摘要:

J.C.S. Faraday I, 1980,76, 196-208Solution of Hydrogen in 1\Jon-stoichiornetric Pd& --x CompoundsBY TED B. FLANAGAN,* GEORGE E. BIEHL AND JOHN D. CLEWLEYDepartment of Chemistry, University of Vermont,Burlington, Vermont 05404, U.S.A.ANDSTIG KUNDQVIST AND YVONNE ANDERSONInstitute of Chemistry, University of Uppsala, S-75 121 Uppsala, SwedenReceived 20th March, 1979The solution of H2(D2) in a series of Pd3P1-x compounds has been investigated for x between0 and 0.20, H2 solubility is significant for compounds with a large degree of non-stoichiometry andnegligible for the stoichiometric compound. A lattice parameter anomaly occurs at x = 0.08 forthe H-free phosphides ; the crystallographic axes expand differently with increase of x above andbelow the anomaly.HZ solubility falls to very small values for x values corresponding tothose above the anomaly, i.e., with x less than 0.08. The maximum solubility observed isH/Pd3P0.s0 = 0.15 (296 K, 93.3 kPa); there is no indication of hydride phase formation. Thelattice expansions accompanying H solution have been measured and these are anisotropic, i.e., theb-axis expands less than half as much as the Q and c-axes for all of the Pd3P1-x compounds. Thevolume change per H is smaller than usually observed for solid-H systems.With the aid of information on H positions derived from a recent neutron diffraction study ondeuterated Pd3P0.807 a model has been proposed which describes the pressure-H content data over awide range of H contents. This model is based on H occupation of interstices in two of the quadri-lateral faces of trigonal prisms not occupied by central P atoms and the assumption that occupationof one interstice in the prism by H blocks the other interstice in the same prism for H occupation.Therelative partial enthalpy (standard) for Hz solution in Pd3PI-x with x = 0.20 is - 16.3 kJ(mo1 €€)-Iand the relative (to H2 at 1 atm) partial entropy (standard) is - 48.7 J(mo1 H)-I K-l compared to cor-responding values of - 9.8 kJ(mo1 H)-l and - 56.5 J(m01 H)-l K-l for Pd-H.Palladium-H 1. has been the most studied of all metal-H systems since its dis-covery in 1866 by Thomas Graham.3 Little is known, however, about the solutionof H2 into Pd when the metal is in chemical combination with non-metal elements.A variety of Pd phosphides have been prepared and characterized by X-ray diffraction.If significant amounts of H2 were to dissolve into these phosphides it would be ofconsiderable interest because, for example, the Pd to P ratio varies in these phos-phides and the effects of different ratios on H2 solubility could be determined.Franzen et aL4 have determined the solubility of H2 in Ta,S and Nb21S8 up to H2pressures of 50.6 kPa at temperatures 2 573 K.From presswe-composition-temperature, p-c-T, relationships, partial thermodynamic parameters for H desorp-tion were determined; these values were not very different from those of the parentmetals. Hence they concluded that the thermodynamics of H2 solution (or desorp-tion) were not very dependent upon the structure of the solid but were controlledmainly by metal-H bonds which are independent of structure.n et aL4 were derived fromthe temperature dependence of the slopes of plots of p$ against mol fraction Hdissolved in the solid.For small H contents such as employed by Franzen et aL4 molThe thermodynamic parameters determined by Frar19FLANAGAN, BIEHL, CLEWLEY, RUNDQVIST A N D ANDERSSON 197fraction H is nearly the same as the (H-to-mol solid) ratio. Their derived values werestated to correspond to the process H(dissolved, standard state) -+ 1 H,(g, 1 atm).Flanagan and Oates have pointed out that the derivation of thermodynamic para-meters from plots of pkt against n = (mol H/mol solid) or XH = (mol fraction H)can lead to spurious values for ASZ because it is implicitly assumed that thestructurally-limiting value of n (appropriate for the fugacity range employed) is unity.The standard state for H in interstitial solutions generally is accepted to be theinfinitely dilute state without the contribution of the partial configurational entropy 6*or equivalently, the solute species in the ideal state at half of the interstices occupied.Therefore the partial entropies derived by Franzen et aL4 from their Sieverts’ lawplots cannot be considered to be standard values unless fortuitously there is one inter-stice to be occupied per in01 of Ta,S and Nb21S8, which seems unlikely.Generallythen, close agreement between relative partial molar entropies derived from Sieverts’law plots, where corrections have not been made for the number of interstices avail-able per mol of solid, and relative partial standard entropies for the pure parent metalsis not of significance.The maximum solubilities observed by Franzen et aL4, expressed as H/mol solid,were xO.1 (50.6 kPa, 573 K).There was no evidence of plateau pressures whichwould suggest the formation of hydride phases. For V,C and Nb2C Franzen et al. *reached similar conclusions about the structure-insensitivity of H2 solution. Onepurpose of the present study was to determine if their findings are of general validitywith regard to the similarity of the partial enthalpy of H2 solution in complex struc-tures of the metal with non-metal elements and in the pure parent metals.have determined the structure of Pd3P to be of thecementite (Fe,C) type.The structure can be described in terms of triangular pris-matic building-blocks, with six Pd atoms at the corners of the prism and a P atom atthe centre. The prisms are packed in such a way that there is a Pd atom outside eachof the three quadrilateral faces of the prisms. For x > 0, P atoms are randomly missingfrom the prisms in the Pd3P1--x compounds. This non-stoichiometry affords a usefulhandle for H2 solubility studies since the role of empty prisms m y be assessed.Most importantly some of the present authors lo have recently carried out neutrondiffraction studies on deuterated Pd3P0.80. The D atoms occupy interstices near themidpoints of two of the quadrilateral faces of the empty triangular prisms, i.e., thoseprisms where the central P atom is missing.The D atoms are thus coordinated byfive near neighbour Pd atoms in a distorted square pyramidal configuration. Squarepyramidal interstices are indeed unusual for H occupation and it should be pointed outthat there are also four octahedral and twelve tetrahedral (distorted) interstices perunit cell of Pd3P and, of course, for non-stoichiometric Pd3P1-x compounds thecentral prismatic interstices, where P atoms have vacated. In compounds of metalsand non-metals such as Pd3P there are a variety of possible interstices which I3 atomscould, in principle, occupy and at this early stage of research on such structurallycomplex systems it seems that the only reliable way of determining which of thevarious types of interstices are occupied is by neutron diffraction studies.Rundqvist and GullmanEXPERIMENTALPREPARATION AND CHARACTERIZATION OF THE Pd3PlPx COMPOUNDSThe Pd3PlWx compounds were prepared from Pd (Johnson, Matthey and Co., 3N5) andred P(Koch-Light, 5N).The elements were reacted in silica tubes after prior evacuation at990 K for 24 h. The resulting phosphides were crushed into powder and annealed for 4-6days at 1020-1040 K. This sequence of crushing, followed by annealing, was repeated up t198four times for each of the phosphides in order to obtain completely homogeneous, singlephase samples. Between each annealing step, the powder samples were stress-relieved byheating at 920-950K for 30min and then examined by X-ray diffraction using a PhilipsXDC-700 camera and CuKal radiation.The final samples of the Pd3Pl-, compounds werevirtually free from other Pd-P phdses judging from their powder patterns. The unit celldimensions of the soichiometric compound Pd3P were a = 5.983, b = 7.442 and c = 5.166 Awhich corresponds closely to the values reported earlier.' 'SOLUTION OF HYDROGEN I N Pd3P1--xX-RAY MEASUREMENTSMeasurements of the lattice parameters of the unit cells of H-containing P ~ ~ P I - ~ com-pounds were made with a focusing Guinier-type powder camera. l2 Stress-relieved powdersamples were mounted in a vacuum chamber which had an inlet fitted for H2. CuKairadiation was passed through beryllium windows for the diffraction experiments.Zone-refined silicon was mixed with each phosphide in order to provide an internal calibrationfao(%) = 5.431 065 A 13]. Diffraction patterns were recorded at room temperature (296 K)both before and after equilibration with Hz (93.3 Wa). Separate experiments indicatedthat equilibrium H contents should have been achieved during the times allotted for equilibra-tion with the possible exception of the phosphide with x = 0.06. In addition, as a furthercheck on the attainment of equilibrium, lattice parameters were remeasured after 24hfurther equilibration with no changes observed. It is quite certain, therefore, that equi-librium H contents (93.3 kPa, 296 K) were obtained for all phosphides with x > 0.06. Thein situ technique employed was advantageous because relative lattice expansions due tointerstitial H could be determined without removal of the sample and therefore the data5-0 7.5557 7505.6 7.455.155.105.0500 0.1 0.2 03x in Pd3Pl-xData taken from ref.(11).FIG. 1 .-Changes of crystallographic axes with P content (298 K). A, a-axis ; CI,b-axis ; 0, c-axisFLANAGAN, BIEHL, CLEWLEY, RUNDQVIST AND ANDERSSON 199should be free from systematic errors and the reported standard deviations should closeiyreflect the true errors. The data were evaluated using a computer program written byErsson l4 which includes least square refinements of the unit cell dimensions.H2 SOLUBILITY MEASUREMENTSEquilibrium H2 solubilities were determined in several vacuum systems which werecapable of attaining ultra-high vacuum and these were naturally free from mercury or greasecontamination.The samples were stressrelieved as described above but were otherwiseuntreated following their preparation. They were sealed into the vacuum systems in smallPyrex tubes and evacuated to w 1 x Pa at 373 K. Ultra-pure H2(D2) was employedfor solubility determinations and detailed p-c-T data were determined volumetrically withthe aid of a series of electronic diaphragm gauges (M.K.S.). These gauges were connectedto a strip chart recorder so that equilibrium could be unambiguously established.RESULTS AND DISCUSSIONLATTICE PARAMETERS OF H-FREE Pd3P1--x COMPOUNDSThe lattice parameters for Pd3P1--x (€3-free) compounds are shown in fig. 1 [fromref.(ll)]. Pnteresthgly the a- and c-axes expand with added P (decrease of x) whilstthe b-axis contracts to x = 0.08. For values of x from 0.08 to 0, the c-axis alsocontracts with added P whilst the a-axis expands at a greater rate than below theanomaly. The unit cell volume is shown in fig. 2 as a function of x and, despite theanomaly in the crystallographic axes expansions, the unit cell volume is only affectedi' I10.610.410.210.0L I 1 10.0 . 0.1 02 0.3x in PdsP1--xHX 103A V X lo3/ Y PdsP1--x75 1550 1025 50 0FIG. 2.-n, H-free unit cell size [ref. (ll)] ; 0, H-free density [ref. (ll)] ; 0 , H/Pd3P1-, at 133.3 Pa(1 torr) and 323 K : A, relative unit cell expansion A V/V corresponding to equilibrium H contentobtained at 93.3 kPa (296 K)200to a minor extent.The partial molar volume of P in Pd3P1-X can be calculated fromthese data as 7.3 xSOLUTION OF HYDROGEN I N Pd3PlWxm3 (mol P)-lEXPANSION OF Pd,P1-, COMPOUNDS INDUCED BY INTERSTITIAL HFig. 2 and table 1 show the expansion of the three crystallographic axes and theunit cell volumes with H (93.3 kPa, 296 K) for the non-stoichiometric phosphides as afunction of x. Fig. 2 shows that the volume expansion of the unit cells correspondingto the equilibrium H content obtained at 93.3 kPa, 296 K decreases in a nearly linearway with decrease of x to reach essentially zero at x = 0.08.TABLE 1 .--RELATIVE CHANGES OF CRYSTALLOGRAPHIC AXES AND UNIT CELL VOLUME RESULTINGFROM INTERSTITIAL H CONTENTS CORRESPONDING TO 93.3 kPa, 296 K- - x 0.06 -0.000 14(10) -0.000 08(15) -0.000 lO(11) - 3(4)x = 0.12 0.000 83(7) 0.000 27( 13) 0.000 72(8) 18(3) 0.061 1.10x = 0.17 0.002 2205) 0.001 18(19) 0.001 87(17) 53(6) 0.140 1.27x = 0.20 0.002 96(7) 0.000 78(9) 0.002 30(8) 6 W ) 0.151 1.33(standard deviations are shown in parenthesis)The equilibrium H contents (column 6, table 1) were determined by separate experi-ments in a vacuum apparatus under conditions as similar to those employed for theX-ray measurements as possible.At least two repetitions were performed for eachphosphide sample. The partial molar volumes computed from the unit cell expan-sions and the equilibrium contents (there are four formula units of Pd3PI-X per unitcell) are shown in column 7 of table 1. The expansion appears to be non-linear withH/uni t cellFIG.3.-Volume expansion of unit cell plotted against H/unit cell (296 K)FLANAGAN, BIEHL, CLEWLEY, RUNDQVIST AND ANDERSSON 201x (fig. 3). A single H atom in an interstice increases the volume of the solid by anaverage of 2.0 A3 for the phosphides listed in the last three rows of table 2, i.e., thosewhere significant absorption of H occurs. From the “ master ” correlation given byBaranowski et al. l5 for the lattice expansions of f.c.c. metals and alloys with H andfrom recent data for b.c.c. systems, 16* Peisl I8 has concluded recently that H has aunique size of w2.9 A3 independent of the nearest neighbour coordination numberand the original size of the hole into which H is to be inserted. Octahedral intersticesare occupied in f.c.c.systems and tetrahedral ones in b.c.c. systerns.l The volume of Hin the Pd3PI-x compounds is therefore smaller than the usual volume of 2.9 A3. Thesmall partial molar volume of I3 in Pd3P1--x may be related to the unusual intersticesoccupied by the H and to the anisotropic lattice expansions, but this cannot be arguedquantitatively because there is no explanation as yet for the usually observed volume ofH in solids,l ti i.e., why should the partial volume be unrelated to the original hole size ?The ratio of Vp/ VH is w 5.5, which is smaller than would be expected from the usuallyemployed “ sizes ” of these atoms.20RATES OF EQUILIBRATION OF H2 WITH Pd,P1-, COMPOUNDSEquilibration of H2 for the phosphide (x = 0.20) was reasonably rapid, e.g.,several hours were required for equilibrium to be attained for a typical dose of H2 at323 K.The sample was in the form of a stress-relieved coarse powder. The timesfor equilibration increased markedly as x decreased. This is a valid generalizationbecause this behaviour was noted independently during the X-ray studies (at Uppsala)and during the thermodynamic studies (at Vermont) and because the rates for all of40-30-r(N 2+s \a20-I10I401 I 1 10.01 0.02 0.03 0.04HIPd3Pr-xFIG. 4.-Equilibrium H2 pressures as a function of H content (323 K). Numbers on straight linesrefer to the following x values : (l), x = 0.06 ; (2), x = 0.12 ; (3), x = 1 (pure Pd) ; (4), x = 0.17 ;(9, x = 0.20. The straight lines are drawn through the data at low H contents where Sieverts’ lawof ideal solubility is obeyed202 SOLUTION OF HYDROGEN IN Pd3P1+the Pd phosphides were found to be quite reproducible.Exposure of the samples tothe atmosphere before evacuation did not appear to affect the rates of H2 absorption.These observations suggest, but do not prove, that surface steps are not rate-control-ling. It is tentatively suggested that the slow step in H2 absorption (desorption) isinternal diffusion. This would lead in a natural way to a decrease in rates as x de-creases, since it might be expected that the jumping of H atoms through the lattice uiainterstices other than the favoured ones (those about the empty prisms) constitutesthe highest potential barrier and it then follows that the fewer empty prisms there are,the slower will be the diffusion.H2(D2) SOLUBILITIES AND THE ISOTOPE EFFECTFig.4 shows typical solubility data for the P C I ~ P ~ - ~ compounds at 323 K. Forsmall H contents ideal solubility obtains, i.e., Sieverts’ law is obeyed, n = (H/Pd3P,-,)K&,, which proves that H2 is dissociatively absorbed when it enters the lattice. Thesolubility decreases markedly with decrease of x until it reaches essentially zero for thestoichiometric compound. This is expected because, as x decreases, P progressivelyoccupies more of the triangular prismatic holes and consequently blocks the squarepyramidal interstices for H occupation. The measured solubilities for x < 0.12 areH/Pd$’o.80FIG. 5.-Equilibrium Hz (D2) pressures plotted against H(D)/Pd3P0.80 at various temperatures.0, H data ; e, D dataFLANAGAN, BIEHL, CLEWLEY, RUNDQVIST AND ANDERSSON 203not very accurate because of the long times required for equilibrium.Data forx = 0.12 are not shown because these were determined at elevated temperatures andthe plot shown for this sample in fig. 4 has been calculated from these higher tempera-ture data. It is of considerable interest that the solubility of H2 is greater in thephosphides with x = 0.20 and 0.17 than in pure Pd at these low H2 fugacities.Solubility in pure Pd has been calculated on the basis of H/Pd3. It can be seen thatthe deviations from ideal behaviour occur in opposite directions for the phosphidesand for pure Pd.The latter’s deviations are caused by attractive H-H interactiom2120 iH/Pd3Po.s3FIG. 6.-Plot of equilibrium pressures against H(D)/Pd3P0.~3 (323 K). D, filled symbols ; H, opensymbols. Different symbols represent different runs carried out at different times on differentvacuum systems.Solubilities at 133.3 Pa (1 Torr) and 323 K are plotted in fig. 2. At these lowpressures, solubilities closely follow the behaviour of the lattice expansions althoughthe expansions reported in table 1 result from greater H contents. The H2 solubilitiesfall to almost zero at the point of the anomaly.Fig. 5 shows a series of low pressure solubility plots for H2 and D2 for the phosphidewith x = 0.20. The solubility of D2 is seen to be much less than that of H2 ; similarbehaviour is found for Pd-H.2 Data were found to be quite reproducible (fig.6) andif, for example, H is removed by evacuation and data redetermined the data agree.Absorption and desorption data also agree. In fig. 6 are shown data for Pd3P0.83obtained in one vacuum system and similar data obtained six months later in anothe204 SOLUTION OF HYDROGEN I N Pd3P1--xvacuum system following reweighing the sample into another vessel and remountinginto the different system having different dosing volumes, etc. This reproducibilityindicates, e.g., that H2 is not reacting with P to form PH3 and also indicates the longterm stability of the phosphides when exposed to the atmosphere.Differences in isotopic solubilities have been used as a means for obtaining OH,the Einstein temperature, which corresponds to the optical mode of H in solids and isgenerally well-separated from the acoustic modes of the lattice. In the ideal case oflocalized H interstitials the H can be treated as an Einstein oscillator.The differencein isotopic solubilities can therefore be used to estimate the optical mode frequency ofthe interstitial H from the known thermodynamic properties of the gaseous speciesand the assumption that 0 H = ,,t&.21s 22 A convenient way to express the isotopeeffect is by the function iRTh(pD,/pH,) which is evaluated at a fixed low solubility.At 323 K the values for this function are 1930 and 2050 J(mo1 H)-l for the phos-phides with x = 0.20 and 0.17, respectively, and for Pd-H the value of this function is1775 J(mo1 H)-1.23 All of these values for the isotope function are quite close to oneanother.In contrast, Pryde and Tsong 24 find a value of - 1325 J(mo1 H)-l for thisfunction for Ta-H at 357 K, i.e., the sign is different from that observed here and forPd-H. The inverse isotope effect for Ta-H arises because of the large value of 0,,reflecting the fact that H is held very strongly in the interstice. It is surprising that theisotope effect is so similar for H in the octahedral interstices of Pd and in the squarepyramidal interstices of Pd3P1--x compounds. Using the average value of the isotopefunction for the two phosphides a value of 8, = 930 K is derived compared with thevalue of 795 K for Pd-H as determined from inelastic neutron ~cattering.~~ (Oatesand Flanagan 23 have pointed out for the special case of pure Pd, the assumption ofOH = d%D may not be valid so the value of 8, determined from neutron scattering isemployed for comparison).The value of 0, from isotopic solubility differences forthe phosphides is therefore slightly higher than for pure Pd. Despite the markedlydifferent environments of the interstitial H in the two cases the isotope effect derivedfrequencies do not appear to differ very much. The slightly larger value of 0, forthe phosphides may reflect stronger binding in the interstice as compared to Pd andthis is consistent with the results from thermodynamics (see below). On the otherhand, if the H is delocalized within a given prism, i.e., if it can readily hop from oneinterstice to the other, the localized Einstein model upon which the isotopic analysis isbased will not be valid so that the derived optical frequency must be regarded astentative.PARTIAL THERMODYNAMIC PARAMETERS FOR H2 SOLUTIONRelative partial thermodynamic parameters for H2 solution can be derived fromp-c-Tdata, e.g., fig.5, from the relation~hip,~ which applies for each H content,RT In p k , = AGH = AHH-TASHand the temperature dependence of AGH gives A S H from the slope and AHH from theintercept. Excess quan-tities are given byA S H = SH--$Sk2 (1 atm) and AHH = HH-&Hk2(1 atm).under conditions where " blocking " is not a factor (see below) which means at lowvalues of n, where n is the H to formula unit ratio and p is the number of intersticeFLANAGAN, BIEWL, CLEWLEY, RUNDQVIST AND ANDERSSON 205per formula unit which can be occupied by H in the fugacity range employed.infinite dilution eqn (2) reduces toAtAGS = AH;- TAS; (3)where AGA and AS; could equally well be designated by AHH+o and AS;"-,, respectively.Eqn (2) and (3) show that p must be known before AGA and AS; can be evaluated.The values at infinite dilution, standard state values, can be determined by extra-polation of plots of AGE against n to n = 0 (table 2).TABLE 2.-RELATIVE PARTIAL THERMODYNAMIC PARAMETERS AT INFINITE DILUTION OF H FORTHE SOLUTION OF +Hz IN Pd,P1-, COMPOUNDS (323 K)x = l -9 800(2)x = 0.12 -10 900x = 0.17 - 15 250x = 0.20 - 16 300- 56.5(2)- 58.3- 56.7- 56.3- 56.5(2)- 46.4- 47.7-48.7ASH+o has been calculated assuming that p = 1 for each formula unit and AS;has been calculated using p = 2x.The data for the phosphide with x = 0.12 wereobtained in a temperature range from 340 to 460 K and the values given in the tablewere adjusted from the experimental values to 323 K using the temperature-depend-ence for Pd-H.2 This is admittedly an arbitrary procedure, but the corrections aresmall for AHA and E -2 J(mo1 H)-l K-l for the partial entropies. The values ofAS; are all similar but considerably less negative than for Pd-H. The reason for thisis unknown but may be related to delocalization of H in the prisms.STATISTICAL MECHANICAL MODEL FOR H I N Pd3P1--xIt is proposed that only one H can occupy an interstice on the face of an emptyprism, i.e., the occupation of one interstice by an H atom blocks the occupation of theother interstices within the sameprism.The H-H distances would be too close if morethan one H were to occupy a given prism. Speiser and Spretnak 26 developed a statist-ical mechanical model based on the idea of blocking which can be called a large hardcore model without overlap. If (2-1) neighbouring interstices are blocked for oc-cupation the configurational partition function is given byZ"@N,)!N ! -Nm- ! (: N ,(4)where N, is the number of formula units and pNm is the number of interstices whichcan be occupied at the fugacities in question ; /?Nm is 2xNm for the Pd3P1--x compounds.N is the number of interstitial H atoms.The chemical potential of dissolved H cor-responding to this configurational contribution is&/RT = In [n/(p-Zn)] ( 5 206 SOLUTION OF HYDROGEN IN Pd3P1--xwhere n = N/N,. Fig. 7 shows an experimental plot of pkZ against n for the phos-phide with x = 0.17 (323 K) up to relatively high H contents where blocking becomesmore important. Theoretical plots of p i 2 against n are also shown as calculatedfrom eqn (6) with 2 = 2RTln p& = AH&- TASG+RT In0 .o 2 .o 4 .O 6 .o 8 0. IH/P&Po. 8 3FIG. 7.-Equilibrium H2 pressures plotted against H/Pd3P0.83 (323 K). A, experimental data ;solid line, calculated data using eqn (6) with blocking allowed for ; dashed line, calculated data witheqn (6) without allowance for blocking.The values of standard parameters used in the theoretical calculation [eqn (6)] werechosen for best fit with the experimental data ; these standard values are within experi-mental error of those given in table 2. (A small change in AG; has a significanteffect on ph2 at large n values).It can be seen that the blocking model gives goodagreement with experiment whereas, without the 2 factor, agreement with experimentis poor even if AGg is selected to achieve agreement with experiment at a medium nvalue. The success of the large hard core model without overlap is noteworthy.While the model was originally devised for metal-H systems, Qates et aLZ7 pointedout that it fails for these systems at all but the lowest H concentrations because ofoverlap effects, i.e., if next nearest 13 interstitials each block the same interstice, eqn(6) is not valid.The success of the fit of the data to the calculated model with blocking,but without overlap, suggests the correctness of the model. The model is quitereasonable because most prisms where P atoms are missing will be surrounded byprisms where P atoms axe present and, therefore, the chance of overlap is small. Theagreement to the model is noteworthy because, as mentioned above, pure metal-FLANAGAN, BIEHL, CLEWLEY, RUNDQVIST AND ANDERSSON 207systems do not follow this model because of overlap effects and therefore these non-stoichiometric Pd3P1--x compounds are the first example where this simple model isadequate. The success of the fit also implies that non-ideality in these systems, i.e.,deviations from Sieverts’ law, arise from configurational effects ; AH, is thereforerelatively independent of H content.The energetics of H2 solution are not affected bythe amount of H already dissolved in the lattice; overall electronic factors such asfilling of the d-band of the phosphides by H appear to be of minor importance.The equilibrium H contents of 0.151(x = 0.20), 0.140(x =0.17) and0.061(x=0.12)found at 93.3 kPa, 296 K axe not inconsistent with the blocking model since the limitingH contents expected for the blocking model correspond to the respective x values. Thevalues for x = 0.20 and 0.17 are only slightly lower than the x values and higher pres-sures would presumably yield values closely equal to x.Much higher pressures arerequired for the phosphide with x = 0.12 to attain its limiting content since the thermo-dynamics of solution axe less favourable.The absence of hydride phase formation is related to the lack of a strong H-Hattractive interaction such as observed in the Pd-H system (fig. 4). It should benoted that the solubilities of H, calculated per Pd atom, in these phosphides exceedsthe concentration at which Pd-H would have already formed its hydride phase. Arelatively long-ranged elastic interaction is believed to be the principal cause of theH-H interaction in pure metal-H systems 28 which results in mean field behavioursince, on an average, a given H feels the influence of all the other H atoms because oftheir homogeneous distribution and diffusive motions.Such an interaction is appar-ently either missing from or negligible in these phosphide systems. Perhaps thereason for this is that the H atoms are located only in discrete sites (the P-emptyprisms) which are fixed in space and may be rather fax, on an average, from the otherM atoms. More importantly it is difficult to imagine how hydride formation couldoccur physically unless the prisms which are devoid of P atoms could aggregate andapparently this does not occur.Although the solubilities and lattice expansions fall to approximately zero near theanomaly (fig. 2), this behaviour can be misinterpreted. There appears to be no sud-den change in the thermodynamic parameters near the anomaly. The solubility at agiven low H content is related to 116;; by a modified Sievert’s lawn = (p&,i/?) exp - (AGg/RT).(7)Therefore for a given value of p i 2 the logarithm of the solubility is related to- (AGg/RT- In p ) and if In n values are plotted against x there is a smooth change asa function of x from x = 0.2 to x = 0.06 (the smallest solubility value determinedwith any reliability), on the other hand, as seen earlier, plots of n itself against x leadto an apparent sudden transition near the anomaly, i.e., x = 0.08.CONCLUSIONSThese phosphides represent the first example of an investigation of H2 solution bycompounds of an H,-absorbing metal and a non-metal where detailed attention hasbeen given to the specific interstices which are occupied. A variety of different typesof interstices are available which could, a priori, be occupied by H atoms, i.e., octa-hedral, tetrahedral, central interstices of empty triangular prisms and distortedsquare pyramidal interstices.Neutron diffraction studies have shown that the latterinterstices are occupied. OFranzen et aL4 concluded from their studies of H2 solution in Ta6S and NbzlS208that H2 solution was not strongly structure dependent since the thermodynamic para-meters for H2 solutions in these compounds at infinite dilution were similar to thosefound for the pure metals. (In the present research it has been stressed, however,that some information must be known about which interstices are occupied in orderto evaluate the standard partial entropy of dissolved H for comparison with the puremetals’ thermodynamic behaviour).In striking contrast, hydrogen solution inPd3P, --x provides an extreme example of structure dependence where the interplaybetween P vacancies in the centre of triangular prismatic interstices and H occupationof adjoining square pyramidal sites largely dictates the solubility behaviour. Further-more, the relative partial enthalpy for H, solution in the pkosphides with x = 0.17and 0.20 is almost twice as exothermic as for pure Pd, so in this sense also solution ofH2 is structurally dependent.SOLUTION OF HYDROGEN I N Pd3PIAxT. €3. I?. thanks the N.S.F. for financial support of his research in the area of solid-Hsystems. Y.A. and S.R. are grateful for financial support from the SwedishNatural Science Research Council.F.A. Lewis, The Palladium Hydrogen System (Academic Press, London, 1967).T. B. Flanagan and W. A. Oates, Adv. Chem. Ser., 1978, 167,283.T. Graham, Phil. Trans. Roy. SOC., 1866, 156,415.H. F. Franzen, A. S. Khan and D. T. Peterson, J. Solid State Chem., 1976, 17, 283.T. B. Flanagan and W. A. Oates, Ber. Bunsenges. phys. Chem., 1972,76,706.H. Brodowsky, 2. phys. Chem. N.F., 1965,44,129.C. Wagner, Acta Metallurg., 1971, 19, 843.S. Rundqvist and L. 0. Gullman, Acta Chem. Scand., 1960, 14,2246.Chem., to be published.* H. F. Franzen, A. S. Khan and D. T. Peterson, J. Solid State Chem., 1976, 19, 81.lo Y. Anderson, S. Rundqvist, R. Tellgren, J. 0. Thomas and T. B. Flanagan, J. Solid Statel 1 L. 0. Gullman, J. Less Common Metals, 1966, 11, 157.l2 6. Hagg, N. 0. Ersson, G. Rudenholm and B. Sellberg, J. Appl. Cryst., 1979,12,221.l3 R. D. Deslattes and A. Henins, Phys. Rev. Letters, 1973, 31, 972.l4 N. 0. Ersson, Institute of Chemistry, University of Uppsala, unpublished.Is B. Baranowski, S. Majchrzak and T. B. Flanagan, J. Phys. F, 1971, 1,258.l6 Mz Pfeiffer and N. Peisl, Phys. Letters, 1977, 60A, 363.l7 A. Magerl, referred to by H. Peisl, in Hydrogen in Metals, ed. G. Alefeld and J. Volkl (Springer-l8 H. Peisl, in Hydrogen in Metals, ed. G. Alefeld and J. Vojkl (Springer-Verlag, Berlin, 1978),l9 A. Magerl, N. Stamp, H. Wipf and G. Alefeld, J. Phys. Chem. Solids, 1977, 38, 683.2o H. J. Goldschmidt, Interstitial Alloys (Butterworth, Chichester, 1967).21 Y. Ebisuzaki, W. Kass and M. O’Keeffe, J. Chem. Phys., 1967,46,1373.22 P. T. Gallagher and W. A. Oates, J. Phys. Chem. Solids, 1971, 32,2105.23 W. A. Oates and T. B. Flanagan, J.C.S. Faraday I, 1977, 73,407.24 J. A. Pryde and I. S. T. Tsong, Trans. Faraday SOC., 1971, 67,297.25 W. Drexel, A. Muriani, D. Tocchetti, W. Kley, I. Sosnowska and D. K. ROSS, J. Phys. Chem.26 R. Speiser and J. W. Spretnak, Trans. Amer. SOC. Metallurg., 1955, 47, 493.27 W. A. Oates, J. A. Lambert and P. T. Gallagher, Trans. Metallurg. SOC. A.T.M.E., 1969, 245,28 G. Alefeld, Ber. Bunsenges. phys. Chem., 1972,76, 746.Verlag, Berlin, 1978), vol 1, p. 53.vol. 1, p. 53.Solids, 1976, 37, 1135.47.(PAPER 9/460

