1.68 and 2.19 More surprisingly, marked asymmetry of the 15.3 A. G. Orpen, A. V. Rivera. E. G. Bryan. D. Pippard. G. M. Sheldrick. and K. D. Rouse. J. C'hcwi. Sot,.. C'iioii. C'oiiiiiiwi., 197X. 773; A. G. Orpen. D. Pippard. G. M. Sheldrick. and K. D. Rouse. A(./(/ Cql..r/..1978, B34, 2466. IS4 A. G. Orpen. J. Orp/muw/.C'/icwi., 1978. 159. C'I : ./. C/iw7.Sot,. , Du//o/i Twis., 19x0. 2509. 15s J. Roriere. J. M. Williams, R. P. Stewart.J. L. Petersen. and L. F. Dahl, J. An7. C/iom. Sor.. 1977. 99, 4497. I56 J. P. Olsen, T. F. Koetzle. S. W. Kirtley. M. Andrew. D. L. Tipton. and R. Bau. J. Am. C'/ic/~i.So(,., 1974. 96. 662 I. 157 R. A. Love. H. B. Chin. T. F. Koetrle. S. W. Kirtlcy. €3. R. Whittlescy, and R. Bau. J. Am. C'/wii. Soc., 1976. 98, 449 1 . 158 M. Y. Darensbourg. J. L. Atwood. R. R. Rurch, W. E. Hunter. irnd N.Walker. J. Am. C'hcwi. So(.., 1979. 101. 2631. 44 1 Hydrido Complexes of the Transition Metals M M Figure 6 ‘Closed’M-H-M bond (a) orbitul rcprescwtution (b) symbol W-H-W linkage in the ‘symmetrical’ complex anion [W,H(CO), O]p (W-H, 1.718 and 2.070 A) has been found by low temperature (14K) neutron diffraction.’ 59 This situation has been further complicated by the discovery that the [W,H(CO), 0]-anion can adopt ‘linear’/eclipsed (Figure 7a) or ‘bent’/staggered (Figure 7b) structures depending upon the nature of the cation present.160 In addition to the mounting number of binuclear species containing ‘unsupported’ hydride bridges there are numerous examples of H atoms occupying edge-bridging sites in larger polynuclear structures.Indeed, the growing body of diffraction data indicate that edge bridging is the most common form of attach- ment for hydride ligands in metal clusters. Figure 7 Structures uck)ptccl by [W,H(CO), ,,I union, (a) 1incrrr:cclipsed und (b) bcntlstuggercd (ii) Dihydride Bridged Species M-(p-W,-M. Binuclear structures containing the M-(p-H),-M unit including Re,H,(C0),161 and [W,H2(C0)J2 -(re#: 162) were first characterized by diffraction methods in the early 1970’s.the first hetero- bimetallic structure of this type [(CsHs)2W(p-H),Rh(PPh3)2]’was reported in 1979.163 Dihydride bridges have also been observed in a small number of metal cluster species including Os,H,(CO), o.l 64 In contrast to the hydrogen atoms in single bridged complexes which are found to lie ‘outside’ the intersection point of the tvctns-ligand vectors, the hydrogen atoms in M-(P,-H)~--M systems lie within the intersection region of the ligand-metal vectors (Figure 8). The arrangement is usually discussed in terms of a four centre, four electron (4c,4e)bond.164 lsvD. W. Hart, R. Bau. and T. F. Koetzle, unpublished work cited in ref: 120.loo R. D. Wilson, S. A. Graham, and R. Bau, J. Or,gurronio/. (‘h~wi.,1975. 91, C49. Ibl M. J. Bennett, W. A. G. Graham, J. K. Hoyano, and W. I,. Hutcheon, J. Am. Chm. Soc.. 1972, 94. 6232. M. R. Churchill and S. W.-Y. Chang. Iriorg. Clim.,1974, 13. 2413. lb3 N. W. Alcock, 0.W Howarth, P. Moore, and G. E. Morris. J. C‘lrtw. So(..,C‘hcwi. C‘onrniuri.. 1970, 1160. R. W. Broach and J Moore and Robinson Figure 8 M(P~-H)~Mlinkuge; locution of hydrogcvi atoms within the intcrswfionregion of thc metuI-ligund vwtors (iii) 7rihydride Bridged Species M-(p,-H) 3-M. Complexes containing this grouping are still relatively rare, the first structurally characterized examples [Fe,(p2-H)3(triphos),I[PF6] and [Co,(p,-H),(triars),I[BPh,] were reported in 1973.' 65 Other more recently characterized structures include [Ir,(p2-H)3(C,Me,)2][BF4],'66the isoelectronic species [Ir2(p2-H),H,(PPh3),] [PF6],I6' and the heterobimetallic complex [(Et,P),Ir(p2-H)3Rh(dppe)][BF4].168 In each case the bridging hydrogen atoms lie 'within' the intersection region of the metal ligand vectors.(iv) Tetrahydride Bridged Species M-(p2-H14-M. The M-(,u2-H),-M core has been structurally characterized in Re,H,(PEt,Ph), (Figure 9)' 69 -the HK1 Figure 9 The skeleton of' the Re2H,(PEt2Ph), ~~oniplt~s,vicwd approsiniur<>ljidong the Re-Re usis 'agnohydride' [ReH,(PEt,Ph),], of Chatt and Coffey.' 70 The bonding has been rationalized in molecular orbital terms. The only other example proposed to date has been postulated for Ta,CI,H4(PMe3)4.172 The existence of the lb5 P. Dapporto.G.Fallani. S. Midollini, and L. Sacconi. J. Am. Ch~tn.Sot,., 1973,95,2021; P. Dapporto, S. Midollini, and L. Sacconi, ltiorg. C'hotn., 1975, 14, 1643. lb6 R. Bau, W. E. Carroll. D. W. Hart. and R. G. Teller. Ah. CIIP~I.Scr.. 1978. 167, 73. R. H. Crabtree. H. Felkin, and G. E. Morris, J. Orgumvie/. Clicwi.. 1977, 141, 205.'" A. Albinati, A. Musco, R. Naegeli. and L. M. Venanzi. Atigm. Chmi.,Int. Ed. EtzgI. 1981, 20, 958. R. Bau, W. E. Carroll, R. G. Teller, and T. F. Koetzle, J. Am. Clic~ni.Soc... 1977, 99. 3872. ITo J. Chatt and R. S. Coffey. J. Clicvn. Soc. (A), 1969, 1963. IT' A. Dedieu. T. A. Albright, and R. Hoffmann. J. Am. Chm. Soi,., 1979, 101, 3141.R. B. Wilson. A. P. Sattelberger, and J. C. Huffman, J.An?.Clicrti. Soc.. 1982, 104. 858. 443 Hydrido Complexes of' the Transition Metals M-(P,-H)~-M core has been speculated upon171 but examples have yet to be found. (v) M-M and M--H Bond Distances. Bridging M--H bonds tend to be significantly longer (ca.0.15-0.20 8,) than their terminal counterparts, a situation parallel to that found in boranes. Typical values are CLC. 1.6-1.9A for first row metals and ca. 1.7-2. I 8, for second and third row metals. In the absence of other bridging groups, the metal-metal distances in M(p,-H)M structures are L'U. 0.1to 0.4 8, longer than the M-M bonds in comparable non-bridged systems. Addition of further p2-H bridges decreases the M-M distance.Thus the observed trend in M-M distances is M(p,-H)M >M-M >M(p2-H),M >M(p2-H),M > M(p,-H),M. The introduction of non-hydridic (4-electron donor) bridging groups (i.e.Cl-, OR-, SR-) leads to lengthening of M-M distances.173 The short M-M distances in M(p,-H),M (n =2, 3, or 4) linkages have been rationalized in terms of M-M multiple bonds.162, 16s*166 (iv) Aciditj. of'Bridging (p,) Hydrid~s.Bridging hydride ligands in transition metal clusters are considerably more acidic than their terminal counterpart^.^ This observation, though rather anomalous at first sight, parallels that made for termi- nal and bridging hydrogen atoms in boranes.'74 In an early quantitative study hie be^-'^^ established the acidity order Fe,H,(CO), > Fe,H,(CO), > FeH,(CO),; later have shown that Fe,H,(CO),, is fully ionized in solution.Deprotonation is observed to occur with reagents of low nucleophilicity and high basicity (i.e.OH-, OMeC, or amines); very strong bases often cause degradation of the cluster. H. Chemical Reactivity.--Bridging (p2)hydrides are generally less reactive than their terminal counterparts. However, the expanding organometallic chemistry of metal cluster complexes includes many reactions involving hydride ligands. Since these are beyond the scope of the present article, interested readers are referred to recent more specialized reviews.' (i) Djwarnic Equilibriu. In polynuclear cluster structures fluxional behaviour involving rapid intramolecular exchange of hydride ligands between p2 and p3 or terminal sites is often Examples include Re,H,(PEt,Ph), (Figure 9) where terminal and bridging hydride ligands are equivalent on the n.m.r.time scale at temperatures down to 273 K,' 69 [Ir,H,(Ph,PCH,CH,CH,CH,PPh,),][BF,], which undergoes rapid intramolecular scrambling of the p, and p3 bridging hydride ligands,' 77 and Ru,(~,-H)~(CO)~,{P(OMe),) in which p2-hydride ligands in non-equivalent sites participate in fast exchange down to below 173 K.178In binuclear systems dynamic isomerization processes involving cis-trans inter-M. R. Churchill and R. A. LashewycL. 1tiorg. C'hl~r??..1979. 18, 1926. 3261; M. R. Churchiil and s. A. Julis. Inorg. CIIP~I..1977. 16, 1488. E. L. Muetlerties. 'Boron Hydride Chemistry'.Academic Press. New York. 1975. W. Hieber and G. Brendel. Z. Atiorg. AlIg~wi.Cliwi., 1057. 289. 324. H. A. Hodali. D. F. Shriver, and C. A. Ammlung, J. Am. C'liwi SIC,.,1978. 100. 5239. H. H. Wang and L. H. Pignolet. 111org.C/?iwl.,1980, 19, 1470. S. A. R. Knox and H. D. Kaesz. J. At77. Ch~t??.Soc.. 1971, 93. 4594. Moore and Robinson changes (equation 37) and ligand migrations (equation 38) are also encoun-tered. 40 PEt3 PEt-r H\l!/H\ (37)Et3P'I'H' PEt, PEt3 + ----z (38) trans -CPtH(solvent) (PCY,)~I' (ii) Hydrogen-Deuterium Exchange. Although bridging hydride ligands display considerable thermodynamic acidity they are frequently slow to undergo H-D exchange with D,O or D,PO,. Exchange is much faster if terminal hydride ligands are also present or can be introduced.The catalysis of H-D exchange in these systems by chromatographic supports (Florisil) has been attributed to formation of terminal hydride ligands by nucleophilic attack on the cluster (equation 39).'79 5 Face Bridging (p,) Hydrides The face bridging mode of hydrogen binding, first reported in 1968,' ** is relatively rare and has only been fully established in metal-cluster chemistry where about twenty examples involving hydride ligands located above the triangular faces of metal or metal-boron polyhedra have been identified. There is also evidence that hydrogen chemisorbed onto metal surfaces exists as atoms bound in p3 manner to triangular groups of adjacent metal atoms.' 81 A. Preparative Methods.-Very few p3-bridged hydrido complexes have been syn- thesized by design and no simple systematic preparative routes are known at present.Some typical synthesis are shown in equations 4042. B. Characterization.-Hydride ligands bound in p,-bridging mode are rather inert and cannot therefore be readily detected by chemical means. Characterization by I" M. A. Andrew, S. W. Kirtley. and H. D. Kaesz, Adv. C'hei~r.SPI-.,1978. No. 167, p. 215. IHo 0. S. Mills and E. F. Paulus. J. OrgimomoI. C'hmi., 1968, 11. 587. ]*I R. G. Teller, R. D. Wilson, R. K. McMullan. T. F. Koetzle, and R. Bau. J. Am. C'h~ni.Soc.. 1978. 100, 307 I. Hydrido Complexes of the Transition Metals spectroscopic techniques is difficult, and location by diffraction methods is not without problems.C. Spectroscopic Properties.-(i) Vibrational Spectra. The rather weak broad ab- sorptions associated with the vibrational modes of ,u3-bridging hydride ligands occur in a crowded region of the spectrum (1200-600cm-1), consequently they have attracted relatively little attenti~n.,~ However, in a recent study on carbonyl hydrido clusters assignments were made on the basis of deuteriation experiments and a vibrational model relating the ratio v(M-H),,,,/v(M-H),,, to the angular disposition of the three M-H bonds was successfully developed.* 82 (ii) Nuclear Magnetic Resonance Spectroscopy. Relatively few data are available since the small number of ,u3-bridged hydrido complexes known include several which for various reasons do not display high field proton resonances.Where high field signals have been recorded their positions range from ca.z 15 for Re,H,(CO),, to ca. z33 for Co,H,(C,H,), (refs. 183 and 184). In several instances fluxional behaviour has been detected by n.m.r., thus l3C n.m.r. spectra for the anion [Os,H(CO),,]- reveal rapid migration of the p,-H ligand between all eight faces of the octahedron above 188 K.18sRapid exchange of ,uz and p, hydride ligands has been noted earlier (Section 4H). (iii) Mass Spectroscopy. Face bridging hydride ligands tend to be firmly bound, species containing this grouping usually afford parent molecular ions and fre- quently lose all other ligands in stepwise fashion to leave the naked metal hydrido cluster ion.',' D.Diffraction Methods.-Most known face bridged hydrido complexes have been characterized by X-ray diffraction methods. Usually the p,-hydride ligands cannot be located directly because of the adjacent heavy metal atoms and their positions have to be inferred from spectroscopic or structural evidence. Early examples solved in this way include Rh,H(C,H,),, where the location of the hydride4gand centred over one face of the Rh, triangle was deduced from the high field proton n.m.r. quartet pattern' and Ru,H,(CO) where carbonyl groups distorted away from enlarged Ru, faces betrayed the positions of the ,u,-hydride ligands.*86 However, location of p3-hydride ligands in the highly symmetrical pseudo- tetrahedral Re,H,(CO), cluster was achieved directly from X-ray diffraction data by an image enhancement process involving superimposition of electron density functions calculated for all six mirror planes of the tetrahedron to yield composite difference maps. * 87 Other clusters in which ,u,-hydride ligands have been located by X-ray or neutron diffraction methods include Co,H,(C,H,), (re$ 188), IB2J.A. Andrews, U. A. Jayasooriya, I. A. Oxton, D. B. Powell, N. Sheppard, P. F. Jackson, B. F. G. Johnson, and J. Lewis, Inorg. Chem., 1980, 19, 3033. Is3 R. Saillant, G. Barcelo, and H. Kaesz, J. Am. Chem. Soc.. 1970, 92, 5739. Is4 J. Miiller and H. Dorner, Angew. Chem., Inf. Edn. Engl., 1973, 12, 843. Is5 C. R. Eddy, B. F. G. Johnson, and J. Lewis, J. Chem. SOC.,Chem. Cornmun..1976, 302. Is6 M. R. Churchill and J. Wormald, J. Am. Chem. SOL..,1971, 93, 5670. R. D. Wilson and R. Bau, J. Am. Chem. Sol.. 1976, 98, 4687. G. Huttner and H. Lorenz, Chem. Bet-., 1975. 108, 973. Moore and Robinson Ni,H,(C,H,), (ref. 18Y), and FeCo,H(CO),(P(OMe),} (ref. 181). Finally, face- bridging hydride ligands have been located in metalloborane clusters including FeH,(Me,C,B,H,), (H on FeB, face) and Co,(B,H,)(C,H,), (H on Co,B face). 90 E. Nature of Face Bridging Hydride.-Face bridging (p3)hydride ligands are generally located ca.0.9-1.08, above the centre of a triangular M, face,'91 metal-hydrogen distances range from ca. 1.55 8, (Ni-H) to 1.77 8, (Re-H),Iz0 metal-metal distances are close to the inter-atomic separations in the bulk metal. The metal hydrogen interaction can be described in terms of four centre, two electron (4c,2e) bonds; a rationalization of bonding in tetrahedral metal clusters containing p, and p3 hydride ligands has been given by Hoffmann and co-worker~.~~~Since the triangular M, groups involved in ,u3-hydride binding possess dimensions similar to those encountered in certain faces of solid metal crystals (notably (1 1I} or (01 l} surfaces of CCP or HCP metals, respectively) several authors have drawn attention to possible analogies between ,us-hydride ligands bound in metal cluster complexes and hydrogen atoms chemisorbed on metal surface^.