摘要:

Chemical Society Reviews Vol 6 No 21977 Page CENTENARYLECTURE Light Scattering in Pure Liquids and Solutions 109By W. H. Flygare The Binding of Heavy Metals to Proteins By T. L. Blundell and J. A. Jenkins 139 The Chemistry and Binding Properties of Aluminium Phosphates By J. H. Morris, P. G. Perkins, A. E. A. Rose, and W. E. Smith 173 The Organic Chemistry of Superoxide By E. Lee-Ruff 195 Immobilized Enzymes By C. J. Suckling 21 5 The Polymerization and Copolymerization of Butadiene By D. H. Richards 235 The Chemical Society London Chemical Society Reviews Chemical Society Reviews appears quarterly and comprises approximately 25 articles (ca. 500 pp) per annum. It is intended that each review article shall be of interest to chemists in general, and not merely to those with a specialist interest in the subject under review.The articles range over the whole of chemistry and its interfaces with other disciplines. Although the majority of articles are intended to be specially commissioned, the Society is always prepared to consider offers of articles for publication. In such cases a short synopsis, rather than the completed article, should be sub- mitted to The Editor, Books and Reviews Section, The Chemical Society, Burlington House, Piccadilly, London, W 1V OBN. Members of the Chemical Society may subscribe to Chemical Society Reviews at E5.00 per annum; they should place their orders on their Annual Subscription renewal forms in the usual way. Non-members may order Chemical Society Reviews for E14.00 ($30) per annum (remittance with order) from: The Publications Sales Officer, The Chemical Society, Blackhorse Road, Letchworth, Herts., SG6 1 HN, England. 0Copyright reserved by The Chemical Society 1977 Published by The Chemical Society, Burlington House, London, WlV OBN Printed in England by Eyre 83. Spottiswoode Ltd, Thanet Press, Margate

ISSN:0306-0012    

DOI:10.1039/CS97706FP005

出版商:RSC    

年代:1977

数据来源: RSC

摘要:

CENTENARY LECTURE* Light Scattering in Pure Liquids and Solutions By W. H. Flygare NOYES CHEMICAL LABORATORY AND THE MATERIALS RESEARCH LABORATORY, UNIVERSITY OF ILLINOIS, URBANA, ILLINOIS 61801, U.S.A. 1 Introduction and Development of the Theory We start by writing some general relations between measurements of electro- magnetic radiation in the time and frequency domains. The total intensity of light (electromagnetic radiation) is written as a long-time average of the instantaneous intensity, I(t), where the brackets indicate the long-time average. The instantaneous intensity is given in terms of the electric fields of the electromagnetic radiation according to I(t) = 477 E*(t)E(t) , where we use the complex form for the instantaneous electric field, E(t).We also define the spectral distribution function, I(w), which has units of energy (area x t x w)-l. Thus, the total intensity which is equal to Iin equation (1) is given by integrating Z(w) over all frequencies : I fa3 --co = 5 I(w)dw T+cO 2T -T = lim {L 1’’I(t)dt) - (3) Substituting equation (2) into equation (3) gives I = 9 ’ I(w)do -a = (Z(t)) = (E*(t) E(t)) ,47r (4) where the brackets again indicate the long-time average. We now define the correlationfunction for the electric field,] C(T)= lim L-{LT+o3 47r E*(t)E(t + 7)dt] = (E*(t)E(t + 7)) ,477 (5) * Based on the Centenary Lectures of the Chemical Society given in January, 1976, and on a short course on Light Scattering given at Colorado State University in June, 1976.The measurement of the correlation function from optical electromagnetic fields is discussed in B. Chu, ‘Laser Light Scattering’, Academic Press, New York, 1974. Light Scattering in Pure Liquids and Solutions where we note from equations (4) and (5) that It is also easy to show that C(7)and I(w)are Fourier transform pairs related by C(T)= I(w)exp (-iw7) dw5:: 1 f+oo I(w) = v -Re C(T)exp (iw7) d7 , (7)5: where the last step indicates the real part of the integral. This last important K step is easily proven given that I(w) is real. It is now evident that if we can measure either C(T)or Z(w), we can use the Fourier transform to obtain the other member of the pair.2 The above definitions and equations are general and can be used to analyse the spectrum of a given distribution of electromagnetic fields.We will now introduce the more useful ensemble average over all positions and momenta which is essential to evaluating the correlation function for a system of scattering atoms or molecules. First, we define a stationary system which requires that the time average in equation (5) leading to the correlation function is independent of the origin in time. Thus, for a stationary system we can write Now for a stationary system, the ergodic hypothesis states that each scattering system in the ensemble of particles will pass through all values accessibIe to it, given a sufficiently long time. Thus, the time average is essentially the same for all systems of the ensemble.The result is that for a stationary ergodic system, the time average is equivalent to the ensemble a~erage.~ When the brackets in equation (5) indicate an ensemble average, the relation between C(7) and I(w) discussed above is called the Wiener-Khintchine theorem.4 We will find it useful to examine more carefully the properties of the correlation functions and spectra as related to a scattering experiment. Starting with zy-polarized radiation travelling along the y axis, as shown in Figure 1, we can write the complex dipole field scattered from thejth scatterer into a detector at a distance R from the origin of the scattering system according to5 a ‘Photon Correlation and Light Beating Spectroscopy’, ed. H. Z. Cummins and E.R. Pike, Plenum Press, New York, 1973. F. Reif, ‘Fundamentals of Statistical and Thermal Physics’, McGraw-Hill, New York, 1965. C. Kittel, ‘Elementary Statistical Physics’, Wiley, New York, 1958. See, for instance, P. Lorraine and D. R. Corson, ‘Electromagnetic Fields and Waves,’ W. H. Freeman and Co., San Francisco, 2nd Edn., 1970. Flygare where Uyis the unit vector along the polar angle y. wo and EO are the frequency and amplitude of the incident radiation, respectively, aj(t)is the polarizability tensor of the jth scatterer in the laboratory-fixed axis system (which is time incident radiation parallel to z I i pinri7ation 1 clnalyzer Scattering Vector k,. -k,s K = .cl-rrn sin(8,/2)10 Figure 1 Basic scattering diagram showing plane-polarized z axis (I!) incident radiation which is scattered along the line shown to the detector at angles y and BS.Choosingy = n/2,we note either polarized (z axis, 1) or depolarized (xyplane, I)scattered light designated by IvV and Ihv, respectively. koand ksare the incident and scattered wave vectors, n is the refractive index in the scattering medium, A, is the vacuum wavelength of the radiation, and K is the scattering vector. dependent due to the rotation of a non-spherical scatterer), K = ko -ks is the scattering vector, which bisects the angle between the incident (ko)and scattered (ks)radiation, and rj(t) is the centre of mass (cam.) position of thejth scatterer from an arbitrary origin in the scattering system.rj(t)is, of course, time dependent if the scatterer has translational freedom (liquid or gas). Figure 1 shows the basic scattering diagram where the incident light, travelling