ISSN:0300-9599    

DOI:10.1039/F19807600196

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

J.C.S. Faraday I, 1980, 76,209-219Solubility of Synthetic Calcium HydroxyapatitesBY RQNALD M. H. VERBEEK,": HERMAN STEYAER,j- IlARRY P. THUN ANDFRANS VERBEEKLaboratorium voor Amlytisclae Scheikunde, Universiteit Gent,J. Plateaustraat 22, B-9000 Gent, BelgiumReceived 30th April, 1979The solubilities of synthetic calcium hydroxyapatites [Ca(OH)A], Ca5(P04)30H, are determinedin the ternary system Ca(OH)2+H3P04+H20 at 25°C in the pH range 5-7 by equilibration withdilute &PO4 solutions. The thermodynamic solubility product Ksp of precipitated Ca(OM)A,prepared by titrating a boiling aqueous suspension of Ca(OH), with M3P04, is (3.2s k 0.29) x11101~ dm-27 . Ca(OH)A.prepared by solid state reaction of CaC03 and CaHP04 and precipitatedCa(0H)A heated at high temperatures in a water vapour atmosphere showed no constant Ksp.Afterbeing washed, these samples equilibrate according to the usual solubility product principles with a Kspvalue of (3.d4+0&) xThe influence of heat treatment and washing on the solubility behaviour is discussed. Deviationsfrom solubility behaviour which accords with the usual solubility product principles are caused byreactions occurring at the surface of the solid. The latter are induced by a high free energy of thesolid and/or the presence of a fraction with high solubility associated with the Ca(0H)A sample.and (2.6,-, k0.21) x mo19 dm-27, respectively.The complexity of the formation and remineralisation processes of biological hardtissues, such as bone and teeth, has intensified the search for a less complicated proto-type for these materials.Its close resemblance to the main mineral phase has givenhydroxyapatite, Ca5(P04)30H [hereafter referred to as Ca(OH)A)] its special im-portance as the prototype of these tissues.l However, reports on its physicochemicalproperties, in particular solubility behaviour and thermodynamic solubility product,appear to be contradictory.2 -6Much of the confusion on this subject has been removed by the work of Brownand his coworkers who adequately described the solubility behaviour of Ca(0H)Aand its relations with several calcium phosphates in the ternary Ca(OH), + &PO4 +H20 system.' Even so, some uncertainty surrounds the thermodynamic solubilityproduct of Ca(OH)A, which still seems to depend on the method of preparation.8-1For this reason the solubility behaviour was examined on two differently preparedCa(0H)A samples.EXPERIMENTALFor the preparation of solutions and samples analytical reagent grade chemicals wereused throughout. Only water freshly generated by a Milli-Q-system (Millipore) and keptfree of COz was used.MATERIALSCa(0H)A samples were prepared by precipitation or by solid state reaction.SAMPLE ACaC03 and CaHP04 were weighed in a ratio to conform to the stoichiometry of Ca(0H)Aand niixed and ground as an acetone slurry in an agate ball mill. After evaporation of thej- Astronomisch Observatorium, Universiteit Gent, Belgium.20210 SOLUBILITY OF SYNTHETIC CALCIUM HYDROXYAPATITESacetone, the reaction mixture was pressed into bars under 2 x lo7 N m-2 to ensure a bettersolid state reaction. Open platinum crucibles containing the bars were inserted into a tubefurnace and slowly heated from room temperature to 1200°C.A stream of nitrogen bubbledthrough boiling water was led through the tube, thus removing evolved C02 and providingthe water necessary for the solid state reaction.12 After 10 h at 1200°C the samples werequenched in air, crushed, powdered and homogenized in an acetone slurry in an agateball mill. After drying they were again pressed into bars and heated at 1280°C under watervapour for at least 8 h. This homogenization and sintering cycle was repeated three times.At the end of the last sintering the temperature was kept at 900°C for 4 h.Finally thesamples were slowly cooled in a water vapour atmosphere to 4OO"C, quenched in air, crushed,powdered and homogenized in an agate ball mill.Chemical analysis yielded 39.77 wt % Ca, 18.48 wt % P, and a molar Ca : P ratio of1.663. The uncertainties in the analysis of the solids were determined as 0.08 wt %,0.04 wt % and 0.005 units, respectively.SAMPLE BIA boiling calcium hydroxide slurry was titrated with phosphoric acid solution accordingto the procedure followed by Avnimelech et aZ.13 and McDowell et aL8 Further, after beingwashed with a dilute phosphoric acid solution at boiling temperature, the solid was stirredfor one day in water at room temperature under nitrogen and the supernatant solutiondecanted. This washing was repeated three times.Finally, the solid was filtered on asintered glass filter, washed on the filter with acetone, dried at 50°C overnight and at 110°Cfor several days.Chemical analysis yielded 40.01 wt % Ca, 18.37 wt % P and a Ca : P ratio of 1.683.SAMPLE BIISample BI was pressed into a bar and sintered at 1200°C for one day. Then, afterbeing held at 900°C for 5 h, the sample was slowly cooled to 400°C and treated further asdescribed for sample A. All heat treatments were done under a water vapour atmosphereas described previously.Chemical analysis yielded within experimental error the following analytical compositionof sample BI : 40.07 wt % Cay 18.39 wt % P and Ca : P = 1.684.SAMPLE c A N D DSamples A and &I, respectively, were washed and further treated as described for sampleChemical analysis yielded 39.87 and 40.00wt %Ca, 18.51 and 18.42wt %P and aBI.Ca : P ratio of 1.665 and 1.678, respectively.SAMPLE EThe treatment as described for sample BIT was followed, except that the sintering at1200°C was prolonged for one week and the sample was quenched in air after one day ofsintering at 900°C.The sample was washed and further treated as sample BI.Chemical analysis yielded 39.94 wt % Cay 18.44 wt % P and Ca : P = 1.674.SAMPLE FPrepared as described for sample A, except that the solid was heated at 900°C in a streamChemical analysis yielded 39.84 wt % Ca, 18.41 wt % P and a Ca : P ratio of 1.672.For all samples the characteristic X-ray diffraction pattern and i.r.spectrum show sharpand well resolved peaks characteristic of well crystalline solids. The pattern and spectrumof sample Bl are slightly broadened in comparison with those of the other solids. Moreover,in the i.r. spectrum of this sample some bap-ely detectable absorptions indicate the presenceof nitrogen, washed in water at 5°C and then quenched in airVERBEECK, STEYAERT, THUN AND VERBEEK 21 1of a small carbonate impurity. The appearance of small peaks attributable to CaO andCa(0H)2 in the X-ray diffraction pattern aid i.s. spectrum of sample BII confirms that Cog-substitutes to some extent for PO:- in the lattice of BI.14 To a smaller extent the presenceof CaO and Ca(OW)2 is also seen in the pattern and spectrum of D. In samples E and Fthis presence is barely detectable.Phosphoric acid solutions used in the equilibrations were prepared from a 1 mol dm--3&Po4 stock solution.The concentration of these dilute solutions was checked regularlyand the uncertainty in the concentration was estimated as 1 % maximum.SOLUBILITY DETERMINATIONS0.25 g Ca(0H)A was added to 250 cm3 C02-free phosphoric acid solution in 500 cm3polyethylene flasks. The flasks were slowly rotated end-over-end in a water bath at(25.00+ 0.02)"C. After 14 days equilibration, the suspensions were filtered through Milli-pore filters (0.22pm pore size). Samples of the clear filtered solutions were taken for pHmeasurement and determination of calcium and phosphorus. All manipulations of sus-pensions and filtered solutions were done at 25°C under nitrogen atmosphere.Calcium was determined by atomic absorption spectrometry.Lanthanum was added tothe samples and standards to suppress interference by phosphate. Phosphorus was deter-mined spectrophotomefrically as orthophosphate using a slight modification of the methodof Brabson et The uncertainty is 3 and 1 %, respectively, on the amounts ofcalcium and phosphorus analysed. The glass- and silver-silver chloride electrode pair usedin the pH determination was calibrated with certified N.B.S. buffer standards.16 The uncer-tainty in the pH measurements is estimated as 0.020 pH units.CALCULATIONSIon activity products for Ca(0H)A defined bywere calculated from the molar activities ax of the species X in solution by the proceduredescribed by Moreno et a1." The presence of hydroxyl ions, all ionic phosphate speciesand calcium phosphate ion pairs was also taken into account.The dissociation constantsat 25°C for phosphoric acid and for water were taken from Smith and Martell.18 Thestability constants for the ion pairs CaH2PO$ and CaHP04 are given by Gregory et aZ.19The iteration was continued until the ionic strength, I, remained constant within 0.01 %.The activity coefficient yx of a zx-valent ion X was calculated from the Debye-Hiickelequation-logy, =1 -t- Ba&where A and B are temperature dependent constants 2o and & represents the distance ofclosest approach for the X ion.21Whenever I= remains constant over a broad range of pH values and total molar calciumand phosphate concentrations, the solution can be considered to be saturated with respect toCa(0H)A and thus I= equals the activity solubility product Ksp of that phase.In that case,Ksp was estimated by a generalized least-squares procedure simultaneously adjusting theobserved values of the observations.8* 9 9 l 9 This procedure, described in greater detailelsewhere,19 consists of minimizing the weighted sum of squares of residuals of four cb-servations per equilibration (Le., total molar concentration of calcium, CCa, total molarconcentration of phosphorus, cp, initial molar phosphoric acid concentration, c;, and 1pH-I)subject to three condition functions given b2 12andSOLUBILITY OF SYNTHETIC CALCIUM HYDROXYAPATITESwithD, = -- ‘‘a , the dissolution ratio,(c, - ‘3In eqn (3), defining the deviation from electroneutrality Ek, the summation includes allionic species present in the Ca(OH),+ H3P04+ H20 Eqn (3)-(5), respectively,require for each equilibration that the conditions of electroneutrality, saturation withrespect to Ca5(P04)30H and congruent dissolution are satisfied simultaneously.Theobservations were weighted according to Gregory et al.RESULTSThe compositions of solutions equilibrated with the Ca(0H)A samples aresummarized in table 1 together with the calculated quantities I, Ek, DR and IHA. Thetable shows that E& is within the anticipated experimental error, indicating no grosserror in the analytical figures, calculations and the model used. The dissolutionratio DR equals the stoichiometric 5 : 3 ratio within experimental error for sampleseries A, C , D and E.For the other series DR tends to increase with increasing pHand at high pH values becomes significantly greater than the value the Ca(0H)Astoichiometry requires.TABLE 1 .-COMPOSITION AND CHARACTERIZATION OF SOLUTIONS EQUILIBRATED AT 25°C WITHTHE Ca(0H)A SAMPLESsolution compositionequilibration &/lo - 4 cC,/10-4 cP/1o-4 z/io-3 ~ k 1 1 0 - 4 iaAi 10 - 56no. moldm-3 pH moldm-3 moldm-3 moldm-3 moldm-3 DR 111019 dm-27A-12345678910Br-12345678910BII-123456789102.033.054.075.096.108.149.1510.220.340.72.033.054.075.096.108.149.1510.220.340.72.033.054.075.096.108.149.1510.220.340.76.5976.4076.2606.1325.9885.8495.7885.7095.3314.9786.4136.2926.1555.9995.8985.7915.7225.7105.4155.0776.6146.3626.1525.9935.8945.7215.6215.2524.9115.5462.122.783.354.194.936.337.217.5115.128.81.792.383.253.804.595.796.687.4515.029.61.792.613.194.034.616.166.897.5614.929.73.264.666.037.549.0011.913.414.729.457.83.014.365.837.158.5911.413.014.529.258.42.954.445.837.378.7711.813.314.729.258.40.6890.8831.061.301.511.932.172.294.508.570.5700.7651.011.181.411.792.042.274.478.760.5950.8270.9991.241.421.872.082.284.448.770.2730.1580.1460.2330.1000.438-0.2340.303-0.567-0.0460.093-0.1420.120-0.0510.110-0.379-0.172-0.1490.1050.248-0.0400.1400.0060.181-0.0250.0790.1140.0330.0960.6621.741.721.711.711.701.671.711.671.671.691.831.811.851.841.841.751.731.731.701.681.961.871.821.761.731.691.681.671.681.6711469.137.929.011.19.008.633.911.570.5972.654.926.092.041.732.132.003.685.623.3648.123.15.312.661.761.030.4960.3040.41 10.24C-12345678910I)- 12345S789YOE-12345C78910F-123456789PO2.033.054.075.096.108.149.1510.220.340.72.033.054.075.096.108.149.1510.220.340.72.033.054.075.096.108.149.1510.220.340.72.033.054.075.096.108.149.1510.220.340.7VERBEECK, STEYAERT, THUN AND VERBEEKTABLE 1-continued6.2136.0946.0015.8825.7745.66,5.6025.2395.5584.9186.2746.1065.9455.8765.7665.6665.5695.5175.2124.9366.3126.0485.9435.8325.7345.6875.5835.5195.2114.9406.9626.6646.5066.3806.2796.1176.0545.9975.6295.3401.942.523.184.154.496.136.617.4514.830.11.602.503.143.894.896.406.807.4814.929.31.692.413.183.884.706.206.987.6714.929.52.423.033.544.394.926.557.2 17.9415.229.83.164.565.997.508.8711.813.114.729.558.22.974.495.897.3 18.9811.913.614.529.357.93.034.505.957.388.8411.813.314.729.358.13.304.665.967.458.7511.713.414.829.258.10.5900.7820.9881.251.391.862.022.264.458.840.5170.7740.9681.191.471.922.062.254.448.660.5420.7510.9791.191.421.872.102.304.458.710.8230.9901.151.391.552.012.232.444.558.830.3900.124-0.0250.409-0.2680.065-0.302-0.198-0.242I .64-0.1040.1330.0450.0900.4280.5120.0940.1350.325-0.0200.0020.0530.0310.2250.1940.2790.3060.1350.4450.2260.212-0.0440.182-0.0520.273-0.080-0.028-0.0130.0730.2131.721.681.661.721.62I .651.661.641.621.711.721.741.731.751.701.711.661.721.671.701.711.671.691.691.721.781.681.701.661.691.921 .a91.871.861.851.831.711.721.721.712130.2230.3590.5540.5730.2430.4280.2930.3500.3400.2830.1900.3950.2080.3570.3330.5300.2040.1790.2190.3280.4630.1370.2200.1820.1600.6170.2990.2150.2200.3643 .3 2 ~ 1044 . 2 6 ~ 1031 . 6 2 ~ 1031.46X 1038.56X 1026 . 2 2 ~ 1025.32 X 1024.55 x 1021.75 X 1022.24 x 102From table 1 it is seen that for series BI, C, D and E the ion activity productremains constant over the concentration range studied. For each of these series anestimate of KSB was made with the generalized least-squares method.The resultsare summarized in table 2. Using the three condition functions defined by eqn (3)-(5)unacceptable corrections to the observations were found for series BI. X2-Statistics2show also that the data apparently do not belong to a population characterized by thethree conditions. Dropping the congruent dissolution condition [eqn (5)] in theTABLE 2.-RESULTS OF THE GENERALIZED LEAST SQUARES-METHOD FOR THE INDIVIDUAL SERIESnumber of ~ ~ ~ / 1 0 - 5 9series conditions B>b s2 x 108b 111019 b - 2 7BI 2a 1 9 1.11 32.5+2.9bC 3 29 1 . 3 4 3.44+ 0 . 3 6D 3 29 0.85 2.69+ 0.22E 3 29 0 . 8 1 2 . 6 0 k 0 . 2 1a Electroneutrality and saturation ; b degrees of freedom Q, variance of the adjustment s2 andstandard deviation on Ksp from the adjustment as defined in ref.(19)214 SOLUBILITY OF SYNTHETIC CALCIUM HYDROXYAPATITEScalculations results in the Ksp estimate given in table 2. The largest adjustments forseries BI with only two conditions were 3.3 and 0.6 % for c, and c,, respectively, and0.04 units for pH. A variance analysis on the IHA values for series C, D and E givenin table 1 reveals that the variance among these three series is uniform and that thereis no significant difference in IHA values, thus confirming the trend shown in table 2.Therefore an estimation of K,, was made from the combined data of series C, D and E.The overall variance based on 89 degrees of freedom (4) was 1.03 x and thecritical values of x295% (89), 64.8 and 117.0 bracket the value of 4, indicating thegoodness of fit.lg* 23 The resulting estimate of the activity solubility product Ksp is(2.89 k0.15) x m019 dm-27.The adjustments on the observations are mostlywithin experimental error. With the exception of equilibration C-1 and D-1, wheresomewhat larger adjustments on cCa, c,, c; and pH are found, the largest correctionon these observations is, respectively, 5.6, 0.8 and 0.8 % and 0.04 units.-2.5- 3.6- 3.:5.0 5.5 6.0 65PHcurves represent the calculated solubility isotherms at 25°C.FIG. l.--Log cCa (filled symbols) and log cp (open symbols) plotted against pH for sample BI. TheUsing the best estimates of Ksp, 3.25 x * and 2.89 x m019 dnr2’, thetheoretical solubility isotherms were calculated for sample BI and for the set of C, Dand E samples.These isotherms are drawn in terms of the logarithms of cCa (and cp)against pH in fig. 1 and 2 for the respective samples. These figures show that theexperimental points scatter around the theoretical isotherms, indicating the adequacyof the estimated Ksp and of the condition functions used in the adjustment procedure.The solutions equilibrated with sample A, BII and F show no constant ion activityproduct and are predominantly supersaturated with respect to Ca(0H)A. The I,VERBEECK, STEYAERT, THUN AND VERBEEK 21 5values progressively decrease as the solution pH decreases and apparently tend to aconstant value of and m019 d~n-~', respectively, in the equilibration ofsample F and of samples A and BII.!aPHFIG.2.-Log cCa (filled symbols) and log CP (open symbols) plotted against PH for samples C(O),D(n) and E(A). The curves represent the calculated solubility isotherms at 25OC.DISCUSSIONTheory 24 predicts that the equilibrium condition between a calcium phosphatesolid phase and an aqueous solution in the Ca(OH), + H3P04 + H20 system is definedby the so-called potential line-log (ad3 (ap04> = e+q log (aod2' (6)The coordinates in eqn (6), i.e. the logarithms of (ad3 (apo4) and (aca) areproportional to the chemical potentials of the components Ca(OH), and H3P04 insolution and the absolute values of the slope q equals the molar calcium to phosphorusratio in the saturating solid. Weighted least-squares analysis of the data calculatedfrom table 1 gives a value for q of (1.667+0.050), (1.653&0.034), (1.6663-0.040) and(1.677+0.053), respectively, for series BI, C, D and E.These values agree with thestoichiometry of the samples and of Ca5(P04)30H (1.669, indicating that the satur-ating solid is a Ca(0H)A phase. This confirms that BI, C, D and E equilibrate ac-cording to the usual solubility product principles. Moreover, this implies that sampleBI congruently dissolves although eqn (5) cannot be applied to the equilibration dat216 SOLUBILITY OF SYNTHETIC CALCIUM HYDROXYAPATITESand its DR values in table 1 suggest an apparent nonstoichiometric dissolution. Thisnonstoichiometry is mainly caused by the contribution of sorption phenomena on thesurface of the solid to the overall dissolution reaction as demonstrated by Avnimelechet aZ.I3 and C h ~ o n g .~ ~ Although the rather high DR values observed in the equi-libration series BII and F (see table 1) also can be explained by these sorption pheno-mena, precipitation reactions, as discussed further, certainly contribute to these highvalues.The Ksp of BI agree with values reported by Clark3 and Bell et aZ.26 for precipitatedCa(0H)A stabilized by prolonged boiling. However, it is slightly larger than thevalues (2.8-6.3) x mo19 d r r Z 7 obtained by others.8* 1 3 9 25 Sintering the pre-cipitated Ca(0H)A with subsequent washing promotes crystallinity, stoichionetryand purity of the solid, thereby decreasing its free energy as reflected by the lowersolubility product for sample D.Prolonged sintering does not affect the solubilityproduct, but seems to promote the removal of the CaO second phase by the washingprocedure (sample E). Further, the heat treatment produces solids of a quality equalto those obtained from very pure CaCO, and CaHPO, by solid state reactions (sampleC). This indicates that impurities present in the heat treated precipitates (sample IDand E) are small and do not influence the solubility significantly. Moreno et aZ.’report a Ksp of 3.7 x and 2.5 x m019 dm-27 for precipitated Ca(0H)Awashed after heating at 1000°C in an atmosphere of water vapour and in air, re-spectively. These values are larger than the 2.89 x ni019 dm-27 obtained herefor heat treated Ca(0H)A and are probably a reflection of a higher degree of dehy-dration in their Ca(0M)A.Partial dehydration increases the free energy of the solidas indicated by the experiments of Semter.12 The agreement between the solubilityproduct of the heat treated and washed samples C, D and I3 and the values of (2.8-6.3) x mo19 dm-27 obtained on precipitated Ca(0H)A 8 v 1 3 9 25 is a possibleindication that the degree of dehydration is small for the former.Although these slight but significant variations in Ksp among different samples canbe partly ascribed to experimental error and to differences in experimental and cal-culation procedures, they nevertheless indicate that a variation in the free energycontent of the solid also contributes to this solubility product variability. Considerabledifferences in the free energy content of the Ca(0H)A phase can be introduced, e.g ., bya variation in impurity content, in solid state defects (such as vacancies on Ca2+ andOH- lattice positions), in degree of disorder in the OH--positions and in the surfaceenergy.” 27 However, the high IHA values observed in the equilibration series F andfor the air heated sample of Moreno et al. indicate that variations in free energy can-not account completely for this persistent and high supersaturation. Neither can thesupersaturation and pH dependence of IHA in series A, BII and F be explained as areflection of a recrystallisation process involving the bulk of the solid. In that casenearly the same Ksp and solubility behaviour as for sample BI could have been ex-pected after washing samples A and BII.Moreover, the chemical and structuralproperties of the solids would have to change more drastically upon washing (samplesC and D).This variation of the ion activity product of a Ca(0H)A sample in solution is quiteoften observed 4* 6 * 9-11 and several theories, mainly surface oriented, have beenadvanced to account for this aberration.2* 4 9 5 * lo However, none of these theoriesadequately explains the different solubility behaviour and serious criticisms havemade them unac~eptable.~-~ It is now generally accepted that the difference in the dis-solution and precipitation rate of Ca(0H)A is the main cause for the observed pheno-mena. So, Francis et aZ.28 demonstrated that as a result of this difference thepresence of DCPD (dicalcium phosphate dihydrate, CaHPQ4.2H20) as a surfacVERBEECK, STEYAERT, THUN AND VERBEEK 217coating on the Ca(0H)A crystals controls the solubility behaviour of Ca(0H)A inacidic to neutral solutions. Narasaraju and R ~ o ~ ~ even found that a surface coatingof DCPD controlled the solubility behaviour of their solid state prepared Ca(0H)Ain solution up to pH 8.40.This indicates that the free energy of their preparationand/or the amount of second phases, as discussed further, was extremely high.However, the presence of more acidic calcium phosphates cannot account for thesolubility behaviour of samples A, BII and F. In fig. 3 the negative logarithms ofX \I '\. \271\'\ \ \ O \ '.\X0I I t I I18 19 20 21-log (ac,) (a&"FIG. 3.-Potential lines of Ca(OH)A, TCP and DCPD in the system Ca(OH)2+H3P04+H20 at25°C. Points represent experimental data for sample A( 0), BII( 0 ) and F( x ).(ac,) and (apo4) axe plotted for experiments A, BII a d F. In the figure thepotential lines defined by eqn (6) for Ca(OH)A,DCPD and /3-tricalcim phosphate(TCP) axe also drawn. The former was obtained by weighted least-squares fitting ofthe combined equilibration data of samples C, D and E, while the latter two werecalculated from their respective thermodynamic solubility pr0duct.l 9* 30 Weightedleast-squares analysis shows that the data fall on a straight line with a slope q of(1.312f0.012), (1.306&0.038) and (1.350&0.037), respectively, for series A, BII andF.These slopes approach the theoretical value of 1.333 for Ca8H2(P0&5H2O(octscalcium phosphate, OCP) very well. Since this calcium phosphate can gro218 SOLUBILITY OF SYNTHETIC CALCIUM HYDROXYAPATITESepitaxially on Ca(OH)A,31 the possibility that a surface layer of OCP controls thesolubility behaviour in these equilibrations cannot be ruled out. However, fromfig. 3 it is seen that all points are lying between the potential lines of Ca(OH)A, TCPand DCPD. Since TCP is less soluble than OCP7 this means that for all equilibra-tions the solutions are really undersaturated with respect to all calcium phosphatesmore acidic than Ca(0H)A. Therefore, the presence of OCP controlling the solubilitybehaviour is not a plausible explanation.Nancollas et al.[see, e.g., ref. (32)] and others 3 3 9 34 have shown that precipita-tion and growth of Ca(0H)A crystals from stable supersaturated solutions startsalmost immediately upon seeding with Ca(0H)A crystals. Moreover, formationand growth is controlled only by reactions at the surface of the Ca(0H)A seeds. Athigh supersaturation, these surface reactions are quite analogous to the reaction occur-ring during the de novo formation and maturation of Ca(0H)A [see, e.g., ref. (35)]. Atlow supersaturation precipitation seems to tend toward a process involving predom-inantly a simple crystal growth of Ca(OH)A.32* 34 So, if during the equilibrationthe solution becomes momentarily supersaturated with respect to Ca(OH)A, theconsequent precipitation-dissolution reactions will be quite similar to those observedwith the seeded growth of Ca(0H)A.Since the peculim solubility behaviour ob-served for samples A and BII disappears after the washing procedure, it is probablycaused by the presence of a labile fraction associated with the solids ; one which can beremoved or at least strongly reduced by washing. This fraction may be micro-crystalline Ca(0H) A with high surface energy, a surface layer [e.g., defective Ca(OH)A]and/or second phases (e.g., pyrolysis products such as CaO or TCP) having a higherfree energy and consequently a higher solubility than the bulk of the Ca(0H)Acrystals. In contact with an aqueous solution such fractions preferentially and morerapidly dissolve than the bulk of the solid phase, provoking an initial state of super-saturation with respect to Ca(0H)A.The degree of supersaturation is then deter-mined by the kind and amount of solid contaminating the Ca(0H)A phase, the solid-to-solution ratio used in the equilibration and the composition of the equilibrationsolution. Ca(OH)A, free of contaminations but having an abnormally high freeenergy, also causes an initial state of supersaturation. In general, the degree ofsupersaturation will be high under conditions where only a small amount of thethermodynamically most stable Ca(0H)A must dissolve to establish the saturationcondition. Therefore, during equilibration the consequently precipitating calciumphosphate gradually coats the dissolving apatite and forms a dissolution barrier untilno further dissolution of Ca(0H)A OCCUTS.In the end, the resulting surface layercontrols the solubility behaviour.The present interpretation is consistent with the experiments of Bell et al.26s 36Their results point toward the existence of a surface layer of variable compositionwhich readjusts according to the solution composition following bulk dissolution.This surface layer has a higher solubility than the bulk of the Ca(0H)A phase andcan be removed by successive equilibrations at the same pH. Further, a higher IHAvalue (two to three orders of magnitude) was invariably found when the solid, whichhad been equilibrated at low pH value, was subsequently seequilibrated at much higherpH value.The difference of 5.7 kJ mol-1 (1.4 kcal mol-l) in standard free energy offormation of their sample with respect to the present samples C , D and E seems toaccount for this behaviour.The exact composition of the surface layer is less certain because the earliermentioned sorption phenomena disturb the interpretation^.^^. 36 In the light ofpresent knowledge from structural and chemical studies on the de nouo formation ofCa(0H)A and on the seeded growth from supersaturated solutions, it can be said thaVERBEECK, STEYAERT, THUN AND VERBEEK 219the final surface layer is probably defective Ca(0H)A which may gradually changeinto a more perfect apatite depending on the solution composition. Whether thischange is thermodymn<cally and/or kinetically controlled and whether hydrolyticreactions and/or continuous dissolution-precipitation (known as Ostwald ripening)occw is not yet clear.These changes axe very slow and give rise to stationary solu-tion compositions over long periods of time.So, even though the equilibrations of samples C, D and E give a reliable value Ksp,considering the constancy of Ksp for one particular sample and the Ksp variabilityamong samples, it is still not certain whether the value of 2.89 x m019 dm-27is a good approximation of the real thermodynamic solubility product for Ca(0H)A.Further investigation of the solubility product and of the formation from solution ofCa(0H)A is necessary.R. A. Young, Clinical Orthop., 1975, 113,249.W . F. Neuman and M. W.Neuman, Chem. Rev., 1953,53,1.J. S . Clark, Canad. J. Chem., 1955,33,1696.H. M. Rootare, V. R. Deitz and F. G. Carpenter, J. Colloid Interface Sci., 1962, 17, 179.V. K. La Mer, J. Phys. Chem., 1962, 66,973.G. J. Levinskas and W. F. Neuman, J. Phys. Chem., 1955,59,164.W. E. Brown, in Environmental Phosphorus Handbook, ed. E. J. Griffith, A. Beeton, J. M.Spencer and D. T. Mitchell (John Wiley, New York, 1973).H. McDowell, T. M. Gregory and W. E. Brown, J. Res. Nut. Bur. Stand., 1977, 81A, 273.E. C. Moreno, T. M. Gregory and W. E. Brown, J. Res. Nut. Bur. Stand., 1968,72A, 773.S. Larsen and A. E. Widdowson, Experientia, 1971,27,483.lo D. R. Wier, S. H. Chien and C. A. Black, Soil Sci., 1971, 111, 107.l2 A. M. J. H. Seuter, Reactivity of Solids, Proc. 7th Int. Symp., 1972, p. 806.l3 Y. Avnimelech, E. C. Moreno and W. E. Brown, J. Res. Nut. Bur. Stand., 1973,77A, 149.l4 R. Z. Legeros, 0. R. Trautz, J. P. Legeros and E. Klein, Bull. SOC. chim. France, 1968, 1712 ;J. C. Labarthe, G. Bone1 and G. Montel, Ann. Chim. (France), 1973, 8,289.J. A. Brabson, R. L. Dunn, E. A. Epps, W. M. Hoffman and K. D. Jacob, J. Assoc. Ofic. A@.Chem., 1958,41,517.I6R. G. Bates, J. Res. Nat. Bur. Stand., 1962,66A, 179.l7 E. C. Moreno, T. M. Gregory and W. E. Brown, J. Res. Nut. Bur. Stand., 1966,70A, 545.l8 R. M. Smith and A. E. Martell, Critical Stability Constants (Plenum Press, New York, 1976),l9 T. M. Gregory, E. C. Moreno and W. E. Brown, J. Res. Nat. Bur. Stand., 1970, 74A, 461.2o R. S. Robinson and R. H. Stokes, Electrolyte Solutions (Butterworth, London, 2nd edn, 1965)." J. Kielland, J. Amer. Chem. SOC., 1937, 59, 1675. '' W. E. Brown, Soil Sci., 1960, 90, 51.23 W. E. Deming, Statistical Adjustment ofData (Dover Publications, New York, 1964)."J. McGregor and W. E. Brown, Nature, 1965,205,359.25 R. Chuong, J. DentaZ Res., 1973,52,911.26 L. C. Bell, H. Mika and B. J. Kruger, Arch. Oral Biol., 1978, 23, 329." F. C. M. Driessens, Caries Res., 1973, 7, 297.28 M. D. Francis, Ann. New York Acad. Sci., 1965, 131, 694; M. D. Francis, J. A. Gray andW. J. Griebstein, Adv. Oral Biol., 1968, 3, 83.29 T. S. B. Narasaraju and V. L. N. Rao, 2. phys. Chem. (Leipzig), 1974,255,655.30 T . M. Gregory, E. C. Moreno, J. M. Patel and W. E. Brown, J. Res. Nut. Bur. Stand., 1974,31 B. Dickens, L. W. Schroeder and W. E. Brown, J. Dental Res., 1973,52,84.32 G. H. Nancollas and B. Tomazic, J. Phys. Chem., 1974,78,2218.33 E. D. Eanes, Calcif. Tiss. Res., 1976, 20, 75.34 E. C. Moreno, R. T. Zahradnik, A. Glazman and R. Hwu, CaZcif. Tim. Res., 1977, 24, 47.35 E. D. Eanes and J. L. Meyer, Calcif. Tiss. Res., 1977, 23, 259.36 H. Mika, L. C. Bell and B. J. Kruger, Arch. Oral Bid., 1976,21,697.vol. 4.78A, 667.(PAPER 9/675