^^^^ 94 The suggestion that hydrogen atoms chemisorbed on the Pt( 1 1 1) face occupy threefold bridging sites has been supported by electron- energy-loss spectroscopy, which gave frequencies of 1230 and 550 cm- ' for v(Pt-H),,y, and v(Pt-H),,,, respectively.' 94 F.Chemical Reactions.-The few known ,u3-hydride complexes tend to be rather inert thus CO,H,(C~H,)~ withstands heating to 300 "C under Nz.Is4 In several instances intramolecular exchange of ,u2 and p3 hydride ligands has been observed but few chemical reactions involving p3-hydride ligands have been reported. 6 Encapsulated (Interstitial) Hydrides Transition metal hydride chemistry came full circle in 1967 with the character- ization of {(Nb6H18)16,,} the first non-metallic solid containing an interstitial hydrogen atom.' More recently, the upsurge in metal carbonyl cluster chemistry has revealed a growing range of polynuclear molecular complexes incorporating hydrogen atoms encapsulated or semi-encapsulated by four, five, or six metal atoms.These structures provide hydrogen environments akin to those found in binary metallic hydrides and promise to throw new light on the behaviour of hydrogen occluded in metals. lay T. F. Koetzle, J. Muller, D. L. Tipton, D. W. Hart, and R. Bau, J. Am. Chem. SOC.,1979, 101, 5631. 190 J. R. Pipal and R. N. Grimes, Inorg. Chem., 1979, 18, 252 and 263. l9I T. F. Koetzle, R.K. McMullan, R.Bau, D. W. Hart, R. G. Teller, D. L. Tipton, and R. D. Wilson, Adv. Chem. Ser., 1978, 167, 61. 19* R. Hoffmann, B. E. R. Schilling, R. Bau, H. D. Kaesz, and D. M. P. Mingos,J. Am. Chem. Soc., 1978, 100, 6088.193 S. G. Louie, Phys. Rev. Lett., 1979, 42, 476. 194 A. M. Baro, H. Ibach, and H. D. Bruchmann, Surf: Sci., 1979, 88, 384. Hydrido Complexes of the Transition Metals A. Preparative Methods.-Encapsulated hydrides are often prepared by pro- tonation or hydrogenation of metal carbonyl clusters. Typical syntheses for specific complexes are illustrated in equations 4345. [CO,(CO),,]~-+ H+__* [Co,H(CO),,]-(43) B. Characterization.-Although the spectroscopic properties of interstitial hy- drido complexes have been examined they do not at present afford a reliable means of characterization. Encapsulated hydride ligands are therefore usually detected using structural data provided by diffraction studies. C. Spectroscopic Properties.-(i) Vibrational Spectra.Very little work has been done in this area. However, Ru-H/D stretching absorptions have been observed at low temperatures in the infrared spectra of the ions [Ru,H(CO),,]- and [Ru,D(CO) ,I-near 825 cm- and 600 cm-respectively. The very low Ru-H stretching force constant (21 Nm-l) is attributed to the fact that the H to Ru bonding is equally shared between six metal atoms. (ii) Nuclear Magnetic Resonance Spectroscopy. Chemical shift values for encap- sulated hydride ligands range from ca. -13zto +41 zindicating that interstitial hydride environments differ substantially from structure to structure. Early ~uggestions’~~that the extremely low chemical shifts found for [Ru,H(CO), ,I-(-6.5 z) and [Co,H(CO),,]- (-13.2 z) were indicative of hydrogen bound to carbonyl (formyl CHO group) or H-bonded to oxygen (0.. .H . .. .O) have largely been refuted (for the solid state at least) by subsequent neutron diffraction 198 The high chemical shift values found for [Ni, ,H(C0),,]3 -and [Nil ZH2(C0)21]2 -(z34.0 and z 28.0 respectively) are similar to those observed for some face-bridged (p,) hydride ligands and there is evidence from diffraction studies that the hydride ligands in the& structures are offset toward one internal face of the Ni, octahedron and may indeed be bound in this manner.199 D. Diffraction Studies.-Despite the obvious difficulties encountered in detecting, by diffraction methods, a single hydrogen atom invested by four or more heavy metal atoms, advances in knowledge of encapsulated hydride ligands have depended heavily on X-ray and, in particular, neutron diffraction studies. In some Iy5 I.A. Oxton, S. F. A. Kettle, P. F. Jackson, B. F. G. Johnson, and J. Lewis, J. Chem. SOL‘.,Clzem. Commun., 1979, 687. IgbP. Chini, G. Longoni, S. Martinengo, and A. Ceriotti, Adv. Chem. Ser., 1978, 167, 1 and references therein. Ig7 D. W. Hart, R. G. Teller, C.-Y Wei, R. Bau, G. Longoni, S. Campanella, P. Chini, and T. F. Koetzle. Angew. Chew., Int. Ed. Engl., 1979, 18, 80. Iy8 P. F. Jackson, B. F. G. Johnson, J. Lewis, P. R. Raithby, M. McPartlin, W. J. H. Nelson, K. D. Rouse, J. Allibon, and S. A. Mason, J. Chem. SOL‘.,Chem. Commun., 1980, 295. 199 R. W. Broach, L. F. Dahl, G. Longoni, P.Chini, A. J. Schultz, and J. M. Williams, Adv. Chem. Ser., 1978, 167, 93. 448 Moore and Robinson instances the hydride ligand has been located directly, in others its position has been deduced from structural evidence. Thus a slight enlargement of the metal atom polyhedron is often symptomatic of the presence of an interstitial hydride19'. 199 and the absence of localized distortions can often eliminate alterna- tive (p2or p3)modes of hydride co-ordination. E. Nature of Encapsulated Hydrides.-The introduction of a hydride ligand into an interstitial site in a metal cluster usually leads to a small expansion of the metal polyhedron-metal-metal distances increase by 0.02 to 0.07A-similar to that found in metallic hydrides. 99 Conversely, diffraction studies on some octahedral M,H clusters have revealed that the hydride ligands are located off-centre in the interstitial site and may have room to 'rattle around'.The conclusion is supported by the observation that in 'symmetrical' M,H octahedra the M-H distances are rather long (i.e.Co-H, 1.82A; Ru-H, 2.04 A). Marked similarities in the interatomic distances and interstitial site dimensions in nickel metal, NiH,., and [Nil ,H(CO), -emphasize the close structural relationship between metallic hydrides and interstitial complex hydrides. 99 Finally there is clear evidence (see below) that encapsulated hydride ligands have the ability to migrate within the larger metal clusters and, in solution, to leave the cluster entirely. These observations may have important implications for our understanding of hydrogen migration in metallic solids.(i) M4(y,-H) Clusters. Although hydrogen atoms occupy tetrahedral interstitial sites in metallic hydrides of the early (larger) transition elements, the presence of such an arrangement in a molecular complex was unknown until very recently. Early suggestions that FeCo,H(CO), ,(ref. 147) and [Fe,H(CO), 3]-(ref. 200) contained M4(p4-H) groups were subsequently refuted. However, the first fully authenticated example [Os,,H(C)(CO),,]- was reported in 1982 and more recently a similar structure with four M4(p4-H) groups has been proposed for [OS,,H,(CO),,]~ -(ref.201). (ii) M,(yu,-H) Clusters. This arrangement in which the hydride ligand sits at the base of a square pyramid and is semi-encapsulated by five metal atoms has recently been proposed for the twinned cubo-octahedral clusters [Rh, 3H5 -,,(CO),,]" -(n = 2, 3, or 4) (Figure 203 Variable temperature n.m.r.has shown that the hydride ligands are migrating rapidly within the metal cluster203 and chemical studies have revealed facile, reversible removal of protons by base.202 (iii) M,(p6-H) Clusters. Enclosure within an M, octahedron is the most common form of hydride encapsulation encountered in polyhedral metal carbonyl clusters. The hydride ligand may be centred in the M, cavity as in [Ru,H(CO),,]- (ref. 198) and [Co,H(CO),,]- (ref. 197) or may be offset, as if bound in p3 fashion to one 1 200 K. Farmery, M. Kilner, R.Greatrex, and N. N. Greenwood, J. Chem. Soc. (A), 1969, 2339. 201 D. Braga, J. Lewis, B. F. G. Johnson, M. McPartlin, W. J. H. Nelson, and M. D. Vargas, J. Ckem. SOL..,Chem. Commun. 1983, 241. P. F. Jackson, B. F. G. Johnson, J. Lewis, M. McPartlin, W. J. H. Nelson, ibid., 1982, 49. 202 V. B. Albano, G. Ciani, S. Martinengo, and A. Sironi, J. Chem. Soc., Dalton Trans., 1979, 978. 203 S. Martinengo, B. T. Heaton, R. J. Goodfellow, and P. Chini, J. Chrm. Soc., Chem. Commun., 1977, 39. 449 Hydrido Complexes of the Transition Metals Figure 10 Rhodium cluster in [Rh,,H, -(CO),,]" -(n = 2,3, or 4)anions; hydride ligunds are thought to he located within the square fraces internal face of the M, octahedron, as observed for [Nil,H(C0)2,]3- (ref.199). Those structures in which the hydride ligand is centred display remarkably low proton n.m.r. resonances (ca.-15z to -6 z). These signals which disappear revers- ibly when the sample is warmed have been taken to indicate that in solution the hydride ligand is able to migrate from the cluster and associate with solvent Protonation and H-D exchange reactions also indicate that the interstitial proton can leave its 'cage' with considerable ease.Ig7 7 Hydrides of the 'f-Block Metals Binary and ternary metallic hydrides of the lanthanide and actinide elements have been actively studied for many years. In contrast the first molecular hydrido complexes are of very recent origin and their chemistry is largely unexplored. A. Lanthanide Hydrides.-The synthesis of cerium(1v) cyclopentadienyl hydrides, CeH(C,H,), and CeH,(C,H,),, was claimed in 1974.204However, character- ization of these products as hydrido complexes rests heavily on the presence of a band at ca.2040 cm-in their infrared spectra and has not been confirmed. The first fully authenticated lanthanide hydrido complexes were prepared in 1982 using the reactions outlined in equations 46 and 47.,l 2 LnBu'(C,H,),.THF + 2H, -[LnH(C,H,),(THF)J, + 2 Bu'H (46) (Ln = Lu or Er) 3 ErBu'(C,H,),.THF + LiC1- [Li(THF),][Er,H,Cl(C,H,),] (47) All the complexes reported are extremely air and moisture sensitive. An X-ray diffraction study on the trinuclear erbium complex [Li(THF),][Er3H,C1(C~H5),] has established a triangular structure with p2 and p3hydride ligands (Figure 11); p2-hydride structures have been proposed for the binuclear species.2o5 Spec- troscopic features include infrared modes v(Ln-H) at ca. 1350-1 200 cm-and proton n.m.r.resonances z(Ln-H) at ca.5-9 z. B. Actinide Hydrides.-N.m.r. evidence for the formation of hydridic inter- mediates during the reduction of UCl(C,H,), with LiAIH, was reported in 204 S. Kapur, B. L. Kalsotra, and R. K. Multani, J. Inorg. Nuclear Chem., 1974, 36, 932. 20s H. Marquet-Ellis and G. Folcher, J. Organomet. Chem., 1977, 131, 257. 450 Moore and Robinson Figure 11 Structure determined by difraction methods ,for the union in the salt [Li(THF),I[Er,H,Cl(rI-C, H 5)61 1977,205the first isolable actinide hydride complexes [MH,(C,Me,),], (M = Th or U) were obtained the following year by a hydrogenolysis reaction (equation 48).20 2 MMe,(C,Me,), + 4H, -[MH,(C,Me,),], + 4CH, (48) The thorium complex was subsequently shown by neutron diffraction methods to possess a hydride bridged structure [Th(y,-H)H(C,Me,),],.zo6Terminal and bridging hydride ligands show infrared v(Th-H) modes at 1406, 1361 cm-' and 1215, 11 14 cm- respectively. A low field (-9.25 z)n.m.r.signal which remains a singlet down to 183 K is taken to indicate rapid exchange of bridging and terminal hydride ligands.,O Mononuclear thorium and uranium hydrides MH(N(SiMe,),), have been prepared by treatment of the corresponding chlorides with Na[N(SiMe,),]. For the thorium compound v(Th-H) and z(Th-H) occur at 1480cm-' and 9.1 z respe~tively.~~'All of these complexes react readily with chlorinated hydrocarbons to generate the corresponding chloridesz0* ,07 and un- dergo rapid H-D exchange.207* 208 The high affinity of the actinide elements for 0-donor ligands is reflected in the formation of dimeric enediolate derivatives [ThOC(H)=C(H)OTh] by facile insertion of CO into Th-H bonds at -78 cC.209 8 Conclusion The developments of the past 50 years, summarized in this review, clearly demon- strate that the hydride anion, the smallest and simplest possible ligand, possesses quite unique chemical and physical properties.In addition to arousing great academic interest, it has established an exceptional position as a participant in many important homogeneous catalysis reactions.Continued interest in the field 206 R. W. Broach, A. J. Schultz, J. M.Williams, G. M. Brown, J. M. Manriquez, P. J. Fagan, and T. J. Marks, Science, 1979, 203, 172. 207 H. W. Turner, S. J. Simpson, and R. A. Andersen, J. Am. Chem. Soi'.,1979, 101, 2782 and 7728. *O* P. J. Fagan. J. M. Manriquez, E. A. Matta, A. M. Seyam, and T. J. Marks, J. Am. CIiem. Soc., 1981. 103. 6650. 209 P. J. Fagan, K. G. Moloy, and T. J. Marks, J. Am. Clzeni. Soc., 1981, 103, 6959. 45 1 Hydrido Complexes of the Transition Metals of transition-metal hydrido complexes seems assured for the foreseeable future and must surely be rewarded with many more exciting and important discoveries. New work reported in the short period since the original typescript was submitted testifies to the pace of current progress.Novel complexes recently described include new paramagnetic h ydrides M H ,Cl,(Me, PCH ,CH,PMe,), (M = Nb, Ta),210 [Re,H,(PPh,),][PF,], [Re2H,(PPh3),(CNBu')][PF6], and [ReH(NCMe),(PPh,),(py)][PF6]3;2 tetrahydrido-bridged ruthenium com-plexes Ru,H,(PPh,), and Ru,H6(N,)(PPh3),;212 the trinuclear salt [H,(MePh,P),Re(p-H),Cu(p-H),Re(PMePh,),H,][PF,]; a related species [(Re2H,(PMePh,),), Cu,][PF,], containing an unprecedented planar rhomboidal (Re,Cu,) metal array2' and the remarkable molecufur hydrogen complex W(q2-H2)(CO),(PPrj), .2 l4 New reactions reported include facile inter-molecular activation of aromatic C-H bonds by lutetium hydride complexes,2* oxidative addition of dihydrogen reversibility to a metal cluster in [Os,Pt(p-H),(CO), ,(PCy,)12 and irreversibly across a Ta=Ta bond in [Ta,C16(PMe,),],21 and the reversible conversion of Cr(CH,)(CO),(C,H,) into CrH(CH,)(CO),(C5H,).218 2io M.L. Luetkens, W. L. Elcesser, J. C. Huffman, and A. P. Sattelberger, J. Chem. Soc., Chcvn. Commun.. 1983, 1072. 211 J. D. Allison and R. A. Walton, J. C'hem. Soc., Chrm. Commun., 1983, 401. *I2 B. Chaudret, J. Devillers, and R. Poilblanc, J. Chrm. Soc., Chem. Commun., 1983, 641. *I3 L. F. Rhodes, J. C. Huffman. and K. G. Caulton, J. Am. Chem. Soc., 1983, 105, 5137. 214 P. J. Vergamini and G. J. Kubas, Cliem. Eng. N~M's,28th Mar. 1983, 4. 215 P. L. Watson, J. Clirm. Soc., Chern. Commun., 1983, 276. *16 L. J. Farragia, M. Green, D. R. Hankey, A. G. Orpen, and F. G. A. Stone, J. Chem. Soc., Chem. Commun., 1983, 310. 2i7 A. P. Sattelberger, R. B. Wilson, and J. C. Huffman, Inorg. Chem., 1982. 21, 4179. *I8 K. A. Mahmoud. A. J. Rest. and H. G. Alt, J. Chrm. Soc.. Cliom. Comrnun., 1983, 1011.