along the y axis, is plane-polarized in the vertical z direction (ExO = Ego = 0) and the scattered light is observed along a line which is at an angle of y with respect to the z axis. The scattering angle 8, is between they axis and the line of observation. Of course, we could also start with x-polarized incident radiation. Of the large number of incident and scattered polarizations and values y and &, we will usually choose a set of values which corresponds to those which are most often used for experi- mental studies. These correspond to zy-polarized incident radiation and obser- Light Scattering in Pure Liquids and Solutions vation in the xy plane (y = 5712) of either parallel (11) or perpendicular (I) polarized scattered radiation.These choices are also shown schematically in Figure 1. A polarization analyser selects the scattered radiation which is polarized parallel to either the z( 11) or y(I) axes). These xy in-plane fields are given from equation (9). W0211, E2j(t)= -Ezoazzj(t)exp (-hot) exp [iK-rj(t)] RC2 W02I,E1/j(t)= 2Ezomzyi(t)exp (-iwr) exp [iK*rj(t)] . (10)RC We can now write the total fields scattered in phase into the detector by summing over allj scatterers. Substituting this sum into equation (8) gives the correlation functions : c11, Cz(7)= -(Ez(O)*Ez(7)) = A (C azzi(0)c~(T)exp (-iom)47r i.i exp {-iK-[ri(O) -Q(T)]}) (11) I,~(7)477 = exp (-iwo7)= C (E,(O)*~~(7))A azyi(o)azl/i(T> exp { -iK-[ui(O) -Yj(T)]} where the sums over i and j are independent with all terms in i at t = 0 and all terms in j at t = T being included. I0 is the incident intensity for the plane- polarized radiation. If the incident radiation is unpolarized, we write COO*A=-c4R2 (5) (1 + COS2 8,) . The correlation functions in equations (1 1) and (12) are written6 in terms of the initial t = 0 and later t = T positions and orientations and the brackets indicate the time average or the equivalent ensemble average according to the ergodic hypothesis. The orientational correlation is contained in the azzi(0)azzj(T)andazl/i(0)CX~~~(T) termsand the translationalcorrelation iscontainedin theexp { -iK-[ri(O)-Q(T)] } C,(T), which indicates the correlation function for y-polarized scattered light, is valid at any angle 6, in the xy plane (y =7r/2).In this expression we use CV(7)=[c/(47r)] <Ey(f)Eu*(f+T)>.However, C,(T) =[c/(47r)] (Ez(t)Ez*(t+T)) is equally valid. We can write the scattered field at angle 8, in Figure 1 as a linear combination of E,(t) and Ey(t),Eos = sin &Ex + cos &E,, where<EeS(t)Ees*(f+ T)) = sin2 0, (E,(t)E,*(t + T)) + cos2 OS <E,(t)E,*(t+T)>,where (E,(t)E,(t + T)> = 0. Now, it is easy to show that <EZ(t)Ez*(f+ T)> = <EV(r)Ey* (I + T)) and, therefore, (Ep(t)Ep*(t + 7)) = (Ez(t)Ex*(t + 7)) = <Ey(t)Ey*(t+ 7)). Flygare phase factors.The i = j terms are called the self terms and the i # j terms are called the distinct terms. In gases, it is reasonable to assume that the dominant contributions will arise from the self terms. Of course, in liquid crystal-like systems we would expect significant i # j contributions and in solids we expect even larger i # j contributions. We will now examine the nature of the polarizability tensor elements which enter the correlation functions in equations (11) and (12). We remember that in the presence of an electromagnetic field with frequency w, the system (molecule) is described by the time-dependent function y(r,t) in terms of the stationary states, $&), and time-dependent coefficients, Br(t), according to Y(r,t) = Bi(t)$&), where Bi(t)= exp (-iEit/h) Ci(t).Assuming that the perturbation is small and that the system is initially in the kth state, Ci(t)4 G(t)z 1.0 and we write the time-dependent function as Y(r,t) = $k(r) exp (-iEkt/h) + F'C&) exp (-iEit/h) $i(t) , (14) where the prime on the summation excludes the kth term.The value of G(t)is given from perturbation theory for the electric dipole interaction Hamiltonian perturbation, X' = -E-D, between the molecular dipole moment operator, D, and the incident radiation electric field, E = 2Eocos(k*r -ot), We can think of the kth state as a vibration-rotation state in the ground elec- tronic state and the sum over i is over the excited electronic (vibration-rotation) states, We now calculate the average value of the D operator to first order in Ci(t)for the time-dependent system described in equations (14) and (15) to give Light Scattering in Pure Liquids and Solutions where Dki = #k*(Y) D$t(r) d V and the denominator is unity to first order in the coefficients, C&).We note from this expression that k and k’ can represent two different vibration-rotation states in the ground electronic state and the primes on the summation omit the i = k and i = k’ terms, respectively. Remembering that Wik -~ik.= 1/fi (Ei -Ek -Ez + Ek!) = 1/fi(Ek#-Ek) = ~k.k= -w~R,,where Wik is an electronic transition frequency and Wkkt is a vibration-rotation or rotational transition frequency in the ground electronic state, we can rewrite this equation to give Eo exp [i(wk + w)t] + exp [i(Wk.k -~)t]-k -c’Dkti Dfk Wik, -cr) Wik.