ISSN:0300-9599    

DOI:10.1039/F19807600209

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

J.C.S. Frraday I, 1980,76,220-223Reviews of BooksSolid Electrolytes: General Principles, Characterization, Materials, Applications. Ed. by P. HAGEN-MULLER and W. VAN GOOL. (Academic Press, N.Y., 1978). Pp. xvii-l-549. Price E33.80.Over the last decade, there has been an increasing interest in solids with unusually high ionicconductivities. In these materials, variously referred to as " solid electrolytes " or " superionicconductors " or " fast ion conductors ", the conductivities can be comparable to those found inionic solutions or melts. Probably the two best known examples are or-silver iodide and sodium-P-alumina. The sodium-sulphur battery, using a p-alumina solid electrolyte phase, is being developedby a number of companies and it offers possibilities for electric traction.Currently, solid electrolytesare the subject of an intensive, world-wide research effort with the major areas of interest being thenature of the ion transport, the preparation of new materials and the development of technologicalapplications. Over the last few years, several international conferences have been devoted specificallyto solid electrolyte research and more are being planned. In this book, the editors have aimed atproducing a broad, introductory text which covers the major areas of solid electrolyte research. Intheir preface, they express the hope that the book will give the student and research worker a firstimpression of the Various aspects of the field.The book is divided into three sections: theory and experimental methods, materials and applica-tions.There are thirty-two articles by forty-three contributors. The list of contributors containsthe names of many leading authorities in this field. The multi-disciplinary character of solid electro-lyte research is fully covered. This is exemplified by the diverse nature of the titles of some of thearticles; " Transport Mechanisms and Lattice Defects ", (Y. Haven), " Interface Phenomena ",(E. Bergman and H. Tannenberger), " Organic Ion Conductors ", (T. Takahashi), " Zeolites ",(L. V. C. Rees), " Solid State Potentiometric Gauges for Gaseous Species ", (M. Gauthier et a/.).The quality of the articles, as would be expected in such a large number, is varied although on thewhole a good standard has been maintained in the first section.Most authors have attempted toreview their particular topic. There are on average only about sixteen pages per article, as a con-sequence the degree to which the authors have successfully coped with this limitation varies consider-ably. A few of the articles read like research papers from a conference proceedings or a journal.I found the lack of coordination between the articles rather disconcerting. For example, the intro-duction to some of the articles show too many similarities. One problem of a book of this type, ofwhich the editors are aware and mention in their preface, is that much of the information presentedwill soon be out-of-date. This is certainly true, one particular example being the dramatic re-appraisal of the ideas concerning defect formation and migration in fluorites.I would recommend this book to workers new to the field of solid electrolytes.To the extentthat the book provides a reasonable first impression of the field and its breadth the editors could beconsidered as achieving their aim. I regard much of the book as equivalent to the proceedings of aconference held on this subject in 1977. It provides a very valuable source of references and anindication of the direction that research was moving at that time. As with conference proceedings,I expect that the major purchasers will be libraries and not individual scientists.A. V. CHADWCKReceived 4th April, 1979Comprehensive Chemical Kinetics. Vol. 20. Complex Catalytic Processes. Ed. by C.H. BAMFORDand C. F. H. TIPPER. (Elsevier, Amsterdam, 1978). Pp. xiis-413. Price $SS.OO/Dfl. 195.00.This volume consists of only three chapters. G. Webb writes on hydrogenation, K. van derWiele and P. J. van den Berg on oxidation processes and L. Berhnek and M. Kraus on heterogeneouseliminations, additions and substitutions. The preface by the editors describes the contents of thepresent volume but does not indicate their intentions about the coverage to be given to other complexcatalytic processes such as reforming, Fischer-Tropsch synthesis and related reactions and hydro-desulphurization. One presumes that some of these topics may be covered in later vollumes inthe section on " Kinetics and Technological Processes " but it would be helpful to be given, at thisstage, more of the rationale for the division of subjects between this volume and later volumes.The chapter on hydrogenation is interesting and clear but it is definitely not as comprehensiveas one would expect it to be.There is a detailed account of the hydrogenation of alkenes, alkynes22REVIEWS OF BOOKS 221and alkadienes but very little on the hydrogenation of aromatic hydrocarbons and nothing on thehydrogenation of molecules containing hetero-atoms or on the hydrogenation of larger unsaturatedmolecules of industrial importance. Nevertheless, Dr. Webb has an interesting kinetic story to tellwith respect to the hydrogenation, isomerisation and exchange with deuterium of the C4- and C5-olefins and he tells it well.The authors of the chapter on oxidation processes have reduced a large amount of publishedmaterial to a readable and balanced presentation.They made the decision to concentrate mainlyon papers published in the period 1970-76 because reviews had covered the earlier material. Itcould be argued that a preferable alternative would have been to adopt a more ruthless attitude torecent work and produce an account which covered the development of the subject of heterogeneouscatalytic oxidation over a longer time scale. The references in this chapter are listed in alphabeticalorder of the authors whereas in the other two chapters the references are numbered according totheir first mention in the text.Drs. Beriinek and Kraus have brought together a number of subjects, often considered separately,in the final chapter on eliminations, additions and substitutions.They have stressed the parallelbetween heterogeneous reactions taking place on acid-base sites on solid surfaces and the correspond-ing reactions in homogeneous solution and their chapter is particularly strong on the correlation ofthe results in heterogeneous catalysis with concepts widely used in physical organic chemistry. Thischapter will certainly prove to be of general interest.The final impression is that, while the volume contains material of high standard with chaptersby recognized experts, one is disappointed that the coverage of the whole subject is not a little morecomprehensive. But, on the other hand, a wider coverage would have entailed a less detailedtreatment.The price is not cheap.C. KEMBAUReceived 1 8 th April, 1979Synthesis and Properties of Low Dimensional Materials. Ed. by J. S. MILLER and A. J. EPSTEIN.(New York Academy of Sciences, N.Y., 1978). Pp. 828. Price $80.00.This book contains the papers and discussion of a conference on the synthesis and properties of lowdimensional solids, arranged by the New York Academy of Sciences in June 1977. The term " lowdimensional " is used for solids which are highly anisotropic in their electrical, optical and magneticproperties and sometimes in their mechanical properties. Solids loosely described as " onedimensional " usually have chain-like or linear structures and typical members are K.C.PTTF-TCNQ, (SN), polymer and its halogen derivatives, polydiacetylene, polacetylene and itshalogen derivatives and compounds such as NbSe3 and Hg,.,, as F6.There is widespread interestin these materials and most readers will now be familiar with the abbreviations used. The layermaterials are referred to as " two dimensional " and again typical members are Ti&, NbSe2, MoS,.Sadly there is no account in this book of solids which might be called " zero dimensional " and whichwould include the cluster systems. These systems are of considerable theoretical and practicalimportance.The book consists of a short introduction by J. S. Miller and is followed by sixty papers gatheredinto four groups. These are (i) general papers (16), (ii) organic materials (17), (iii) inorganic materials(17) and (iv) covalent materials (10).The papers lean heavily to what might be called the more" chemical aspects ", by which I mean the preparation, and a bewildering number of new materialshave been synthesized, the characterisation and structure determination. Having said this, there aresome accounts of the optical, electrical and magnetic properties. It is perhaps not fair to single outindividual papers for special mention, since all the papers are good. I particularly enjoyed the papersby Chaikin who reviewed the physics of low dimensional systems, Epstein, Conwell and Miller whodiscussed charge transport in molecular conductors and Yoshimura and Murukami who gave apaper on solid state reactions in organic conductors and their technological applications.All thesewere on one-dimensional systems. The layered solids were mainly covered by Gamble in an excellentpaper, in which he brought out not only some of the basic ideas in this field, but also the possibleapplication of solids such as Ti& as a cathode material in high energy density lithium batteries.This selection of papers, however, reflects my own interests.This book should have a strong appeal to solid state chemists. It is packed with information onthe preparation and structure of many interesting new materials. It also deals with some of the basicphysical ideas in a fairly simple way. Many readers will also want to read about some of the possiblepractical applications of these solids.A. D. YOFFEReceived 23rd April, 197222 REVIEWS OF BOOKSTopics in Surface Chemistry.Ed. by E. KAY and P. S. BAGUS. (Plenum Press, New YorIc, 1978).Pp. viiiS-408. Price 526.77.This book contains the fifteen papers presented at a conference organized by the I.B.M. corpora-tion, at Bad Neunahr, West Germany, in September 1977. It is the ninth volume in the I.B.M.Research Symposia series and the first to deal with surface science. The stated aim was to “ bringtogether key scientists who appear to be asking related questions about the role of surface structureand phenomena ”. The published volume offers snapshots of a number of topics, each of which isinteresting and important. The areas considered are often so diverse, however, that there appearto be few points of contact between groups of papers.There are three contributions in each of five sections.Under the general heading of “ SurfaceStudies in Electrochemical Systems ”, the book opens with Memming’s lucid account of reactions onsemiconductor electrodes. Emphasis is placed on the band structure of the electrode. Sass’discussion of photoemission into electrolytes emphasizes the need for an interdisciplinary approach.Topics covered include interfacial plasmons, energy dissipation in polar liquids and solvent de-composition by photoholcs. Bewick and Fleischmann consider the formation of surface compoundson electrodes. The study of electrochemical kinetics, which is treated first, is now well establishedand has provided much basic information. More recent methods for examining the structure ofelectrodeposits and phase transformations in 2-dimensional systems are considered.Mobius’ article on monolayer assemblies opens the section on “ Ordered Arrays of OrganicMolecules at Surfaces and Interfaces ”.He reviews techniques for production and manipulation ofmixed monolayers such as oscacyanine and thiacyanine and considers electrical conduction in suchsystems. Gratzel considers the application of laser photolysis and pulse radiolysis to the study ofrapid processes in micellar assemblies. He suggests that functional surfactant micelles may beapplied to the conversion of light into chemical energy. Tiecke and Wegner report on the polymer-ization of various diacetylene monocarbonic acids in monolayer and niultilaysr configurations.Conditions for optimum stability of the layers and the methods of characterization are detailed.The section on atomic and molecular scattering from surfaces demonstrates the substantialrecent progress in this area.Wilsch provides a detailed account of the theory of elastic gas-solidinteractions together with some experimental details. Cavallini’s article on reactive scatteringconcentrates on carbon monoxide oxidation at palladium surfaces. He reviews other recent studiesand describes his own ambitious approach, combining a twin supersonic beam system with secondaryion mass spectronietery (SIMS). Gerlach-Meyer and Hulpke compare theoretical and experimentalcurves for the impulsive scattering of lithium ions from nickel and tungsten single crystals.Under the general heading “ Aspects of Surface Chemical Bonding ”, Bradshaw and Menzelhighlight the respective strengths of X-ray and ultraviolet photoelectron spectroscopy in providingchemical information.Ednionds and McCarroll give an impressive list of examples to show thatthe techniques of surface physics can assist in the understanding and development of catalysts. Theyconcentrate on the action of promoters, in particular alkali metals. Andersson’s article on “ Spectro-scopy of Surface Vibrations ” describes the high resolution electron energy loss technique, currentlyone of the most exciting developments in surface science. Results on the adsorption of hydrogen,deuterium, oxygen and carbon monoxide are presented.Luth opens the final section, on optical excitations at surfaces, with an exposition of surfacephotovoltage spectroscopy.Applications to the study of germanium, silicon and gallium arsenidesurfaces are discussed and the study of semiconducting organic films illustrated. Philpott’s articledemonstrates the power of reflection spectroscopy in the study of organic single crystals such asTCNQ, phenanthrene and anthracene. Lastly, Hendra and Fleischmann review the rather dis-appointing progress made in the study of surfaces by Raman spectroscopy. This article has nobibliography .The standard of the contributions is high both in content and presentation. There are manyreferences to original literature and there is a well prepared subject index. The range of subjectmatter is, however, so broad that the whole is rather less than the sum of its parts.The scope is toowide to commend the book to specialists, while the treatment of the subject matter is often verydetailed for the general reader.R. W. JOYNERReceived 3rd May, 197REVIEWS OF BOOKS 223Chemistry and Physics of Carbon, Vol. 14. Ed. by P. L. WALKER and P. A. THROWER. (MarcelDekker, hTew York, 1978). Pp. ixS-309. Price $34.75.The present volume in this series continues the task of earlier volumes to help its readers to becomeand to remain well informed about carbon and graphite. One general comment is that the SubjectIndex seems disproportionately short as a guide for casual consultations to such a rich field ofinformation. Indeed, a comprehensive Subject Index to the series might be of real value havingregard to the complex terrain.Chap.I (82 pp., 37 fig., 96 ref.) gives general illustrations of the astonishing advances in resolvingpower achieved by electron microscopy. As is well known, " aromatic " resonance bonding andother types of bonds between carbon atoms are remarkably adaptable. They are both strong andstable and produce some of the toughest solids known. Distinctive methods of electron microscopyhave been developed for studying general texture as well as for identifying elementary structures incarbons. Lattice imaging of individual sheets of carbon atoms using defocusing procedures is nowa common achievement. With these techniques a recurrent question is how far " images of ostensibIyisolated sheets such as appear in carbon fibres and vitreous carbons " are reliable quantitatively.The authors (G.R. Millward and D. A. Jefferson) provide a very fair view of what can be done, butdo not present any very critical instances of what are possible limitations of these methods.Chap. I1 (83 pp., 30 fig., 95 ref.) surveys the formation of carbon filaments on solids, througli thedisproportionation of carbon monoxide, or catalysed decomposition of hydrocarbons. Somebeautiful illustrations of unusual carbon filaments and useful descriptions of techniques of studyare given by the authors (R. T. K. Baker and P. S . Harris). Despite the technical significance of thisform of carbon deposition-, e.g. in certain uses of catalysts, our existing knowledge about mechanismsand kinetics of formation remains disappointingly superficial.Chap.I11 (126 pp. 38 fig., 280 ref.) deals with the formation of carbon blacks. This is one of themajor fields of carbon science and technology, and the present account (by J. Lahaye and G. Prado)links up with a 1965 review by Palmer and Cullis in vol. 1 of this series. An abiding source of com-plications is that " many carbonaceous deposits consist of a mixture of surface-formed and gas-formedcarbons ". Modes of growth (and of nucleation in the case of gas-formed carbons) are discussed ininformative detail. Inevitably, carbon blacks formed by way of reactions in flames must passthrough viable routes of flame chemistry and flame stability. This consideration may narrow downwhich chemical regimes are considered as precursor to the separation of carbon blacks.Otherprecursor regimes which by-pass flames may give rise to blacks with interesting differences of textureand structure, for example " carbon flocs " formed behind detonation waves in fuel rich mixtureswith oxygen. Not all such variants are included in the present text, though chap. I11 gives a usefulview of many novel developments since the 1965 survey.A. R. UBBELOHDEReceived 21st May, 1979Principles of Solution and Solubility. By K0z6 SHINODA. Translated from the original Japanese byPAUL BECHER. (Marcel Dekker, New York, 1978). Pp. x+222. Price S.Fr.42.Those who are already familiar with the books on solutions and solubilities of non-electrolytesand " regular " solutions by Joel H. Hildebrand (who contributes a foreword to this one by his formerpupil) and his colleagues will not find much that is new to them here, except some of chap. 9 onsolutions of surfactants and some of chap. 10 on " solubilization ", fields to which Professor Shinodahas made independent contributions. The rest, though no doubt useful enough in its originalJapanese, hardly justifies this translation into English. The chapters are brief and introductory,but introduce only one approach to the theory of solutions; an approach that is at best old and atworst outdated. Whatever one thinks of this kind of theory, when Professor Hildebrand expoundsit he always allows experimental results to dominate his discussion; Professor Shinoda is moreinclined to wave his arms about. The book has been translated into readable English; it is reproducedfrom " camera-ready " typescript.The best advice I can give to anyone who wants to read about " solubility-parameter " theory andthe like is to buy instead Regular and Related Solutions by Joel H. Hildebrand, John M. Prausnitzand Robert L. Scott (Van Nostrand Reinhold, New York, 1970) which is only six pages longer andat least comes from the (now 98 years old) fountain-head himself.M. L. MCGLASHANReceived 24th May, 197