ISSN:0306-0012
DOI:10.1039/CS9831200415
出版商:RSC
年代:1983
数据来源: RSC
|
7. |
High resolution laser spectroscopy |
|
Chemical Society Reviews,
Volume 12,
Issue 4,
1983,
Page 453-504
G. Duxbury,
Preview
|
PDF (3287KB)
|
|
摘要:
High Resolution Laser Spectroscopy By G. Duxbury DEPARTMENT OF PHYSICS, THE UNIVERSITY OF STRATHCLYDE, 107 ROTTENROW, GLASGOW G4 ONG 1 Introduction The rapid development of lasers since their first use as molecular spectroscopic tools in 1964, has resulted in very high resolution spectrometers becoming available for the wavelength region from the infrared to the ultraviolet, i.e. from 1 mm to 300nm. The availability of laser sources has completely transformed atomic and molecular spectroscopy by providing a sensitivity, resolution, and selectivity several orders of magnitude greater than previously available. The development of laser based systems parallels that of the earlier NMR and ESR spectrometers. The first decade after the discovery of the physical principles of the devices was primarily the province of physicists, who developed the techniques and the theoretical understanding of the processes involved in the interaction of coherent radiation with atomic and molecular systems, whereas in the second decade chemists have been applying these new radiation sources to systematic molecular spectroscopy.The parallel between laser spectrometers and magnetic resonance systems extends to a comparison of particular experiments, since many of the laser-based experiments are the optical analogues of those previously developed in the microwave and the radiofrequency regions. One area of particular interest in laser spectroscopy has been the study of free radicals such as NH, and CH,, in order to obtain a detailed knowledge of their potential energy surfaces in different electronic states.Other topics that have been treated in some detail are the experimental investigation of the variation of molecular dipole moments with electronic, vibrational, and rotational state, and the monitoring of state selective processes, where the definition of a ‘state’ includes the possibility of velocity group monitoring, which is precluded in the low frequency spectroscopic region of most microwave spectrometers, ten to forty GHz. In this article we will consider both the armoury of techniques now available for carrying out high resolution spectroscopy and also the use of combinations of the methods for studying small polyatomic molecules, which are of considerable interest because of their role in chemical reactions, or in testing current models of vibration-rotation interaction and of dipole moment variation. 2 Principles of Sub-Doppler Spectroscopy In order to appreciate the reasons for the high resolution and the high sensitivity of laser spectroscopy it is first of all necessary to consider the details of the High Resolution Laser Spectroscopy interaction between monochromatic high-power coherent sources and molecular absorbers.This comprises both the mechanisms of spectral line broadening in low pressure gases, and the perturbations of the line shape associated with the strong laser radiation field. The various types of available spectrometers can then be classified not only from the point of view of the resolution and the selectivity available, but also from the standpoint of analogies between laser spectrometers and magnetic resonance or microwave spectrometers currently in use in university chemistry departments. A.Line Broadening in Low-pressure Gases, Homogeneous, and Inhomogeneous Line Shapes.-Line broadening mechanisms in low-pressure gases can be classified into two types, homogeneous and heterogeneous. A similar classification was originally developed for line shapes in magnetic resonance spectroscopy. 1* Homogeneous processes are associated with the uncertainty principle, some decay process prevents the molecule remaining in a specified energy state for longer than a time interval At on average, and the line width y = l/d t.The principal contributions to this linewidth in gases are collisional broadening, when y is proportional to pressure; power or saturation broadening, which depends upon the rate at which the molecules are transferred between the upper and the lower energy states by the radiation field; and radiative decay of the excited-state level leading to natural line broadening. The natural contribution to the linewidth is particularly important when electronically excited states are involved in a transition. In laser spectroscopy a further mechanism becomes operative, transit-time broadening, which is caused by the molecules spending only a finite time in the radiation field as they pass through the laser beam. The general form of a homogeneously broadened line is Lorentzian, with the absorption coefficient given by where I‘, the power broadened line width is given by: and y, the half width at half height in the absence of saturation, is the sum of contributions from natural linewidth, pressure broadening, and transit time effects.1p121is the transition dipole matrix element between levels 1 and 2, which we will abbreviate as p, E is the electric field amplitude of the incident radiation field, and S1is the integrated absorption coefficient for the line. Heterogeneous line broadening in gases is associated with the Doppler effect. If a moving molecule emits radiation, the emitted frequency is not centred at the resonant frequency of the molecular transition, but is shifted by an amount which J.I. Steinfeld, ‘Molecules and Radiation: An Introduction To Modern Molecular Spectroscopy’, Harper and Row, 1974.* W. Demtroder, ‘Laser Spectroscopy’, Springer Verlag, 1981. Duxbury depends on the component of the molecular velocity in the direction of the emitted radiation. The Doppler effect occurs because the moving wave fronts of the emitted radiation are compressed in the direction of the molecule’s motion, and are expanded in the opposite direction, giving rise to a shift in wavelength and hence in frequency. The apparent emission frequency is increased if the molecule moves towards the observer, and is decreased if the molecule moves away. A similar effect occurs if we consider the absorption of radiation by a molecule.If we choose the +z direction in the laboratory to coincide with the direction of propagation of a laser beam, the frequency v1 of a laser in the laboratory frame of reference appears in the frame of reference moving with the molecule as: v’ = Vi(l -UJC) (3) where vz is the component of the molecular velocity along the z direction in the laboratory frame. The molecule can only absorb radiation if v’ coincides with its absorption frequency vo, i.e. v1 = vo(l + u,/c). The absorption frequency of the laser radiation is thus higher than the centre frequency if u, is positive, i.e. if the molecule moves parallel to the wave propagation direction, and is lower than the centre frequency if u, is negative, i.e. the molecule moves in the opposite sense to the direction in which the laser beam is propagating.These effects are shown schematically in Figure 1. The components of the molecular velocities along a fixed direction in a gas at thermal equilibrium at a temperature T obey the Maxwell distribution, w(uz)= (I/uJn)exp( -oZ2/u2) (4) where u2 = 2kT/m,rn is the molecular weight and k is Boltzmann’s constant. This gives rise to the Gaussian lineshape where the half width parameter S is given by The half width at half height is related to 6 by S(ln2)), and that at maximum slope by 8/42. Line shapes in gases are governed by both the homogeneous and the heterogeneous contributions. In the infrared region the line shape may be considered to be made up from a large number of Lorentzian curves, one for each group of molecular velocities.This is shown schematically in Figure 1. The resultant line shape is a convolution of the Gaussian and the Lorentzian contributions, and is known as the Voigt line shape. In the spectra of many molecules observed at wavelengths shorter than 20 pm, the predominant source of line broadening is the Doppler effect, with the homogeneous contribution to the overall linewidth being at least a factor of one hundred times smaller than the heterogeneous Doppler contribution. The partial or almost complete elimination High Resolution Laser Spectroscopy I I Frequency * (d) Figure 1 Influence of the Doppler efect on the width of a spectral line. (a) Homogeneous (Lorentzian) line shape, centred at v,,, if particle is moving perpendicular to the laser beam direction, z.(b) Doppler-shifted line position for one velocity component along z. (c) Envelopeof the line shapes of (a) and (b) produced when particles can move in all directions. (d) In a molecular beam, the spread of velocities along the z direction is small, area (a) is almost recovered of the Doppler contribution to line broadening thus results in a very great enhancement of the resolution. The oldest form of sub-Doppler spectroscopy is atomic or molecular beam spectroscopy, in which a collimated beam of atoms or molecules is directed at right angles to the radiation field. The reduction in Doppler broadening obtainable is then dependent upon the spread of the beam.The main problems encountered are associated with the low pressures in molecular beam spectrometers, which limits the sensitivity of the system, and with the difficulties inherent in producing intense, well collimated beams. T he more recent methods of sub-Doppler spectroscopy make use of the properties of the radiation field itserf to pick out a particular velocity group from the random set of molecules in the bulk gas sample in the cell. Duxbury This results in a higher sensitivity than is generally achievable in molecular beam experiments, it also results in much higher resolution than can usually be obtained using molecular beam spectrometers in the infrared and the visible spectral regions. B, Saturation-Effects and Sub-Doppler Linewidths.-If we now consider the effect of a strong saturating field produced by a laser on a predominantly heterogeneously broadened molecular absorption line, we can see from the discussion in Section A that the effect will be confined to a narrow region centred upon the velocity component of the group of molecules whose absorption is Doppler shifted into resonance. This produces a ‘hole’ in the population of the lower state and a ‘spike’ in the population of the upper state, resulting in a ‘hole’ in the absorption profile.This ‘hole burning’ model is due to Bennett3 and is shown schematically in Figure 2 for interaction of molecules with a running wave propagating in the +z direction. Since the molecular velocity component at resonance will have a homogeneously broadened lineshape, the ‘hole’, which depends upon the partial saturation of the Lorentzian line centred at u,, will also have a Lorentzian shape.Thus the ‘hole’ can frequently be treated as an absorption feature with a negative absorption coefficient. The width of the hole is determined by the factors governing homogeneous line broadening, i.e. The various methods of saturation spectroscopy rely on detecting this narrow hole by differential saturation effects, and hence achieving considerable resolution enhancement. Hole burning is really a description of the effect of the strong fields on the energy level populations and can be described quantitatively using a rate equation model, but it omits coherent excitation processes, which are sometimes very important.Some of the principal schemes for experiments based on velocity selection using the properties of the laser radiation field are sketched in the remainder of this section. C. Lamb Dips.-If the molecules are subjected to a standing wave field, holes are burned on both sides of the centre frequency as shown in Figure 2. At line centre the holes coalesce, corresponding to one group of molecules with zero velocity component along the direction of the standing wave field interacting with both running waves’ components. This double saturation interaction results in a sharp dip in the absorption coefficient at line centre. This effect was first seen as a dip in the gain of atomic gas lasers4 and slightly later as a dip in the absorption of an intracavity sample, giving rise to an increase of laser power at the centre of the saturated absorption line.It is for this reason that the absorption feature was initially known as an inverted Lamb dip.s The change in the absorption coefficient 3 W. R. Bennett, Phys. Rev., 1962, 126, 580; Comments on Atomic and Molecular Physics, 1970, 2, 1. 4 R. A. McFarlane, W. R. Bennett, and W. E. Lamb, Appl. Phys. Lett., 1963, 2, 189. 5 P. H. Lee and M. C. Skolnick, Appl. Phys. Lett., 1967, 10, 303. 457 High Resolution Laser Spectroscopy 0 Upper level "res Y Lower Level Level population difference 0 Figure 2 Saturation of an inhomogeneously broadened line. (a) Changes in particle velocity distribution under the action of a laser of frequency v.The z component of velocity of the particles interacting with the light wave is v,,, = c (v -vo) /vo.A Bennett hole is generated in the lower level and a population peak in the upper level. (b) Changes in population diflerence induced by a standing wave of resonance, v # vo. (c) On resonance, v = vo, a double saturation eflect is observed at line centre, the Lamb dip Duxbury between that which is observed using a standing wave field, and that which would be observed using a running wave over the same path length, can be denoted as the ‘effective absorption coefficient’ of the Lamb dip. This is given by 4’)= a(v)standing -a(v)running: (r+ ri)2 ~da(v) = -2a0 A (2~-2~+ (r+ )rOz~ (7) Where r and r’ are the power saturated half-widths of the ‘hole’ for the forward and the backward running wave fields, and A is a ‘saturation parameter’ dependent on the level of saturation of the absorption line.-4 ”2 -2 (a) (b) (c) Figure 3 Schematic energy level diagrams illustrating (a) single resonance, (b) three-level double resonance and (c) .four-level double resonance. The wavy arrows in (c) representcollision induced transitions (Reproduced by permission from J. Chem. Phys., 1975, 62, 1488) D. Velocity-tuned Three- and Four-level Resonances.-When two transitions share a common level and their transition frequencies overlap to within about two Doppler widths of each other, three-level ‘crossover’ resonances can be seen in addition to the ‘normal’ Lamb dips.6 A schematic energy level diagram is given in Figure 3.The effects of saturating one of the transitions can be probed via the second transition. The velocity matching condition for the velocity group, v,, for which this occurs is given by sZ( 1 + u,/c) = v1 and sZ( 1 -u,/c) = vt or sZ( 1 + v,/c) = v2and sZ( 1 -u,/c) = v1 (8) That is the oscillating fields E, and E-of the laser, frequency 52, affect molecules with the same velocity urn by means of different transitions, v1 and v2. This causes centre dips at the frequency 52 = (vl + v2)/2, and H.R.Schlossberg and A. Javan, Phys. Rev., 1966, 150, 267. High Resolution Laser Spectroscopy The intensity of the ‘centre dip’ is related to that of the adjacent Lamb dips, and to the degree of Doppler mismatch by da(v) = (dalda2)+exp -(9)-[:;I where da, and da, are the intensities of the Lamb dips, and dv = v1 -v,.If pairs of two-level systems are coupled by collisional population transfer, where the collisions ihvolve very little change in molecular velocities, four-level resonances may be observed, as first described by Shoemaker eta!. and by Johns et al. This situation is shown schematically in Figure 3. The frequency-matching condition is identical to that for three-level resonances, but the intensities are different, since not all collisions are effective in coupling the levels within the four-level system without change in velocity, and may couple to other energy levels in the system, the heat bath.If we denote the rate of dip transferring transitions by k, and the rate of transitions connecting the four-level system to the heat bath as k,, the intensity of the four-level resonance is given by The forces responsible for this effect are long range dipole-dipole interactions. The effects have mainly been observed as MJ changing collisions in Stark spectroscopy, and have been described as angular momentum tipping, or re- orientation, collision. The range over which they must occur has been estimated as ca. 2081.8 E. Optical-Optical Double Resonance and Level Crossing.-If a pair of transitions sharing a common level and overlapping to within about two Doppler widths interact with two co-propagating laser beams of different frequencies, a non-linear response will occur for the resonance condition9* -vv52, -G2= v1 -v2 or 4,-v1 = Q2 -v2 = 1 (1 1) where 52, -52, is the difference frequency between the lasers, v1 -v2 is the difference frequency between the molecular transitions, and v -(vl -v2)/2.The energy level patterns for which this will occur are the ‘vee’ scheme of Figure 3, or its inverse the inverted ‘vee’. These experiments are usually carried out by keeping SZ, -SZ, fixed, and by varying v, -v2 by the application of either an electric or magnetic field. Since the signal depends on a particular velocity group, uZ, being saturated, the linewidths are similar to those observed for Lamb dips. If two or more overlapping transitions sharing a common level become accidentally degenerate at some value of an applied electric or magnetic field, a ’ R.L. Shoemaker. S. Stenholm, and R. G. Brewer, Phys. Rev. A, 1974, 10, 2037. J. W. C. Johns, A. R. W. McKellar, T. Oka, and M. Romheld, J. Chem. Phys., 1975, 62, 1488. R. G. Brewer, Phys. Rev. Lett., 1970, 25, 1639. Duxbury Doppler free non-linear signal can be seen at high laser power. In the two level case this is just the degenerate case of OODR, when two transitions become coupled at one laser frequency. This type of signal is most frequently seen at zero field when it is known as the zero field level crossing, or Hanle, signal. At zero field all the MJ components are degenerate, or crossed, and hence can all be saturated. Once a finite field is applied the MJ degeneracy is broken, and only some of the non- degenerate components of the transition can be saturated.Thus a differential saturation signal can be seen centred on zero field. This signal is seen in all cases in which OODR signals are observed, and helps to locate lines which are suitable for study. Since the position of the signal, whether at zero or at a finite field, depends only on the degeneracy of energy levels and not on the absolute transition frequency, its position remains invariant to laser drift. The zero field level crossing signal line shape is rather complicated because of the problems involved with the multiple degeneracy of the crossing levels.10 Linear level crossing signals were observed in pre-laser spectroscopy in the fluorescence excited by atomic 1amps.l In the linear example a change in the angular dependence and the polarization characteristics of the emitted radiation occur, so that the fluorescence is monitored at right angles to the exciting beam.However, in the linear Hanle effect the total intensity radiated over all directions remains unchanged. F. Anti-crossing.-If avoided crossings between Zeeman or Stark sub-levels occur, sub-Doppler spectra can be observed that resemble those obtained in level crossing experiments. When anti-crossing signals are observed in absorption it can be shown that the terms which account for the differential saturation detection of the avoided crossing are similar to those responsible for the observation of non-linear level crossing signals.Additional terms arise, however, from the variation of the transition dipole matrix elements from the common level to the two interacting levels. The matrix elements of the dipole moment operator have a very rapid variation in the crossing region, owing to the fact that at the point of nearest approach the wavefunctions of the two interacting states are almost a 50:50 mixture of the unperturbed functions. Anti-crossing signals have been observed both in laser Stark' and in laser magnetic resonance' experiments. The terms in the molecular Hamiltonian responsible for the production of the weak avoided crossings are usually higher-order terms that are neglected in the standard treatments of the Stark and Zeeman effects. G.Doppler-free Two-photon Signals.-If we consider the interaction of a molecule with velocity u, interacting with a standing wave field of frequency vl, in the frame of reference of the molecule the molecule interacts with two oppositely travelling lo M. S. Feld, A. Sanchez, A. Javan, and B. J. Feldman, in 'Methodes de Spectroscopie San Largeur Doppler de Niveaux Excites de Systemes Moleculaires Simples', CNRS No. 217 1974, p. 87. J. Sakai and M. Katayama, Appf. Phys. Letr., 1976, 78, 119. l2 H. Uehara and K. Hakuta, Chern. Phys. Lett., 1978, 58, 287. 46 1 High Resolution Laser Spectroscopy waves of frequencies vl(l -VJC) and vl(l +UJC) respectively. If the molecule can be excited from the ground to the excited state by the absorption of two photons, one from each of the oppositely directed travelling waves, then at resonance the following condition is fulfilled, -UJC)+~~(1(E' -E"j/h =~~(1 +VJcj =2v1 (12) i.e.the resonance is independent of the velocity of the molecule, ~,.2,13?14In practice, two-photon transitions will be induced by the absorption of photons from travelling waves that are co-propagating, as well as the signal described above that results from the absorption of the two photons from counterpropagating waves. The Doppler-free two-photon signal will therefore be superimposed upon a broad Doppler broadened base, as shown in Figure 4.A fundamental difference exists between the narrow signals seen in two-photon experiments of this type and the narrow signals described in the previous sections.The two-photon signal arises from all molecules that absorb one photon from each of the counter-propagating beams, whereas the narrow signals described in the previous sections depend on the contribution of only one particular velocity component of the group of molecules to the signal. 3 +vz /c) -vz Ic) 2v =v,z Figure 4 (a) Compensation for the Doppler sh$t in the simultaneous absorption of two photons from two counter-running waves. (b) Absence of compensation from two unidirectional waves. (c) Shape of the narrow resonance in two photon absorption H. Optical-microwave Double Resonance.-A final way of reducing the Doppler broadening in the infrared and visible regions is to shift the detection from the optical to the microwave region, where the Doppler widths are very small.These double resonance methods involve either saturating the optical transition and detecting the effects of this on the microwave absorption, or vice versa. The l3 V. S. Letokhov, Science, 1975, 190, 344. l4 V. S. Letokhov and V. P. Chebotaev, 'Nonlinear Laser Spectroscopy', Springer Series in Optical Science, Springer Verlag, Berlin, 1977, Vol. 4. Duxbury observation of double resonance signals requires either that the optical and microwave transitions share a common level, or that the levels are strongly coupled by collisions that obey well defined selection rules. A variation of optical-microwave double resonance is two-photon spectroscopy, where the optical and microwave photons are absorbed simultaneously, so that the sum or difference of the frequencies of the optical and microwave photons is equal to the frequency of the transition being probed.Two-photon transitions have been observed in molecules subjected to a standing-wave field in which case ‘two-photon Lamb dip’ signals are seen. Since both the two-photon and the Lamb dip processes are non-linear in the radiation field, the theory is rather complicated.15 Several rather complete reviews of the subject matter of this section have recently appeared, and the reader is referred to them for further information.l6v l7 3 Experimental Techniques A. Resonance Methods.-Resonance methods that rely on ‘tuning a molecule’ rather than a laser were the first techniques to be developed for Doppler-limited and for saturation spectroscopy, since lasers with a good broadband tuning range were not available until the early 1970’s.The first two methods to be developed relied on the use of electric or magnetic fields to tune a particular set of transitions into resonance with a stable fixed-frequency gas laser. These methods are known as ‘Laser Stark Spectroscopy’ in the case of electric field tuning, and ‘Laser Magnetic Resonance’, (LMR), when magnetic field tuning is used. Both methods suffer from the disadvantage that a conversion from field sweep to frequency has to be made. The position of the zero-field transition frequency must therefore be inferred from the field tuning pattern.Despite this drawback they are still widely used owing to their high sensitivity, particularly of LMR, which has led to the detailed study of a wide range of free radicals and semi-stable species. Other resonance techniques that employ fixed-frequency lasers, or tunable lasers locked in frequency, are beam methods: ion-beam and bolometric spectroscopy. Both of these methods detect the absorption of radiation via some change in the properties of the ion or molecule, such as enhanced dissociation or heat capacity, rather than by detecting changes in the intensity of the radiation field itself. (i) Laser Stark Spectroscopy. The use of electric modulation methods in spec- troscopy dates from its introduction in microwave spectrometers in 1948.’ * These methods allow a considerable enhancement in signal to noise ratio over video detection methods, since the only signal detected is due to molecular modulation of the microwaves.F. Shimizu, Phys. Rev. A, 1974, 10, 950. l6 H. Jones, ‘Modern Aspects of Microwave Spectroscopy’, ed. G. W. Chantry, Academic Press, 1979, p. 123. T. Oka, ‘Frontiers in Laser Spectroscopy’, ed. R. Balian etal., North Holland, 1977, p. 531. la C. H. Townes and A. L. Schawlow, ‘Microwave Spectroscopy’, McGraw Hill, 1955. High Resolution Laser Spectroscopy In the infrared and visible regions it is possible to use parallel plate Stark cells in which very homogeneous electrostatic fields can be generated, and hence precise values of electrical properties of molecules, such as the dipole moment, can be measured.These methods were pioneered by Shimoda's Group in 1967,19 and in their present form by Shimizu in 1970.20,21 The original type of apparatus used a single CO, or CO laser, and this method of electric resonance spectroscopy has become known as Laser Stark Spectroscopy. In laser Stark spectroscopy, fixed-frequency laser radiation is passed through a parallel plate Stark cell. At certain values of the applied electrostatic field-specific electric-field components of the molecular vibration-rotation, or pure rotation, transitions of the absorbing gas are brought into resonance with the laser fre- quency giving rise to an electric resonance spectrum. In microwave spectroscopy it is conventional to use a waveguide Stark cell in which the electric vectors of the static electric field and of the radiation field are parallel, giving rise to the selection rule AM, = 0.In the infrared region the short wavelength radiation propagates through the Stark cell as a free-space wave and hence the electric vector of the radiation field can be both parallel and perpendicular to the static electric field, leading to both AM, = 0 and AM, = & 1 selection rules in the observed spectra. Since the method relies on the use of an electrostatic field for tuning, it is necessary to generate high uniform-fields and to study molecules with appreciable Stark tuning coefficients. In order to generate high electrostatic-fields, which may approach 90kVcm-', narrow plate spacings of from 1 to 4mm are commonly used. With such narrow gaps the plates must be flat to within one or two fringes of visible light, and be held accurately parallel.The gas pressure used must also be restricted to the low-pressure region below 100 millitorr. A useful rule of thumb is that a potential difference of 3000 volts may be sustained without electrical breakdown across any gas at a pressure of IOmillitorr and below. The Paschen curve of breakdown voltage against pressure for a particular gas can then be used to extrapolate to a different pressure/voltage regime. It should be noted that the minimum in the Paschen curve occurs at ca. 1 torr pressure, and hence that very restricted Stark tuning is available at higher pressures.The detectors used are quantum limited liquid nitrogen cooled semiconductor devices, PbSnTe or CdHgTe in the 10pm region and Au doped Ge in the 5pm region. Modulation frequencies of 5 to 100kHz are used to get above the principal noise region of the 5 and 10pm gas lasers. Intracavity** or multiple-pass absorption cells of the Shimizu type23 are gener- ally used, as described in Section 2C. In both types of cell Lamb dips can usually be seen due to the presence of a standing-wave field. The main disadvantage of the lg K. Sakurai, K. Uehara, M. Takami, and K. Shimoda, J. Phys. SOC.Jpn., 1967, 23, 103. 2o F. Shimizu, J. Chem. Phys., 1970, 52, 3572. 21 F. Shimizu, J. Chem. Phys., 1970, 53, 1149. 22 J. W. C.Johns and A. R. W. McKellar, J. Chem. Phys., 1977, 66, 1217. 23 G. L. Caldow, G. Duxbury, and L. A. Evans, J. Mol. Spectrosc., 1978, 69, 239. 464 Duxbury 10 20 30 40 E(kV/cm) I I I I I I 1 I 4.2 4.8 10.0 11.0 0.8 1.0 E(kV/cm) Figure 5 (a) Observed AM = & 1 transitions for the Q-branch series of "CH,F. The laser line used is the CO, 9pm P(18) line. The upper trace is a computer calculated band contour and the lower picture shows observed oscilloscope traces. The features on the shoulder belong to the Q (3,2) pattern. The sample pressure was about 5 m torr, and the time constant for detection was 10 ms. (b) Q(1,l) and Q(2,2) and Q(3,3) in l3 CH,F transitions with Lamb dip resolution (Reproduced by permission from J. Mol. Spectrosc., 1974, 53, 38) intracavity cells is that the spot size of most infrared gas lasers restricts their use to the 5 pm region, so that only low fields can be achieved as the plate spacings must be ca.3 to 4mm. The transitions seen in the v3 band of CH,F form a useful vehicle for discussion of typical laser Stark patterns.24 The 9 pm P( 18) line of the CO, laser lies very close to the v3 vibrational band origin. In perpendicular polarization the lowest J QQ transitions to be seen with AM, = 1 selection rules are shown in Figure 5. The spectrum consists of a set of QQ transitions (AKA4 which are brought into resonance at differing values of the static electric field. The first derivative shape is due to the use of small amplitude sinusoidal modulation, and is similar to that 24 S. M.Freund, G. Duxbury, M. Romheld, J. T. Tiedje, and T. Oka, J. Mol. Spectrosc., 1974, 38, 52. 465 High Resolution Laser Spectroscopy M 3 C .o -2+i-4 aJ IA x II V 54 0 -2 -L I I I I 1 0 2 4 6 8 10 Electric field kV cm-' Figure 6 Stark energy level diagram for the Q( 1,l) transition of "CH3 with AM = & 1. Since the Stark shifts of the excited and ground states are very similar, the splitting of the Q(1,l)resonance indicated in Figure 5 is caused by the small diflerence between the dipole moments in the upper and lower states of CH,F(Reproduced by permission from J. Mol. Spectrosc., 1974, 52, 38) observed in e.s.r. spectra.25 The origin of the patterns can be seen using the energy level diagram of Figure 6.It can be seen that if the dipole moment of the molecule is the same in the two vibrational states, and if the Stark tuning approximates to first order, all the resonances for a particular vibration-rotation transition will be degenerate. Since the Stark effect in many of the rovibrational transitions of CH,F is nearly first order, the centre of the patterns is determined by the average value of the dipole moment in the two states, and the spread of the pattern by the difference. The number of the components is given by 2J, and hence gives the J value of the transition. It should be noted that if the conventional microwave arrangement of parallel polarization is used, these resonances are seen only at very high values of the electric field since the transitions then tune as the dipole moment difference, ca.0.05D. The rather asymmetric pattern of the QQ (3, 2) transition shows that at high values of the Stark field the second and higher order effects can be very important. This means that the usual perturbation treatment of the Stark effect'* is inadequate for low J values, and that methods involving the diagonal- ization of truncated infinite matrices must be used in order to fit the ~pectra.'~ 25 A. Carrington, 'Microwave Spectroscopy of Free Radicals', Academic Press, 1974. 466 Duxbury The line widths observed in the spectra often provide clues to the assignment of spectra. If the Stark shifts are approximately first order, the apparent widths of the lines in terms of electric field are related to the Doppler width by the relationship where d V is the width of the resonance, and V is the central voltage.dv, is the width between the points of maximum slope, 2 8/42, and [vlaser -vo] is the fre- quency offset between the laser and the zero-field line. This relationship was particularly useful in the analysis of the spectrum of an unstable species such as HNO or HzCS, where the conventional gas-phase spectrum had not been obtained at the time of the original analysis, and where the band origins were obtained from matrix isolation data at low resolution. Other characteristic patterns are found for P and R branches of parallel bandsz5 and for Q branches of perpendicular bands.26 Even though the Doppler-limited spectra of light molecules such as NH, and CH,F can often be resolved, it is frequently impossible to resolve the structure in the crowded regions of the spectrum.Indeed, with heavier molecules such as 1,l-difluoroethylene, CH,CF,, it becomes very difficult to find regions in which simple patterns can be clearly seen. The types of saturation spectroscopy discussion in Section 2 have been widely applied in Stark spectroscopy to give greatly en- hanced resolution. This has enabled the structure of crowded spectral regions to be resolved, and has allowed the measurement of small changes in the electric dipole moments with vibrational states, as well as the measurement of small vibrationally or rotationally induced dipole moments, and the effects of weak intermolecular forces via the observation of four-level resonance signals.B. Laser Magnetic Resonance Spectroscopy (LMR).-LMR relies on the same idea as laser Stark spectroscopy, namely the magnetic field tuning of specific Zeeman components of pure rotation or vibration-rotation transition into resonance with a fixed frequency laser. The apparatus is more bulky than the Stark system, since a large electromagnet is usually required to generate the magnetic field. Since its introduction by Evenson and his colleagues in 1968,’ pure rotational LMR has proved to be one of the most powerful means of studying small para- magnetic species. The key to the success of the method was the adoption of the intracavity absorption cell pioneered by Wells and Evenson.28 This allows the small changes in the absorption in the intracavity cell to be detected via the non-linear amplification of the laser medium, giving rise to a very long effective path length.The advantages of Zeeman spectroscopy over Stark spectroscopy stem principally from the restriction in pressure imposed in Stark cells, owing to the necessity of avoiding electric breakdown between plates, so that the pressure in the Stark cell is well below the optimum for radical production. The narrow gap 26 Ci. Duxbury and S. M. Freund, J. Mol. Spec(rosc.,1977, 67,219. 27 K. M. Evenson, H. P. Broida, J. S. Wells, R. J. Mahler, and M. Mizushima, Phys. Rev. Lett., 1968, 21, 1038.28 J. S. Wells and K. M. Evenson, Rev. Sri. Instrum., 1970, 41, 227. High Resolution Laser Spectroscopy between the plates also precludes routine intracavity operation of Stark cells, except in the 5pm region. Finally there is the destruction of the species on the surface of the Stark plates. Intracavity Zeeman spectrometers resemble the Stark system of Figure 1 la if the Stark plates are replaced by a magnet, In general the energy level patterns seen in LMR spectra are more complex than their electric field equivalents, mainly as a result of two additional complications: the interplay of the various magnetic moments associated with the unpaired elec- tron and the nuclear spins, and the effects of spin-uncoupling at high magnetic fields, the Paschen-Back effect.This is illustrated in Figure 7. At low field the angular momentum associated with the unpaired electron is tightly coupled to the nuclear frame and hence Jand Fare good quantum numbers. At the low-field limit the magnetic sub-levels of the vibration-rotation or pure rotation transition have different tuning rates, resulting in a differential tuning into resonance similar to that discussed for Stark spectra. However, when the electron spin is uncoupled from the nuclear frame at high fields, the main tuning observed is that of the free electron and the differential tuning is lost. This limits the molecules with large Zeeman tuning coefficients to those with large spin-rotation interaction constants, such as NH,.,' The spin-rotation interaction is responsible for the large zero-field splitting seen in Figure 7.Thus in LMR the spin-rotation constants, principally E,,, play an analogous role to the electric dipole moment p in Stark spectroscopy in determining the tuning range, since E,, represents the magnetic moment about the a-axis which tends to align the spin with the molecular frame. LMR has a final and more subtle advantage over Stark spectroscopy in that the magnetic field does not mix levels of opposite parity, and hence information about asymmetry splitting in near prolate asymmetric tops such as H,CS and HO, that is often lost in Stark spectroscopy, is preserved in LMR. In pure rotation LMR the original lasers used, electric discharge HCN and H,O devices, have largely been replaced by optically pumped lasers.In the mid-infrared region CO, and CO lasers are mainly used, in general with intracavity LMR cells. (i) Ion Beam Spectroscopy. Much of the highest resolution ion-beam spectroscopy is carried out using fixed frequency lasers as described in a recent review by Carringt~n.~'It is a resonance technique in that the molecules are tuned into resonance with the laser, but it differs from LMR and laser Stark spectroscopy in that direct Doppler tuning of the molecular ion is accomplished by changing the ion velocity through the potential applied to a drift tube. The ions are generated in a mass spectrometer source giving a beam flux of 10" -lo4 s-with an effective ion concentration of lo4-lo6 ions cm-3.Light ions with a mass in the range 2 to 20, can be accelerated to a velocity of ca. lo5ms-with an acceleration potential of a few keV. In the drift tube the ion beam is colinear with that of a laser, which is reflected so that positive and negative Doppler shifts can be employed. Detection of the absorption of radiation is observed by modulating the laser beam and 29 P. B. Davies, D. K. Russell, B. A. Thrush, and H. E. Radford, Proc. R. Soc. London, Ser. A, 1977, 353,299. Duxbury MJ MN -3 2 1 1 2 0 -2-1 -1 -2-3 -1 --1 02 1-2 1 0 2 4 6 Flux density /kgauss Figure 7 Electron spin uncoupling in the levels of a polyatomic molecule caused by application of an external magnetic field (the Paschen-Back efect).At low fields the spin is coupled to the molecular frame and the Zeeman splitting is linear in M,. At highJield, spin is oriented with respect to the magneticfield direction in the laboratory frame, and the Zeeman splitting is linear in M,. The lower set of levels correspond to M, = -3,and the upper set to M, = +4.Note that-kJ = M + M, _I (Reproduced %y permission from ‘Chemical and BiologicaI Applications of Lasers’, ed. C. B. Moore, Academic Press, 1980, P.95) detecting the effect of this on some property of the ions that changes with ex-citation. The three methods most commonly employed have been charge exchange, predissociation, and muitiphoton-dissociation.30 One of the virtues of the ion beam method is that very high resolution is obtained owing to the kinematic compression of the velocity spread.This arises because 30 A. Carrington, Proc. R. SOC.London, Ser. A, 1979, 367,43. High Resolution Laser Spectroscopy although the energy dispersion of the ions when formed is several tenths of an electron volt, this forms a small proportion of the total energy once the ion beam has been accelerated with an energy of a few kilovolts, and hence there are very large reductions in the Doppler width. The observed Doppler widths range from 10MHz in the infrared to 100 MHz in the visible. (ii) Bolometric Spectroscopy. In this method the infrared spectrum of a molecular beam is obtained using a 2K doped silicon bolometer to measure the energy imparted to the beam by an appropriately tuned infrared laser,31 or by tuning the energy levels of the molecules in the beam using the Stark effect.32 This is a very sensitive method for obtaining accurate spectroscopic data on stable molecules and Van der Waals’ clusters.