ISSN:0300-9599    

DOI:10.1039/F19807600220

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

J.C.S. Faraday I, 1980,76, 225-235Investigation into the Existence of Edge-Face CoagulatedStructures in Na-Montmorillonite SuspensionsBY BRIAN RAND* AND ERDOGAN P E K E N ~ ~Department of Ceramics, Glasses and Polymers,University of Sheffield, Elmfield, Sheffield S10 2TZANDJAMES w. GOODWIN AND ROBERT w. SMITHSSchool of Chemistry, University of Bristol,Cantock’s Close, Bristol BS8 1TSReceived 8th March, 1978The rheological and coagulation characteristics of Na-montmorillonite suspensions have beeninvestigated as a function of pH and NaCl concentration to ascertain whether edge-face coagulationoccurs.Contrary to the current hypothesis of particle interactions in this system, dilute suspensions(< 1 %w/w) were Newtonian liquids and showed no tendency to coagulate over the pH range 4-11at NaCl concentrations below 5 x mol dm-3.At higher NaCl concentrations, however, co-agulation and Bingham flow behaviour were observed. There is no evidence for edge-face inter-action in this pH range ; the coagulation produced by NaCl additions is considered to be initially aresult of the edge-edge type of particle interactions with face-face coagulation occurring at NaClconcentrations >0.3 niol dm-3. The extrapolated shear stresses of the salt contaminated systemsexhibited a minimum in the pH range 7-9.When the concentration of Na-xnontmorillonite was increased above 1 % WIW, at NaCl con-centrations below rnol dm-3, the suspensions became pseudoplastic and a shear modulusbecame measurable. The shear modulus and the extrapolated shear stress both increased expo-nentially with clay concentration.These effects are attributed to the interpenetration of electricaldouble layers because the addition of small amounts of NaCl, by compressing the electrical doublelayers, markedly decreased the magnitude of both parameters. However, increasing the NaClconcentration above 5 x mol dm-3 brought about edge-edge coagulation demonstrated by anincrease in the shear modulus, the appearance of a yield stress in the flow curves and the onset ofthixotropic character.The concept of electrostatic attractive forces between the positively charged edgesof clay particles and the negatively charged faces, at pH values below the edge iso-electric p ~ i n t , l - ~ has been applied successfully to kaolinite dispersions to explain theinfluence of electrolytes on the rheological characteristics4’ and has also led to amethod of estimating values of the edge isoelectric points of different kaolinitesample^.^ However, the ability of this model to account for the behaviour of sodium-montmorillonite suspensions at low ionic strengths has recently been questioned byCallaghan and Ottewill who support the contention of Norrish that the gelation ofthis particular type of clay may be due to the mutual repulsion of the particles as aresult of interactions between their electrical double layers.van Olphen 1-3 showed that the extrapolated shear stress of a Na-montmorillonitedispersion is large at low ionic strengths.It decreases sharply on addition of a smallamount of NaCl, supposedly as the edge-face linkage is weakened, and then increasest Present address : Eczacibasi Ceramic Company, Istanbul, Turkey.Present address : Matthey Printed Products Ltd, Burslem, Stoke-on-Trent.1-8 22226 EDGE-FACE INTERACTIONS IN PLATELET SUSPENSIONSagain at higher electrolyte concentrations as a result of edge-edge coagulation.However, in kaolinite suspension^,^ at pH values below the edge surface isoelectricpoint, it has been shown that, once the edge-face structure is destroyed by electrolyteadditions and the electrical double layers at the edges are so compressed that there isno impediment to edge-edge coagulation, the extrapolated shear stress remainsThis is because the interparticulate forces which bring about edge-edge coagulationare van der Waals in nature rather than the stronger electrostatic forces in the edge-face coagulated system.There appears to be no satisfactory explanation for thedifferent behaviour of the montmorillonite. Indeed, although a number ofstudies 5 9 9-11 have established the effects of pH and electrolyte on the rheologicalcharacteristics of kaolinite dispersions, no detailed studies of the pH dependence ofthe behaviour of carefully prepared Na-montmorillonite systems have so far beenreported, Natural bentonite clays have been investigated l2 and exhibit a minimumextrapolated shear stress in the mid pH range, which has not been adequately ex-plained either. These latter materials contain appreciable quantities of electrolytesand are thixotropic.The present investigation was undertaken to ascertain whether edge-face struc-tures do exist in homoionic Na-montmorillonite suspensions, by studying the effectof NaCl on the rheological behaviour at different pH value^.^ It was anticipatedthat if this technique confirmed van Olphen’s model it would also enable the edgeisoelectric point to be determined. Measurements were carried out at solids con-centrations from 0.7 to 5.0 % w/w.EXPERIMENTALMATERIALSTwo batches of homoionic Na-montmorillonite were prepared from WyomingBentonite samples by the method of van Olphen,l following the removal of coarse mineralimpurities by sedimentation and the oxidation of any organic matter by H202 treatment.The cation exchange capacities of the two samples, A and B, were respectively 93 and100 meq.per 100 g dry clay. The batch A sample was investigated by microelectrophoresisat 25°C and an ionic strength of mol dm3. The batch B sample was similar to thatstudied by Callaghan and Ottewill.6CONTINUOUS SHEAR MEASUREMENTSThe dependence of shear stress on rate of shear (flow curve) for the batch B suspensionsin the range of solids concentrations 0.9-5 % w/w was studied at a pH of 7.3k0.2 and aNaCl concentration of 4.6 x mol dm-3, using a Weissenberg rheogoniometer fittedwith 10 cm diameter Mooney type combined concentric cylinder, cone and plate platens.A range of shear rates of 2-700 s-l was employed.Suspensions of the batch A clay were investigated at solids concentrations of 0.74 and3.3 % w/w over a wide range of pH and NaCl concentration using the Contraves Rheomat30 viscometer.A cone and plate attachment (angle 31’59”, maximum shear rate 3950 s-l)was used to study the 3.3 % samples, whereas the flow curves for the dilute suspensions wererecorded using a hollow cylinder rotating in a coaxial annulus (maximum average shear ratein the 1 mm gaps was 1750 s-l). The ascending and descending branches of all flow curveswere recorded automatically on an X- Y recorder, allowing a period of 4 min for each branchto be traversed. Suspensions were first mixed under high shear conditions to fully dispersethe montmorillonite; they were then adjusted to the required pH value using NaOH orHCl, remixed and allowed to equilibrate for 24 h before recording the flow curve.NaCl wasthen added in increments, flow curves and pH values being recorded again after equilibrationat each concentration. Care was taken to minimise structural disruption whilst placing thesamples in the measuring systemB . RAND, E . PEKENC, J . w. GOODWIN AND R . w. SMITH 227SHEAR MODULUS MEASUREMENTSElasticity measurements of gelled samples were made using a parallel disc "pulseshearometer "l3, of similar design to that used by van O l ~ h e n . ~ Rapid, but small(z rad), electrically induced torsional displacements of one disc relative to the otherinitiated shear waves which travelled through the gel at a wave velocity that was measuredwith an oscilloscope. The disc separation was maintained constant and the torsional dis-placement was too small to produce measurable structural disruption.The wave frequencywas ~ 2 0 0 Hz. Since the storage part of the dynamic modulus was much greater than theloss component, the high frequency limit of the shear modulus, G, was calculated from :G = pv2where p is the density of the gel and is the velocity of the shear wave. Thixotropic sampleswere measured as a function of time after thorough stirring.COAGULATION EXPERIMENTS0.055 % w/w suspensions of batch A clay were made up at various pH values and thestate of coagulation observed over a period of two weeks. NaCt was then added to bringthe concentration of Naf to 0.06 mol dmW3 and the experiments repeated.RESULTSThe electrophoretic data were similar to those of Swartzen, Allen and Matijevic,lshowing a constant negative mobility in the pH range 4-9 with a sudden increase inmagnitude (40-50 %) at higher pH values.Table 1 summarises the coagulation dataon the dilute suspensions. Before the addition of electrolyte they were stable overthe pH range 5-1 1 , but exhibited coagulation and sedimentation at pH values below 2and above 12. However, at a NaCl concentration of 0.06 mol dm-3 rapid coagulationand sedimentation was observed at all pH values.TABLE STATE OF COAGULATION OF (0.05 % w/w) MONTMORILLONITE SUSPENSIONSpH of suspension pH of suspension(NaCl conc initially state of (NaCl conc state ofmol dm3) coagulation 0.06 mol drr3) coagulation~~1.42579.3511.012.012.6rapidcoagulationslowcoagulationstablestablestablestableslowcoagulationrapidcoagulation2 rapid5 rapid7 rapid9.35 rapid11.0 rapid12.0 rapidcoagulationcoagulationcoagulationcoagulationcoagulationcoagulatio228 EDGE-FACE INTERACTIONS I N PLATELET SUSPENSIONSThe rheological measurements at low NaCl concentrations showed that the0.74 % w/w batch A clay suspensions were Newtonian in the pH range 4-11.However, at pH values below 4 the flow curves resembled the ideal Bingham curvewith a small thixotropic loop, whereas at pH 11.5 an anti-thixotropic effect wasapparent and the initial, ascending shear rate, branch was shear thickening.It wasalso difficult to obtain reproducible results at these high pH values.If sufficientNaCl was added, the flow curves obtained at all pH values below 11 could be trans-formed to the Bingham type, but again the behaviour above pH 11 was anomalousshowing shear thinning character and a wide hysteresis loop.PHFIG. 1.-Effect of pH and NaCl on the extrapolated shear stress of 0.74 % w/w sodium montmoril-lonite suspensions (batch A). 0, ; 0,0.017 ; 0, 0.034 ; A, 0.12 ; A, 0.20 ; M, 0.30 mol dmd3NaCI.Since most of the curves are good approximations to the Bingham system, theycan be characterised in terms of the extrapolated shear stress and plastic viscosity.The extrapolated shear stress increases with increasing NaCl concentration up to0.034 mol dm-3 (fig. 1) and is at a minimum in the pH range 7-9.Successive ad-ditions of electrolyte gradually lowered the pH value and it may be significant, asdiscussed later, that those points marked “ a ” in fig. 1 had all previously been madeup at pH values of 11 or greater. At pH values below 11, the variations in plasticviscosity were < = 10 %, regardless of NaCl concentration.Increasing the clay concentration at very low NaCl concentration changes theflow behaviour from Newtonian character to pseudoplastic [i.e., a time independentdecrease in viscosity with increasing rate of shear and no definite yield stress, fig. 2(A)and (B)]. As is discussed later, it is doubtful whether extrapolating the high shearrate part of these flow curves to determine the extrapolated shear stress has any realsignificance as a measure of the extent of coagulation.Nevertheless, this operationhas been performed (a) in order to demonstrate the deviations from Newtoniancharacter (i.e., at a clay concentration > 1 % w/w for batch B clay) and (b) to make adirect comparison with the results of van Olphen’s experiments. Fig. 3 shows thatboth the extrapolated shear stress and the plastic viscosity of batch B suspensionsincrease exponentially with clay concentration along with the shear modulus, whicB . RAND, E. PEKENC, J . w. GOODWIN AND R. w. SMITH0shear stress/N m-240 802290 40 0 40 0 40shear stress/N m-2FIG. 2.-Typical flow curves of 3.3 % w/w sodium montmorillonite suspensions (batch A). pH = 7.(A) ; (B) 4 x ; (C) 0.046 ; (D) 0.6 mol dm-3 NaCl.I 1 I I 1 I 10 2 4 6clay concentration/ % w/wFIG.3.-Variation of rheological properties of sodium montmorillonite suspensions (batch B) onsolid concentration at pH = 7.3 and NaCl concentration = 4.6 x mol dm-3. A Extrapolatedshear stress ; Ay plastic viscosity ; 0, shear modulus, Go23 0 EDGE-FACE INTERACTIONS IN PLATELET SUSPENSIONSwas also independent of time at this NaCl concentration. This latter parameter andthe extrapolated shear stress show almost identical dependence on concentration, bothincreasing more rapidly than the square relationship observed for kaolinite suspensionsand other systems.11* 1 6 p l7 An order of magnitude change is observed for a 1 %increase in concentration.I I I t I0 4 8NaCl concentration/m.tnol dm-3FIG.4.-Effect of pH and NaCl on the extrapolated shear stress of 3.3 % w/w sodium montmorillonitesuspensions (batch A) : 0, pH 5 ; A, pH 7 ; 0, pH 10.N 1 1i3zs.- 588.um -.r( + - 5 aI I I I I I I0 4 8NaCl concentration/mol dm-3FIG. 5.-Effect of pH and NaCl on the plastic viscosity of 3.3 % w/w sodium montmorillonitesuspensions (batch A) : 0, pH 5 ; A, pH 7 ; 0, pH 10.The effect of NaCl on the flow curves of 3.3 % w/w suspensions of batch A claywas investigated at pH values of 5, 7 and 10. Initially, as the NaCl concentration isincreased, the extrapolated shear stress decreases to a minimum (at 5 x mol dm-3)and then increases again (fig. 4) as was observed by van Olphen. The effect is ap-parent at all three pH values and, at any particular NaCl concentration in this range,the extrapolated shear stress is greater the lower the suspension pH.However, overthe same range of NaCl concentration the plastic viscosity decreases continuously(fig. 5 ) ; it remains constant between and 0.4 mol dm-3 (fig. 7), decreasing sharplyagain at higher concentrations. A definite yield stress (z,) is developed at NaCB. RAND, E. PEKENC, J . w. GOODWIN AND R . w. SMITH 231mol dm-3 [fig. 2(C)] and the degree of curvature of concentrations above 5 xthe flow curve is reduced. The data can be fitted to the power lawand fig. 6 shows that n decreases from 2 towards unity as the NaCl concentration isincreased (i.e., the flow behaviour tends towards Bingham flow). In contrast to theextrapolated shear stress, which is comparable to van Olphen's apparent Binghamyield stress, this yield stress (7,) is initially zero. It rises to a maximum at 0.1-0.3 moldm-3 (fig.7) and then falls to a further constant value. At NaCl concentrationsbelow 0.5 rnol dm-3, much higher yield stresses are exhibited at pH 5 than at pH 7,but as the NaCl concentration is further increased they become identical at the twopH values. Such a wide range of NaCl concentration was not investigated at pH 10.Z-z,, = kj"3 . 43 . 0*?-M 0c(2 . 62.2I I I I 1 I1.2 1.6 2.0log (7 - 4FIG. 6.-Application of the power law to 3.3 % w/w sodium montmorillonite suspensions (batch A) :0, pH 5 ; A, pH 7 ; a, pH 7 ;4 x V, pH 7 ;0,010 rnol dm-3 NaCI; n = 1.66. 0, pH 7 ; 0.036 mol dm--3 NaCl ; n = 1.26.0, pH 10;mol dm-3 the clay suspensions becomethixotropic. A hysteresis loop appears in the automatically recorded flow curves ofbatch A suspensions and shows a sharp increase in area at NaCl concentrationsgreater than 0.4 mol dm-3 (fig. 2). The shear modulus, G, of batch B suspensionsalso becomes time dependent under these conditions. It can be described by theempirical expressionwhere a is a constant and Go is the value obtained by extrapolation to zero time.mol dm-3 NaCl ; n = 2.0. mol dm-3 NaCl ; n = 2.0.mol dm-3 NaCl ; n = 1.60. mol dm-3 NaCl ; n = 1.76. 0, pH 7 ; 6 xmol dm-3 NaCl ; n = 2.0.At salt concentrations greater thanG = Got232 EDGE-FACE INTERACTIONS I N PLATELET SUSPENSIONSTable 2 shows that, like the extrapolated shear stress, Go also falls from its maximumvalue to zero as the NaCl concentration is increased, but increases again when thethixotropic character appears.0 0 .4 0.8NaCl concentration/mol dm-3FIG. 7.-Dependence of yield stress and plastic viscosity of sodium montmorillonite suspensions(batch A) on NaCl concentration : 0, yield stress, pH 7 ; A, plastic viscosity, pH 7 ; 0, yield stresspH 5 ; A, plastic viscosity, pH 5.TABLE 2.-RESIDUAL SHEAR MODULUS, Go, OF Na-MONTMORILLONITE SUSPENSIONS(BATCH €3) AS A FUNCTION OF CLAY CONCENTRATION ANDNaCl CONCENTRATION AT pH 7.3 5 0.2clay concentration NaCl concentration GO/%wlw /mol dm-3 /N m-’ a4.984.804.864.824.834.1 14.464.825.154.6 x 10-41 .ox lo-’3 .0 ~6 . 0 ~1.ox 10-16 . 0 ~ lo-’6 . 0 ~ lo-’6 . 0 ~ lo-’6 . 0 ~15970288378386152247378716000.0480.1000.1140.0970.1170.1000.10B. RAND, E . PEKENC, J . w. GOODWIN AND R . w. SMITH 233DISCUSSIONCallaghan and Ottewil16 have pointed out that the invariant mobility in thepH range 3-9 is due to the predominance of the pH independent surface chargedensity at the basal faces, since the edge areas are small in comparison. It is un-likely, therefore, that the marked increase in mobility observed at high pH values isdue to hydroxyl adsorption at the edges as has been suggested.l5 An alternativehypothesis, which can be used to explain the rheological effects at high pH, is thatthe zeta potential at the basal face becomes pH dependent above a certain pH value.This could come about via an ionization of surface siloxane groupings or there maybe some dissolution from the edge surfaces releasing silicate and aluminate specieswhich subsequently are specifically adsorbed on the basal faces.Over a wide pH range no coagulation can be detected except after addition ofNaCl; that which occurs at very low and very high pH values may also be due tothe increased ionic strength.However, in dilute kaolinite suspensions under con-ditions where the edge-face type of particle interaction is proposed (pH <6), co-agulation is observed even at very low ionic strength. There is, therefore, no evidencefor edge-face coagulation in the pH range 5-11 from these coagulation experiments.In the pH range 4-11 the 0.74 % w/w suspensions exhibit Newtonian flowcharacteristics, again contrary to expectation if edge-face structures are assumed toexist.However, after the addition of NaCl the flow curve approximates to theBingham model, exhibiting a definite extrapolated shear stress. This is doubtless aresult of the normal action of NaCl in compressing the electrical double layers andallowing coagulation to occur, probably in the edge-edge orientation as van Olphen lm3suggests. If an extrapolated shear stress can be measured for the system coagulatedby NaCl it should also have been observed initially, if the system was coagulated inthe edge-face mode. It must therefore be concluded that in the important pHrange 4-11 there is no evidence for edge-face coagulation in these suspensions ofNa-montmorillonite.Fig. 3 shows that the extrapolated shear stress in the presence of NaCl is de-pendent upon the suspension pH and, if the coagulation is due to the edge-edge typeof interaction under these conditions, it would appear that the edge surface doesbehave like that of an oxide, as van Olphen proposed.A lower surface potential,e.g., at low pH, would lead to a higher extrapolated shear stress.16 It may be thatthere is no evidence of edge-face coagulation because the edge surface isoelectricpoint lies at a pH value below 4. If this is the case then, since the edge isoelectricpoint is determined by the proportions of siliceous and aluminous sites, which in turnare determined by the previous history of the clay,5 it is possible that other mont-morillonite samples might show edge-face coagulation at the pH values investigatedhere.However, an alternative and more likely explanation is that of Ottewill l8 whosuggests that, because of the small area of the edge, the electrostatic attraction betweenedge and face is small compared to the repulsion between the two faces when twoparticles approach perpendicular to each other. This repulsion probably does notdominate the net interaction energy in the case of kaolinite particles with their sub-stantial areas of edge surface.Although the dilute suspensions show no evidence of coagulated structures underconditions where edge-face interaction is to be expected, non-Newtonian characterdevelops at higher clay concentrations.Both the continuous shear results withbatch A clay and the shear modulus experiments with batch B clay show similaritieswith those of van Olphen’s earlier experiments (fig. 4 and 5). The observation thatboth the shear modulus and the extrapolated shear stress of the suspensions at ver23 4 EDGE-FACE INTERACTIONS IN PLATELET SUSPENSIONSlow NaCl concentration show similar dependence on clay concentration suggests thatthey are both determined by the same type of particle interactions. However, theminimum in the extrapolated shear stress at 5 x mol dm-3 occurs at all threepH values studied and, if this is attributed to the destruction of edge-face linkages, itis necessary to come to the unlikely conclusion that the isoelectric point of the edgesurface is greater than pH 10.It is more likely that, as the clay concentration in-creases, the particles and their electrical double layers fill the available space and,therefore, the shear modulus is a measure of the repulsive force between particles dueto their interpenetrating electrical double layers. This force is inversely proportionalto the exponential of the interparticle distance, accounting for the strong dependenceof Go on clay concentration. Similar effects have been observed in some concen-trated polymer latex gels in which there was no coagulation.13* l4The marked shear thinning shown in the continuous shear experiments can alsobe attributed to electroviscous effects and the progressive orientation of the defloccu-lated montmorillonite platelets in the direction of flow as the rate of shear is increased.Thus, the extrapolated shear stress also varies exponentially with clay concentration.When the electrical double layer at the basal face is compressed by the addition ofNaCl, the contribution from electroviscous effects is reduced.Hence the flow curveshifts to lower shear stresses, the plastic viscosity and extrapolated shear stress de-crease, whilst the exponent, n, in the power law is unaffected. The shear modulusalso falls to zero as the repulsive forces between particles are decreased.The effect of increasing the NaCl concentration above 5 x mol dm-3 is tochange the nature of the system markedly. A definite yield stress appears in theflow curves of the 3.3 % w/w batch A clay suspensions ; it increases in magnitude upto a NaCl concentration of 0.1-0.3 mol dm-3 (fig.7) and thixotropic character de-velops (table 2). The shear modulus also increases again in this region. Theseeffects can be attributed to edge-edge coagulation as discussed earlier. They are alsoconsistent with studies of gelation kinetics of bentonite suspensions in which it hasbeen shown that the lower the pH value and the higher the NaCl concentration, themore rapid is the gelation pr0cess.lFig. 1 shows that the extrapolated shear stress increases markedly at pH valuesabove 9 in the presence of NaCl. This is the region where the electrophoreticmobility increases also. It is tentatively suggested that the effect is the result of anincrease in the Stern potential of the basal surface, as discussed earlier, leading toincreased swelling, producing more platelets in suspension.When these coagulateafter the addition of electrolyte, the extrapolated shear stress is higher than in theneutral pH range. The irreversible shear thickening observed at pH 11.5 could alsobe due to the separation of platelets being facilitated by shear. However, it is stronglyemphasised that this explanation has not yet been rigorously tested, since the majoraim of this work was to investigate the existence of edge-face coagulation in the neutraland slightly acid pH range. Further work at high pH is envisaged.van Olphen showed that at very high NaCl concentrations the Bingham yieldstress decreases due to the onset of face-face coagulation which reduces the number ofplatelets in suspension. In accord with his observations, fig.7 shows that at NaClconcentrations around 0.25-0.3 mol dm-3 the yield stress and plastic viscosity bothdecrease ; this is precisely the concentration region below which osmotic swellingoccurs as established by the X-ray measurements of Norrish7 and Foster et uL20The present work has therefore fully established the existence of face-face and edge-edge coagulated structures in the regions of NaCl concentration proposed by vanO l ~ h e n , ~ but no evidence has been found for the edge-face mode of coagulation atpH values >4B . RAND, E. PEKENC, J . w. GOODWIN AND R. w. SMITH 235E. P. acknowledges financial support from the Eczacibasi Ceramic Co., Turkeyduring the period when this work was carried out.H. van Olphen, Introduction to Clay Colloid Chemistry (Wiley and Sons, New York, 2ndedn, 1977).H. van Olphen, Disc. Faraday SOC., 1951,11, 82.H . van Olphen, Proc. 4th Nat. Con$ Clays and Clay Min. (Nat. Res. Council Nat. Acad. Sci.,Washington, D.C., 1956), p. 204.R. K. Schofield and H. R. Samson, Disc. Faradzy SOC., 1954,18, 135.B. Rand and I. E. Melton, J. Colloid Interface Sci., 1977, 60, 308, 321.I. C. Callaghan and R. H. Ottewill, Faraday Disc. Chem. SOC., 1974, 57, 110. ' K. Norrish, Disc. Faraday SOC., 1954, 18, 120. * BS 5168: 1975, Glossary of Rheological Terms.A. S. Michaels and J. C. Bolger, Ind. Eng. Chem. Fundamentals, 1964,3, 14.lo A. W. Flegman, J. W. Goodwin and R. H. Ottewill, Proc. Brit. Ceramic SOC., 1969, 13, 31.l1 S. K. Nicol and R. J. Hunter, Austral. J. Chem., 1970,23,2177.l2 W. F. Rogers, Composition and Properties of Oil Well Drilling Fluids (Gulf Publishing Co.,l 3 J. W. Goodwin and R. W. Smith, Faraday Disc. Chem. SOC., 1974,57, 126.l4 J. W. Goodwin and A. M. Kidher, Colloid and Interface Science, ed. M. Kerker (AcademiclJ S. L. Swartzen-Allen and E. MatijeviC, J. Colloid Interface Sci., 1975, 50, 143.l6 B. A. Firth and R. J. Hunter, J. Colloid Interface Sci., 1976, 57, 257, 266.l7 R. J. Hunter and S. K. Nicol, J. Colloid Interface Sci., 1968,243, 250.l8 R. H. Ottewill, Faraday Disc. Chem. SOC., 1974,57, 129.l9 E. PekenC, Ph.D. Thesis (University of Sheffield, 1976).2o W. R. Foster, J. G. Savins and J. M. White, Proc. 3rd Nat. Con$ Clays and Clay Min., 1954,Texas, 1948).Press, N.Y., 1977, vol. 4).p. 296.(PAPER 8/423

ISSN:0300-9599    

DOI:10.1039/F19807600225

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

J.C.S. Faraday I, 1980,76,236-243Structure of the Lamellar Lyotropic Liquid Crystal,Tetramethylammonium Perfluor o-oct anoat e + W ater ,Determined by Neutron DiffractionBY ROBYN M. HEDGE AND ROBERT K. THOMAS"Physical Chemistry Laboratory, South Parks Road, Oxford OX1 342ANDMICHAEL MORTIMERPhysical Chemistry Department, University of Leeds, Leeds LS2 9JTANDJOHN W. WHITEInstitut Laue-Langevin, X156 Centre de Tri, Grenoble, 38042 Cedex, FranceReceived 23rd October, 1978The tetramethylammonium perfluoro-octanoate + water system forms a lamellar liquid crystallinephase at room temperature. The stability of this phase with respect to temperature and compositionhas been investigated and structural parameters have been measured for several samples. Thelamellar phase is shown to be of the single-layered lamellar type.Samples up to 0.5 mm thick weresuccessfully oriented by shearing between calcium fluoride plates.Addition of water to fatty acid soaps has long been known to cause the formationof a wide range of lyotropic liquid crystalline and gel phases, the resultant phasedepending on the temperature and composition of the mixture (see Ekwall for areview). The lamellar phase of such a liquid crystal consists of layers of aqueoussolution alternating with bilayers of the soap, the interfaces being occupied by thepolar head groups of the amphiphilic soap molecules.Although the amphiphile is mobile, its rate of diffusion is usually two orders ofmagnitude smaller than that of any species in the aqueous layer, making the surfaceof the amphiphile relatively static.The structure of the surface is well defined andbecause the amphiphilic layer is essentially only one molecule thick the surface areaof the interface is very large ( ~ 5 0 0 m2 g-l). Most of this surface can be orientedalmost perfectly in a particular direction. Froin the point of view alone of studyingadsorption from solution and properties of the electrical double layer, the lamellarphase of a lyotropic liquid crystal is therefore an attractive model system. In addition,the surface chemistry of this type of interface is expected to play an important part indetermining the properties of biological membranes. We have used neutron scatteringto study both the structure of the lamellar phase of a lyotropic and the dynamicalproperties of species in the interface region.Neutron scattering is a technique well suited to the study of the dynamical proper-ties of molecules adsorbed at an interface.The space and time scale of the experiment issmall enough to follow the diffusive motion of water over distances smaller than the23HEDGE, THOMAS, MORTIMER AND WHITE 237interlamellar distance. Thus, motion within the aqueous layers can be separated fromlonger range diffusion and there can be no complication in the interpretation of anyanisotropy of the motion. Proton-containing molecules are highlighted by their largecross-sections for incoherent scattering of neutrons. For this reason we have chosen toinvestigate the system water + tetramethylammonium perfluoro-octanoate (TMAPFO)which offers the possibility of using isotopic substitution to study the dynamicalproperties of water and counter ions separately.Also, for this particular compound,there is practically no incoherent scattering from the amphiphilic layer.To interpret neutron scattering experiments on the dynamics of the aqueous layerit is necessary to have a complete set of structural data for the lamellar phase over awide range of concentration and temperature. For this we have used neutron diffrac-tion which has the special advantage, for our purposes, that it can readily be used toinvestigate the orientation of the relatively thick samples (0.5-1 mm) between calciumfluoride plates that were used for experiments on the dynamics of the system.Thispaper describes the diffraction experiments.EXPERIMENTALTMAPFO was prepared by direct neutralisation of perfluoro-octanoic acid (Fluorochem,purity greater than 97 %) with the appropriate amount of tetramethylammonium hydroxidesolution (B.D.H.). Neutralisation was complete in 24h. The TMAPFO was dried overphosphorus pentoxide.A known amount of dry TMAPFO was placed in a Pyrex tube with a narrow constrictionand the required amount of demineralised water added with a syringe. After weighing tocheck the amount of water added, the tube was temporarily sealed and the contents centri-fuged to the bottom. The tube was then sealed permanently. The mixture was centrifugedback and forth through the constriction 20 times, sonicated for 15 min at 70°C and left toequilibrate for at least five days.The samples were clear, colourless, viscous liquids. Afinal check on the composition of the samples was made by analysing them by loss of weighton drying after each experiment has been completed. The composition of samples is giventhroughout this paper by their percentage weight of water.Teflon spacer(1/z mm)I0I \ II \ ings I// lc a l c i u m fluoride50mm - FIG. 1 .-Cell for conducting neutron scattering experiments on oriented lyotropic liquid crystals.Diffraction experiments were done on unoriented samples contained either in sealedsilica tubes or in flat aluminium sachets made of thin aluminium foil. Oriented sampleswere prepared by shearing the liquid crystal between optically flat calcium fluoride platesand then sealing the plates in the frame shown in fig. 1.The diffractometer used for these experiments was the Guide Tube 7H2R at the A.E.R.E.Harwell.This has an incident wavelength of 4.76 rf.238 LAMELLAR LYOTROPIC LIQUID CRYSTALSRESULTS AND DISCUSSIONPHASE DIAGRAM OF TMAPFOA typical diffraction pattern from an unoriented sample is shown in fig. 2. Thesharp peaks at low angles are in positions corresponding to lattice spacings of d,4 2 , d/3, characteristic of 001 reflections from a lamellar structure. At higher anglesthere is a broad weak peak from a fluorocarbon chain-to-chain reflection. Its widthis consistent with liquid like packing in the fluorocarbon layer. The liquid like diffrac-tion peak indicates that the system is in a lamellar rather than a gel phase.0 20 40 60291"FQG.2.Neutron diffraction pattern of an unoriented sample of tetramethylperfluoro-octanoate +water (26 % HzO by weight). ho = 4.76A.Diffraction patterns were measured over the concentration range 19-60 % byweight of water. At all concentrations except 60 % water, there are two 001 re-flections in positions expected for the lamellar phase. There is also a smooth varia-tion in the total spacing of the structure over the whole concentration range (fig. 5).At room temperature the lamellar phase is therefore the stable one. If any intermediatephases exist it is only over very narrow ranges of concentration.Diffraction patterns were also measured for a single concentration (33 % water)over the temperature range 23.5-52°C.At 23.5 and 42°C two peaks were againobserved in positions appropriate for a lamellar phase. The position of the first peakvaries smoothly over most of the range of temperature covered suggesting that thestructure is also lamellar at 52°C.The full range of stability of the lamellar phase has been determined independentlyby Boden and Jackson using optical microscopy and n.m.r. and confirms thatTMAPFO +water is lamellar over the range of temperature and concentration that wehave studied.Most soap +water systems have several different stable phases but previousstudies of perfluoro-soaps have also shown that the perfluoro-compounds have ratheHEDGE, THOMAS, MORTIMER AND WHITE 239Fluorocarbon chains are more rigid than hydrocarbons simple phase diagram^.^^and this may favour the formation of lamellar phases over hexagonal phases.ORIENTATION OF TMAPFO BETWEEN CALCIUM FLUORIDE PLATESThe lamellar phase is readily oriented by shearing between silica plates, the layersfinishing parallel to the plates.In the quasielastic scattering experiments describedin the next paper in this series, silica plates would have conttibuted too high a back-ground. Calcium fluoride, on the other hand, becomes more or less completelytransparent to neutrons with a wavelength X 6.5 A. TMAPFO was found to orienteven better between calcium fluoride plates than between silica plates.00 14 12 20 28scattering angle, 20/"FIG. 3.-Neutron diffraction pattern (0, 20 scan) of 30 % HzO+TMAPFO shear oriented betweentwo calcium fluoride plates.A. = 4,76 A.Fig. 3 shows the diffraction pattern of a shear oriented sample between calciumfluoride plates. The pattern (showing three peaks characteristic of a lamellar phase)is a great improvement on those from unoriented samples.Fig. 4 shows a rocking curve taken by keeping the detector fixed at the angle of the001 reflection and changing the orientation of the sample about the half angle betweenbeam and detector. The lamellar phase of a lyotropic liquid crystal is an assembly ofdomains, the bilayers being parallel to one another within each domain. The orienta-tion of each domain is defined by the director (normal) to the planes. A rockingcurve measures the angular distribution of the directors about their mean. The fullwidth at half maximum of the rocking curve is only 0.8" which shows that the domainswhich are oriented have a very low angular spread of orientation. Moreover, thelow background level shows that practically all the domains are oriented.It was foundpossible to orient samples of up to 0.5 mm in thickness with a spread in orientation ofthe domains of 1" or less240 LAMELLAR LYOTROPIC LIQUID CRYSTALSlb 1.3 103.3 10 5.3 107.3angle of sample to beam, O/'FIG. 4.-Rocking curve on (001) reflection of 30 % H20+TMAPF0 shear oriented between twocalcium fluoride plates. f.w.h.m. = 0.8".STRUCTURE OF THE LAMELLAR PHASEThe position of the 001 reflection gives the total repeat spacing of the lamellarstructure which is just the sum of the thickness of one fluorocarbon and one aqueouslayer.The division of the total spacing between the two layers could, in principle, bedetermined from the intensities of the different 001 reflections in the diffraction pattern.Unfortunately only the intensity of the first reflection can be measured with anyaccuracy, the second and third being very weak. Even by making use of differentH20/D20 ratios to vary the contrast it seems unlikely that anything but a crudeestimate could be made of the thickness of the individual layers using diffraction peakintensities.Better estimates of the structural parameters can be made using the total repeatspacing in conjunction with the concentrations and partial specific volumes of thefluorocarbon and aqueous layer, respectively.' The accuracy of this method dependson the accuracy of estimation of the partial specific volumes of the two components.The density of tetramethylammonium chloride solutions in water varies from0.9983 to 1.0037 g ~ m - ~ for a change in concentration of 0-60 % by weight of salt.8A reasonable estimate of the partial specific volume of the aqueous layer, Va, istherefore 1.0 cm3 g-l. The estimation of the partial specific volume, V,, of thefluorocarbon component is more difficult.Tiddy has measured the density of solutions of ammonium perfluoro-octanoate(APFO) at different concentrations.By extrapolating to zero water content he ob-tained a value of 0.61 JrO.001 cm3 g-l for the specific volume of APFO.What isneeded for the present calculation is the volume of the perfluoro-octanoate anion only.Unfortunately, there is no experimental method of separating the contributions ofcation and anion. From the ionic radius of the ammonium ion its contribution tothe specific volume is z 15 cm3 mol-l. Taking this figure with an uncertainty oHEDGE, THOMAS, MORTIMER AND WHITE 241100 % leads to a molar volume of the perfluoro-octanoate ion of 247f 18 cm3 inol-I.This figure has been used to calculate the structural parameters given in table 1 anddisplayed in fig. 5.The thickness of the fluorocarbon layer varies between 13.8 and 16.1 A, decreasingwith increasing water concentration. To maintain a constant density of the fluoro-carbon layer the surface area per head group has to show a corresponding increase.Similar calculations were made of the variation of the structural parameters withtemperature. They are given in table 2 and fig.6. As the temperature increases sodoes the area per head group. These calculations have an additional uncertainty tothose for varying concentration because the variation of the specific volume of theperfluoro-octanoate ion with temperature is not known.60-4020TABLE 1 .-SUMMARY OF STRUCTURAL PARAMETERS OF TMAPFO+ H20 AT ROOM TEMPERATURE----weight mol% H20 fraction H20 dtot/A @f df/A dw/A S/A’1921.725.630.333.735.339.340.540.943.147.649.054.360.00.8640.8820.9030.9220.9320.9370.9460.9480.9490.9540.9610.9630.9700.97628.3 k 0.229.7 +_ 0.230.8k0.232.5k0.234.6k0.236.0+ 0.238.3k0.239.2f0.340.0k0.340.8k0.343.5 & 0.445.2+ 0.750.7f 1.058.95 1.30.568+0.0160.542+0.0160.506+0.0160.46440.0160.435+ 0.0160.422 k 0.01 60.389+0.0160.379 f 0.01 60.376k0.0160.359k 0.0150.32450.0150.314+0.0140.275+ 0.01 30.2355 0.01216.1k0.616.lf0.615.650.615.140.615.1f0.615.2k0.714.9+ 0.714.950.7l5.0+ 0.714.6k0.714.1k0.814.2+ 0.913.9-1-0.913.85 1.012.22 0.813.6f0.815.2f0.817.4f 0.819.5k0.820.8k0.923.4k0.924.3 k 1 .O25.0f 1.026.2+ 1.029.4$. 1.231.0$- 1.636.8-t- 1.945.1 k 2.351+551+553+654+ 655+ 654+ 655+ 655+ 6553- 656+ 658+758+759+ 860+ 8t d,01 1 I0.85 0.90 0.95mol fraction waterFIG.5.-Structural characteristics of the water+ tetramethylaminonium perfluoro-octanoate liquidcrystal as a function of mol fraction of water. S = surface area occupied by one headgroup (A”),d~ = total repeat spacing (A), dw = thickness of aqueous layer (A) and d: = thickness of fluoro-carbon layer (A)242 LAMELLAR LYOTROPIC LIQUID CRYSTALSTABLE 2.-TEMPERATURE VARIATION OF STRUCTURAL PARAMETERS OF A 33.5 Wt. % H2OfTMAPFO SAMPLET/"C dt,Jli df/A dw/A S/A2(k0.2) (k0.6) (k0.8) (&6)23.5 34.6 15.1 19.5 5532 34.4 15.0 19.4 5542 34.1 14.8 19.3 5552 30.6 13.3 17.3 6160402c0-0-0 /so'-oyd: 0-30 40 50temperature/"CFIG. 6.-Structural characteristics of the water + tetramethylammonium perfluoro-octanoate systemas a function of temperature.Symbols are the same as in fig. 5.05.3A-"ta + 5.3 aa 0 a 0FIG. 7.-Body centred square packing arrangement of head groups in the HzO+TMAPFO liquidcrystal. 0, headgroup ; 0, tail of interdigitated chains. Area per headgroup = 56 A2. Chain tochain spacing = 5.3 A. First reflection = 5.3 AHEDGE, THOMAS, MORTIMER A N D WHITE 243The extended fluorocarbon molecule is = 13 A long. This is comparable with thethickness of the amphiphilic layer, suggesting either that the molecules are interdigitedin this layer or that the molecules are coiled up and pack end to end to form a bilayer.The weak equatorial reflection from the fluorocarbon chains (fig. 1) occurs at an anglecorresponding to a spacing of 5.3 A.This could result from either a square array withan interchain distance of 5.3 A or a hexagonal array with an interchain distance of6.1 A. The area occupied per head group would then be either 28 or 32 A2, abouthalf that observed. This indicates that the fluorocarbon molecules are both ex-tended and interdigited. The simplest arrangement is then a square array as shownin fig. 7 with an area per head group of 56A. It is not possible to have an ordeiedinterdigited array with hexagonal symmetry but it is possible that the interdigitedamphiphiles form a disordered two dimensional array with approximate hexagonalpacking. This would also be consistent with the broad equatorial reflection shownin fig. 1.found in theliquid crystal. This suggests that the chains pack almost at right angles to the layersbecause any tilt would lead to a lower interchain spacing. This is also consistent withthe thickness of the layers which corresponds to about the length of a single amphiphile.The interchain spacing in Teflon is 5.66 A,8 larger than the 5.3R. M. H. thanks CSIRO (Australia) for a grant. We also thank the S.R.C. for agrant.P. Ekwall in International Review of Science, PhysicaZ Chemistry (Butterworth, London, 1975),series 2, vol. 7 .B. C. G . Haywood and D. L. Worcester, J. Phys. a, 1973,6,568.N. Boden and P. H. Jackson, personal communication.G. J. T. Tiddy, J.C.S. Faraday I, 1972,68,608.E. Everiss, G. J. T. Tiddy and B. A. Wheeler, J.C.S. Faraday I, 1976,72,1747.D. L. Worcester and N. P. Franks, J. MoZ. Biol., 1976, 100, 359.Press, New York, 1968).R. Penninchx, Compt. Rend 27e Congr. Int. Chem. Ind. (Brussels, 1954), p. 3.' V. Luzzati in Biological Membranes-Physical Fact and Fiction, ed. D. Chapman (Academic(PAPER 8/1860