(iii) Optical-Optical Double Resonance and Level Crossing. As we saw in Section 2, if a molecular energy level splitting is tuned by the Stark or Zeeman effect, a non-linear response occurs for the resonance condition, AQ = Av, where 452 = Ql -Q,, the difference frequency between the laser beams co-propagating through the cell, and Av = v1 -v2 is the difference frequency between the coupled transitions that share a common energy level. In the earliest experiments, carried out by Bre~er,~ two very stable COz lasers were used to generate An.However, it was subsequently realised that the use of a single-amplitude modulated laser removed the requirement for frequency locking two separate lasers, and also allowed the frequency difference to be set directly, as it corresponds to the fre- quency of the modulator radiofrequency drive oscillator. As the sidebands move in phase with the carrier, the difference frequency is independent of the frequency drift of the laser, depending only on the stability of the RF drive oscillator. Two methods have been used, ele~tro-optic~~ and acousto-optic modulation. 34 Al-though electro-optic modulation was the first technique to be used to study the dipole moments of polar and non-polar acousto-optic modulation produces a single sideband only and hence will be used in the following discussion.A block diagram of a single sideband OODR spectrometer is given in Figure 8. By use of a half-wave plate/linear polarizer combination, overall selection rules of AM, (or AM,) = 0, & 1, _+2can be produced. For the selection rule )AM,I = 2, either beam 1 excites AM, = +I and beam 2 AMJ = -1, or vice versa. The resonance condition for a symmetric rotor is34 *As2 = QRF = If the overall selection rule is lAM,l = 1, the resonance condition for a symmetric rotor becomes 52 --PEK RF -hJ(J + 1) 31 T. E. Gough, R. E. Miller, and G. Scoles, Furaduy Discuss. Chem. Soc., 1981, 71, 77. 32 T. E. Gough, personal communication. 33 B. J. Orr and T. Oka, Appl. Phys., 1980, 21, 293.34 D. J. Bedwell and G. Duxbury, Chem. Phys., 1979, 37, 445. 470 Duxbury c 4 %-- c' Figure 8 Block diagram of a microcomputer controlled OODR spectrometer 47 I High Resolution Laser Spectroscopy j’ iI’ d 35 B. J. Orr and T. Oka, J. Mol. Spectrosc., 1977, 66, 302. 36 G. Duxbury, H. Kato, and D. Robinson, J. Mol. Sfruct., 198?, 80, 371. 37 G. Duxbury, H. Kato, M. L. Le Lerre, J. McCombie, and J. C. Petersen, in ‘Quantum Electronics and Electro-Optics’, ed. P. L. Knight, Wiley, 1983, p. 191. 38 G. Duxbury and H. Kato, Chem. Phys., 1982, 66, 161. Duxbury This method of double resonance, using a modulated source, is related to a method introduced by Dodd and Series39 called ‘resonances in a modulated light beam’.This is usually carried out using a weak light source and detecting the change in polarization of the emitted fluorescence as in the spontaneous emission level crossing experiments. The main difference from the experiments described above is that the detection is made at the RF beat frequency in the detector, as has been described by Lehman.40 This requires a detector with a wide bandwidth, (up to 30 MHz has been used), whereas in the experiments described in this article, a low modulation frequency of up to 100 kHz is used, putting a less severe constraint on the detector and the subsequent amplifier and associated electronics. Finite field level crossing and anti-crossing experiments have been carried out using both Stark and Zeeman tuning.The positions of the crossings or the avoided crossing are determined, for example, by the ratio of dipole moment matrix elements to the quadrupole matrix elements in the signals seen in CDJ, and are therefore outside the control of the spectroscopist. For this reason level crossing or anti-crossing is a less flexible tool than OODR, although it has been exploited successfully in several systems. In all these experiments the information obtained is the ratio of the electric dipole moment to the rotational hyperfine, or quadrupole splittings, or of the magnetic moment to the hyperfine constants. Experiments such as OODR are necessary to fix one or other of the variables so that absolute values of parameters can be obtained from these measurements.C. Frequency Swept Experiments.-These methods have been coming into general use over the last ten years, mainly due to the development of diode lasers, which provide coverage of most of the infrared region from 16 to 3 pm, and the devel- opment of dye lasers covering much of the near infrared, visible, and near ultra- violet regions.’ In the infrared region the main method used is absorption spectroscopy, since the spontaneous emission probability is low. In the visible and near ultraviolet regions, where the spontaneous emission probability is high, laser- induced fluorescence is the most sensitive method of detection. For molecules that do not readily fluoresce, indirect detection methods can be used.The most com- monly used methods are the opto-acoustic and opto-galvanic effects. (i) Sources. Most of the tunable laser sources commonly used, dye lasers, lasers based on non-linear frequency mixing, and semiconductor diode lasers have been described in some detail in a previous Chemical Society Review.41The main lasers which have been developed in the intervening period are colour-centre lasers and waveguide CO, lasers. Colour-centre lasers using F,(II) centres in KC1:Li and in RbC1:Li provide 10mW of tunable single mode power in the 2.3-2.8 and the 2.5-3.3pm re-gion~.~~Since these lasers are pumped by ion lasers in the same way as dye lasers, 39 J. N. Dodd, G. W. Series, and M. T. Taylor, Proc. R. SOC.London, Ser. A, 1963, 273, 41. 40 J.C. Lehman, Rep. Prog. Phys., 1978, 41, 1609. 41 J. K. Burdett and M. Poliakoff, Chern. SOC.Rev., 1974, 3, 293. 42 A. S. Pine, ‘New Techniques In Optical and Infrared Spectroscopy’, ed. G. W. Series and B. A. Thrush, Philos. Trans. R. SOC.London, Ser. A, 1982, 307, 481. 473 High Resolution Laser Spectroscopy they exhibit considerable amplitude fluctuations arising from the fluctuations in the ion-laser output. A commercial version of this type of laser is available, but requires considerable skill in operation. By constructing optical waveguides of high conductivity, materials such as beryllia, boron nitride or alumina, high pressure operation of CO, laser systems has become possible. These lasers are compact devices with power output of several watts.For high power operation their tuning range is restricted to ca. & 250 MHz from line centre, but up to & 1000 MHz can be obtained from short lasers in which output power has been sacrificed to obtain a wider tuning range. Very stable lasers of this type have been constructed for use in non-linear spectroscopy. A detailed review of all of the sources mentioned above, and also of some potentially useful devices, has recently been given by M~oradian.~~ (ii) Optogalvanic and Opto-acoustic Spectroscopy. In both of these techniques the absorption of laser radiation is detected as a change in the bulk properties of the medium. Opto-acoustic spectroscopic signals are obtained when internal excitation is degraded into translational energy to produce sound waves.If the source of radiation is chopped, an AC sound signal is detected by an intracavity microphone. One graphic example of the sensitivity of this method is the observation of the opto-acoustic spectrum of the electronic spectrum of H,CS44 in a lOcm long cell. The original experiments using conventional long-path absorption spectroscopy with a high resolution grating spectrometer required a pressure/path-length com- bination at least a factor of lo4 greater.45 Opto-acoustic spectroscopy is now widely used for trace-gas analysis and is reasonably well understood theoretically. The other indirect technique, optogalvanic spectroscopy, is also under active study but the basis of the effect is still the subject of some controversy, although its usefulness in some circumstances has been conclusively dem~nstrated.~~’~~ Optogalvanic signals arise when the impedance of a low-pressure discharge changes in response to the absorption of laser radiation by an atomic or molecular species within the discharge.The laser optogalvanic (LOG) spectrum is recorded by scanning the wavelength of the laser probing a DC excited discharge and monitoring the change in current through the discharge tube. The main require- ment is that the discharge generates as little electrical interference in the detection system as possible. In some cases radiofrequency excitation is more suitable than the direct current method and, in that case, the impedance changes are detected by the reaction of the oscillator.(iii) Optical-Optical Double Resonance using Fixed Frequency and Tunable Lasers. One method of combining the virtues of tunable and fixed frequency lasers to achieve sub-Doppler double resonance has been demonstrated by Weber and Terh~ne.~*.~’They carried out Stark-tuned double resonance experiments using 43 A. Mooradian, Rep. Prog. Phys., 1979, 42, 1534. 44 R. N. Dixon, D. A. Haner, and C. R. Webster, Chem. Phys., 1977, 22, 199. 45 R. H. Judge and G.W. King, J. Mol. Spectrosc., 1979, 74, 175. 46 A. I. Ferguson, Ref. 44,p. 645. 47 C. R. Webster and R. T. Menzies, J. Chem. Phys., 1983, 78, 2121. 48 W. H. Weber and R. W. Terhune, Optics Lett., 1981, 6, 455. 4q W. H. Weber and R.W. Terhune, J. Chem. Phys., 1983, 78, 6437.474 Duxbury a CO laser as a pump and a diode laser as a probe. For example, in the first series of experiments, the CO laser was locked to one of the Stark components of the aRR (9, 9) v4 transition of NH,, and the diode laser probed the aQ (9, 9) v, line. Complex narrow sub-Doppler features were seen, line narrowing effects and collision-induced resonances. This method should be generally applicable to the wide range of systems which lie within the tuning range of CO, and CO lasers. D. The Experimental Detection of sub-Doppler Signals. -Although the principles of sub-Doppler spectroscopy described in Section 2 apply equally to experiments carried out in the infrared and the visible wavelength regions, the methods used to detect the sharp resonant signals are usually rather different.In the infrared region the spontaneous emission probability of the excited state is usually rather low and hence the saturated absorption signal is detected via the change in the absorption coefficient. In the visible region, however, the spontaneous emission probabilities are usually quite large, and hence the saturation of the transition is often mon- itored via spontaneous emission from the excited-state energy levels. In level crossing and anti-crossing experiments the differences between experiments in the two regimes are more marked. Linear-level crossing and anti-crossing signals can be observed via fluorescence detection but cannot be observed in direct absorption. The signals observed via laser induced fluorescence can therefore be due to mix- tures of the linear and non-linear signals, whereas in absorption only the non- linear, or stimulated, signals can be seen.Another distinction that can be made in the detection of signals induced by the interaction of molecules with a standing wave field is whether the standing wave field is produced by a two-beam or a multiple-beam system. Two-beam systems are frequently used in electronic spectroscopy, a typical arrangement being shown in Figure (loa). The finite crossing angle of the beams results in an incomplete removal of the Doppler broadening, but the separation of the pump and the probe beams allows various tricks to be employed to enhance the selectivity of the system. Two of the most commonly employed are intermodulated fluorescenceSo and polarization spectroscopy.51 In intermodulated fluorescence the pump and the probe beams are chopped at frequencies o1and 0,,the fluorescence signal at either the sum frequency w1+ o2or the difference frequency o1-0,is detected. Since both the pump and the probe beams are necessary to detect the Lamb dip, only the dip is modulated at the sum or the difference frequency, the background being modulated at either ml or at o2is rejected by the phase-sensitive detection system. A similar scheme has been used to record intermodulated optogalvanic and opto- acoustic spectra.46 In polarization spectroscopy experiments the pump beam is circularly polarized and the probe beam linearly polarized, as shown in Figure lob.The circularly polarized beam induces an optical anisotropy in the sample, so that the atoms or molecules ‘dressed’ by the circularly polarized field rotate the plane so M. S. Sorem and A. L. Schawlow, Optics Commun., 1972, 5, 148. s1 C. Wieman and T. W. Hansch, Phys. Rev. Lett., 1976, 36,1170. High Resolution Laser Spectroxopy Crossed polarizcr (b) Figure 10 Two-beam saturation spectroscopy arrangements. (a) IntermocIulatedpuoresc.ence.(b) Polarization spectroscopy (Reproduced by permission from Opt. Commun., 1972, 5, 148; Phys. Rev.Lett., 1976, 36, 1170) of polarization of the probe beam. The molecule interacting with a strong ‘handed’ field thus behaves like a chiral molecule and the rotated probe beam can then pass through the analysing polarizer, which is crossed for the original direction of linear polarization.Since only the molecules with u, = 0 can be both pumped and probed in this way only Lamb dip signals are detected. Various versions of these methods such as polarization-intermodulated excitation, POLINEX, have been used sub- sequently. In the infrared region the principal methods used for sub-Doppler spectroscopy have involved the use of multiple-beam systems. These consist of either multiple- pass cells in the Shimizu c~nfiguration~~ or intracavity cells.22 These are shown in Figures 1la and 11 b. An intracavity cell can be considered as an example of a Duxbury FIXED ~4R;k' *44k* MOD e ref in I e--e x-Y m--DVM RECORDER (b) Figure 11 (a) Schematic diagram of an intracavity laser Stark spectrometer.PSD stands for phase sensitive detector, DVM for digital voltmeter, HV for high voltage, and mod for modu- lation source (Reproduced by permission from J. Chem. Phys., 1977, 66, 1217)(b) Mirror conjguration of a crossover White's cell ofthe Shimizu type with N = 4. The solid lines show beams propagating from left to right, and the broken lines from right to left. The circules are the crossing points between solid and broken lines. The angle 9 represents the$nite crossing angle of the beams. The Stark electrodes are parallel to the paper and occupy the area shown by the broken line rectangle (Reproduced by permission from J.Mol. Spectrosc., 1978, 69,239) multiple-beam system since the Fabry-Perot cavity of the laser acts as a multiple- pass system for the generation of the laser signal, and hence the absorber can be considered to affect the gain for each pass. In multiple-pass systems the forward and the backward waves are equal in intensity, and selective modulation of the pump and the probe beams is not practicable. However, the detection systems commonly used do in fact allow the sub-Doppler signals to be enhanced relative to the background. The main detection schemes use either a small amplitude electric or magnetic field modulation, or a small amplitude frequency modulation High Resolution Laser Spectroscopy of the laser. This results in the signal from the phase-sensitive detection system being detected as a derivative signal.’, Since it is also possible to detect signals at harmonics of the modulation frequency, further selectivity is possible.Detection at the second harmonic of the modulation frequency results in a second derivative signal, which resembles an inverted absorption signal. Since the Doppler-broadened background is a slowly varying function of either field or frequency, second-derivative detection results in an almost complete suppression of the back- ground if the modulation amplitude is of the order of magnitude of the linewidth of the sub-Doppler signal as is shown in Figure 9. Detection at the third harmonic is sometimes employed, particularly in laser stabilization systems.The third- derivative signal resembles a sharpened first-derivative signal but with far better background suppression. Third-derivative signals are, however, rather weak and hence the method is only suitable for molecules that are strong absorbers. Tests carried out with similar modulation methods in the visible region have shown that high frequency modulation methods give a signal to noise ratio that is only slightly inferior to that provided by methods such as intermodulated fluorescence. Multiple-beam cells are particularly useful for the detection of free radicals and of trace constituents since, for a limited volume cell, the total absorption coefficient is greatly enhanced over a two-pass system. The final differences that we consider, between saturation spectroscopy in the infrared and the visible wavelength regions, depend upon the type of laser used.In the infrared region most of the tunable lasers that have been developed, in particular diode lasers, generate insufficient power for saturation experiments. Most experiments so far have therefore been carried out with CO or CO, lasers. These lasers are line tunable but possess only a limited frequency tuning range about the centre frequency of each emission line. Most of the sub-Doppler experi- ments therefore depend upon the use of resonance methods. In the visible region the principal lasers used are dye lasers, which do allow sufficient power to be generated for saturation experiments, hence most experiments are carried out in the frequency swept regime.4 Stark Spectroscopy of Small Molecules Stark spectroscopy has been used for two main purposes, as a high resolution spectroscopic method for measuring the absorption spectra of low pressure gases, particularly of short lived species, and for the accurate measurement of electric dipole moments. It soon became evident that although the absolute precision of dipole moment determination by laser methods was little better than that achieved in the best microwave spectrometers, and was inferior to that of microwave molecular beam systems, the laser based spectrometers provided a more accurate wave of mea- suring changes in dipole moment than is possible in most microwave spectroscopy experiments. Furthermore, for some ‘non-polar’ molecules only laser techniques 52 D. H.Whiffen, ‘Spectroscopy’, Longmans, 1966, p. 55. were suitable for the measurement of the small vibrationally and rotationally- induced dipole moments. Most of the experiments so far have been carried out using fixed frequency, or frequency locked, lasers. A. Electronic, Vibrational, and Rotational Dependence of Dipole Moments.-The variation of the dipole moment of a molecule with its state is associated with the distortion of the electronic charge distribution. For changes in the electronic state it is easy to visualise a large effect occurring. The best known examples of this are the large changes of dipole moment that have been measured in aldehydes, partic- ularly formaldehyde, between the ground and the first excited state.These large changes are associated with the promotion of an electron from a non-bonding orbital to an antibonding 7c orbital on electronic excitation. The changes induced by vibration and rotation are usually much smaller, and are associated with the distortion of the nuclear frame from the equilibrium geometry. The formal treat- ment of these effects is similar to that applied to vibration-rotation interaction, where the vibrational and the rotational distortion effects are observed via the vibrational dependence of the rotation constants and the need to invoke centrifugal distortion constants. (i) Vibrational Eflects. The dipole moment of a polyatomic molecule may be expanded as a power series in the vibrational quantum numbers, and to first order in u may be written as ru = Pe + dPs0, (16) where pe is the equilibrium dipole moment and us is the vibrational quantum number of the s’th normal mode.Two terms contribute to dps,one involving first derivatives of the dipole moment with respect to changes in the vibrational co- ordinates, and one involving the second derivative with respect to the s’th co- ordinate, e.g. for the v3 vibration of CH,F, the CF stretching vibration, hp3 takes the where v3 and k,,, etc. are the vibration frequencies and cubic anharmonic force constants, qs is the s’th normal co-ordinate (in dimensionless units). It should be noted that the sum only includes the dipole moment derivatives for totally symmetric vibrations in the case of C,, molecules such as CH,F, and for the in plane vibrations for C,, molecules such as H,CO.A number of problems arise in the comparison of the experimentally determined dipole moment variations with those expected on the basis of the theory presented above. In very few molecules is the anharmonic force-field well determined, nor is the absolute sign of the dp/dqsknown in most cases, since the observables are the 53 M. Toyama, T. Oka, and Y. Morino, J. Mol. Spectrosc., 1964, 13, 193. 479 High Resolution Laser Spectroscopy infrared intensities, which depend upon the square of the dipole moment deriva- tive. However, it is often possible, by the use of simplified model^^^.'^ and of information on the intensity asymmetry of Coriolis coupled vibration-rotation bands,” to obtain some insight into the reasons for the patterns of dipole moment variation found.One of the most interesting effects of dipole moment variation occurs in mole- cules such as methane. At its equilibrium geometry it posesses no electric dipole moment, but under the influence of non-totally symmetric vibrations the ‘vibrationally averaged’ dipole moment is non-zero. This means that the ‘vibrationally’ averaged structure is no longer tetrahedral, a type of vibrational Jahn-Teller effect. The original model, which was due to Mizushima and Venkate~warlu,’~has been extended by Mills et aLS7 Vibrational changes in the electric polarizability tensor are also associated with the size of the dipole moment derivatives.However, unlike the changes in dipole moment mentioned above, the polarizability changes are directly related to the infrared intensities, i.e. to the square of the derivatives. Changes in the polar- izability are usually difficult to measure, as the polarizability plays a very minor role in the Stark shifts seen in most spectra. They have recently been observed in CO, which has no permanent dipole moment.’* (ii) Rotational Ed’ects. The theory of the rotational dependence of dipole moments is, in its present form, largely due to Wat~on.’~,~~ The effect can be most easily visualized as caused by the centrifugal distortion affecting the geometrical struc- ture. This effect can be expressed as: where the Qkare the normal co-ordinates. In the molecule fixed co-ordinate system, the dipole moment can be written as: pa = pa(e’+ C,, O,”’J, J, (19) where J,, the molecule-fixed components of the total angular momentum, are in units of h, and the coefficient of the dipole moment, QY, which is in units of electric dipole moment is given by: 54 G.Duxbury, S. M. Freund, and J. W. C. Johns, J. Mol. Spectrosc., 1976, 62, 99. 5s C. di Lauro and I. M. Mills, J. Mol. Spectrosc., 1966, 21, 386. 36 M. Mizushima and P. Venkateswarlu, J. Chem. Phys., 1953, 21, 705. s7 I. M. Mills, J. K. G. Watson, and W. L. Smith, Mol. Phys., 1969, 16, 329. 58 T. E. Gough, B. J. Orr, and G. Scoles, J. Mol. Spectrosc., 1983, 99, 143. 59 J. K. G. Watson, J. Mol.Spectrosc., 1971, 40,536. 6o J. K. G. Watson, unpublished notes. 480 Duxbury (a) Figure 12 Centrijiugal distortion-induced dipole moment in CH,(Reproduced by permission from 'Molecular Spectroscopy: Modern Research', ed. K. N. Rao, Academic Press, 1976, 11.) where Bs is the rotational constant for the b-axis, and dlp,/aQ, is the distortion of the By component of the inertia tensor with the excitation of the kth vibrational mode, which is of frequency Wk. From the above it can be seen that in order to calculate the vibrational and the rotational changes in dipole moment it is necessary to have a detailed knowledge of the changes in the dipole moment associated with excitation of particular normal vibrational modes in a molecule.The most interesting effect of rotational distortion is probably the production of a rotationally induced dipole moment in non-polar molecules, particularly those of tetrahedral equilibrium geometry. The distortion of a tetrahedral frame needed to produce such an induced moment is shown in Figure 12. Examples of both vibrationally and rotationally induced changes in dipole moment will be given in the following sections. B. Experimental Measurement of Dipole Moment Variation.-(i) Vibrational and Rotational Dependence in Polar Molecules. The dipole moment variation with vibrational state has been studied in some detail for three small polyatomic molecules, ammonia, methyl fluoride, and formaldehyde. In some of these mole- cules there is also some evidence for a rotational dependence of the dipole moment.Ammonia was the first molecule to be studied by laser Stark spectroscopy.20*2' It was discovered that a very large change in dipole moment, 0.22 D, occurred on the excitation of one quantum of the bending vibration, v2, and that there was some indication of a rotational dependence of the dipole moment as well.6i Since the 61 K. Shimoda, Y. Ueda, and J. Iwahori, Appf. Phys., 1980, 21, 181 481 High Resolution Laser Spectroscopy inversion barrier in ammonia is rather low, about two vibrational quanta of v2,a large change in dipole moment is to be expected, since if ammonia were planar the dipole moment would be zero. Since ammonia is inverting very rapidly, the ‘dipole moment’ of a state is really the expectation value of the dipole moment operator connecting the two components of an inversion doublet.Above the barrier to planarity the inversion doubling eventually becomes equal to half the vibrational spacing, and hence for high vibrational levels the vibrational spacing appears to be halved. Once the inversion doubling approaches the size of the vibrational spacing the vibrational transition moment and the ‘dipole moment’ become almost identical. This has been demonstrated by a recent experimental study of the 2v2 -v2 band of NH3,62and by observations of the 2v2 band by Doppler free two photon spectro~copy.~~ Another set of recent measurements on NH364have shown some of the pitfalls in the laser Stark approach.The effective dipole mo- ments of some of the inversion levels of the v4 state of ammonia were measured. v4 lies close to 2v2 and many of the levels are coupled via Coriolis interaction. One set of levels of the v4 state is not coupled to levels of the 2v2 state, and the effective dipole moment of these levels is within 1% of that of the vibrational ground state. This is as expected from the theory of vibrational changes in dipole moment given previously since v4 is a perpendicular vibration, and hence should produce little effect on the average value of the dipole moment along the symmetry axis. The other levels are strongly coupled and the ‘dipole moments’ measured ranged from 0.793 D to 1.326 D, compared with a ground state dipole moment of 1.47 D.The laser Stark method is therefore only really suitable for studies of unperturbed states where the molecular dipole moment can be treated as almost constant. Methyl fluoride was one of the first molecules to be studied at both Doppler limited and sub-Doppler resolution. A series of measurements using both the laser Stark and OODR method^,^^,^^,^^,^^ tied to the accurately known dipole moment of the ground state, has allowed the variation of the dipole moment to be measured in some detail. It has also allowed the isotopic dependence in the series 12CH,F, l3CH,F, and ”CD3F to be established. The results are summarized in Table 1. It can be seen that the dipole moment change on excitation of the v3 vibration of CD,F is approximately half that in CH,F.This has been explained by a ‘vi- brational mixing’ model.54 A small rotational dependence of the dipole moment has also been e~tablished.~~ Formaldehyde, H2C0, is possibly the best characterized of the systems, since the dipole moment has been measured in almost all of the vibrational states of H2C0, and in many of those of D2C0, using CO and CO, laser Stark spectro~copy.~~ -67 The dipole moments in excited vibrational levels involving v2 and v4 have been measured via the electronic spectrum, using the stimulated emission PUMP and 62 M. Takami, H. Jones, and T. Oka, J. Chem. Phys., 1979, 70, 3557. 63 W. K. Bischel, P. J. Kelly, and C. K. Rhodes, Phys. Rev. A, 1976, 13, 1829. 6A W. H. Weber and R. W. Terhune, J. Chem. Phys., 78, 6422. 65 J.W. C. Johns and A. R. W. McKellar, J. Chem. Phys., 1975, 63, 1682. 66 M. Allegrini, J. W. C. Johns, and A. R. W. McKellar, J. Mol. Spectrosc., 1977, 67, 476. 67 D. Coffey, C. Yamada, and E. Hirota, J. Mol. Spectrosc., 1977, 64, 327. 482 Duxbury Table 1 Dipole moments (D) of CH,F, I3CH3F, and CD,F "CH,F Ref 13CH3F Ref. "CD,F Ref. ground state 1.8585(5) a 1.8579(6) b 1.8702(21) c v3 1.9054(6) b 1.9039(6) b 1.8964(15) c v5 1.8751(21) e '6 1.859(5) f 1.8771(7) d 2v3 1.9519(20) b 1.951(4) 1.9170(5) e '3'6+ 1.909(5) f 1.932(7) f "M. D. Marshall and J. S. Muenter, J. Mol. Spectrosc., 1980, 83, 279. bS. M. Freund, G. Duxbury, M. Romheld, J. T. Tiedje, and T. Oka, J. Mol. Spectrosc., 1974, 52, 38."G. Duxbury, S. M. Freund, and J. W. C. Johns, J. Mol. Spectrosc., 1976,62,99. G. Duxbury and S. M. Freund, J. Mol. Spectrosc., 1977, 67, 219. "G. L. Caldow and G. Duxbury, J. Mol. Spectrosc., 1981, 89, 93. IG. Duxbury and H. Kato, Chem. Phys., 1982, 66, 161. Table 2 Dipole moments (Debyes) for formaldehyde and thioformaldehyde in various vibrational and electronic states State H,CO Ref. H,CS Ref. 'A 1 ground state v3 = 1 (CS)v, = 1 co 2.33 15(5) 2.3470(5) a b 1.6491(4) I .6576( 12) a C v4 = 1 2.3086(5) a 1.622(3) e V6 = 1 v3= I (CO) 2.3285(5) 2.3250(25) a d 1.642(5) e v5 = 1 2.2844(47) d v, = 2 2.3605( 20) d v, = 4 2.2723(86) f v2 = 2 v4 = 2 vz= 1 v4=4 2.3222(47) 2.2825( 33) f f a" 3A2 A" 'A, I .29(3) 1.56(7) g g 0.79(4) C State D,CO d D,CS e ground state 2.347 l(5) 1.658(3) v3 = 1 (CS) v, = 1 (CO) 2.3672( 15) 1.66l(3) v3 = 1 (CO) 2.319(10) V6 = 1 2.347(4) B.Fabricant, D. Krieger, and J. S. Muenter, J. Chem. Phys., 1977, 67, 1576. C. Brechignac, J. W. C. Johns, A. R. W. McKellar, and M. Wong, J. Mol. Spectrosc., 1982, 96, 353. ' D. J. Bedwell and G. Duxbury, J. Mof. Spectrosc., 1980, 84, 531. M. Allegrini, J. W. C. Johns, and A. R. W. McKellar, J. Mol. Spectrosc., 1977,67,476. G. Duxbury, H. Kato, and M. L. Le Lerre, Faraday Discuss. Chem. SOC., 1981,71, 97. P. H. Vaccaro, J. L. Kinsey, R. W. Field, and H. L. Dai, J. Chem. Phys., 1983, 78, 3659. R. N. Dixon and C. R.Webster, J. Mol. Spectrosc.. 1978, 70, 314.DUMP scheme of Field in a Stark-tuned system.68 The dipole moment variation with v4 is much larger than would usually be expected for an out of plane vibration because of the large amplitude motion in this particular co-ordinate. The changes P. H. Vaccaro, J. L. Kinsey, R. W. Field, and H. L. Dai, J. Chem. Phys., 1983, 78, 3659. High Resolution Laser Spectroscopy (a) t 1 I 1 I 0 10 20 30 -E/kV cm-’ --LA I 1 L-50 60 70 80 90 Figure 13 OODR signals in CH,D using electro-optic modulation. The resonunces labelled 2v occur between the sidebands only, and those by v between the sidebands und the carrier. The polarizers are set so that both signals can be observed. (a) Signals due to the$rst-order Stark shift of the J = 9, K = 2 level of the v6 state, SZ,, = 1.139 MHz, P = 4m torr.(b)Signalsdue to the first-order Stark shift of the J = 10, K = i level in the ground state, a,, = 0.5722 MHz, P = 4m torr (Reproduced by permission from J. Chem. Phys., 1979, 70,5376) in dipole moment are attributed to the smaller vibrationally projection of the CO and CH bond moments in the bent configuration upon the planar molecule configuration. The dipole moment variation in the other states is similar to that seen in CH,F, with the main change being associated with the excitation of the CO stretching mode, v2. This latter change is almost identical in v2 of D2C0since the CO stretching mode is almost an ‘isolated’ vibration in formaldehyde, unlike the situation when methyl fluoride is deuterated.These data are summarized in Table 2. (ii) Dipole Moments of Non-polar Molecules. The principal experimental work in this area has been by Oka and his colleague^.^^ They have studied the dipole ‘’T. Oka, ‘Molecular Spectroscopy: Modern Research’, ed. K. Pi.Rao, Academic Press, 1976,II, p. 229. Duxbury moments induced by rotation in the tetrahedral molecules SiH,70 and GeH,71 by a variety of methods including, infrared-microwave double resonance, optical-optical double resonance, and laser Stark-Lamb dip spectroscopy. The results from experiments on the latest molecule in the series, CH,D, will be described in some detail, since they exemplify many of the important aspects of the work. CH,D has a small dipole moment of ca.5.6 x D which is due to the breakdown of tetrahedral symmetry. In order to observe OODR spectra very small RF frequencies and very high values of the applied electrostatic field are required as shown in Figure 13.72 In a C,v molecule there are only four independent 8 parameters, I$‘, I$‘, $‘,and $it,where [ is taken along the C, axis, the ( axis on a 0,plane and the q axis is perpendicular to them. Since the Stark effect depends upon the projection of the molecule fixed axes onto axesfixed in the luborutory, the space fixed axes, only certain linear combinations of the 8 parameters are deter- mined. The matrix elements which give rise to the first-order Stark shift are obtained by calculating the expectation value of the dipole moment operator in the space fixed axis system.This gives rise to an ‘effective dipole moment operator’ of: When the 8tyare expressed in terms of the internal co-ordinates S,, rather than the normal co-ordinates, the parameters in CH, and CH3D can be related. In particular it could be shown that the parameters of CH,D could be predicted from those of CH,, which can be represented by a single parameter, 82.59*73 (iii) Polarizability Changes in Nan-polar Molecules. A recent experiment on CO, has provided information on an important property, the vibrational dependence of the molecular polarizability tensor. An F-centre laser has been used to excite the v1 + v, band of CO, at 37 15 cm- in a supersonic nozzle beam and the laser Stark spectrum bolometrically detected.” In order to measure the Stark splitting, fields of up to 230 kV cm-’ were required.This is much higher than the fields used in ‘conventional’ Stark spectroscopy, where the maximum field rarely exceeds 80 kVcm-.’. Measurements of the Stark splittings of the R(0)and P(2)transitions have allowed the evaluation of the isotropic part, (a,,), and the anisotropic part, (da,,),of the static polarizability in both the excited (n = v) and the ground (n = 0) states. The direct observables are da,/da, = 1.021,8 (ao-a,)/da, = 0.012,6 and da, = 2.65 The result for da, is 12% greater than that predicted for the static polarizability anisotropy, and the reason for this is not well understood. The other parameters are well predicted by the model for the vibrational dependence of c1 which is discussed in detail by Gough et al? (iv) Dipole Moment Variations with Electronic State.The change in dipole moment with electronic excitation is expected to be large compared with most vibrationally induced effects. Dipole moment changes with electronic excitation were first mea- ’O W. A. Kreiner, T. Oka, and A. G. Robiette, J. Chem. fhys., 1978, 68, 3236. ’l W. A. Kreiner, B. J. Orr, U. Andresen, and T. Oka, fhys. Rev. A., 1977, 55, 2297. 72 J. K. G. Watson, M. Takami, and T. Oka. J. Chem. Phys., 1979, 70, 5376. l3 I. Ozier, Phys. Rev.Lett., 1971, 27, 1329. 48 5 High Resolution Laser Spectroscopy sured using high resolution grating spectrographs and spectrometers,74- 76 but laser methods allow a much higher resolution to be achieved and hence higher precision.They also permit the characterization of perturbations via the changes in the effective dipole moment, as we saw in the vibrational example, i.e. NH,.64 As an example of the use of laser methods for the measurement of dipole moment changes the visible OODR spectrum of HNO observed by Dixon and his colleagues will be used.77 They measured the Stark effect of several transitions in the JIA”---flA‘spectrum of HNO using OODR with fluorescence detection. The tunability of the dye laser was exploited to select the desired zero-field coincidences, rather than the chance method using fixed frequency lasers. The amplitude mod- ulation of the RF carrier by a second RF oscillator produced frequencies of 52, -52, & 52,, where a,, a,, and 52, are the frequencies of the laser, carrier, and modulator respectively.Thus the double resonance was between 52, -52, & 52, and 52,, or between (52, -52,) + 52, and (52, -52,) -a,, with 52, suppressed. As 52, can be made very small, continuous tuning at low frequency can be achieved. Examples of the M resolved double resonance produced in this way are shown in Figure 14. As the ground state of HNO has been studied in some detail by laser Stark spectroscopy,22 the electric field was calibrated using the known ground state dipole moments. Using this method the dipole moment, pa of the A state of HNO was evaluated as 1.08 001 D, compared to a value of 0.996 D in the ground state. Since the ‘A’’ state is perturbed and predissociated, the OODR method is being used to characterize the perturbations and hence elucidate the mechanism of the effects.C. Effects of Collisional Energy Transfer.-Much of the work on collisional energy transfer has stemmed from the pioneering work of Oka,78 who pointed out that many collisional processes between dipolar molecules obey electric dipole selection rules, i.e. that the molecule after a collision possesses a ‘memory’ of its state before the collision. In the early work there was little evidence for velocity selection effects, since the majority of the experiments were carried out in the microwave region where the broadening processes are largely homogene~us.~~ However, in a set of infrared experiments using two photon pumping and probing Oka and his col- leagues showed that many dipolar collisions are ‘soft’.79 This means that, although the molecules change their rotational quantum state following electric dipole selection rules, the velocity component in the direction of the laser field is un- changed.Following this difficult experiment it was realised that these effects could be seen ‘routinely’ in laser Stark Lamb dip, and in OODR experiments as ‘four- level resonances’. 74 D. E. Freeman and W. Klemperer, J. Chem. Phys., 1966, 45, 52. ‘Is N. J. Bridge, D. A. Haner, and D. A. Dows, J. Chern. Phys., 1968, 48, 4196. 76 A. D. Buckingham, D. A. Ramsay, and J. Tyrrell, Can. J. Phys., 1970, 48, 1242. 77 R. N. Dixon and M. Noble, Chem.Phy.~.,1980, 50, 331. T. Oka, Adv. At. Mol. Phys., 1973, 9, 127. 79 S. M. Freund, J. W. C. Johns, A. R. W. McKellar, and T. Oka, J. Chem. Phys., 1973, 59, 3445. Duxbury i 0 50 100 150 Vo\ts (a 1 +200 r 1M1 41 \ 0 50 100 Volts+200 I (b)Figure 14 MJ resolved OODR spectra of HNO using laser inducedfluorescence, with the difference frequency generated using an amplitude modulated acousto-o tic modulator. (a) Mixedpolarization spectrum of the RR3(3) line of the A'"' band with (000)A'.! (100)--X a,, = 15 MHz, and a 3 mm spacing between the Stark electrodes. ES, excited state resonances and GS, ground state resonances. (b) Energy level diagram showing the Stark splitting and resonant intervals corresponding to the spectrum in (a) (Reproduced by permission from Chem.Phys., 1980, 50, 31 1) High Resolution Laser Spectroscopy 5-4 4-5 7.3 7.5 7.7 4-3 3-4 12.0 12.2 Electric field kVcm-1 Figure 15 Lamb dip spectra of the Q (53) and the QQ(4,4) transitions of the v3 band of H,CS using second derivative presentation.?he four-level collisionally transferred resonances seen between Lamb dips of the QQ(5,5) transition are approximately 50% of the intensity of the Lamb dips. The time constant for detection was 300 ms and the sample pressure about 5 m torr (Reproduced by permission from J. Mol. Spectrosc., 1980, 84, 531) Methyl fluoride was the first molecule in which four-level collisionally trans- ferred resonances were clearly identified,7,8 although subsequently they have been recognized in many systems.The example of these signals in H,CS, Figure 15, shows that they may approach at least 50% of the intensity of a three-level resonance. This demonstrates that in molecules in which the dipole moment is directed entirely along the principal near symmetric rotor axis, Q or c, the AM = 4 1 angular momentum tipping collisions are extremely selective. However, in molecules such as CH,OH and CH,NH80,81 where there is also a b component of the dipole moment, the four-level resonances are much less pronounced, indi- cating that collisional coupling to other rotational levels is now much more proba- ble. In some of the transitions seen in H,CO changes of AM = 4 or 6 have been observed.8 80 D.J. Bedwell, G. Duxbury, H. Herman, and C. A. Orengo, Infrared Physics, 1978, 18, 453. 81 G. Duxbury, H. Kato, and M. L. Le Lerre, Furuduy Discuss. Chem. Soc., 1981, 71, 97. 488 Duxbury D. Level Crossing and Anti-crossing Spectroscopy.--CD,I provides nice examples of Stark-tuned level crossings and of anti-crossings of M, levels associated with the quadrupole hyperfine structure.’ 