ISSN:0300-9599    

DOI:10.1039/F19807600236

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

J.C.S. Faraday I, 1980, 76, 244-255Solvation of Silver Ions in Mixed Methyl Cyanide+Waterand Methyl CyanideSMethanol SystemsRadiation and Electron Spin Resonance Study'BY COLIN K. ALESBURY AND MARTYN C. R. SYMONS"Department of Chemistry, The University, Leicester LE1 7RHReceived 8th March, 1979Exposure of solutions of silver perchlorate in methyl cyanide to 6oCo y-rays at 77K gave Agocentres with a reduced silver hyperfine coupling indicative of electron delocalisation on to solventmolecules. Extra splitting into 9 features separated by 6 G established the presence of four equi-valent solvent ligands coordinated via nitrogen. This shows that Ag+ is solvated by four MeCNmolecules and that electron addition does not cause desolvation at 77 K. A similar pattern of 14Nhyperfine features for the AgII centre establishes the presence of four strongly coupled 14N nucleiand the form of the g and A(lo9Ag+ lo7Ag) hyperfine components showed that a square planararrangement of ligands was probably achieved at 77 K.Addition of water or methanol had little effect on the e.s.r.spectra up to w 0.5 mole fraction (MF)of protic solvent, showing that Agf is solvated preferentially by methyl cyanide. However, in the0.45-0.90 MF region (D,O) a new species exhibiting hyperfine coupling to only one 14N nucleus anda greatly increased lo9Ag/'O7Ag hyperfine coupling was detected, whose concentration increased onwarming above 77 K. This is thought to be formed from Ag+(H20),(MeCN) ions by loss of waterfollowing electron addition. Also, an atom-like Ago centre was detected in this mole fraction range.Other species detected by e.s.r.spectroscopy included a species with A('09Ag) = 180 G, thoughtto be [Ag-C(Me)k], and a species with A(lo9Ag) = 130 G, thought to be [AgACH20H]+.We have recently shown that when solutions of silver nitrate in methyl cyanideare exposed to 6oCo y-rays at 77 K, Ago and Ag" centres are formed, in addition tovarious solvent radical^,^ and NO:- +NO, are formed from the anions. The resultswere remarkable in that there was no evidence of superhyperfine coupling to 14N,which had been expected from MeCN ligands, especially for the AglI centre. Wetentatively concluded that these centres were formed in clusters, the ligands beingNO; rather than MeCNe2 However, we have shown in a preliminary report thatwhen the nitrate is replaced by the perchlorate both centres displayed well definedhyperfine coupling to four equivalent I4N nuclei showing that Ag(MeCN)z ions hadgained and lost electron^.^We conclude that the tendency for ClO, ions to co-ordinate to silver is very muchless than that for NO; ions and hence that clustering is also less important in suchsolutions.Having discovered a system in which well defined ligand hyperfine coupling wasapparent for the Ago and Ag" centres, it seemed of interest to study mixed solventsystems and in this report we describe our results for methyl cyanide+water andmethyl cyanide + methanol.The results reveal the presence of various discretesolvates.This is unusual in mixed solvent studies, which commonly yield only shiftsbetween two limiting value^.^ Whilst these shifts can be interpreted in terms ofpreferential solvation,6 the significance of such interpretations is not always clear.'Only in a few cases have well defined spectra for separate solvates been detected inmixed solvent systems.24C. K . ALESBURY AND M . C . R. SYMONS 245EXPERIMENTALThe experimental procedures were as desctibed previously.2 Samples were frozen assmall glassy beads and exposed to 6oCo 7-rays in a Vickrad source for periods up to 2 h at adose rate of z 1.7 MRad h-l.E.s.r. spectra were recorded on Varian E3 or El09 spectrometers at 77 K. Occasionallyspectra were also measured at Q-band frequencies using a spectrometer constructed byJ. A.Brivati. Samples were annealed in the insert Dewar flask and spectra were continuouslymonitored. Samples were recooled at 77 K whenever significant changes were detected inthe e.s.r. spectra and the new spectra were then recorded. The El09 spectrometer, equippedwith a Hewlett Packard 5255 A frequency converter, was calibrated with a Bruker n.m.r.-oscillator model B-H 12E. This was used for accurate spectral calibrations.RESULTS AND DISCUSSIONMETHYL CYANIDE 4- WATERA typical e.s.r. spectrum is displayed in fig. l ( a ) (CD,CN), showing features as-signed to Ag" and Ago centres. (In general, we used CD3CN since the hyperfine com-ponents were narrower. Otherwise, results were the same using CH3CN.) Hyper-fine features from Io9Ag, Io7Ag and 14N are well defined for both centres.TheIo9Ag and Io7Ag components for the Ag" centre are superimposed because of thesmall A-values. Apart from variation in resolution the initial spectra in the D20range 0-0.5MF were all closely similar. The stick diagram in fig. l(a) gives the best re-construction for the low-field (parallel) features for the Ag" centre, but we remainunsure about the best assignment for the perpendicular features because several werealways obscured by overlap with lines from other centres. However, after slightannealing resolution was improved and the data in table 1 are, we believe, reliable.The hyperfine coupling constants and g-values are normal for Agrr centres. Identi-fication of this centre as Agrr is thus unambiguous.There are 9 14N components inthe parallel region, indicating four equivalent methyl cyanide (MeCN) ligands.TABLE 1 .-E.s.R. PARAMETERS FOR SILVER SPECIES FORMED IN CD,CN+D,O MIXTURESCONTAINING AgC104 Cg-valuesa hyper fine interact i ona/G[Ag(MeCN)4I0 1.997 lo9Ag -532[Ag(MeCN),I2+ 2.313 Io9Ag -46.5 bAg2' 1.981 - 267species A 2.003 lo9Ag - 679species B 2.002 Io9Ag -721species C 2.002 IogAg - 175I4N + 614N +20 I4N+2614N +7a Breit-Rabi corrected : G = T ; g-values are m +0.0005 ; lo9Ag A values M 4 2 G ; I4NA values M -f- 1 G. b g~ and A 1 values unobtainable due to overlap of lines with central features.C Data are consistent throughout mole fraction range246 SILVER ION SOLVATION I N MIXED SOLVENTS13100G (9.100G Hz)I h -1 -2 -3 -4(a)3150G(9.094 GHz)}*_,I-- Ag2+(-l) Ag *; J(b) As;(+ 11FIG.1.-First derivative X-band e.s.r. spectra for solutions of AgC104 in methyl cyanide (CD3CN)after exposure to 6oCo y-rays at 77 K, (a) showing features assigned to [Ag(MeCN),I2+ and[Ag(MeCN),]O centres, (6) as (a) after annealing to give features characteristic of Agz ions. Thenumbers given here and in subsequent figures are the MI quantum numbers for 14N and Ag. loC . IS. ALBSBURY AND Mgain x 10032503250G (9.1400 MHz) I247FIG. 2.-First derivative X-band e.s.r. spectra for solutions of AgC104 in DzO+CD3CN (0.25 MF)after exposure to 6oCo prays at 77 K, (a) showing features for species A and B, (b) showing thegrowth in A and loss of B on annealing248 SILVER ION SOLVATION IN MIXED SOLVENTSThere are also 9 14N components for the Ago centre, again indicating 4 equivalentMeCN ligands, although the 14N hyperfine coupling is much smaller (w 6 G for Agoand w21 G (I\), 26 G ( 1_) for Ag").The outer (MI = 14) features are appreciablybroadened, suggesting slight anisotropy, but we are unable to estimate its extent. Itis worth noting that if the bonding to nitrogen involves an sp hybrid orbital, in accordwith simple concepts, then the maximum difference between 11 and 1. componentsshould be w0.5 G. Since for a Td complex there is no direction in which all the ligandscan contribute maximum anisotropy, this small effect would never be detected inpowder spectra.On annealing above 77 K, the Ag" features broadened reversibly, indicating theonset of some thermal relaxation or partial averaging.The Ago features partlydecayed irreversibly and were partly converted into new features assigned to Agi[fig. l(b)]. This assignment rests upon the appearance of new central features[F, MF = 1,0 and 0,O where F = T+_+] and on the fact that the derived g and Avalues based on this assignment are reasonable compared with those previously as-signed to A&.' However, the features are broader than those previously re-corded.2* '* lo There is some indication of multiplet splitting in addition to the trip-let expected from 07Ag + O 'Ag combinations and it therefore seems that some MeCNligands have been retained in the Agl unit and that the unpaired electron is not totallyconfined to the two silver atoms.Also, the extent of anisotropy seems to be less thanusual for Agz centres. Further annealing did not lead to detectable features forAg:+ or Ag$+ aggregates.lO* l1initial addition of water (DzO) to AgC104 in methyl cyanide (CD,CN) had littleeffect on the spectra except that the hyperfine features became better defined up to0.375 MF(D20). We are not clear why this narrowing should occur, but suggest thatthe water increases the rigidity of the medium and hence the precision of solvation.Even for concentrations of water in the region up to 0.85 MF(D20), the features forthe Ag0(MeCN), and Ag11(MeCN)4 centres remained well defined. However, from0.4 MF D20 upwards two new sets of features for Ago species appeared [A and B infig.2(a)]. A grew in intensity and became better defined on slight annealing whilst Bwas lost. A comprises two sets of doublets, each line being further split into triplets.We assign these to a (D20),Ago(MeCN) complex where n is not determined. Thefeatures are almost isotropic, and the 6s (Ag) population has increased considerably(Aiso = -679 G) relative to the Ago(MeCN)4 centre (table 1). Since the relaxedAgo(D20)4 centre at 77 K has Ai,,(109Ag) = -530 G, and the Ago(MeCN), centrehas Aiso = -532 G, the increase to -679 G suggests that n < 3 and quite possiblythat n = 0.Centre B has Ai,0(109Ag) = -721 G and hence is effectively the unsolvated atom.That such a species should form in aqueous media is significant in the light of a recentcontroversy regarding the nature of silver atom aquation 12* l 3 and is discussedfurther below.In the 0.5-0.7 MF region a third centre (C) was clearly detected,especially after annealing. Centre C has a far smaller silver hyperfine coupling of = 180 G (table 1) [see fig. 1 of ref. (15)]. There is some anisotropy and a poorlydefined triplet splitting. Apart from being broader, these features were unchangedwhen CH3CN or H20 were used and we therefore suggest that the triplet splitting isdue to 14N nuclei. We suggest that (C) is either the o1 complex (I), or the compleC. K . ALESBURY AND M. C . R . SYMONS 249represented by (11) or (111). The coupling to 14N is large for species (I), since there isno possibility of direct interaction between the unpaired electron and the 14N nucleus.We therefore strongly favour structures (I) or (111).These are limiting structures forby/C= NM eAg\ 0Me 0 /C = N(11) (111)a single species which has some structural features in common with imino radicals,R2C = N(1V). Indeed, the e.s.r. parameters are quite comparable. Thus,(IV)MeHC-h has AN(//) = 33 G, AN (1) E 0 G, and AN(iso) = 11 G, with A(lH) =85 G.14 The estimated spin density on hydrogen is ~ 0 . 1 7 which can be comparedwith that now obtained for silver ( ~ 0 . 2 5 ) . Thus perhaps the most satisfactory limit-ing representation is that given in insert (111). If this assignment is correct, this is oneof the first examples of an organometallic radical of this c1ass.l The structure con-trasts with that proposed by Kasai and McLeod for a silver atom-ethylene complex09studied by e.s.r.spectroscopy (V).16 In this case there is almost no 5s character inthe orbital of the unpaired electron, which rules out the alternative asymmetricstructure (VI).METHYL CYANIDE -t METHANOLWe generally used CD3CN and CD30D as solvent because of better spectral reso-lution. Initial addition of CD30D again had little effect on the e.s.r. spectra. Formole fractions (CD30D) > ~ 0 . 9 , a new species, D, became significant especiallyafter annealing. This species comprised a doublet of broad fairly isotropic lines(fig. 3). Its concentration grew strongly as the central features due to dD20D anddD317 were lost on annealing. Since similar, but broader features grew in withCH30H systems and since dH3 radicals are not trapped therein,17 we assign structur250 SILVER ION SOLVATION I N MIXED SOLVENTS(VII) to this species.The isotropic silver hyperfine coupling of a130 G is com-parable with that for the Ag-H+ species (103 G), which is structurally similar. InHAgACT HOH(VII)the light of this assignment, the similar structure, (I), for species C appears even lesslikely, since the silver hyperfine coupling is much greater for C than for D. Species Dwas also formed in relatively high yield for solutions in pure methanol2 * The dataare given in table 2.0J32cOG (9.137 GHz!1Ag :++As 2+JFIG. 3.-First derivative X-band e.s.r. spectrum for solutions of AgC104 in methanol, showingfeatures assigned to species D.On annealing solutions rich in CD,OD, well defined spectra for Ag; were obtainedand, for relatively concentrated AgC104 solutions, (> x0.2 mol dm-3) Agz+ featureswere also obtained (fig.4). We have previously detected e.s.r. features for Ag:+centres in aqueous sulphuric acid lo and in toluene glasses l1 and have commented onthe remarkable stability of this metal cluster, held together apparently by only a singlebinding electron. Methanol glasses again seem to favour this complex and we wereunable to detect features for the expected A&+ species?STRUCTURAL ASPECTSOur results establish unambiguously that Ag+ ions are preferentially solvated byMeCN molecules even in the presence of 50 % water or methanol.Since MeCNnormally solvates cations weakly,19 we need to postulate a special reason why thiC. K. ALESBURY AND M. C. R. SYMONS 251should be. We invoke the “ back-Chatt ” effect.20 By n-overlap silver can donate4delectrons into the empty 71: and n; orbitals of the CN group : as well as contributingto the overall Ag-N bonding, this back donation makes silver a better a-acceptorand nitrogen a better a-donor, thus enhancing the overall bonding. No such back-bonding is possible for water or alcohol molecules.In the light of this conclusion, it is interesting to return to the problem of Agocentres in water. The aquo-centres, when first formed at 4 K, have a larger spin-density on silver than that of the [Ag(MeCN)J centre, 12* l3 as expected for a complexwith weaker a-bonds.However, on annealing to 77 Ky the isotropic coupling fallsTABLE 2.-E.S.R. PARAMETERS FOR SILVER SPECIES FORMED IN CD~CN-~CDSOD MIXTURESCONTAINING AgC104cg-valuesa hyperfine interactiona/Gcentre 911 91 9av A ll A l Aiso~~ ~[Ag(MeCW4l0 1.997 lo9Ag - 532[Ag(MeCN)4]2+ 2 . 3 1 2 b lo9Ag -46.5 bAg,’ 1.972 - 267 &af 1.971 - 134species C 2.003 - 180species D 2.002 - 13014N +614N +21 I4N 3.26as b9 C S e e footnotes to table 1.3200 G (9.068 GHzIH-2 -1FIG. 4.-First derivative X-band e.s.r. spectrum for AgC104 in CD30D after exposure to 6oCoy-raysat 77 K and annealing to w 120 K, showing features assigned to Agi+ ions252 SILVER ION SOLVATION I N MIXED SOLVENTSfrom 630 G to = 530 G, and it has been argued that this is indicative of an increaseddelocalisation resulting from solvent reorganisation to suit the silver atom rather thanthe i0n.l Since metallic atoms are not expected to be solvated significantly by water,it is difficult to understand why this should arise.In our view,12 the loss of spindensity from silver is apparent, not real, and we propose that spin is simply transferredfrom 5s into 5p orbitals. The observed slight anisotropy in A(Io9Ag) is sufficient toaccommodate this change, and the shift to low g-values is in accord with 5p-population.This change is triggered, in our view, by an outward or rotatory motion of one watermolecule such that the bonding electrons are moved away from the Ag-0 region.This provides a less antibonding region for the excess electron, which can move into itonly by admixture of some 5pz orbital into the wavefunction.If this is correct, thenthere is no need to postulate extra delocalisation and the results for Ago in water andin methyl cyanide are rationalised.THE AgI1 CENTREThe results establish that the unpaired electron is in a dX2-,,2 orbital on silver andhence that the four ligands have moved towards a square-planar configuration. Thefact that an annealing sharpens the features but does not appear to modify the e.s.r.parameters suggests that the planar configuration has been achieved at 77 K. Thisshows that the MeCN ligands are free to move on the surface of silver in these mat-rices.5s[ A~(M~CNI,]*+ [Ag (MeCNIL]FIG. 5 .-Energy level diagram for [Ag(MeCN),I2+ and [Ag( MeCN)l0 complexes.The fact that the 14N hyperfine coupling for the Ag"(MeCN);+ complex is ~ 3 .6times greater than that for the Ago complex is to be expected since the former involvesthe relatively deep 4dx2,,2 orbital whilst the latter involves the shallow 5s orbital : thesituation is shown qualitatively in fig. 5. The 2s-orbital population on the 4 ligandsin the Ag" complex can be estimated to be ~ 0 . 0 4 5 using the usual approximate pro-cedure.21 The total population on all 4 ligands, assuming sp hybridisation for nit-rogen is then ~ 0 . 3 6 . For the Ago centre, the lo9Ag hyperfine coupling leads to anorbital population of ~ 0 . 7 6 on silver and the above procedure for nitrogen leads to atotal ligand population of = 0.1, assuming sp hybridisation and localisation on nitro-gen.The remaining spin-density is probably in silverp-orbitals and, to a lesser extent,on carbon and hydrogen.The structure of the Agi centre formed from Ag(MeCN), also deserves com-ment. The features are much broader than those observed, for example, for Agiinaqueous sulphuric acid.1° Also, there is far less apparent anisotropy than thatclearly evident in sulphuric acid solutions. The species in sulphuric acid is almostcertainly close to being the diatomic d-radical, AgA Ag+. The species now detectedis, as stressed above, almost certainly still solvated, which probably means that thC . K . ALESBURY A N D 14. C . R . SYMONS 253a-bonding is weaker, more closely resembling simply the sum of the silver s-orbitalstogether with suitable ligand orbitals.It is worth noting that the sum of the isotropichyperfine coupling on the two silver atoms, 534 G, is almost exactly equal to that onsilver in Ag(MeCN),.ASPECTS OF MECHANISMIn all cases, the initial reaction is expected to be electron loss from solvent molecules.The electron is captured either by MeCN to give MeCN- or (MeCN); , by methanolicor aqueous glasses to give trapped electrons (e;) or by Ag(MeCN)z to giveAgo(MeCN),. Addition to silver is strongly favoured. The Ag" centres are pre-sumably formed from solvent cations, (MeCN)+ or MeOH+, prior to proton loss togive H,&N or H2eOH radicals. Our results show that on going from MeCN toH20 solvent there is little change in the yield of Ago centres, but that on going fromMeCN to MeOH solutions there is a clear fall by a factor of ~ 1 0 .The methanolicsolutions acquire an intense purple colouration showing that e; formation activelycompetes with Ago formation. This result shows that formation of MeCN- or(MeCN); is a relatively slow process and that e; centres compete more effectively forthe electrons. This piobably reflects the fact that MekN- is strongly bent at the CNcarbon atom, coupled with the probability that trapping is only effective whenbending occurs.Detection of two AgOMeCN complexes is of interest. The " solvate"OAg-NCMe is essentially an Ag atom with a weakly coordinated ligand. InAg\/contrast, the isomer C=6 has its major spin density localised on nitrogen.MeThis species may either be formed by rearrangement :Ag\ */Ag-NCMe + C=N,Meor by attack on MedN- anions on Ag+ :AgAg+ + MeCN- 3 C-N./MeReaction (2) is expected to lead initially to this species rather than to OAg-NCMebecause MeeN- is already bent, with a high spin density on carbon and nitrogen.3* 22Ag\/In the systems in which C-h is detected, it is more stable than OAg-NCMe.M254 SILVER ION SOLVATION IN MIXED SOLVENTSHowever, species C was never detected in pure methyl cyanide and the presence ofwater or methanol (~0.6-0.9 MF) is essential.There seems to be a relatively subtlebalance, with the tetrasolvate, Ag(MeCN)4, so stabilised that rearrangement to give Cis not favoured. when the first coordination shell includes water, these moleculesprovide less stabilisation and are far more readily lost.Probably subsequent to suchloss, rearrangement (1) can occur. Again, reaction (2) may be more inhibited byfour strongly coordinated MeCN ligands than by weakly coordinated water msle-cules.Our inability to detect Ag(MeCN)3 and Ag(MeCN), species probably means thatthe parent cations, 1Ag(MeCN)3(S)]+ etc., are never major constituents in the solutionsstudied. However, the features due to Ag(MeCN)4 broaden considerably in the0.7-0.9 MF(HzO or MeOH) region and this may well be caused by overlap from otherunresolved features from Ag(MeCN), and Ag(MeCN), species present in low relativeabundance. We tentatively explain this result as follows : addition of water to MeCNgives, initially, MeCN - - - HOH - - - NCMe Since these hydrogen bonds arevery weak, the oxygen is only weakly basic and cannot compete with MeCN for Ag+.However, as water-water bonding sets in with high water concentrations, waterbecomes both a stronger hydrogen bond donor and acceptor.This means that thereaction(3) [Ag(MeCN),]+ + 8H20 + [Ag(H20),]+ + 4MeCN - - - HOHrapidly becomes more favourable provided there is enough water to solvate fully thereacting water molecules in (3). Thus a large excess of water is required, but once thefirst stage of (3) is favourable, so are the subsequent three stages, and hence the inter-mediate mixed solvates are never major constituents. This remains true for the mono-MeCN complex : its features are detected despite its low abundance only because theyare narrow and are well removed from those for Ag(MeCN)4.Finally, we should comment on the ready formation of Agz+.This cluster was notformed significantly in pure MeCN solutions, nor in the MeCN+H,O systems, butit was the major product on annealing the pure methanol and methanol rich systems.The Ag; cation seems generally to be a precursor to Ag:+, but we were unable todetect Ag2,+. This result is puzzling, since the logical mode of formation is stepwiseaddition of Ag+. Features for Ag$+ are well defined l1 and hence we can argue thatA&+ is never a significant species in these thermal reactions. We conclude that in themethanolic systems, clusters of ions containing at least two Ag+ are important, andthat the stability of Ag:+ is considerably greater than that of As$+.We thank the S.R.C.and Kodak Ltd for a CASE Studentship (C.K.A.) and Dr. G.Farnell for helpful discussions.Taken as Soloution Spectra, Part 61.D. R. Brown, T. J. V. Findlay and M. C. R. Symons, J.C.S. F'ruu'ay I, 1976,72,1792.R. J. Egland and M. C. R. Symons, J. Chem. SOC. (A), 1970, 1326.D. R. Brown, G. Eastland and M. C. R. Symons, Chem. Phys. Letters, 1978, in press.See, for example, M. Smith and M. C. R. Symons, Trans. Furaduy SOC., 1958,54,338 ; 346.A. K. Covington and J. M. Thain, J.C.S. Furaday I, 1974,70,1879. ' M. C. R. Symons, Ann. Rep. Chem. SOC., 1976,91. * See, for example, R. D. Green and N. Sheppard, J.C.S. Furaduy 11, 1972, 68,821 ;A. D. Covington and A. K. Covington, J.C.S. Furaduy I, 1975,71,831.L. Shields and M. C. R. Symons, Mol. Phys., 1966,11, 57.lo R. S. Eachus and M. C. R. Symons, J. Chern. Soc (A), 1970, 1329.C. E. Forbes and M. C. R. Symons, Mol. Phys., 1974,27,467.l2 M. C. R. Symons, J. Chem. Phys., 1978,69,3443C . K . ALESBURY AND M. C . R. SYMONS 255l 3 L. Kevan, J. Chem. Phys., 1978,69,3446.l4 S . P. Mishra and M. C. R. Symons, J.C.S. Perkin 11, 1973, 394.l5 C. K. Alesbury and M. C. R. Symons, J. Orgunumetul. Chem., 1978,162, C27l6 P. H. Kasai and D. McLeod, J. Amer. Chem. Suc., 1975,97,6603.l7 G. W. Eastland and M. C. R. Symons, J. Chem. Research, ( S ) 1977, 254; (M) 1977, 2901.l8 D. R. Brown, Ph.D. Thesis (Leicester University, 1976).l9 I. D. Kuntz and C. J. Cheng, J. Amer. Chem. Soc., 1975,97,4852.2o J. Chatt, Suppl. Ricercu Scient., 1958, 28, 130.21 P. W. Atkins and M. C. R. Symons, The Structure uf1nurgunic Radicals (Elsevier, Amsterdam,22 K. D. J. Root, M. C. R. Symons and B. C. Weatherly, MuZ. Phys., 1966,11,161.23 A. Le Narvor, E. Gentric and P. Saumagne, Cunud. J. Chem., 1971, 49,1933.1967).(PAPER 9/380