1*23,82 The level crossing signals observed in the 2171,2state of NO are associated with magnetic hyperfine structure,83 and POF, exhibits an extensive range of anti-crossing signals, which are due to weak avoided crossings between levels that differ in k by +3.84 In these experiments the information obtained is the ratio of the electric dipole moment to a hyperfine, rotational, or quadrupole splitting, or of the magnetic moment to the hyperfine constants.Experiments such as OODR are necessary to fix one or other of the variables so that absolute values of parameters can be obtained from these measurements. 5 Studies of Semi-stable Molecules One of the principal advantages of laser spectrometers is their sensitivity for the detection of small quantities of short-lived species. The sensitivity of laser Stark spectrometers has been discussed by Freund etaZ.,24 and by Johns and M~Kellar.~’They estimate that in the intracavity spectrometer at pressures of 4 m torr in a 20 cm absorption cell, only about lo6 molecules are responsible for the Lamb dips seen on a particular vibration-rotation transition.This is close to Shimoda’s estimate of the limiting sensitivity of a laser spectrometer.8s This sensi- tivity has been exploited in the study of small ‘semi-stable’ molecules which, in Kroto’s recent definition,86 have lifetimes of the order of seconds under the conditions of the gas phase experiments. For example, the laser Stark spectroscopy method is suitable for studying molecules such as HN0,22 CH,NH,8’*87 and H2CS,88 but not most free radicals since the metal surfaces of the Stark plates catalyse the decomposition of the unstable molecules. The shortest-lived species studied in this type of cell is HC0,89 which has been shown to be on the limit of the detectivity of this type of spectrometer. Possibly the most interesting of the semi-stable molecules to be studied by laser methods is thioformaldehyde, H2CS.Until quite recently thioformaldehyde was known only as a trimer, the first evidence for the existence of the monomer being in mass spectrometric studies. Interest in the species developed following the observation of the microwave spectrum of H2CS in 1970,90followed by its sub- sequent detection in the interstellar dust cloud^.^' The first high resolution infrared spectrum of the 3pm bands was obtained in 1971,92and required a high gas 82 J. Sakai and M. Katayama, Chem. Phys. Lelt., 1975, 35, 3. 83 A. R. Hoy, J. W. C. Johns, and A. R. W. McKellar, Can. J. Phys., 1975, 53, 2029. 84 T. Amano and R. H. Schwendeman, J. Mol. Spectrosc., 1979, 78, 437. K. Shimoda, Appl.Phys., 1973, 1, 77. 86 H. W. Kroto, Chem. SOC.Rev., 1982, 11, 435. M. Allegrini, J. W. C. Johns, and A. R. W.McKellar, J. Chem. Phys., 1979,70, 2829. D. J. Bedwell and G. Duxbury, J. Mol. Spectiox., 1980, 84, 53. 89 B. M. Landsberg, A. J. Merer, and T. Oka, J. Mol. Spectrosc., 1977, 67, 459. 90 D. R. Johnson and F. X. Powell, Science, 1970, 169, 679. 91 M. W. Sinclair, N. Fourikis, J. C. Ribes, B. J. Robinson, R. D. Brown, and P.D. Godfrey, Ausf.J. Phys., 1973, 26, 85. 92 J. W. C. Johns and W. B. Olsen, J. Mol. Spectrosc., 1971, 39, 47. High Resolution Laser Spectroscopy pressure and a long path length, and was with a diffraction grating instrument a 'tour de force'. Similar difficulties were encountered in the recording of the W'A,-%'A, visible absorption in 1975, when path lengths of ca.100m were req~ired.'~ Laser techniques were first employed in the infrared region when the 10 ,um band system was observed using a laser Stark spe~trometer.~~~~~ The thioformaldehyde was produced by the pyrolysis of dimethyl disulphide.88 The pyrolysis products, which include CH, and CH,SH, flowed continuously through a multiple-pass Stark cell. A very high signal-to-noise ratio was obtained (Figure 15) with short time constants of 100 to 300 m s in the detection system, even though the total gas pressure in the cell was only 10 m torr. In the 10 pn region three fundamental vibration-rotation bands were observed, v3, vq, and v6, rather than the two v3 and v, expected from the matrix isolation spectroscopic result^.^' Two of the bands, v, and v6, are nearly degenerate and hence their energy level pattern resembles that of a degenerate state of a symmetric rotor.Owing to the high resolution and sensitivity of the Lamb dip method in a crossover multiple-pass cell, the vibrational charge in dipole moment could be measured very accurately, and strong signals due to collisional energy transfer observed, as shown in Figure 15. Stark spectroscopy, therefore, played a useful role in sorting out the vibrational analysis of the infrared spectrum of H,CS,23 the details of which were subsequently confirmed by Fourier Transform Spec-tro~copy.~~Results from OODR experiments have been combined with laser Stark88 and molecular beam dataQ7 to give a very detailed picture of the vi-brational dependence of the dipole moment in the ground electronic state.The observed variation of p is compared in Table 2 with that observed in formaldehyde. Subsequent laser experiments have been aimed at obtaining a better under- standing of the structure of the electronically excited states. Initial experiments were carried out using opto-acoustic dete~tion,~, but most recent ones have used laser induced fluorescence.98- loo The one exception to this was the determination of the dipole moment change with electronic excitation in the 4', band of the 2-2 system which was carried out using absorption spectroscopy."' Studies of the fluorescence excited by pumping a single rotational level of a given vibronic state have allowed the final ground state fundamental vibrational fre- quency, v,", to be determined.98 They have also shown that the collisions between electronically excited species and other molecules produce very efficient self- quenching, so that fluorescence is only observed from molecules that have not 93 R.H. Judge and G. W. King, Can. J. Phys., 1975, 53, 1927. 94 D. J. Bedwell and G. Duxbury, XXF Colloque International d'dstrophysique., 1977, p. 434. 95 M. E. Jacox and D. E. Milligan, J. Mol., Spectrosc., 1975, 58, 142. 96 P. H. Turner, L. Halonen, and I. M. Mills, J. Mol. Spectrosc., 1981, 88, 402. 97 B. Fabricant, D. Krieger, and J. S. Muenter, J. Chem. Phys., 1977, 67, 1576. 98 D. J. Clouthier, C. M. L.Kerr, and D. A. Ramsay, Chem. Phys., 1981, 56, 73. 99 D. J. Clouthier and C. M. L. Kerr, Chem. Phys., 1982, 70, 55. loo T. Suzuki, S. Saito, and E. Hirota, J. Chem. Phys., 1983, 79, 1641. lo* R. N. Dixon and C. R. Webster, J. Mol. Spectrosc., 1978, 70, 314. 490 Duxbury undergone collisions since being excited. This contrasts with the behaviour ob- served when the gas phase phosphorescence of the ci3A,-f'A1 is excited.99 In triplet state excitation, vibrational relaxation was found to play a very important role. The frequencies of several vibrational fundamentals of the triplet state were also measured in this work. Some of the singlet-triplet perturbations have also been characterized in a recent 'non-laser' experiment that measured the magnetic rotation spectrum of the A'A,-f'A, system.'02 Finally the laser excitation and microwave-optical double resonance, (MODR), spectra of the 3', of the 2-2 system have been obtained.'" The analysis of these data has given precise values of the rotational constants, centrifugal distortion constants, the spin-spin and spin-rotation interaction constants and for the band origin.It has also allowed the difference frequency, v,' -v6', to be measured. As a result of this recent activity the vibrational and rotational constants of the triplet, a", state are almost as well characterized as for the 2 and 2 states. Since ab initio calculations of the vibration frequencies of H,CS in the ground and excited electronic states have been made,'03,'04 it provides one of the best examples of molecules containing second-row atoms, for which a detailed com- parison of experimental and theoretical data can be made as shown in Table 3.In fact thioformaldehyde can now join its prototype, formaldehyde, as one of the best characterized tetra-atomic molecules. It is not a coincidence that many of the bands in the electronic spectrum fall within the range of the most commonly used dye in dye lasers, Rhodamine 6G! 6 Spectroscopic Studies of Free Radicals and Molecular Ions The development of laser-based methods for the study of free radical spectra has paralleled that for stable and semi-stable molecules. The first methods to be developed were of the fixed frequency type, in particular, laser magnetic resonance (LMR), the laser analogue of gas-phase electron paramagnetic resonance spec- troscopy.Following the development of tunable visible and infrared lasers, many free radicals have been studied via their electronic and vibration-rotation spectra. In this section we will consider the applications of the experimental techniques in approximately the historical order of their widespread use. In view of several recent reviews of this field'05 -we will concentrate on some of the most recent experi- ments on chemically interesting free radicals such as CH,, NH,, CH,, and HCO, and of ions such as H3+ and PH,'. A. Laser Magnetic Resonance Spectroscopy (LMR).-In a series of recent papers the far-infrared (FIR)'09 and the mid-IR"' LMR spectra of the Z3B, state of Io2 D.J. Clouthier, D. C. Moule, D. A. Ramsay, and F. W. Birss, Can. J. Phys., 1982, 60,1212. Io3 R. Jaquet, W. Kutzelnigg, and V. Staemmler, Theor. Chem. Acta., 1980, 54, 205. lo4 J. D. Goddard and D. J. Clouthier, J. Chem. Phys., 1982, 76, 5039. lo5 K. M. Evenson, Ref.83, p. 7. A. R. W. McKellar, Re$ 83, p. 63. lo' E. Hirota, 'Chemical and Biological Applications of Lasers', ed. C. B. Moore, 1980, V, 39. lo8 E. Hirota, J. Phys. Chem., 1983, 87, 3375. 49 1 High Resolution Laser Spectroscopy Table 3 Comparison of observed and calculated vibrational fundamentals of thioformaldehyde with those of formaldehyde, and with those in the A" 'A, and a" 3A, states of thioformaldehyde Frequencylcm -Description H2COa H,CS,X,' A, H,CS,A",'A, H2CS,i,3A2 Symmetric C-H stretch a 2782 297 1 3034 -2782 3057 C 2937 -2962 Symmetric H-C-H bend a 1500 1439 1310 1320 1503 1440 C 1464 -1346 C-S (C-0) stretch a 1746 1059 820 861.6 1709 1058 --C 1053 -792 Out-of-plane bend 1167 990 37 1 356a 1161 1065 --C 1029 -383 Anti-symmetric a 2843 3025 308 1 -C-H stretch 2884 3054 --c 3023 -3078 H-C-H wag a 1249 99 1 799 762.3 1245 989 C 968 -761 * Experimental-H CO-ref.b H,CS D. J. Bedwelt and G. Duxbury, J. Mof. Spectroso., 1980,84, 53. W. B. Olsen and J. W. C. Johns, J. Mol. Specirosc., 1971,39,47. D. J. Clouthier, C. M. L. Kerr, and D. A. Ramsey, Chem. Phys., 1981,56, 73. R. H. Judge and G. W. King, J. Mol. Spectrosc., 1979, 74, 175. Theoretical frequencies from R.Jaquet, W. Kutzelnigg, and V. Staemmler, Theor. Chem. Acta., 1980, 54, 205. Harmonic fre- quencies reduced by 5.15% to give correct weighting for H,CO, for justification for scaling see ref. c. 'Theoretical frequencies from J. D. Goddard and D. J. Clouthier, J. Chem. Phys., 1982, 76, 5039. 10.5% scaling of calculated harmonic frequencies applied. CH, have been obtained, and combined with a large amplitude vibrational model of the ground state."' Previous data on the triplet state were derived from the analysis of the pre-dissociated electronic spectrum,' l2and from the e.s.r. spectrum of matrix isolated CH2.'13 From the pure rotation LMR spectrum in the FIR laser region,Io9 accurate ground-state rotational constants have been obtained, as well as spin-spin, spin-rotation, and hyperfine constants. As an example of the sensitivity and resolution attainable on the FIR and mid-IR regions, Doppler and sub-Doppler spectra produced using both methods are presented in Figure 16.Note that triplet patterns can be seen without the benefit of saturation spectroscopy in the FIR region because of the much smaller Doppler width of the lines. The mid-infrared data on 12CH2and 13CH,' 'O has shown that the bending vibration v2, has a much lo9 T. J. Sears, P. R. Bunker, A. R. W. McKellar, K. M. Evenson, D. A. Jennings, and J. M. Brown, J. Chem. Phys., 1982, 77, 5348. T. J. Sears, P. R. Bunker, and A. R. W. McKellar, J. Chem. Phys., 1982,77, 5363. A. R. W. McKellar and T. J. Sears, Can. J.Phys., 1983, 61, 480. Duxbury . I I I I 1 0 0.05 0.10 0.15 0.20 Magnetic field (TI (a) I I 1 I . 0.2 0.4 0.6 Magnetic field (kG) (b) Figure 16 LMR spectra of CH,, using FIR and mid-IR lasers, showing triplet hyperjne structure. (a) Spectra obtained using the 144.1 pm (69.38765 cm-’) laser line of CD,OH in x and r~ polarizations. The resonances observed are for various Zeeman components of the 413-404 rotational transition. (6) Spectrum observed with the ‘2C’602P(34) laser line at 93 I .001 cm-’ due to the 2,, -I transition. The triplet structure is revealed as a result of the enhanced resolution of saturation spectroscopy (Lamb dips) (Reproduced by permission from J. Chem. Phys., 1982, 77, 5348 and 5363) lower frequency than was previously assumed in modelling the negative ion photo- detachment spectrum of CH,-, casting doubt on the details of the interpretation proposed for-the vibronic structure.’ 10*114 The non-rigid bender Hamiltonian’ ’’ has enabled the equilibrium geometry of the 23B, state to be determined, and a height of the potential barrier to linearity for the v2 vibrational state to be derived.These are compared in Table 4 with analogous parameters recently derived for the ;‘A, and the b”’B, states of CH,. The singlet-triplet separation in CH, has been a source of controversy for some 111 P. Jensen, P. R. Bunker, and A. R. Hoy, J. Chem. Phys., 1982, 77, 5370. 112 G. Herzberg, Proc. R. SOC.,London, Ser. A, 1961, 262, 291.113 R. A. Bernheim, H. W. Bernard, P. S. Wang, L. S. Wood, and P. S. Skell, J. Chem. Phys., 1970, 53, 1280. 493 High Resolution Laser Spectroscopy Table 4 Geometrical parameters qf the electronic stutes of CH, Barrier height/cm 1877.6 a' 945 I I193 2032 Equilibrium bond angle/" 133.88 101.7 141 133.4 Equilibrium bond length/8, 1.075 a 1.1 14 1.083 1.085 ' Non-rigid bender-P. Jensen, P. R. Bunker, and A. R. Hoy, J. Chem. Phys., 1982,77, 5370. Semi-rigidbender-T. J. Sears, P. R. Bunker, and A. R. W. McKellar, J. Chem. Phys., 1982, 77, 5363. Semi-rigidbender and Renner-Teller coupling-G. Duxbury, J. Chem. SOC.,Furuduy Truns. 2, 1982, 78, 1433. 'Singlet-triplet splitting d-X = 2994 30 cm-', single-singlet splitting 6-fi = 8258 f100 cm-' -A. R.W. McKellar, P. R. Bunker, T. J. Sears, K. M. Evenson, R. J. Saykally, and S. R. Langhoff, J. Chem. Phys., 1983, in the press. time, and even more so recently since the value deduced in' l4 of ca. 0.85eV is approximately twice the currently accepted value from other measurements and calculations. Perturbation allowed transitions, which are due to the interaction between excited vibrational levels of the triplet state and the ground vibrational state of the lowest electronic singlet, 6'A, state, have been observed in the FIR LMR spectra of the Z state of CH,.l15 The detailed analysis of these transitions has shown that the singlet-triplet splitting derived from the photodetachment spectra is far too large, and has given the first unambiguous and accurate value for the singlet-triplet splitting.This is found to be 3 165 t20 cm-for the zero-point energy difference, and 2994 f30 cm-' for the separation between the minima of the potential wells.'15 This has led to a reappraisal of the analysis of the CH2- spectra from which it has been concluded that the effective temperature of the negative ions is far higher than had been assumed in modelling the envelope of the p ho t odetatc hmen t spectrum. An interesting observation in the high field LMR spectra of some free radicals has been the presence of strong resonant signals whose positions do not vary in field strength when the laser frequency is detuned. These signals have subsequently been identified as anti-crossing signals and have been observed in the spectra of several radicals including C10212and DOz' 16.B. Tunable Infrared Laser Spectroscopy of Free Radicals and Ions.-In many of the free radical experiments recently reported the radicals have either been generated by atom reactions or by DC or RF discharges within the absorption cells. Since the method does not depend upon Stark or Zeeman tuning for anything other than 114 P. C. Engelking, R. C. Corderman, J. J. Wendolski, G. B. Ellison, S. V. O'Neil, and W. C. Lineberger, J. Chem. Phys., 1981, 74, 5460. 115 A. R. W. McKellar, P. R. Bunker, T. J. Sears, K. M. Evenson, R. J. Saykally, and S. R. Langhoff, J. Chem. Phys., 1983, in the press. T. J. Sears and P. R. Bunker, J. Chem. Phys., 1983, in the press.'16 H. Uehara, J. Chem. Phys., 1982, 77, 3314. Duxbury modulation, metal-free cells can be used to prevent radical recombination, and pressures of ca. 1 torr can be used as there is no risk of electrical breakdown. For paramagnetic species such as BO, two methods of detection have been used, Zeeman modulation and source modulation. Zeeman modulation is accomplished by winding a Zeeman coil around the glass pipe of the absorption cell. With Zeeman modulation only paramagnetic species are detected and low J lines are preferentially enhanced, since the modulation efficiency is inversely proportional to J. With source modulation, on the other hand, all species are detected. In systems where there is no ambiguity in assignment source modulation allows much higher J lines to be detected.As an example of this, spectra of BO, '" using both types of modulation are shown in Figure 17. In addition to BO,, which is very difficult to detect in the microwave region since it has no permanent electric dipole moment, several other interesting free radicals and ions have been detected using diode, F-centre, or difference frequency spectrometers. The first ion to be detected in this way was H, + by Oka. l8 The spectrum of this ion was finally identified after a long search in a high-current low-temperature discharge in hydrogen. The geometry and other molecular parameters of H,+ are close to those predicted by ab initio calculation and serve as a severe test of theoretical methods.One electronic spectrum which happens to occur in the infrared region, that of the 2C--217 transition of C,H, was detected at 3772cm-' using an F centre laser."' It is possible that transitions between other close lying electronic states of polyatomic molecules will be detected using tunable infrared lasers. One molecule of particular interest that has been studied recently is CH,. Spectra have been obtained of transitions involving the bending vibration, v,, using diode lasers,'20.'21 and of the v, region using a difference frequency spec- trometer.' 22The study of the v, vibrational ladder has shown conclusively that the equilibrium geometry of the CH, radical is planar in the gas phase and has allowed a potential function for the bending well to be developed.The v, study has characterized the asymmetric stretching vibrational band, and together with the v2 data, yields a good approximation to the equilibrium structure of the radical. The v, region lies in an atmospheric window, and hence these transitions may be useful for searching for lines of interstellar CH,. The most recent measurements on CH, are of the transition dipole moment of the v, band.',, This transition moment is very large, 0.28 D, and comparable to that in NH,. Knowledge of this quantity has allowed reaction kinetics of the CH, radical to be carried out in the gas phase.'24 11' K. Kawaguchi, E. Hirota, and C. Yamada, Mof. Phys., 1981, 44, 509. 11* T. Oka, Phys. Rev. Lett., 1980, 45, 531. 119 P. G. Carrick, J.Pfeffer, R. F. Curl, E. Koestre, F. K. Tittel, and J. V. V. Kasper, J. Chem. Phys., 1982,76, 3336. 120 C. Yamada, E. Hirota, and K. Kawaguchi, J. Chem. Phys., 1981, 75, 5256. Iz1 E. Hirota and C. Yamada, J. Mol. Spectrosc., 1982, %, 175. lz2 T. Amano, P. F. Bernath, C. Yamada, Y. Endo, and E. Hirota, J. Chem. Phys, 1982, 77, 5284. lZ3 C. Yamada and E. Hirota, J. Chem. Phys., 1983, 78, 669. lz4 G. A. Laguna and S. L. Baughcum, Chem. Phys. Lett., 1982, 88, 568. High Resolution Laser Spectroscopy 6.5 5.5 4.5 3.5 2.5 1.5 1 1 I 1 I 1 11 I I 1 1 1 3 2.5 3.5 4.5 5.5 6.5 7.5 8.5 1 I I I 1279.1 1278.8 cm-1 (b) Figure 17 (a) Q-branch transitions of the v3 band of "B02 in the 'l7+ state, obtained by Zeeman modulation. (b) Analogous transitions in ''B02 obtained by frequency modulation of the diode (Reproduced by permission from Mol.Phys., 1981, 44, 509) Duxbury This method has greater potential than the previous gas phase method for de- tecting methyl which relies on measurement of the vacuum ultraviolet absorption spectrum, 25 particularly since this spectrum is partly predissociated. C.Velocity Modulated Spectra of Ions.