ISSN:0300-9599    

DOI:10.1039/F19807600244

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

J.C.S. Faraday I, 1980, 76,256-265Spectroscopic Studies of Water-Aprotic-solvent Interactionsin the Water-rich RegionBY MARTYN C. R. SYMONS,* JOHN M. HARVEY AND SUSAN E. JACKSONDepartment of Chemistry, The University, Leicester LE1 7RHReceived 30th April, 1979Although the fundamental infrared spectrum for HOD in D20 in the 0-H stretching regionshows little change as basic aprotic solvents such as dimethylsulphoxide or dimethylformamide areadded, the first overtone @OH) shows a marked fall in the 7120 cm-l band normally assigned to“ free ” 0-H oscillators. This is interpreted in terms of hydrogen bonding to the base resulting inscavenging of free 0-H groups.N.m.r. studies of the water proton resonance on adding such basic cosolvents show little changeor a small down-field shift in the 0-0.05 mole fraction region prior to an upfield shift which is a functionof the basicity of the cosolvent.This region of insensitivity is explicable in terms of the scavengingof the free OH groups detected by overtone infrared spectroscopy. The concepts of water structuremaking and breaking are discussed in the light of these results.Since the concept of “ Structural Temperature ” was introduced by Bernal andFowlerY2 it has become customary to discuss the effects of solutes on the properties ofwater in terms of structure “ making and breaking ”, corresponding to a fall or risein the structural temperature. It has been used, for example, in discussing changesin the first overtone of the 0-H stretching frequency for water.3* We contendthat this is often too facile a concept that embraces many possible changes, and thatunder certain circumstances different measurements can lead to opposite conclusionsif this concept is used.5* Hence one of our present endeavours is to produce moreexplicit structural descriptions of aqueous solutions that are qualitatively in accordwith the data.As part of this study, we have turned our attention to aqueous solutions of someimportant basic aprotic solvents such as dimethylsulphoxide (DMSO) and dimethyl-formamide (DMF), but excluding the alcohols, and protic amides and amines whichdiffer in that they can act as proton-donors as well as acceptors.Because specificwater structure effects are only to be expected in the water-rich region, when themedium still resembles bulk water, we concern ourselves primarily with the 0-0.1 MF(mole fraction) range.Despite the recent mass of publications concerned with water and aqueoussolutions of non-electrolytes, there have been relatively few studies in which infrared,Raman or n.m.r.spectroscopies have been applied specifically to such solutions. Brinkand Falk7 studied the fundamental 0-H stretching region of the infrared spectrumof HOD in D20 on adding DMSO, and concluded that no significant changes occurred.Such spectra are, in fact, very broad, and small changes are difficult to detect. Theyconcluded that DMSO “ does not change the average value of the width or distri-bution of hydrogen-bond energies ”. Results such as these must have dissuadedchemists from using this technique.a paper by Paquette and Jolicoeur hasappeared8 which reports the effect of a range of basic aprotic solvents on one of theSince this work was largely comp1etedy5*25M.C. R . SYMONS, J . M. HARVEY A N D S . E . JACKSON 257overtone bands of water which is assigned to "free" 0-H oscillators. Wherethey overlap, their results are comparable with ours, but their conclusions differ fromours, as we show below. They do, however, utilise several of the novel concepts putforward in our preliminary note.5The major n.m.r. studies of the water proton resonance in the presence of basicaprotic solvents are those of Glew et aL9 and of Wen and Hertz.1° The formerreported small down-field shifts in the proton resonance on adding acetone or tetra-hydrofuran (THF), which were interpreted in terms of " structure making ".g Thelatter confirmed the down-field shift for THF, but not for acetone.lo They dividedthe shifts for aprotic solutes into " polar " and '' non-polar " effects, but did notspecify the structures envisaged in the polar effect, nor the way in which the protonshielding was affected, their main concern being to achieve a separation of the twophenomena.EXPERIMENTALMATERIALSWater was doubly distilled from alkaline permanganate under nitrogen. D28 was usedas supplied by Merck, Sharp and Dohme.Other solvents, of analytical grade, were storedover molecular sieves. Whenever water was detectable by n.m.r. or infrared spectroscopy,these solvents were dried with appropriate reagents, refluxed and fractionated.PREPARATION OF SOLUTIONSFor overtone infrared studies, in order to maintain a constant concentration of 0--Hoscillators, stock solutions of HOD in DzQ were utilised, 1.000 g being taken in a fixedvolume of 5.00 cm3 of solution.These solutions were measured against similar solutionsin pure DzO.SPECTRAOvertone infrared spectra were measured with a Unicam SP.700 spectrometer, for whichthe cell holders were thermostatted between 5 and 75°C to +O.l"C. Spectra in the 6000-8000 cm-l region were obtained using 2 cm stoppered glass cells.N.m.r. spectra were measured on a Jeol 100 MHz instrument, using TMS as an internalmarker. Thus, strictly, the shifts are relative to TMS, but we are satisfied that in the0-0.05 MF region, the lH resonance of TMS is shifted insignificantly.This was establishedfor a few selected systems using an external marker with coaxial tubes, the bulk magneticsusceptibility being estimated from the n.m.r. spectra.Low temperature infrared spectra were measured with a Unicam SP.100 double-beamspectrometer using a prism-grating combination. A SPECAC variable temperature cell wasused for cooling. Cell windows were SPECAC Irtran-2, and were separated by Teflonspaces1RESULTS AND DISCUSSIONOur results are discussed separately under fundamental infrared, overtone infraredand n.m.r. spectroscopies, the infrared data being utilised to explain the n.m.r. shifts.FUNDAMENTAL INFRARED STUDIESWe confirm that very little change can be observed in the fundamental OHstretching region of the spectrum of HOD in D20 in the 0-0.1 MF region, whenaprotic solvents are added.' However, certain of these mixtures could be frozen toglassy solids apparently without significant phase separation, and their spectra haverevealed some details that are concealed in the liquid-phase spectra.This technique,1-258 WATER-APROTIC-SOLVENT INTERACTIONSpreviously utilised in a study of electrolytes in water and methanol'l is sometimeseffective because of band narrowing. We make the simplified assumption that glassesare infinitely viscous liquids and hence that their structure resembles the instantaneous,or " static", structure of the liquid. A typical spectrum for aqueous DMSO isshown in fig.1. Pure ice (HOD in D20) at - 125°C has a relatively narrow band at3280 cm-l. This is lost in the spectra of the DMSO glasses, being replaced by abroader feature in the 3300 cm-1 region. However, a definite shoulder can be seenat ~ 3 4 2 0 cm-1 that is not apparent at room temperature.We suggest that the major 3300 cm-l feature is due to bulk water. The fact thatit is broader than the ice band and shifted to higher frequencies reflects the situationfound in fluid water : a greater range of hydrogen bond lengths and bond angles thanfound in ice, the average bond being slightly weaker than that in ice. The fact thatan extra band is detected for these solutions can be understood in terms of the fol-lowing reaction :, / , ,where (H20)b represents bulk water, B is the basic cosolvent, and the (LP)free symbolrepresents a water molecule having one lone-pair of electrons not participating in ahydrogen bond.Since, in our view, there is an equilibrium (2) involving (OH)freeand (LP)f,ee groups, there will be a buffering effect5, in which the concentration of(OH),,, will fall and the gain in the concentration of (LP)free groups will be less rapidthan that predicted by equilibrium (1)We stress that equilibrium (1) is effectively complete in water at ambient temperatures.If this were not the case, the aprotic solvent would be only sparingly soluble. Thiscontention is supported by our e.s.r. results for aqueous solutions of di-t-butyl-nitroxide, which showed that, despite steric hindrance, effectively every nitroxidemolecule is hydrogen bonded,12 and also by infrared studies of the C=O oscillatorof aqueous a~et0ne.l~ We also stress that we envisage these " free " (OH) and (LP)groups as being quite separate from each other, with eqn (2) obeying the law of massaction.Reaction (1) greatly increases the concentration of (LB)free groups, thereby de-creasing any infrared bands associated with (0H)free oscillators, but increasing bandsdue to 0-H oscillators a and p [in eqn (111.The band associated with 0-Hoscillators bound to the solvent B will have a solvent-dependent frequency. However,those for the a and fl 0-H oscillators will be characteristic of water only. Theseunits should be present in pure water, but when pure water is frozen they are, ofcourse, completely scavenged by ice crystals.There has therefore been no previousinfrared evidence for the formation of such units. Oscillators associated with ahydrogen bonds are predicted to w u r on the high frequency side of the bulk waterband, since these H-bonds will be weaker than average. They are probably respon-sible for the band at 3420 em-' since this was independent of the base used. Hydrogenbonds /3 will be stronger than average, and so 0-H groups forming these bonds shouldhave an absorption band on the low frequency side of the main band. They may bM. C. R . SYMONS, J . M. HARVEY AND S. E . JACKSON 2593230 cm”’3428 cm’ 13.8 3.6 3.4 3.2wavenumber/lo3 cm-lFIG. l.-Infrated absorption in the 0-H stretching region for HOD in D20 containing 0.11 MPdimethylsulphoxide : (a) 25 ; (b).- qrorr.responsible for a broadening on the low frequency side of the main band. Wepostulate that for strong bases such as DMF and DMSO the band for solvent bondedto the base B is concealed beneath the broad band at ~ 3 3 0 0 cm-l. Unfortunately,a systematic study of this type has been precluded by the difficulty of obtaining goodglasses. Nevertheless, the results explain the almost complete lack of shift observedat room temperature for basic solvents, since the two (LP)fr,e bands (for a and /?groups) simply contribute to the width, and the band for water bound to the basicsolvent is close to the bulk water band.OVERTONE INFRARED STUDIESThe distribution of intensities for the range of hydrogen bonded oscillators in thefirst overtone of the 0-H stretching frequency for HOD molecules is quite differentfrom that in the fundamental region.In the latter, strongly bound oscillators con-tribute more strongly than weakly bound oscillators and hence the band profile isbiased towards the low frequency side. In contrast, because of changes in the selec-tion rules, the first overtone profile is biased in favour of weakly bound O-H groupsand hence has its maximum on the high-frequency side of the envelope of oscillators.Since the band maximum, which is found at ~ 7 1 2 0 cm-l at room temperature,clearly correlates with that found at E 7150 cm-l at high temperature,13 it is customaryto assign it to (OH),,, groups.Certainly this band represents the limit of weaklybound oscillators, and since this limit is the free oscillator it must have a contributionfrom these groups. The shift from the high temperature value does not prove thepresence of weak hydrogen bonding since the water molecules responsible still formthree strong hydrogen bonds and the effect of these is to shift the band due to (0H)freOoscillators slightly to low frequencies. For the purpose of the present discussion, w200 WATER-APROTIC-SOLVENT INTERACTIONSassign part of that band to (OH)fr,, oscillators. This part is taken to be that which islost when basic cosolvents are added. We stress that our concept of (OH)free and(LP),,, groups is one in which these entities are quite separate from one another.This differs from a commonly depicted model in which these " free " units are simplyviewed as the limiting situation of very weak, long or bent hydrogen bonds. Thisview implies that the units remain correlated.In our model, once they are separated,they are unable to form normal water-like hydrogen bonds until they collide for-tuitously and hence they obey normal equilibrium statistics. Whilst apart, they maywell participate in weak bifurcated hydrogen bonds. These, when symmetrical,simply represent a transition state for migration of the group concerned :Addition of basic solvents caused a fall in the (OH)fre, region and a far smallerincrease in the bound region. This fall levelled off in the 0.2 MF region (fig. 2), thespectra at this stage showing only a continuously rising absorption in the (OH),,,region.We make the assumption that effectively all (OH),,, groups have beenscavenged at this stage, and hence can calculate the fractional loss in (OH),,, groupsin the 0-0.1 MF range. This procedure was satisfactory for the strong bases DMSO,DMF and Et,N. Results for tetrahydrofuran (THF) were also analysable, but thosefor acetone and cyanomethane were not, because of the concomitant growth of newbands in the 7000 cm-l region. These are assigned to OH oscillators in B- - -HODunits. This procedure does not depend upon a knowledge of the initial concentrationof (OH),,, groups in pure water. We have attempted to obtain this datum byanalysing the spectra with a DuPont curve analyser and with a computer procedure.However, unless we assume that the extinction coefficient is the same as that for themonomer, we cannot obtain a reliable result.This assumption is, in our opinion,unlikely to be correct since the effect of the other three hydrogen bonds is bound to havea considerable effect on the oscillator strength for the (O-H)fr,, groups.groups for a range ofinitial concentrations, as shown in fig. 2. These calculations are based on eqn (1)and (2) and the assumption that equilibrium (1) lies effectively completely to the right.The results show good agreement with the experimental curve for the strong baseswhen an initial concentration of 8 % is used. [This result is 0.08 MF of watermolecules possessing an (OH)free group : the alternative description as a percentage ofthe 0-H bonds gives the result as 4 %.I This result is considerably smaller than thatof R 12 % deduced from the extinction coefficient of the monomer band.This maymean that the oscillator strength of the (OH)free band increases when it becomes partof a hydrogen-bonded unit. The result differs from those of l6 There isa wide range of values in the literature, but we prefer to use the present result since itleads to a consistent explanation of the experimental data. One important conse-quence of our simple model is that the rate of loss of (OH)fre, groups should beindependent of the nature of the base. This seems to be approximately supported byour results, but cf. ref. (8). The two major assumptions needed to give this inde-pendence are that equilibrium (1) is largely to the right and that the basic cosolventInstead, we have calculated the fractional loss iM.C. R. SYMONS, J . M. HARVEY AND S. E. JACKSON 26 1molecules only form one hydrogen bond to water. The latter assumption is supportedby the fact that data for Et,N fit reasonably onto the line defined by other strongbases (fig. 2). Since Et,N has only one lone-pair and access to it is hindered by thealkyl groups it seems reasonable to assume that this molecule behaves as a monobasetowards water. Nevertheless, the results of fig. 2 could be understood in terms of allthese solvents acting as di-bases, and we stress that our results cannot be taken asestablishing conclusively the H-bond basicity of these molecules.0 - 9T 4I I I I I 1 I I I I0.04 0.08 0.12 0.16 0.20 8 01mole fraction of baseFIG.2.-Concentration of (0H)free groups for water (at 25°C) as a function of the mole fraction ofadded base : (a), dimethylsulphoxide ; (b), dimethylformamide ; (c), tetrahydrofuran ; (d), triethyl-amine; (data at 5°C corrected to 25°C) (e), dimethylacetamide. The set of curves are for differentinitial percentages of (Omfree groups expressed in terms of the total number of (OH) groups.Our conclusions differ from those of Paquette and Jolicoeur* in that these workersmade no allowance for absorption by OH- - -B oscillators. Indeed, they stated thatsuch oscillators " should have weak intensities in the near infrared region and thusare not readily identified ".This is demonstrably incorrect for weak bases such asacetone or methyl cyanide. In fact, as the base strength falls, so the band due toOH- - -B oscillators will shift to higher frequencies and increase in intensity, thusmaking a steadily increasing contribution to absorption in the 7000 cm-1 region.Hence, as the base strength falls, there is a steadily increasing contribution from theOH- - -B band and this makes an increasingly positive contribution to the apparentconcentration of (OH),,, groups. Hence the data reported* roughly followed thebase strength of the cosolvents.This effect is so great for weakly basic solvents that there is a net gain in intensityat 7120 cm-l, which becomes very marked after the (OH),,, groups have been sca-venged.This result was interpreted in terms of a " structure-breaking " effect byPaquette and Jolicoeur.* However, since the new bands are at lower frequenciesthan that assigned to (OH)free groups, we feel that the present view is to be preferred.Unfortunately, we cannot obtain good values for the postulated fall in (OH)f,,, con-centration for these systems262 WATER-APROTIC-SOLVENT INTERACTIONSThat the growth in intensity in the 7000 cm-1 region for weak bases is not causedby a gain in the concentration of (OH)free groups is clearly established by our measure-ments in the fundamental OH stretching region (for HOD in D20). If (OH)free groupswere being generated in large amounts a band in the (OH),,, region ( M 3660 cm-l)would have developed.That a band in this region would be detected is establishedby results for aqueous sodium perchlorate, for example.14* l5 There is no indicationof such a band, but growth of bands intermediate between this position and the bulkwater band can be inferred from the spectra. We conclude that no “structurebreaking ” phenomena need be invoked. The fact that a new, relatively high fre-quency band grows in rapidly when weakly basic cosolvents are added stronglysupports our case that weak hydrogen bonding to the base is extensive. We assignthese new bands to --GO-H- - -B units. Since band contours in the fundamentalfavour strongly bonded units and those in the overtone favour weakly bonded units,there is an apparent shift to high frequencies when comparing the maxima of broadbands in the fundamental and overtone regions.We have refrained from expressing our results in terms of AT*, the equivalenttemperature concept,2 for a variety of reasons.One is that concentration plots suchas that in fig. 2 are curved and hence the values depend upon an arbitrary choice ofconcentration. However, our main objection in the present work is that strong basesare assigned a negative equivalent temperature because they cause a fall in the(0H)free band. However, had we been able to monitor a (LP),,, absorption wewould undoubtedly have observed a gain in intensity, and hence would have assignedthe same cosolvents a positive equivalent temperature. Clearly this concept is un-satisfactory in this context.\,/‘WATER PROTON RESONANCE STUDIESOur results for the water-rich region are summarised in fig.3. Except for the veryweak bases there is a well defined plateau region with a dip to low-field in some casesprior to a nearly linear trend to high-field. These results are comparable with thoseobtained previously for acetone and THF.9* lo However, we do not interpret theinitial region in terms of a “ structure making ” effect, but in terms of the loss of(0H)free groups detected in the overtone infrared spectra. Since the protons in suchgroups must resonate close to the value for water monomers, they make a considerableup-field contribution to the normal water resonance. Their loss must contribute asignificant down-field component to the total shift. Our low-temperature infraredresults suggest that protons hydrogen-bonded to basic cosolvents will experience littleshift, but those associated with (LP)rree groups will be shifted both up- and down-field.Since twice as many experience the up-field shift, there is probably a net up-fieldcomponent from these protons.When basic solvents are added to methanol, theOH resonance moves rapidly up-field with no initial plateau region. This fits nicelywith our theory since there are almost no (OH)free groups in liquid methanol at roomtemperature. Unfortunately, we cannot utilise the low-temperature infrared data inattempts to reconstruct the n.m.r. shifts, because we do not know the temperaturecoefficients for the absorption bands.For weak bases such as methyl cyanide, OHprotons bound to the base experience a strong up-field shift and this, coupled withthat for the (LP)free protons, must wtweigh the loss of (0H)free groups. Westress that loss of (OH)free is buffered by equilibrium (2) and hence is less effectivethan the gain of MeCN- - -HO- - - units. Thus in these cases the plateau regionis smallM. C . R. SYMONS, 3 . M. HARVEY AND S . E. JACKSON 2634*7 t4.84.95.003.03.54.04.55.04 . emole fraction of base2.53.03.54.04.55.00 0.04 0.08 0.12 0.160 mole fraction of baseFIG. 3,-&0tOn resonance shift for water as a function of the mole fraction of added base : dashedcurves (u), dimethylsulphoxide ; (b), dimethylformamide ; (c), tetrahydrofuran ; (a, acetone ; (e),methyl cyanide : full curves 5.0 4 1.0 are computed curves described in the text, A, ignoring anyeffect from (LP)iree groups, B, using an effective shift of 3.12 p.p.m.for (LP)iree264 WATER-APROTIC-SOLVENT INTERACTIONSWe have used the data obtained from our infrared spectral studies to deduce a setof predicted shifts for the water proton resonance in the 0-0.1 MF of added bases.We selected values for the shift for protons directly bound to the baseH(the “ base-shift ”) ranging from the water value (5.0 p.p.m. in fig. 3) down to 1 p.p.m.,when the monomer resonance shift is taken as zero. In fig. 3(A) this is the onlyvariable : we make no allowance for the growth in (LP)free units, and assimilate theireffect in the “ base-shift ”.These curves are based entirely on equilibria (1) and (2)and the assumptions used in arriving at the curve in fig. 2. The fit with the experi-mental curves is reasonable, and enables us to assign apparent “ base-shifts ” to thesolvents used (table 1). In fig. 3(B), we have included possible values for the averageeffect for the gain in (LP),,, groups. A range of values were tried, that chosen inthis case being 3.32 p.p.m. As can be judged by comparison with fig. 3(A), the fit isequally satisfactory in the low MF region, but deteriorates later, the experimentalcurves rising less steeply than predicted. This could be taken to mean that the neteffect of the gain in (LP)fr,, groups is negligible. However, in corresponding studieson methanolic solutions, we have found it essential to incorporate shifts of this magni-tude for (LP),,, groups in order to explain the re~u1ts.l~ Clearly, in the MF region> 0.1 the solvent units envisaged herein are becoming modified, and a proper descrip-tion of the shifts must take this into account.Again we must stress that the theoreticalcurves in fig. 3 are derived on the assumption that the bases only form one hydrogenbond per molecule. A satisfactory fit can also be obtained if they are assumed to bedibasic, but there is an extra unknown factor, namely, the rate at which the di-bondedmolecules are converted into mono-bonded molecules. Until we have more definitiveevidence on this problem we prefer to continue using the simple assumptions whichseem to be quite satisfactory for an understanding of our infrared and n.m.r.data.TABLE APPARENT “ BASE-SHIFTS ”‘ (p.p.m.) FOR A RANGE OF BASICAPROTIC SOLVENTS IN AQUEOUS SOLUTIONbase Nil DMSO DMF THF MezCQ MeCNbase-shift Ab 5 .O 3.4 3.3 3 .O 3 .o 1 .obase-shift BC 5.0 4.0 3.8 3.5 3.5 1 .o., c ‘ Shift from the value for monomeric water for B- - -J+Oc- units, w. No allow-ante for the affect of (LP)free groups. cTaking (LP)free as havhg an effective shift of3.12 p.p.m.\We conclude that groups play an important r81e in the dissolution ofaprotic solvents in water, and that the presence of these groups is largely responsiblefor the high solubility of weakly basic species. Proton resonance shifts reflect thescavenging of (0H)free groups by such solvents.Low temperature infrared spectragive some evidence for the gain of (LP)free groups as the groups are lost, andthis supports our contention that the widely used concept of “equivalent tem-perature ” is totally inappropriate as a description of such systenisM. C . R. SYMONS, J . M. HARVEY AND S. E . JACKSON 265We thank the S.R.C. and M.R.C. for grants.' Taken as Solvation Spectra, Part 62.J. D. Bernal and R. H. Fowler, J. Chem. Phys., 1933,1,516.J. D. Worley and I. M. Klotz, J. Chem. Phys., 1944, 45, 2868; P. Philip and C. Jolicoeur, J.Phys. Chem., 1973, 77, 3071; J. Solution Chem., 1975, 4, 3.W. A. P. Luck and W. Ditter, Adv. Mol. Relax. Proc., 1972, 3, 321; W. A. P. Luck, Ber.Bunsenges. phys. Chem., 1965,69,626.S . E. Jackson and M. C. R. Symons, Chem. Phys. Letters, 1976,37,551.S. E. Jackson, E. A. Smith and M. C. R. Symons, Faraday Disc. Chem. SOC., 1977, 64, 173.J. Paquette and C. Jolicoeur, J. Solution Chem., 1977, 6, 403.D. N. Glew, H. D. Mak and N. S. Rath, Chem. Comm., 1948,264; Hydrogen-Bonded SolventSystems, ed. A. K. Covington and P. Jones (Taylor and Francis, London, 1968), pp. 197.lo W.-Y. Wen and H. G. Hertz, J. Solution Chem., 1972, 1, 17.l 1 I. M. Strauss and M. C. R. Symons, J.C.S. Faraday I, 1977, 73, 1796; 1978,74,2518.l2 Y. Y. Limy E. A. Smith and M. C. R. Symons, J.C.S. Faraday I, 1976,72,2876.l3 A. Harvey, J. M. Harvey and M. C. R. Symons, unpublished results.' G. Brink and M. Falk, J. Mol. Struct., 1970,5,27.D. M. Adams, M. J. Blandamer, M. C. R. Symons and D. Waddington, Trans. Faraday SOC.,1971,67, 611.l5 Water-A Comprehensive Treatise, ed. F. Franks (Plenum Press, London, 1973).l6 D. Eisenberg and W. Kauzman, The Structure and Properties of Water (Oxford University'' V. K. Thompson, N. J. Fletcher and M. C. R. Symons, Chem. Phys. Letters, 1979,60,323.Press, London, 1969).(PAPER 9/674