-In a DC glow discharge the ions possess a net drift velocity, ud, as a result of their mobility in the axial electric field of the discharge. The Doppler shift, Av, is then related to the drift velocity by: Avlv = vd/c where c is the velocity of light. For typical discharge conditions dv can be ca. 90 MHz at v = 2500cm-',i.e.comparable to the Doppler broadened line- width. The resultant Doppler shift can be positive or negative, depending on whether ud is in the same or the opposite sense as the laser beam direction. This Doppler shift has been used in two ways, as a means of differentiating the spectra of ions from those of neutral species present in the discharge, and as a way of studying ion transport processes in detail. Selective detection of ion spectra has been obtained by modulating the drift velocity of ions using an AC discharge.126 When the polarity of a glow discharge is reversed at several kilohertz the ion absorption frequencies are alternatively red- and blue-shifted at this rate. Detecting at the field modulation frequency then yields the infrared spectrum of the ionic species.This method has been used with an F-centre laser to detect R-branch absorption transitions of the v1 vibration-rotation band of HCO', and has given a very good signal to noise ratio. In the second type of e~periment'~~ a frequency modulated diode laser beam was split into two parts, which were then sent in opposite directions through a discharge in a mixture of He, Ar, and H,. This allowed the Doppler shifts of absorption lines of ArH', formed in the discharge, to be measured as a function of the pressure and composition of the discharge. Preliminary results obtained suggest a dependence of the ion mobility on vibrational excitation, with u = 1 ions being more mobile than those with u = 0. This is one of the first demonstrable effects of quantum processes in ion transport.D. Optogalvanic Spectroscopy.-The mechanism of the optogalvanic effect is still the subject of some controversy. One proposal is that the main mechanism in the visible and U.V. regions is the difference in the collisional ionization rates of the two states. However, visible signals observed using I,128 are very similar to those recently recorded in the infrared, using diode lasers,47 where the ionization mech- anism is inappropriate. It is therefore proposed that an alternative scheme operates in which the main effect of the laser radiation is to pump energy into the electron gas, hence causing an increase in the kinetic energy of the species. 125 G. Herzberg, 'Molecular Spectra and Molecular Structure Vol. 111: Electronic Spectra and Structure of Polyatomic Molecules', Van Nostrand-Reinhold, New York, 1966.lz6 C. S. Gudeman, M. E. Begemann, J. Pfaff, and R. J. Saykally, Phys. Rev. Left., 1983, 50, 727. lz7 N. Haese, F. S. Pan, and T. Oka, Phys. Rev. Lett., 1983, 50, 1575. lZ8 C. T. Rettner, C. R. Webster, and R. N. Zare, J. Phys. Chem., 1981, 85, 1105, High Resolution Laser Spectroscopy Although in general laser powers of greater than 100 milliwatts have been used to obtain LOG signals, it has recently shown that good signals in the infrared spectra of NH3 and NO, could be obtained using diode lasers with a power level of ca. 1 milli~att.~~ A recent of the predissociation of HCO exemplifies the utility of the LOG approach for studying species produced in electric discharges.The HCO J2A"-T2A' spectrum is weak and predissociated. By using the optogalvanic effect in an RF discharge in the parent species, acetaldehyde, mixed with an argon-helium carrier gas, it has been possible to measure the linewidths in the 0,9', 0-4,01, 0 band using a single mode dye laser with a sensitivity of lo5 over that obtained in conventional flash photolysis experiments. The line widths show a strong dependence on the overall rotational quantum number N, increasing by a factor of 4 when N increases from 4to 16. It was also shown that there is an N-independent contribution. By modelling the behaviour of the line width, it has been shown that the main mechanisms for the predissociation are coupling to dissociative A levels through K-type resonance and perturbations with n levels through Coriolis interaction.The remaining contribution to the broadening is thought to be associated with a homogeneous, spin-orbit, coupling between the excited-state levels and high-lying levels of the ground state. E.Ion Beam Spectroscopy.-Two recent sets of experiments that show the utility of the technique will be cited. The first set used visible lasers to investigate the electronic spectra of H2Sf,13' of NH+, PH+, and SH+I3' and of PH2+13, bY monitoring the fragment ions arising from the predissociated excited electronic state. Since tunable dye lasers were used the spectra could be recorded either by sweeping the velocity of the ion beam, as described previously, or in a broad-band low-resolution scan by sweeping the dye laser frequency.In the case of H2S+ this has shown that the complex vibronic pattern seen in the a2A,-f2B, emission spectrum133 is even more complex than previously supposed, when the greatly enhanced resolution of the ion beam techniques is employed. In the spectra of NH', PH+, and SH+ some interesting examples of proton hyperfine interaction splittings were seen in both electronic states. PH2+ possibly possesses the most interesting spectrum of this group of ions since it is the first example of the high-resolution gas-phase spectrum of this ion. Although this ion is isoelectronic with SiH,, the spectra obtained when monitoring the P+ dissociation fragment bears little resemblance to the absorption spectrum of SiH, obtained by Dub- ois.134.1 35 One possibility is that the ion is produced vibrationally hot, so that 129 R.Vasudev and R. N. Zare, J. Chem. Phys., 1982, 76, 5267. I3O C. P. Edwards, C. S. Maclean, and P. J. Sarre, Chem. Phys. Lett., 1982, 87, 11. 131 C. P. Edwards, C. S. Maclean, and P. J. Sarre, J. Chem. Phys., 1982, 76, 3829. 13* C. P. Edwards, P. A. Jackson, and P. J. Sarre, Paper C18 presented at the 8th Colloquium on High Resolution Molecular Spectroscopy, Tours, 1983. 133 G. Duxbury, Ch. Jungen, and J. Rostas, Mof. Phys, 1983, 48, 719. 134 I. Dubois, Can. J. Phys., 1968, 46, 2485. 135 I. Dubois, G. Duxbury, and R. N. Dixon, J. Chem. Soc., Furuduy Trans. 2, 1975, 71, 799.136 E. Hirota, personal communication. Duxbury sequence band structure dominates the spectrum. Another reason for the difference may lie in singlet-triplet perturbations for which there is some evidence in the A"'Bl-T'A fluorescence spectrum of SiH,. Hirota and his colleag~es'~~ have noted that in the laser induced fluorescence spectrum of SiH, the strongest lines are absent from the absorption spectrum recorded in the same region, and it is the weaker lines that correspond to the lines seen in the absorption spectrum. The strong lines may therefore be evidence for singlet-triplet perturbations, which are known to be important in other systems such as H,CS.'02 The spectrum obtained when detecting the PH' fragment must represent the transitions to a state which undergoes asymmetric dissociation. It does not resemble that obtained when using P+ detection and probably represents a transition which does not involve the ground electronic state of the ion.The fixed frequency method originally used in ion beam studiesJo has recently been extended. Infrared multiphoton dissociation has been used to look at some of the transitions of HD+ near the dissociation limit. Infrared spectra of CH''37 and of H3+138 have also been obtained near their dissociation limits by using photodissociation induced by a CO, laser. In the case of CH+,'37 where the low lying energy levels of states have previously been investigated in some detail, some progress has been made in understanding the behaviour in this region which lies within 1000cm-of the upper dissociation limit of CH' --t C'(2P32)+ H H3+138on the other hand is a much more complex system with a very erratic many line spectrum.Some recent progress has been made by showing that, if the resolu- tion is deliberately degraded by bunching the lines into 'clumps', a regularity which can be associated with rotation of an H, unit loosely bonded to H+ is evident.lJ9 F. NH,-A Spectroscopic Case History.-The free radical NH, has been the subject of extensive high resolution spectroscopic analyses, particularly by means of its A"2A,-f2B, electronic spectrum in the visible region. It is of interest on several counts: as a reactive intermediate, as a constituent of cometary spectra, and as possibly the best known and characterized example of the Renner-Teller effect, the breakdown of the Born-Oppenheimer separation in a triatomic molecule, which has been the subject of several recent theoretical paper^.'^^.^^' Although the original work on NH, used the flash photolysis method to gener- ate the free radical,'42 more recent laser based experiments have used continuous flow systems: either atom reactions, or, more recently, DC electric discharges in NH,.The use of lasers has enabled pure rotation spectra to be observed using 137 A. Carrington, J. Buttenshaw, R. A. Kennedy, and T. P. Softley, Mol. Phys., 1982, 45, 747. 13* A. Carrington, J. Buttenshaw, and R. A. Kennedy, Mol. Phys., 1982, 45, 753. lJ9 R. A. Kennedy, personal communication. 140 Ch.Jungen, K. E. J. Hallin, and A. J. Merer, Mol. Phys., 1980, 40, 25. 141 Ch. Jungen, K. E. J. Hallin, and A. J. Merer, Mol. Phys., 1980, 40, 65. 142 K. Dressler and D. A. Ramsay, Philos. Trans. R. SOC.London, Ser. A., 1959, 251, 553. 499 High Resolution Laser Spectroscopy LMR.'43 vibration rotation spectra of the v, band by LMR,144 and infrared spectra of the v1 and vj bands by difference laser spectr~scopy.'~~ These far and mid-infrared data have led to far more accurate rotational and vibrational con- stants for the electronic ground state than are available from electronic spectra alone. In particular v3 is a non-totally symmetric vibration, and hence it can be observed in the electronic spectrum only by its perturbing effect on vl.As well as providing additional data on the lowest vibrational levels of the electronic ground state of NH,, laser induced fluorescence 14' ha ve provided data on the high vibrational levels of the 'B, state that have aided the construction of accurate potential energy curves used in the Renner-Teller calcu- lation~.~~~When laser induced fluorescence is used in conjunction with sub- Doppler methods, a wealth of detail about the fine and hyperfine interactions in NH, is revealed. Most of the techniques in the sub-Doppler armoury have been employed to study the details of this extremely perturbed spectrum. The methods used included saturation spectroscopy using intermodulated microwave-optical double resonance,14' hyperfine level crossing, 50 optical-optical double resonance using fluorescence dete~tion,'~~and infrared-optical double resonance.'52 The OODR15' experiments of NH, probe the O,, level of the A (090) vibronic state. Since this is the lowest possible level with N = 0 and has a value of K, = 0, it is expected to represent an 'isolated' upper state level of NH,, unaffected by Renner-Teller perturbations. This has enabled values of the nitrogen hyperfine coupling constant, a,, and the ratio of the gJ value to that of the free spin, g,, to be measured. These then provide a yardstick for comparison with the values of a, of perturbed levels, and with the g, values of levels with K, >, 1, some of which have a large admixture of orbital angular momentum.The infrared-optical double resonance experiment' 52 is a hybrid of two powerful techniques, LMR and laser induced fluorescence. The fluorescence cell is placed either within or outside the cavity of a CO,/N,O infrared laser, and the visible laser beam introduced into the cell through the end window of the CO, laser in the intracavity system. The magnetic field is used to tune the molecular energy levels into resonance with the IR laser, and the effect of saturation of the IR transition is detected by the change in fluorescence of the visible emission. In the intracavity apparatus two sets of signals are observed, one set resulting from the visible and 143 P.B. Davies, D. K. Russell, B. A. Thrush, and H. E. Radford, Proc. R. Soc. London, Ser A., 1977, 377, 299.144 K. Kawaguchi, C. Yamada, E. Hirota, J. M. Brown, J. Buttenshaw, C. R. Parent, and T. J. Sears, J. Mol. Spectrosc., 1979, 81, 60. 14s T. Amano, P. F. Bernath, and A. R. W. McKellar, J. Mol. Spectrosc., 1982, 94, 100. 146 M. Kroll, J. Chem. Phys., 1975, 63, 319. 14' M. Vervloet, M. F. Merienne-Lafore, and D. A. Ramsay, Chem. Phys. Lett., 1978, 57, 5. 148 G. W. Hills, D. L. Philen, R. F. Curl, and F. K. Tittel, Chem. Phys., 1976, 12, 107. lQ9 G. W. Hills, C. R. Brazier, J. M. Brown, J. M. Cook, and R. F. Curl, J. Chem. Phys., 1982,76,240. lS0 R. N. Dixon and D. Field, Mol. Phys, 1977, 34,1563. R. N. Dixon, Ref: 44, p. 1563. Is* T. Amano, K. Kawaguchi, M. Kakimoto, S. Saito, and E. Hirota, J. Chem. Phys., 1982, 77, 159.Duxbury IR beam propagating in opposite directions, and the other set from the two beams having the same propagation direction. Interesting effects of the very different Doppler widths in the infrared and visible region manifested themselves in the very different linewidth and intensity behaviour when either the IR or the visible laser was scanned, This technique offers the chance of obtaining laser magnetic reso- nance data on excited electronic states and offers a route to the determination of parameters such as those governing the magnetic quenching of fluorescence. A final use of high resolution laser induced fluorescence of NH, has been the determination of the electric dipole moment in the ground electronic state. 53 This has been accomplished by measuring the Stark effect in the asymmetric molecule NHD, which possesses an a-component of the dipole moment yielding comparably large Stark shifts.The dipole moment determined in this way is 1.82 & 0.05 D, which is very similar to that of water. Because of the large size of p, interstellar rotational emission lines should be strong provided that concentrations of NH, or NHD are not prohibitively small, and a search for such spectra is currently in progress. 7 Ultra High Resolution Spectroscopy Spectra obtained using noble gas and CO, waveguide laser based spectrometers are among some of the highest resolution spectra ever obtained. In order to achieve such high resolution very large diameter cells of up to 38cm have been used, and in order to use low pressures, cell lengths of up to 13 m are necessary.Using cells of this type wavefronts approaching those of a plane wave can be achieved. In one of the pioneering experiments by Hall, Borde, and Uehara154 using a HeNe laser, the recoil splitting of hyperfine components in CH, was observed. When a mole- cule is excited by the absorption of a photon it absorbs the momentum of the photon as well as its energy. Because of momentum conservation, molecules interacting with a monchromatic field of a given direction have to belong to two slightly different momentum classes in the upper and lower levels. The momentum change is equal to that of the light quantum momentum: h2v2+.=(where the sign depends on the direction of the beam with respect to the z axis) The return beam of the saturated absorption spectrometer thus sees two velocities, frequencies of anomalously high transmission.One peak occurs in the usual way by interrogation of the population ‘hole’ in the absorbing lower state. The other peak corresponds to spectrally narrow amplification of the return beam by the velocity-selected and displaced molecules, placed into the excited state by inter- action with the other laser beam. The resultant spectrum is shown in Figure 18. Following the observation of these effects in the 3 pm region, ultra-stable wave- guide CO, lasers have been constructed that have enabled similar resolution to be attained in the 10pm region. Very high resolution spectra of the v3 bands of SF, 153 J.M. Brown, S. W. Chalkley, and F. D. Wayne, Mol. Phys., 1979, 38,152. lS4 J. L. Hall, Ch. J. Borde, and K. Uehara, Phys. Rev.Leu., 1976, 37, 1339. 50 1 High Resolution Laser Spectroscopy -30 -20 -10 0 10 20 30 Kilohertz detuning from reference laser (b) Figure 18 (a) Saturation peak in the output of a He-Ne laser seen in the original intracavity 'inverted' Lamb dip experiment on CH,. The laser frequency was swept twice over the gain profile. (6) Ultra-high resolution derivative spectrum of the three main hyperfine lines of I2CH, showing the recoil doubling (lower curve). Methane pressure 70 p torr; room temperature; modulation of 800 Hz peak to peak. A least-squares fit (solid line) gives a width of 1.27 kHz HWHM and a recoil doublet splitting of 2.150 kHz, the high frequency peak 1YOlarger than the low frequency ones.The upper curve, integrated from a sample of such data, shows that each hyperfine component is spectrally doubled by the recoil efect (Reproduced by permission from Phys. Rev. Lett., 1969, 22, 4; 1976, 37, 1339) Duxbury E0 1 I I -507 0 5 07 Kilohertz detuning (a1 I I I * -33 0 33 Kilohertz detuning (b) Figure 19 (a) Ultra-high resolutiok spectrum of the SF,&, cluster at 28.412582452 THz recorded using a frequency controlled saturation spectrometer based on a waveguide CO, laser. (b) Magnetic hyperfine structure of the R,, A: line of 32SF, at 28.46469125 THz. The upper trace is the observed spectrum, and the lower trace calculated using the scalar spin-rotation interaction W RS = -hc,IJ, with c, = -5 kHz.The resolution is ca. 5 kHz. (Reproduced by permission from 'Laser Spectroscopy IV', ed. H. Walther and K. W. Rothe, Springer Verlag, 1979, 142) High Resolution Laser Spectroscopy and Os, have been observed,' 5,1 56 including signals due to nuclear hyperfine interactions. These result from vibrationally and rotationally produced distortions of the tetrahedral or octahedral geometry.'57 This mechanism is similar to that responsible for the generation of rotationally induced dipole moments. Examples of the spectra of SF, observed in this way are reproduced in Figure 19, where the half-width at half maximum of 5 kHz corresponds to a resolution of about 1 part in 10".Progress in frequency-locking ion and dye lasers has allowed similar spectra to be obtained in the visible region. '58 Particular advances have included phase locking of lasers to stable cavities, '59 frequency-modulation spectroscopy,' 6o and 'Ramsay fringe' 7 Conclusion In this article I have endeavoured to present a picture of the impact of lasers on high resolution gas phase molecular spectroscopy by concentrating on what I regard as certain key areas. This approach, by its very nature, must lead to several topics being either omitted, or treated in a perfunctory fashion. In depth treatment of much of the subject matter has been given in the books of Hollas'62 and Demtroder,2 and in the excellent review articles of Macpherson and Bar-row.' 63*164 In these reviews an almost complete bibliography of experimental work in gas phase molecular spectroscopy from 1978 up to the beginning of 1982 has been given.Acknowledgements. It is a pleasure to acknowledge the debt I owe to Richard Dixon and Takeshi Oka for stimulating my interest in this field. I would particu- larly like to acknowledge my collaborators Gareth Jones, Joelle Rostas, Marcel Horani, Christian Jungen, Sam Freund, John Johns, Gordon Caldow, Mike Ashfold, Henryk Herman, David Bedwell, Michel Le Lerre, Hiroshi Kato, Jan Petersen, June McCombie, and David Devoy, without whose enthusiasm and drive little would have been accomplished. 55 Ch. J. Borde, M. Ouhayoun, A. Van Lerberghe, C. Salomon, G. Avillier, C. D. Cantrell, and J.Borde, in 'Laser Spectroscopy IV', ed. H. Walther and K. H. Rothe, Springer Verlag, 1979, p. 142. lS6 J. Borde, Ch. J. Borde, C. Salomon, A. Van Lerberghe, M. Ouhayoun, and C. D. Cantrell, Phys. Rev. Lert., 1980, 45, 14. lS7 J. T. Hougen and T. Oka, J. Chem. Phys., 1981, 74, 1830. lS8 J. Helmcke, S. A. Lee, and J. L. Hall, Applied Optics, 1982, 21, 1686. lS9 J. L. Hall, L. Hollberg, M. Long-sheng, T. Baer, and H. G. Robinson, J. de Phys., 1981, 12, c8. 160 J. L. Hall, L. Hollberg, T. Baer, and H. G. Robinson, Appl. Phys. Left., 1981, 39,680. 161 S. A. Lee, J. Helmcke, and J. L. Hall, Ref 155, p. 130. 162 J. M. Hollas, 'High Resolution Molecular Spectroscopy', Butterworths, London, 1982. M. T. Macpherson and R. F. Barrow, Annu. Rep. Prog. Chem., Sect. C, 1979, 76, 51. 164 M. T. Macpherson and R. F. Barrow, Annu. Rep. Ptog. Chem., Sect. C, 1981, 78, 221.
ISSN:0306-0012
DOI:10.1039/CS9831200453
出版商:RSC
年代:1983
数据来源: RSC
|
|