ISSN:0300-9599    

DOI:10.1039/F19807600256

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

J.C.S. Faraday I, 1980,76,266-271Dissociation of 1,3-Hexadierne and the Resonance Energyof the Pentadienyl RadicalBY ANTONY B. TRENWITHDepartment of Inorganic Chemistry, School of Chemistry,The University, Newcastle upon Tyne NE1 7RUReceived 9th March, 1979Measurements of the rate at which methane is formed in the pyrolysis of 1,3-hexadiene at tem-peratures over the range 694-759 K and pressures between 25 and 200 Torr show that the dissociationreactionCH~CHZCH=CHCH=CH~ + CH3 *+ *CH2CH=CHCH=CH2 (1)is a homogeneous, first-order reaction under these conditions and the rate expression iswhere 0 = 2.303RT/cal mo1-1 (1 cal = 4.18 J).The experimental activation energy yields for the pentadienyl radical, AH p = 45.9 k 1.0 kcal rnol-1at 298 K and a resonance energy of 18.5 k 1.5 kcal rnol-l, assuming a value of 98.2 kcal mol-l for theprimary C-H bond dissociation energy in alkanes.108 (k1 IS-') = (1 5.92 & 0.17)- (66 390 +_ 840)/0The pentadienyl radical may be expected to have a somewhat higher resonanceenergy than that of the allyl radical in view of the greater extent of the delocalisationof the odd electron in the former. Although experimental values obtained for theresonance energy of the radical support this conclusion, they show marked variations,from 15.4 & 1 to 21.7 f22 kcal mol-l, which may be due partly to the use of incorrectdata in deriving the lower value.3 The most recent estimateY3 obtained from thenitric oxide catalysed isomerisation of 1 ,3,5-heptatrieneY yielded a resonance energyof 18.5 & 1.7 kcal mol-1 which appears to provide the best fit between measured andcalculated rate constants for the isomerisation reaction.Previous studies carried out in this laboratory 4-6 have shown that the resonanceenergy of the allyl radical and of various substituted allyl radicals may be determinedfrom measurements of the methane produced in the thermal decomposition ofbut-1-ene and its derivatives.The work described herein involves a study of thedissociation of 1,3-hexadiene following the same procedure as before. The weakestbond in this molecule must be the terminal C-C bond, the fission of which leads tothe formation of a methyl radical and a resonance stabilised pentadienyl radical. Aportion of the methyl radicals so produced will be converted to methane by hydrogenabstraction from the parent molecule so that measurement of the rate of formationof this product provides a means of determining the rate of the primary dissociationreaction.Hence the activation energy of the dissociation reaction may be obtainedand from this the resonance energy of the pentadienyl radical can be evaluated.EXPERIMENTALMATERIALS1,3-hexadiene (Aldrich) of 99 % purity was transferred to the vacuum system and afterthorough outgassing was used without further purification. This material gave only a singlepeak when analysed by gas chromatography using a 1 m column of Chromosorb P con-taining 20 % di-isodecyl phthalate.26A . B . TRENWITH 267APPARATUS AND PROCEDUREThe apparatus was essentially the same as that described previously except that pressurechanges in the reaction vessel during each run were monitored using a (Texas Instruments)precision pressure gauge.Methane estimations were carried out by the same procedure asbefore, the reaction products being transferred through traps cooled to - 126 and - 198°Cto a Topler-McLeod gauge in order to effect a preliminary separation of condensable fromnon-condensable materials before analysis of the latter by gas chromatography. Duplicateruns showed that methane analyses were reproducible to within < +2 %.RESULTSA series of runs was performed at 713 K for a range of initial pressures of 1,3-hexadiene between 30 and 250Torr, the initial pressures being estimated by extra-polation of pressure-time data to zero time.For each initial pressure, methane yieldswere obtained for various reaction times and the initial rate of methane formationwas determined by computing a third degree polynomial equation giving the best fitwith the experimental data, the initial rate being given by the coefficient of the termof power unity. A double logarithmic plot of initial rate of methane formationagainst pressure of reactant gave a straight line of slope 0.99+0.01, showing thatunder the experimental conditions the reaction leading to the formation of methanewas of first-order. Runs were subsequently performed at eight temperatures over therange 685-758.5 K and at each temperature the methane yields from a fixed pressureof ~ 5 5 Torr 1,3-hexadiene were obtained for various reaction times.The results ofthis work are shown graphically in fig. 1.Rate constants for the dissociation reactionCH3CH2CH=CHCH=CH2 + CH3 + *CH,CH=CHCH=CH2 (1)were obtained using the equationwhere k2 and k3 are the rate constants for the addition and abstraction reactionsbetween methyl radicals and 1,3-hexadiene, respectively. From the data reported byCvetanoviC and Irwin,’ for a given mono-olefin, addition and abstraction reactionswith methyl radicals have similar A factors and activation energy differences of theorder of 0.1-0.2 kcal mol-l. For these olefins, therefore, k2 and k3 may reasonablybe assumed to be equal and this assumption has been made in previous investigation^^-^where data for the particular olefin being studied were not available.From the littleinformation which could be found for addition and abstraction reactions betweenmethyl radicals and dienes, the same assumption would again appear reasonable forthese compounds. The Arrhenius parameters E = 4.1 kcal mol-I and log(A/dm3mol-l s-l) = 8.2 have been obtained for the addition of methyl radicals to1,3-butadiene [using Kerr and Parsonage’s preferred Arrhenius parameters and errorlimits for methyl radical abstraction from 2-methylpropane, * these becomeE = 4.5+ 1.0 kcal mol-1 and log(A/dm3 mol-1 s-l) = 8.3k0.61 and these figures areclose to the values E = 5.5rfi0.3 kcal mol-1 and log(A/dm3 mol-1 s-l) = 8.5Jr0.2for the metathetical reaction between methyl radicals and 1,4-~yclohexadiene.~ Rateconstants for the dissociation of 1,3-hexadiene have, therefore, been derived on theassumption that, under the experimental conditions, k2 = k3.The values obtainedare given in table 1 and these yield the rate expressionlOg(k1ls-l) = 15.92kO.17-(66 390+560)/268 DISSOCIATION OF 1,3-HEXADIENE1.51 .cO . !2 4 6 8timelminFIG. 1.-Plots of methane against time for the decomposition of 55 Torr 1,3-hexadiene. +, packedreaction vessel at 725.5 K ; 0, unpacked reaction vessel : (a) 685, (b) 694, (c) 703, ( d ) 713, (e) 725 ,(f) 737, (9) 748 and (h) 758.5 K.TABLE IN INITIAL RATES OF FORMATION OF METHANE AND RATE CONSTANTPFOR THE DISSOCIATION OF 1,3-€EXADIENETIK (d[CH4]/dt)o/l(l-4 Tom s-l [CGHlO]~/To rr k l / 10-6 s-68569470371 3725725.5a737748758.51.442.815.6923.424.645.592.017755.455.455.655.557.656.356.856.35.2110.120.537.2$. 2.0b85.385.5161324630a Data obtained using a packed reaction vessel.From least-squares estimate of inter-cept of plot of l~g(d[CH,]/dt)~ against logpA . B. TRENWITH 269where 8 = 2.303RTlcal mol-1 and the quoted error limits correspond to the estimatedstandard deviation.Since the activation energy for the dissociation of 1,3-hexadiene is of prime interestin this work and the possibility of a difference of 1 kcalmol-l between E2 and E3cannot be ruled out, the effect of such a difference on the activation energy has beendetermined and it was found to change El by no more than 0.65 kcal mol-l, whilstthe A factor remains at virtually the same value.To take account of the possibleerror in El arising from the assumed equality of E2 and E3, the error limits quotedfor El may, therefore, be more reasonably quoted as -t- 840 cal mol-l.A further assumption involved in the derivation of eqn (2) is that radicals formedfrom addition and abstraction reactions between methyl radicals and 1,3-hexadienedo not subsequently dissociate to reform methyl radicals. As pointed out previously,6for mono-olefins, this assumption is strictly valid only at infinitely high pressures.With dienes, the situation is somewhat different since both addition and abstractionreactions must lead preferentially to the formation of resonance stabilised specieshaving rates of dissociation which, over the experimental pressure range, will benegligible in comparison with those of further addition and abstraction reactions.The addition of methyl radicals to 1,3-hexadiene must occur mainly at carbon atoms1 and 4 and, as with lY3-butadiene, should require a lower activation energy than isgenerally found for the addition of methyl to mono-olefins where resonance stabilisedaddition products are not formed.Although metathetical reactions between methylradicals and both mono-olefins and dienes can produce resonance stabilised radicals,activation energies for reactions with the latter should again be generally lower, sincethese can yield radicals which are stabilised by a greater degree of delocalisation,i.e. over five carbon atoms rather than over three.s-l for kl at725.5 K as compared with the measured value of 85.5 x s-l obtained fromexperiments carried out using the packed reaction vessel at this temperature.Theclose agreement between these two figures indicates that in the decomposition of1 ,3-hexadieneY primary methane is formed by an entirely homogeneous process.The rate expression for kl given above yields a value of 83.2 xDISCUSSIONIn the pyrolysis of 1,3-hexadiene, methane is formed, as with other olefins, by bothprimary and secondary processes; this leads to methane against time curves which attemperatures below 723 K show an increase in the rate of formation of methane withtime. Measurements of the rate of formation of the primary methane from the initialslopes of the methane against time curves indicate that this product is formed by anhomogeneous first-order reaction under the experimental conditions eniployed.The reliability of the assumptions made in evaluating kl, the rate constant for thedissociation reaction, may be checked by comparing the experimental A factor withthat derived theoretically from thermodynamic data.The entropy and specific heatdata given in table 2 were derived using group additivity methods lo together with theprocedures described in ref. (1 1) and these values yield for reaction (1) at 700 K,AS," = 28.09 cal K-l mol-1 for a standard state of 1 mol dm-3.The rate constant for the recombination of pentadienyl radicals has not beenmeasured but may be estimated by assuming a linear relationship between the molecu-lar weight of a free radical and the logarithm of its recombination rate ~0nstant.l~Taking values of 10.42,14 9.9 l5 and 9.6 l6 for log(kr/dm3 mol-l s-l) of CH3, C3H5and C4H,, respectively, leads to a value of 9.4 for log(kr/dm3 mol-1 s-l) forpentadienyl radicals and hence, using the geometric mean rule, log(k-, /dm3 mol-270 DISSOCIATION OF 1,3-HEXADIENEs-l) = 10.2.Assuming reaction (-1) occurs with zero activation energy, ths re-combination rate corresponds to AS: = - 17.5 cal K-l mol-l and hence for the for-ward reaction AS: = 10.8 cal K-l mol-1 at 700 K and the theoretical A factor hasa value of s-l, in excellent agreement with the experimental value. Theobservation that for this, as for most of the other systems studied by the samepr~cedure,~-~ there is good agreement between experimental and calculated A factors,strongly supports the conclusion that the assumptions made in deriving the rateconstants for the dissociation reactions are reasonable.TABLE 2.-THERMODYNAMIC DATA FOR RADICALS AND MOLECULESRELEVANT TO REACTION (1)A%298 si298 ci298 C&OOspecies /kcal mol-l /cal K-l mol-1 /cal K-l mol-1 lcal K-I mol-IC5H8 18.3 76.6 24.4C6H10 13.6 86.4 29.5C& 34.8' 46.4 8.3CsH7 45.gb 74.8 23.544.054.411.843.5(I From value reported by Chupka [ref.(12)] ; see text.The activation energy for the dissociation reaction yields, for reaction (l),AH&8 = 67.1 k0.8 kcal mol-l, again assuming an activation energy of zero for thereverse reaction.This leads to a value of AH: = 45.9+ 1.0 kcal mol-l for thepentadienyl radical at 298 K. Thus, for the reactionCHSCH=CHCH=CH2 H + *CH2CH=CHCH=CH2, (3)= 79.71 l.lkca1 mol-l, which, when compared with the long accepted valueof 98.2k1 kcal mol-l for the dissociation energy of a primary C-H bond inalkane$' yields a value of 18.5+ 1.5 kcal mol-l for the resonance energy of thepentadienyl radical. This value, being in exact agreement with that obtained byEgger and JolaY3 must add confidence to these two estimates.Recently reported values for AH; (t-Bu) which are higher than the previouslyaccepted figure have led to a reassessment of bond dissociation energies in alkanes l8and from this a value of 100.5+ 1 kcal mol-1 has been proposed for the dissociationenergy of a primary C-H bond.Using this value, the resonance energy of thepentadienyl radical becomes 20.8 & 1.3 kcal mol-l, a figure which cannot be directlycompared with earlier estimates, all of which were derived from the lower primaryC-H bond dissociation energy.The author gratefully acknowledges the assistance of Messrs. R. J. Mills andW. Nethercott, who carried out the major portion of the practical work involved inthis investigation, and the Royal Society for an equipment grant.K. W. Egger and S. W. Benson, J. Amer. Chem. SOC., 1966,88,241.H. M. Frey and A. Krantz, J. Chem. SOC. (A), 1969, 1159.K. W. Egger and M. Jola, Int. J. Chem. Kinetics, 1970,2,265.A. B. Trenwith, Trans. Faraday Soc., 1970,66,2805.A. B. Trenwith, J.C.S. Faraday I, 1973, 69, 1737.A. B. Trenwith and S. P. Wrigley, J.C.S. Faraday I , 1977, 73, 817. ' R. J. CvetanoviC and R. S. Irwin, J. Chem. Phys., 1967,46,1694A . B. TRENWITH 27 1J. A. Kerr and M. J. Parsonage, Evaluated Kinetic Data on Gas Phase Hydrogen TransferReactions of MethyZ Radicals (Butterworths, London, 1976).D. G. L. James and R. D. Suart, Trans. Faraday Soc., 1968,64,2735.lo S. W. Benson, Thermochemical Kinetics (Wiley, New York, 1968).l1 H. E. O’Neal and S . W. Benson, Int. J. Chem. Kinetics, 1969,1,221.l2 W. A. Chupka, J. Chem. Phys., 1968,48,2337.l3 L. Bertrand, G. R. DeMare, G. Huybrechts, J. Olbregts and M. Toth, Chem. Phys. Letters,1970,5, 183.l4 R. W. Walker, in Gas Kinetics and Energy Transfer, ed. P. G. Ashmore and R. J. Donovan(Spec. Period. Rep., The Chemical Society, London, 1976), vol. 2, page 311.l5 H. E. van den Berg and A. B. Callear, Trans. Faraday §oc., 1970, 66,2681.l6 W. Tsang, Int. J. Chem. Kinetics, 1973, 5,929.l7 D. B. Hartley and S. W. Benson, J. Chem. Phys., 1963,39, 132.l 8 G. M. Atri, R. R. Baldwin, G. A. Evans and R. W. Walker, J.C.S. Faraday I, 1978,74, 366.(PAPER 9/388

ISSN:0300-9599    

DOI:10.1039/F19807600266

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

J.C.S. Faraday I, 1980, 76,272-279Kinetics and Mechanism of the Gas Phase ThermalDecomposition of Hexachlorodisilane in the Presence of IodineBY ALAN M. DONCASTER AND ROBIN WALSH*Department of Chemistry, University of Reading,Whiteknights, Reading RG6 2ADReceived 12th March, 1979The gas phase thermal reaction of Si2Cls with I2 has been investigated. Product analysis revealsthe formation of SiCI4 and SiC1212. The reaction rate was found to obey the rate equationand values of k were insensitive to changes in surface-to-volume ratio. This evidence suggests thefollowing two step mechanismSi2CI6 +. SiC12 + Sic14 slow (3)SiC12 + I2 -+ SiClzIz fast. (2)From rate measurements over the temperature range 596-655 K the Arrhenius equationlog (kI/s-l) = (13.49kO.12)-(205.9k 1.4 kJ mol-l)/RTIn 10is obtained.These figures are consistent with earlier flow studies of this reaction in the absence ofscavengers. The effects of other scavengers have been briefly investigated.The absence of an I atom displacement process suggests that the Si-Si bond in Si2CI6 is muchstronger than that in SizMe6.As part of a programme of studies to determine Si-H bond strengths in silaneswe have recently investigated the kinetics and mechanisms of the reactions of I, withMe,SiH,l* C13SiH3 and F3SiH.4 In all these systems the reactions are best ex-plained as proceeding via an I atom abstraction mechanism viz. :1. + X3SiH + X3&+ HI.In the reaction of I, with where no Si-H bond is available, reaction proceedsvia H abstraction from a C-H bond.By contrast in the reaction of I, with Me,%,(where abstraction from C-H is also possible), Band and Davidson have accountedfor their results in terms of an I atom displacement process on Si, viz. :I- + Me,SiSiMe, + Me,SiI + Me,$.We have recently confirmed this result.’ Because the analogous displacement step1. + Me& 3 Me,SiI + Me.does not occur we speculated that such displacement might be limited to disilanesonly. In order to obtain further evidence for this kind of process we decided toinvestigate the reaction between I2 and Si2C16. There is no previous study of thissystem.27A. M. DONCASTER AND R. W A L S H 273EXPERIMENTALAPPARATUSThis was as previously de~cribed.l-~MATERIALSHexachlorodisilane (Alfa) was purified by evaporation from an ice bath and was degassedbefore use.Iodine (Kocli Light, 99.998 % pure) was degassed before use. Bromine(Fisons, > 99 % pure), benzene (B.D.H., > 95 % pure) and tetrachlorosilane (B.D.H.,> 98 % pure) were degassed before use arid oxygen free nitrogen (B.O.C.) was used assupplied. Hydrogen chloride, prepared from ammonium chloride (Fisons, > 99.5 %pure) and sulphuric acid (Fisons, = 98 %), was purified by evaporation from a -98°Ctrap and its purity was checked by infrared spectroscopy.PROCEDUREIodine was always admitted to the reaction vessel first followed by hexachlorodisilane.Kinetic runs were performed by following iodine absorbance spectrophotometrically as afunction of time during reaction.Reaction times were from 2.5 min to 2.5 h. Wherenecessary, reaction products were frozen out for further identification. In runs designedto study any effect from added SiC14, C&, HCI, N2 or air, the procedure was identicalapart from addition of one of the above after the Si2C16. The reaction between Brz andSi2C16 was followed in an identical manner at A = 420 nm ( E B ~ ~ = 121 dm3 mol-1 cm-l).PRODUCT IDENTIFICATIONSiC14 and SiC1212 were identified as products by a combination of infrared spectroscopyand mass spectrometry. The gas phase i.r. spectrum of the reaction products (polytheneoptics) showed bands (cm-l) at 620-604 (v. strong), 593+ 3(s), 544+ 3(s), 438&2(s), 423+2(s)and 2753: 1 (v. weak). These bands may all be attributed to known allowed absorptions inthe vibrational spectrum of either SiCl4 or SiCI2I2 (based on a Raman study *). No i.r.bands were found at 519 or 333 cm-l where SiCI3I would absorb.* The i.r.spectrum ofSiC131 was verified by preparation of an authentic sample by reaction of SiCI,H with L3The mass spectrum of the reaction product showed clear cut parent ions for SiC121$ (at352, 354 and 356 u) and Sic12 (at 168, 170, 172, 174 and 176 u) as well as characteristicfragments ions such as SiClzIf, SiClI: and SiCl;. Additionally, there was an ion cor-responding to SiCI3I+. However, this is not necessarily indicative of the presence of SiC131amongst the products since the chlorosilanes show evidence of participation in ion-moleculereactions in our mass spectrometer.Also indicative was the U.V. spectrum of the reaction products, which consisted of abroad band with a maximum at A = 250 nm (E E 1200 dm3 mol-1 cm-l).This is to becontrasted with a measured U.V. spectrum of SiC1,I with a maximum at A = 230 nm(E M 350 dm3 mol-1 cm-I). SiC14 has no U.V. spectrum and therefore the observedreaction product spectrum is attributed to SiC1212.RESULTSPRELIMINARY EXPERIMENTSThe qualitative product identification points to the occurrence of the reactionSi2C16 +I2 -+ SiC1212 + SiC14.This is supported by pressure measurements which indicate no pressure change (towithin 3: 0.3 Torr or better) during reaction and complete freezing down of the reactionproducts at 77 K (again to within k0.3 Torr or better). No secondary reactionbetween SiC14 and SiC1212 was found as evidenced by no systematic change in the U.V274spectrum (apart from intensity increase as the products built up) during a typicalreaction time.SiC14 does not decompose below 1100°C and the absence of ISiC13demonstrates that redistribution equilibria of the typeGAS PHASE THERMAL DECOMPOSITION OF HEXACHLORODISILANESiC14 +12SiC12 + 21SiC13were not important. The initial rates of iodine decay were measured for several runsat 636 K and were found to be independent of 12. For instance, at a constant(Si2C16] of 4.0 Torr, rates of 1.78 x Torr s-l were measured at[I2I0 pressures of 0.84 and 3.55 Torr, respectively.This indicates that I2 is not participating in the rate determining step of the reactionand suggests that its role may be to scavenge an intermediate formed by the de-composition of Si,C16.It also demonstrates that this system is not undergoing thetype of displacement reaction observed in the case of I2 plus Me6Si,.and 1.67 xTABLE I.-SELECTION OF RATE DATA AT 636 K0.840.900.900.911.761.821.993.223.483.554.015.146.253.494.566.552.365.543.904.044.775.716.943.842.593.601.191.721.121.143.994.044.013.973.933.874.253.893.954.15If Si2C16 is present in excess and the reaction is allowed to proceed beyond com-plete consumption of 12, the vessel absorbance begins to increase again due to theformation of an involatile film on the windows of the reaction vessel.Fortunately,if this film is not allowed to build up, it may be removed by reaction with additionaliodine. There is no evidence for any such film being formed if a reaction mixtureis pumped out with I2 still present.KINETIC MEASUREMENTSIn accordance with the above findings, a first order rate expression was adoptedas a working hypothesis. The rate constant k was defined byThis expression was tested in two ways. First, plots were made of the integratedform of the rate equation,(11) In ([Si2C16]J = In ([Si2Cl6l0) - kA. M. DONCASTER AND R. WALSH 275([Si,Cl,], being the instantaneously measured Si2C16 pressure at time t ) . Goodlinear plots were obtained up to 85 % decomposition (the maximum followed) withno curvature, indicating that secondary reactions were of no importance.In thesecond test the invariance of k with different starting pressures of reactants wasestablished. Table 1 shows the values of k obtained under a variety of startingconditions at 636 K.103 KITFIG. 1.-Arrhenius plot for kl.The values are reasonably constant and scatter to a maximum of 3.6 % aroundthe mean, which was typical at all temperatures within the range 596-655 K investi-gated. Although the range of initial pressures was limited by the low vapourpressure of Si2C16 ( x 10 Torr in the heated sampling system), this represents goodevidence in support of the assumed rate equation (I). The addition of 95 Torr ofnitrogen to a reaction produced no effect on the value of k obtained at 636 K, showingthe reaction to be in its high pressure first-order region.Addition of up to 250 Torrof SiC14 also has no effect on the value of k obtained. This demonstrates that thereaction is neither self-inhibiting nor reversible. At 600 K, several runs were per-formed in the packed vessel (S/V 12.4 times greater than the unpacked vessel). Therate constants were again found to be independent of initial reactant pressures andtheir average was slightly lower ( w 10 %) than the average value for the unpackedvessel. This close agreement was taken as showing the reaction to be homogeneousover the temperature range investigated276 GAS PHASE THERMAL DECOMPOSITION OF EIEXACHLORODISILANEThe rate constants k , obtained over the range 596-655 K, were fitted by a linearleast squares procedure to the Arrhenius equation, a plot of which is shown in fig.1.This yields log (kls-l) = (13.49_+0.12)-(205.9&1.4 kJ mol-l)/RTln 10, where theerror limits are one standard deviation.The effects of a number of other additives on the rate of the reaction were alsobriefly investigated. Small quantities of air (m 1 Torr) were observed to reduce therate of consumption of iodine by up to a factor of two (in the presence of 0.9 Torrof 12) at 636 K. pic1 was added, up to 250 Torr, to a reacting mixture ([I2lO =1.60 Torr, [Si2ClG]o = 4.1 Torr) at 603 K and no effect was observed on the rate orrate constant of the reaction. In similar fashion addition of benzene (17 Torr) toa reacting mixture at 642 K also produced no effect.Three further experiments were carried out in which reaction between Br, andSj2CIG was investigated, at 617 K.With [BrJ between 2 and 3 Torr, average reactionrates were between 57 and 107 % greater than the equivalent rates for T2 with Si2C16.Furthermore, if a first order dependence on Si2C16 alone was assumed, integratedrate plots showed curvature in the sense of a decreasing rate constant with time.This suggests some increased complexity in this system, but does not rule out acontribution to the rate from a term purely first order in Si2CI6. These observationsclearly merit detailed and separate study and were, therefore, not pursued furtherhere.REACTION MECHANISM AND DISCUSSIONThe observed product formation and kinetics strongly point to the occurrence ofthe following mechanism :Si2C16 -+ SiC12 + SiC14 (1)SiC12 -+ I2 -+ SiC1212with step (1) slow and rate determining.In the absence of I2 it seems likely that theadditional sequence of steps,n SiC1, -+ (SiCl,), -+ solid polymer (3)occurs. This process, however, cannot be competitive with step (2) when I2 is stillpresent as the kinetics would change. Additionally, step (-l), the reverse of therate controlling step, is not only unimportant under ordinary reaction conditions buteven in the presence of a lage excess of SiCl,. This evidence enables us to place alimit of k,l < k2. There is little doubt that SiC12 is formed in this reactionsince it has been detected directly in the pyrolysis of Si2C16 at higher temperatures.gThe observed rate constant for iodine consumption can be identified with kl andindeed the Arrhenius parameters strongly support this assignment.The magnitudeof A, s-l, is consistent with a transition state involving a bridging C1 atomwhich inhibits Si-Si internal rotation and is of lower symmetry than the reactantmolecule. This mechanism has been proposed previously lo in a kinetic study of thegas phase pyrolysis of Si2ClG, under jet conditions in 75-85 % helium (but in theabsence of scavengers for SiC12), for which the authors quote,log (kl/s-l) = (13.8OkO.36)-(207.5+5.0 kJ mol-l)/RTln 10.The agreement between this and our value is excellent.* Other chlorine containingdisilanes, uiz. Me,Si2C1, MeC12SiSiC1,Me and MeCl2SiSiC1Me2, also decompose* Note added in proofi Another recent study 'O which has just come to our attention yieldslog (k1/s-') = (13.0+0.5)-(200+7 kJ mol-l)/RTln 10A .M. DONCASTER AND R. WALSH 277via 1,2-C1 bridged transition states leading to divalent silicon intermediates. TheA factors and activation energies for these processes axe all of the same order ofmagnitude as that found for Si,C16, suggesting that substituent effects are not veryimportant in influencing these transition states.The experiments with added scavengers provide a pointer to the reactivity ofSiC12. As well as its failure to react with SiC14, in competition with 12, the followingreactions were also not observed,SiC12 + HCl 3 SiHC1, (4)although it should be pointed out that this conclusion rests on the assumption thatthe proposed (likely) products do not react with I, under the reaction conditions.For SiHC1, this is not strictly true;, however, the rate of reaction of I2 with SiHC1,is known and we have calculated that an observable stationary concentration ofSiHCl, should have been formed if step (4) were to occur.From the lack of reactions we obtain the following limits: k4 < 10-,kz,k5 < 0.02k2. These figures provide a constrast between the reaction of SiC1, withiodine and other processes.It is possible that step (2) is amongst the fastest knownfor SiC12, on the analogy of the reactions of free radicals with iodine which occurwith little or no activation barrier.12 It seems likely then that there are sizeableactivation energies for the non-observed processes.This is in accord with qualitativeinformation l3 and relative rate data l4 which point to the fact that SiC12 is agood deal less reactive than other divalent silicon-containing species such as SiH2.The present reaction system offers a way to investigate reactions of SiC12 morequantitatively and we plan to follow up these preliminary results.Our failure to observe the anticipated displacement reaction in this system, viz. :I* + si2c16 + CI,S~I + iicl,,1. + &Me6 -+ Me,SiI + Me,&.(6)(7)indicates that this process is considerably slower than the analogousThe results enable us to set an upper limit for k6 of 31 dm3 mol-l s-l at 596 K byassuming that this process contributes -= 10 % to the overall rate under the mostfavourable reaction conditions (high[I,]).If we further assume that A6 = dm3mol-1 s-I by analogy with the known value of A7 6 s then we can set a lower limitof EG > 81 kJ mol-l. This is considerably greater than the observed value for E7of 35 kJ mol-l. The overall energetics of step (6) are not very well known but likestep (7) it is almost certainly endothermic [since D(C13Si-SiC13) > D(C1,Si-I)*].It seems likely that the chief reason for the difference between E6 and E, lies in astrengthening of the Si-Si bond when Me groups are replaced by C1. If this werethe sole reason then we can write D(C1,Si-SiCI3) - D(Me,Si-SiMe,) 3 46 kJmol-l.Since D(Me,Si-SiMe,) = 337 kJ rno1-l l6 then D(Cl3Si---SiCl3) >/ 383 k3 niol-p.Because of small differences between D(C1,Si-I) and D(Me,Si-I) and the backactivation energies of steps (6) and (7) the actual value of D(C13Si-SiC13) may notbe quite as high as this.We feel, however, that this limit is likely to be close to thetrue value.* Si-I bond strengths are not known with certainty but are believed to be z 322 kJ rn~l-'.~278 GAS PHASE THERMAL DECOMPOSITION OF HEXACHLORODISILANEBond strengthening of a C-X bonds by electronegative substituents is well knownin organic chemistry although the analogous dissociation energies D(C13C-CC13)and D(Me,C-CMe,) are a poor standard of comparison because large non-bondedinteractions exist in these latter compounds owing to the shorter C-C bond length.TABLE 2.-ENTHALPY VALUES FOR THE DISPROPORTIONATION REACTION, 2sic13 $ siclz+ sic14AH0(298 K)/kJ mol-l source reference- 1 9 4 k 1 2 based on AH; values a- 1 2 1 + 2 9 AHo estimates for processes (1) and (8) 17 - 9.3 uncertain 10- 5 1 .4 k 2 . 0 Knudsen cell effusion method 1 8-200+25 AH" estimates for processes (1) and (8) this worka Derived using AH: values from ref. (3) (SiC13) and CATCH (SiC12, SiCl,).One final comment should be made concerning the relative stabilities of the pro-ducts of the alternative pathways of decomposition of SiZCl6, viz. :Si2C16 + SiC12 + SiC14 (1)Si2Cl, -+ 2SiC13. (8)Clearly, process (1) is significantly favoured over process (8) and the argumentspresented in this paper enable us to set an approximate value for AHo of 200 kJ mol-lfor2SiC13 + SiClz + SiCl,.This figure incorporates an assumed of 24 (k 16) kJ mol-1 (based on the earlierdiscussion of L1) and assumed overall uncertainty of +25 kJ mol-l [incorporatinguncertainty in D(Cl,Si-SiCl,)].This value may be compared with other estimatespresented in table 2. It is in excellent agreement with an independent value based onAHf"(SiC13) obtained in our laboratory earlier., Other values are, however, inpoorer agreement. The data sources for the two estimates by Russian workers lo* l7are not stated and therefore these figures cannot be discussed further. However,the figure from an equilibration study in a Knudsen cell by Farber and Srivastava l8is a direct measurement. We cannot reconcile our estimates with this result but wefeel that the latter must be wrong since, if correct, it would imply that either theactivation energy for process (1) is much greater than we have measured or that theSi-Si bond in C13Si-SiC13 is significantly weaker than that in Me3Si-SiMe,,both conclusions in direct conflict with our observations.Further studies of iodinewith disilanes are planned.We thank the S.R.C. for the provision of funds to support this work.R. Walsh and J. M. Wells, J.C.S. Favaday I, 1976, 72, 100.A. M. Doncaster and R. Walsh, J.C.S. Faraduy I, 1979,75, 1126.R. Walsh and J. M. Wells, J.C.S. Faraday I, 1976,72, 1212.A. M. Doncaster and R. Walsh, Int. J. Chem. Kinetics, 1978, 10, 101279 A. M. DONCASTER AND R . W A L S HA. M. Doncaster and R. Walsh, J.C.S. Faraday I, 1976,72,2908.S . J. Band and I. M. T . Davidson, Trans. Furuday SOC., 1970,66,406.A. M. Doncaster and R. Walsh, unpublished results.M. L. Delwaille, M. B. Buisset and M. Delhaye, J. Amer. Chem. SOC., 1952, 74, 5768.0. M. Nefedov, A. K. Mal'tsev and V. A. Svyatkin, Izuest. Akad. Nauk SSSR, Ser. Khim.,1974,958.loV. G. Bykovchenko, V. I. Pchelintsev, N. G. Komalenkova, S. A. Bashkirova and E. A.Chernyshev, Kinetica i Kataliz, 1975, 16, 813.l1 I. M. T. Davidson and M. E. Delf, J.C.S. Furaday I, 1976,9, 1912.l2 D. M. Golden and S . W. Benson, Chem. Reu., 1969,69,125.l3 P. P. Gaspar and B. J. Herold, in Carbene Chemistry, ed. W. Kirmse (Academic Press, New14R. L. Jenkins, A. J. Vandenvielen, S. P. Ruis, S. R. Gird and M. A. Ring, Inorg. Chem.,l5 A. M. Doncaster and R. Walsh, J. Phys. Chem., 1979, 83, in press.l6 I. M. T. Davidson and A. B. Howard, Chem. Comm., 1973,323 ; J.C.S. Faraday I, 1975,71,69.l7 A. M. Mosin and Yu. Kh. Shaulov, Rum. J. Phys. Chem. 1972, 46, 1051.l 8 M. Farber and R. D. Srivastava, J.C.S. Faraday I, 1977,73, 1672.l9 J. B. Pedley and B. S . Iseard, CATCHtabIes for Silicon Comporrndr (University of Sussex, 1972).2o V. F. Kochubei, A. P. Gavriliv, F. B. Moin and Yu. A. Pazderskii, Kineticu i Kataliz, 1978,York, 2nd edn, 1971), chap. 13.1973,12,2968 and refs. therein.19, 1084.(PAPER 91407

ISSN:0300-9599    

DOI:10.1039/F19807600272

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

J.C.S. Faraday I, 1980,76,280-290Reaction Mechanism of Ammonia Decomposition on TungstenBY HITOSHI SHINDO, CHIKASHI EGAWA, TAKAHARU ONISHI*AND KENZI TAMARUDepartment of Chemistry, Faculty of Science,University of Tokyo, Hongo, Bunkyo-ku, Tokyo 1 13, JapanReceived 16th March, 1979The mechanism of ammonia decomposition on a clean tungsten surface was studied by directand simultaneous measurements of the reaction rate and the amounts and behaviour of adsorbedspecies during the course of the reaction in the temperature range 773-1473 K under an ammoniapressure of 10-6-10-3 Pa. The order of the reaction rate with respect to ammonia pressure changedwith temperature from first order to fractional at 1473 and 773 K, respectively. No hydrogen wasadsorbed on the surface in any form above 973 K and an increase in the hydrogen partial pressure(PH~) during the reaction had no effect on either the reaction rate or on the amount of surfacenitrogen (ON).Under higher ammonia pressures, up to 100Pa, thick surface nitride layers wereformed during the decomposition, which decomposed at the same rate in wcuo as in the steady-stateNH3 decomposition, provided the uptake of nitrogen was the same. It was also found that the rateof nitrogen desorption from the surface depended only on ON, being faster with increasing ON,whereas the rate of nitrogen uptake from ammonia decreased with increasing ON and increased withammonia pressure (PNH~). On the basis of these data, it was concluded that the overall reactionproceeds through a reaction mechanism of " dynamic balance " between two consecutive rate-determining steps ; the supply and consumption of surface nitrogen.Ammonia decomposition on tungsten has been studied by many researchers asone of the most fundamental heterogeneous catalytic reactions.Hinshelwood andBurk first reported that in the early stage of the reaction the decomposition rate isindependent of the pressures of ammonia, hydrogen and nitrogen, although in thelater stage of the reaction it becomes dependent on the pressure of ammonia. It wasthus suggested that, during the reaction, the catalyst surface would be fully coveredby some reaction intermediate or adsorbed ammonia, whose decomposition is therate-determining step,reported that an ammoniamolecule adsorbed on a tungsten surface easily releases hydrogen; more than oneH2 molecule is obtained per adsorbed NH3 molecule without any N2 desorption at523 K, the temperature being considerably lower than usual decomposition tempera-tures. Accordingly, it seems likely that the tungsten surface is saturated with nitrogenatoms during the reaction and its desorption is the rate-determining step of the reaction.However, Jungers and Taylor and Barrer found that NH3 decomposed 1.6 timesfaster than ND3 at z 1000 K, which strongly suggests that hydrogen would beinvolved in the rate-determining step.Under these circumstances one of the authors studied the reaction on tungstenpowder by simultaneously measuring the rate as well as the adsorption during thesteady-state of the reaction and also the behaviour of adsorbed nitrogen in thenon-stationary state of the reaction, to elucidate the mechanism.By examiningseparately the kinetic behaviour of both the supply and decomposition of surfacenitrogen under the reaction conditions, he proposed a reaction mechanism of280On the other hand, Frankenburger and HoldeH. SHINDO, C. EGAWA, T. ONISHI AND K. TAMARU 28 1" dynamic balance " between two consecutive rate-determining steps ; supply ofnitrogen from gaseous ammonia into the surface nitride and its consumption bydesorption of N2 molecules.Since the surface of the tungsten powder previously employed has probably beencontaminated, in this report the cleanliness of the tungsten foil surface was constantlyexamined by Auger electron spectroscopy (A.e.s.) and the behaviour as well as theamount of chemisorbed nitrogen was followed to elucidate the mechanism of thereaction.EXPERIMENTALAll the experiments "ere carried out in ultra-high vacuum apparatus (base pressure3 x lo-* Pa) made of stainless steel and evacuated by a 200 dm3 s-l turbomolecular pump.A mass filter (QMS; UTI-100C) was used to measure the reaction rate.The atomiccomposition of the surface or the amounts of adsorbed species were estimated by A.e.s.The mount of adsorption was also measured by a flash desorption method.MATERIALSA polycrystalline tungsten foil (99.95 %, Japan Lamp Industrial Co.) measuring5.3 x 83 x 0.025 mm3 (geometric surface area ; 8.8 cm2) was used as a sample.The samplewas resistively heated by d.c. and the sample temperature was measured by a Pt-Pt/Rh(13 %) thermocouple spot-welded at the centre of the sample and by an optical pyrometer.MEASUREMENTS OF AMOUNTS OF ADSORPTIONThe amounts of adsorbed nitrogen and hydrogen were estimated by A.e.s. and flashdesorption. In the A.e.s. measurements the peak-to-peak height of a differentiated spectrumof nitrogen KLL Auger transition was used as a quantitative measure. This measureproved to be appropriate, until nitrogen coverage ON reached 1 [W(surface) : N(adsorbed) =1 :1 as defined later], by comparing the A.e.s. data with that obtained by flash desorptionespecially above 840 K, as shown in fig. 1. At higher coverage the linearity of the A.e.s.peak intensity with flash desorption results began to deviate, since the escape depth for theAugcr electrons is limited to only a few layers in the surface region.0.5 1& measured by A.e.s.< 770K.FIG.1.-Relation between measured by A.e.s. and flash desorption. 0, at > 840 K ; 0, a282 NH3 DECOMPOSITION ON WFlash desorption spectra were obtained by heating the sample, in a short time, to hightemperatures where no appreciable amounts of nitrogen or hydrogen are adsorbed. Thetime integral of the pressures of desorbing gases and the pumping speed of the system, whichhad been measured for each of N2 and H2, gave the total amount of adsorption.A monolayer coverage of nitrogen atoms on the surface was estimated from NHs and Nzadsorption.If ON = 1 was defined as the nitrogen coverage when one N-atom is allottedto every one surface tungsten atom, the saturation coverage of N from NH3 and N2 adsorp-tion are considered as 6~ = 1 and 6~ = 0.5, respectively, according to Estrup and Anderson.sThe saturated adsorption of ammonia without nitride formation was obtained at 520-670 Kunder P m 3 = Pa, which demonstrated the adsorption of 1 . 2 ~ 10l6 N-atom on thesample. The saturation coverage of NH3 gave an Auger peak of nitrogen about twice asstrong as that of N2.MEASUREMENTS OF REACTION RATESThe rate of ammonia decomposition was measured by a flow method. Fresh ammoniagas was supplied to the reaction vessel while the system was being constantly pumped outfrom the other end, such that P m 3 was kept constant during the reaction.The rates ofN2 and Hz production were determined by the changes in their partial pressures in the systemwhen the sample-was heated to the reaction temperature and cooled to room temperature ;any background reactions which may occur over the filaments of the mass filter, for example,were thus subtracted.RESULTSINITIAL PROCESS OF AMMONIA DECOMPOSITION ON TUNGSTENThe initial process of the decomposition reaction is shown in fig. 2. A cleantungsten surface was cooled from 2000 K in a flow of ammonia gas to the desiredreaction temperature in a few seconds and the uptake of nitrogen by the surface andthe decomposition reaction started. The nitrogen coverage, ON, was monitored byA.e.s. The rates of nitrogen and hydrogen gas formation were followed by q.m.s.The adsorption of nitrogen from gaseous ammonia was rapid at first and then sloweddown to approach its saturation value, while nitrogen desorption was slow at firstand then more rapid, finally getting to its steady rate.Almost all the hydrogen atomsdesorbed at the moment of NH3 adsorption, no appreciable amount of hydrogenbeing left adsorbed on the surface above 870 K.The rate of nitrogen uptake from NH3 onto the surface, which we designate asYi,, and the rate of nitrogen desorption from the surface to form NZ, which isrepresented by Vout, are calculated usiqg the curves in fig. 2. Here, both V,, and1 -8 0.5 -2 4 6timelminFIG. 2.-Initial process of ammonia decomposition on tungsten.PNH~ = 1.6 x0, nitrogen coverage measured by A.e.s. ; 0, rate of N2 production.Pa, T = 973 KH. SHINDO, C . EGAWA, T. ONISHI AND K. TAMARU 283Vout are expressed by the number of N-atoms which get into or come out from thesurface per second per surface tungsten atom. They are expressed as follows :nw dtdnNH3(9) Fn= ---dnN(a) + K u t = --nw dtdnN2(g) Vout = - -nw dt ’Here, nw is the number of surface tungsten atoms and nNH3, rt,, and TI^(^) stand forthe numbers of NH3 and N2 gas molecules and adsorbed nitrogen atoms, respectively.Vout is calculated from the partial pressure of desorbing N2 and the pumping speedof the apparatus. Vi& obtained by differentiating the ON curve in fig. 2 with respectto time and adding Vout to it. Vin and Vout are plotted in fig.3 against ON during thedecomposition of ammonia at 1073 K, PNH3 = (1.7-9.8) x Pa. Vin decreasesconsiderably with the increase in ON and with the same ON value Vin increases inproportion to PNH3. On the other hand, You, depends mainly on ON and temperatureand not on PNH3 as is shown in fig. 3.The intersection of the two curves, Vin and Vout, gives the balancing point betweenthe supply and consumption of surface nitrogen, where a steady-state of the overallreaction is realized, ON staying unchanged with time. A different PNH, correspondsrout increases markedly with increasing ON.f0.1 0.2 0.3 0.4 0.5N(a) coverage, ONFIG. 3.-Rates of NH3 uptake and N2 desorption plotted against N(a) coverage on W. T = 1073 K,= dnN(a)/nw dt+YoUt.PNH3 ; 0, 0, 1.7~ ; A, A, 5.1 X ; m, 0, 9.8 X Pa.0, A, My vh = -dnNH,(g)/nwdt; 0, A, 0, Y&t = 2dnN2(g)lnw dt284 NH3 DECOMPOSITION ON Wto a different point of intersection, which means that the overall reaction rate stronglydepends upon PNH3 in this pressure range. The curve of H2 desorption plotted againsttime is almost the same as that of Vin, which suggests that all the H-atoms desorbedimmediately when the parent NH3 molecule decomposed to give adsorbed nitrogen.ADSORPTION AND RATE MEASUREMENTS I N STEADY-STATEAMMONIA DECOMPOSITIONThe rate of ammonia decomposition measured by the nitrogen production rateat 773-1273 K and PNH3 = 10-6-10-3 Pa is given in fig. 4. Turnover frequency isdefined as twice the number of produced N2 molecule on each surface tungsten atomper second.The dependence of the reaction rate upon PNH3 changes with temperature ;2/3 order at 773 K, M 4/5 order at 973-1273 K and exactly first order above 1473 K.ammonia pressure, */PaFro. 4.-Rate of ammonia decomposition on tungsten. Reaction temperature T : a, 773 ; a,973 ; 0, 973 (H2 rich) ; 0, 1073 ; 0, 1273 K.The Arrhenius plot at PNHs = 1.3 x Pa is given in fig. 5, which is composedof two straight lines which meet at M 1100 K. It is, accordingly, suggested thatabove 1100 K the reaction mechanism may change. The higher value of apparentactivation energy, 87kJmol-l, is approximately the same as that measured onW(1ll) and W(100) by McAllister and Hansen with 0.33-6.7 Pa of NH3.The amount of nitrogen atoms adsorbed on the surface during the course of thereaction axe given in fig. 6, which is measured at the steady-state, mainly by A.e.s.and partly by flash desorption.It was demonstrated that changes considerablywith temperature and slightly with PNH3. Combining the results given in fig. 4 and 6,it is demonstrated that a slight increase in results in a marked increase in the sate ofnitrogen formation, which has been already shown in fig. 3 by the steep increase ofYout. N(a) becomes more unstable at higher ONH . SHINDO, C . EGAWA, T. ONISHI AND K. TAMARU 285From the data shown in fig. 4 and 6, true activation energies for N(a) desorptionat constant nitrogen coverages were estimated. They are 310 kJ mol-l at ON = 0.63and M 190 kJ mol-1 at eN = 1.0.The activation energy decreases with the increasein &, which can be associated with the decrease in the heat of adsorption of nitrogen.As to the adsorption of hydrogen-containing species, the flash desorption experi-ments indicate that above 870 K virtually no hydrogen atoms stay on the surface inany form in the steady-state of the reaction. The surface is covered only by nitrogenatoms, as has been demonstrated for Mc, by high resolution Auger electron spectro-scopy.'-8-9M --1 00.6 0.8 1 *o 1.2lo3 KITFIG. 5.-Arrhenius plot of the rate of ammonia decomposition on tungsten. PNH~ = 1.3 x lW4 Pa,T = 773-1473 K. Apparent activation energy changes at M 1100 K from 87 kJ mol-1 for lowertemperatures to 42 kJ mol-1 for higher temperatures.0.25 11 d5ammonia pressure, PNH~ /PaFIG.6.-Nitrogen coverage during ammonia decomposition on tungsten, measured mainly byA.e.s. ; reaction temperature T : e, 773 ; a, 773 (measured by flash desorption) ; H, 973 ; 0,973(Hz rich) ; 0, 1073 K 286 NH3 DECOMPOSITION ON WThe effect of hydrogen partial pressure on the reaction rate as well as the nitrogenuptake during the reaction were examined at 973 K, by adding hydrogen to thereacting system and changing the PHz/PNHJ ratio from 0.1 (no added H,) to M 20.It was thus demonstrated that there were no changes in the adsorption nor in theoverall reaction rate due to the addition of hydrogen, as shown in fig. 4 and 6.FORMATION OF THICK NITRIDE LAYERS ATHIGHER PRESSURES OF AMMONIAWhen ammonia pressure becomes higher, nitrogen uptake during the reactionincreases markedly such that nitrogen atoms penetrate into the bulk of the tungstenand form multilayers of surface nitride.In these pressure ranges (1-100 Pa), areaction time of 15 min is sufficient to attain a quasi-steady-state. Fig. 7 gives theamounts of surface nitride formed in a closed system in 15 min of reaction plottedagainst PNH3. The amounts of nitride were estimated by flash desorption. Afterthe thick nitride layers were formed, the sample was cooled from the reaction tem-perature to room temperature and the whole system was evacuated at a constantpumping speed. Then flash desorption was carried out to estimate the nitrogenuptake.01 # I *ammonia pressure, PNH~ /Pa1 10 100FIG.7.-Average thickness of nitride layer formed in 15 min of NH3 decomposition on W in a closedsystem. [ON = 1 ; I1N(nitride)/I1W(surface) = 11 sample temperature; U, 973 ; 0, 1073 ; ., 1073 K(27 Pa of Hz added).Under ammonia pressures higher than 10 Pa the surface nitride becomes thickat both 973 and 1073 K. Experiments in the temperature range 973-1123 K showedthat at lower temperatures thicker nitride layers axe formed in the steady-state,although at still lower temperatures the mobility of N-atoms into the bulk becomesSO low that it is difficult to form thick nitride layers in a short time. The process ofnitride formation had been examined during the decomposition of NH3 and ND3and it was demonstrated that NK3 gives higher ON than ND3 does during the reaction.H . SHINDO, C. EGAWA, T . ONISHI AND K . TAMARU 2878 3 0.025-5 .d 270.6 0 . 8 1 .onitrogen coverage, ONFIG. %-Date of decomposition of nitride layers and rate of decomposition of ammonia on tungstensurface at 1073 K; 0, rate of decomposition of nitride; 0, rate of decomposition of ammonia.When the nitride in the surface was decomposed by heating in vacuo, the relationbetween the rate of nitrogen desorption and eN in the later stage of the decompositionwas obtained, which is shown in fig. 8. The rate of N2 production in steady-stateammonia decomposition is also shown in fig. 8 by open circles. The two curvesagree with each other reasonably well. The rate of N2 formation is determined byON and the reaction temperature, regardless of the presence or absence of ammoniain the gas phase.ADSORPTION OF Nz ON TUNGSTENThe adsorption of nitrogen molecules on tungsten was studied in the temperatureand pressure ranges 1073-1473 K and 10-6-10-3 Pa, respectively.Nitrogen pressurewas measured using q.m.s. and a €3-A gauge. The amount of Nz adsorption wasmeasured by A.e.s. At temperatures lower than 1073 K the adsorption equilibriumcould not be obtained.c.I I Initrogen pressure, PN,/PaA, 1173; 0, 1273; 0 , 1 4 7 3 K.FIG. 9.-Log-log isotherms for Nz adsorption on W obtained by A.e.s. measurements. 0, 1073 288 NH3 DECOMPOSITION ON WThe log-log plots of N2 adsorption isotherms on tungsten are given in fig. 9,where each of the isotherms exhibits a straight line.Those lines seem to convergeat a single point as observed by Davis and the point is considered to be the saturationcoverage of NZ(ON = 0.5).If we compare the nitrogen uptake at the steady-state ammonia decomposition(as given in fig. 6) with adsorption equilibrium of N2 (in fig. 9), it is clearly demon-strated that N(a) is not in equilibrium with N2(g) during the decomposition of ammonia.400d300 80.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0nitrogen coverage, ONFIG, 10.-Differential heat of adsorption of nitrogen and activation energy for desorption of nitrogenon tungsten. 0, differential heat of N,(g) 3 2N(a) adsorption ; eD, activation energy for 2N(a) --fN,(g) desorption ; (- - - -), heat of formation of bulk nitride.The differential heat of adsorption of nitrogen is obtained using the data of N2adsorption isotherms given in fig.9 and the Clausius-Clapeyron equation, which isindicated in fig. 10. The heat of adsorption, qe,, decreases with ON, although theslope is not so steep as that given by Davis.g The initial heat of adsorption is 450 kJmol-l, which is a little larger than 410 kJ mol-1 observed by Beek lo and approxi-mately equal to 443 kJ mol-l l1 calculated by the method of Higuchi et al.l2DISCUSSIONIn ammonia decomposition on group VIII metals such as iron and nickel, theTemkin-Pyzhev treatment can be generally applied by asswning that the desorptionof N(a) is the only rate-determining step, all the other processes being in quasi-equilibrium. In the case of tungsten, however, a different treatment is required,since the reaction rate is independent of hydrogen pressures.Even if the zero-orderkinetics at higher ammonia pressures is explained assuming the surface to be fullycovered in its working condition, it does not explain the kinetic isotope effect sinceno hydrogen is being adsorbed at the steady-state of the decomposition. The resultsshown in fig. 4 and 6 demonstrate that the reaction rate and at the steady-statedepend upon PNHB but not upon PHt, which was varied by two orders of magnitude.Accordingly, the formation of surface nitrogen species is not an equilibrated processwith ammonia and hydrogen.McAUister and Hansen assumed the following equilibrium during the reaction3WN(a)+NH3(g) + WZN3H2(a)+WNH(a)H. SHINDO, C.EGAWA, T . ONISHI AND K . TAMARU 289They assumed that the decomposition of W2N3H2(a) on a limited part of the surfacedetermines the overall rate of the reaction. The fact that practically no hydrogen isbeing adsorbed in any form on the catalyst surface means that, if W2N3H,(a) stayson the surface during the reaction, its amount should be very small.In general, if an adsorbed species, N,H,(a), plays a decisive role in determiningthe rate of nitrogen (N2) formation, its activity (aNmHnca)) should be dependent uponthe pressures of NH3 and H2 as in the case of a Temkiiz-Pyzhev mechanism (wheren = 0) provided step (1) is equilibrated. The fact that hydrogen does not affectthe rate of nitrogen formation excludes the partial equilibrium of step (1) or wouldsuggest that N,H,(a) saturates the active part of the surface, keeping its activityconstant for various pressures of ammonia and hydrogen.Since the rate of nitrogenformation is solely determined by [step (3) is rapid enough] irrespective of thepresence or absence of NH3(g) and H2(g), the latter possibility will be excluded andthe value of n should be zero, otherwise U N ~ H , , ( ~ ) would depend upon hydrogenpressure as it does upon & a Thus, step (1) is not equilibrated and the value of nis zero which gives a picture as follows :2NH3(g) 7++ ”1 + 3H,(g) (A)m a ) *-, N2k) (B)The reaction rate is determined by the “ dynamic balance ” between these twosteps; supply and consumption of N(a), Vi, and Vout, respectively, in fig.3. Thismechanism explains all the experimental facts, including the kinetic isotope effect.Step (A) is the process of supply of N(a) which proceeds readily at lower ON but withdifficulty at higher coverage. The probability of NH3(g) sticking as N(a) which isobtained by dividing Vi, by the impinging rate of NH3 molecules onto the surface,is larger than 0.5 at ON E 0.but readilyat higher coverage. The chemical potential, or decomposition pressure of the surfacenitride during the course of the reaction, increases with N(a) and is much higher thanthat of ambient nitrogen gas. Surface concentration of N(a), or the average thicknessof the surface nitride, is associated with the activity of N(a).As ON changes, the rates of step (A) and step (B) increase or decrease in oppositedirections.As a result, under a given reaction conditions, step (A) and step (B) arefinally balanced at a corresponding ON and the steady state is attained. When thebalance is attained at a small ON as in fig. 3, PNHB dependence of the reaction rate isnearly first order. However, when the steady state is obtained at higher wherethe Vout curve becomes flat, an increase in PNH3 does not contribute much to theStep (B), the process of N(a) desorption, proceeds slowly at lower1-1290 NH3 DECOMPOSITION ON wincreased N2 formation rate, although steady-state f3N is increased. This is why thereaction rate dependence upon PNH3 decreases from nearly first order at lower PNH3and higher temperatures, where ON is small, to nearly zero-order at higher PNHJ andlower temperatures, where f3N is large.The kinetic isotope effect, which at first seemed difficult to explain, is clearlyunderstood as the difference between the thickness of the surface nitride formed fromNH3 and ND3,S as was first expected by Tama.r~.~ The difference between the ratesof step (A) with NH3 and ND3 results in nitride layers of different average thicknessesbeing formed during the steady state of the reaction, or in a different chernical potentialof N(a).As the result, NH3 decomposes faster than ND3, for the rate of step (B)depends upon only ON.SUMMARYThe mechanism of the decomposition of ammonia was studied by measuring therate of the overall reaction and the amount of adsorption during the course ofammonia decomposition using a clean tungsten surface in a wide pressure andtemperatwe range.The measurements were performed not only at the steady stateof the reaction but also in non-steady states. The results demonstrated that thereaction proceeds through a mechanism of " dynamic balance " between two con-secutive rate-determining steps ; the supply of surface nitrogen atoms from ammoniaand its consumption by formation of nitrogen molecules, the amount of nitrogenuptake during the reaction being determined by the " dynamic balance ". The changein the dependence of the total reaction rate upon PNH3 is well understood. Thekinetic isotope effect in the decomposition of NH3 and ND3 is also reasonablyexplained by the reaction mechanism, giving different amounts of adsorbed nitrogenin the steady-state decompositions.The present work was partially supported by a Grant-in-Aid for ScientificResearch from the Ministry of Education and the Toyota Research Fund. Wethank Dr. Mitsuyuki Soma of the University of Tokyo for his critical discussionsand valuable advice on the experiments. We also thank Dr . Tamotsu Kondow of theUniversity of Tokyo who helped us construct the A.e.s. system and gave us adviceon experimental techniques.C. N. Hinshelwood and R. E. Burk, J. Chem. SOC., 1925,127, 1105.W. Frankenburger and A. Holder, Trans. Faraduy SOC., 1932,28,229.J. C. Jungers and H. S. Taylor, J. Amer. Chem. SOC., 1935, 57, 679; R. M. Barrer, Trans.Furuday SOC., 1936,32, 490.K. Tamaru, Trans. Furuday Soc., 1961,57, 1410.P. J. Estrup and J. Anderson, J. Chem. Phys., 1968,49, 523.ti J. McAllister and R. S. Hansen, J. Chem. Phys., 1973,59,414. ' T. Kawai, K. Kunimori, T. Kondow, T. Onishi and K. Tamaru, Phys. Reu. Letters, 1974,33,533.H. Shindo, C. Egawa, T. Onishi and K. Tamaru, 2. Nuturforsch., 1979, Ma, 96. (In thisreference, &=1 was defined as the nitrogen coverage when one N-atom is allotted to everytwo surface tungsten atoms.)R. T. Davis, Jr., J. Amer. Chem. Soc., 1946, 68, 1395.lo 0. Beeck, Adv. in Cutulysis, 1950,2, 151.I'D. 0. Hayward and B. M. W. Trapnell, Chemisorption (Butterworths, London, 2nd edn,l2 I. Higuchi, T. Ree and H. Eyring, J. Amer. Chem. SOC., 1957, 79, 1330.l3 M. I. Temkin and V. Pyzhev, Actu Physicochim., U.S.S.R., 1940, 12,327.1964), p. 208.(PAPER 9/437

ISSN:0300-9599    

DOI:10.1039/F19807600280

出版商:RSC    

年代:1980

数据来源: RSC