+ 0 } (17)hi This general expression for the polarizability tensor shows that the polarizability and resultant intensity of the scattering will increase as w + WZk. This condition of resonant Rayleigh (k = k’)7 and resonant Raman (k # k’) scattering is useful in enhancing the sensitivity of the scattering if a suitable light source is avaiIable with a frequency, W, near one of the electronic resonant frequencies, ~ik.Resonant vibrational Raman scattering has been particularly useful in identifying specific local vibrations on a site in a large macromolecule.8 kNow, we note that normally ~i 3-Wkpk because ~i iskthe angular frequency of an electronic transition and wk,k is the angular frequency of a vibrational, vibration-rotation, or pure rotational transition.Thus, it is reasonable to assume that ~i =k WZkt for Raman and Rayleigh scattering. Now if LLI~~II WZ~. in equation (1 7), we can write Thus, the polarizability is given from Dind = a.E to be We can relate the space-fixed polarizability tensor in equation (19), a(xyz), to the corresponding values in the molecular-fixed axes, a(abc), by the direction cosine transformation, C, which gives a(xyz) = C+a(abc)C. According to equations (1 1) and (12), we need ffzz = CzaaaaCaz + Czbabbcbz 4 CzcaccCcz ffzy = CzaaaaCay -k CzbabbCby + czcaccffcy , D. R. Bauer, B. Hudson, and R. Pecora, J. Chem. Phys., 1975, 63, 588. 8T.G. Spiro, Accounts.Chem. Res., 1974, 7,339. Ftygure where Czais the direction cosine between the a and z axes. Now if we choose a cylindrically symmetric molecule with aaa# abb = acc(where a is the symmetry axis) and if we place the a, b, and z axes in a plane (with no loss in generality), we can relate a(xyz) to a(abc)through the standard spherical polar angles, 6and cp. If x + b, y -+ c, and z -+ a when 8 = cp = 0, we can show that azz = a + Q( -~abb) ~P ~ ~e) (~~~ azy= (aaa-abb)cosOsin#sin cp (20) where Pz(c0s 8) = +(3 cos # -+) is the Legendre polynomial of order I = 2 and a = +(aaa+ am + act). If the scattering (molecule) is vibrating, we must also include the vibrational dependence in the polarizability tensor elements by expanding each tensor element in the molecular-fixed axis system about the small-amplitude molecular vibrations according to where aaaois the equilibrium structure polarizability.The change in polarizability with normal co-ordinate, Qi, is evaluated at equilibrium. In summary, we now have the formal expressions to write the correlation functions in equations (1 1) and (1 2) for the general vibrating-rotating molecule. We now return to equations (11) and (12) to examine the correlation functions at very high pressures or in the condensed phase (liquid) where the time between collisions is short relative to a molecular-rotation period. To simplify our dis- cussion we will initially assume a rigid non-vibrating rotor and substitute equation (20) into equations (11) and (12).If the scatterers are randomly orien- tated in space (isotropic distribution), the averages of the aPz(c0s @j and aPz(cos #)i cross-terms will vanish. We also assume that all non-vibrating molecules or scatterers in the gas, liquid, or solid are equivalent and that a and (aaa-am) are time-independent. Acknowledging the above statements, we obtain I,Cy(t)= A exp (-iwt)(aaa -abb)2 x (c(cos o sin 8 sin cp)oi (cos e sin e sin q)j i, j exp { -iK*[ri(O) -rj(t)l1), (23) where 8 and cp are time-dependent due to molecular rotation. The subscript 0 are the t = 0 values in the correlation functions. The terms within the time- Light Scattering in Pure Liquids and Solutions average brackets contain the independent sums over i andj.The i = j are called the self terms and i # j terms are called the distinct terms as mentioned previously. Having defined the concept of the self and distinct correlations, we now rewrite equations (22) and (23) in a form which includes both the self and distinct parts in single terms. Using the ergodic hypothesis, the correlation functions for identical scatterers are given by the following averages over position and orienta- tion: CZ(t)= A exp (-hot) Na2 exp (iK*R) P(R, t) d Vs + vs g Aexp (-iwot) N(aaa -abbl2 47T x dVs d cos 80 d cos 8 dqo dcp (24) x (cos 8 sin 8 sin cp)~(cos 8 sin 8 sin q) x P(R,8, q,t)dVs d cos 80d cos 8 dqo dq . N is the number of scatterers within the scattering volume Vs,and dVs indicates the volume element and corresponding integral over the scattering volume (the illuminated volume which is focused onto the detector).The P(a) functions in these expressions are the space-time and space-time-orientation correlation functions. P(R,t) is the probability per unit volume that if a scattering centre is at position R = 0 at t = 0, there will also be a scattering centre at R at t. Thus, it is evident that P(R,t) contains both the self and distinct terms in both space and time as described following equation (23). P(R,8, cp, t) is the space- time-orientation correlation function which is the probability per unit volume that if a particle's centre of interaction is at R = 0 with orientation 80 and qo at t = 0, there will also be a particle at R with orientation 8 and q at t.The additional factor of [1/(47r)]associated with theP(R, 8, q,t)probabilities assumes normalization of the probabilities to the t = 0 initial conditions of P(R,8, y, t) = 8(R)S(C0S 8 -cos 80) S(Cp -yo). Equations (24) and (25) will be used at various points in the remaining parts of this paper to describe the scattering from rotationally quenched systems. We will use hydrodynamic theories to obtain expressions for P(R,t)andP(R, 8, cp, t). For instance, P(R, t) in the first term of equation (24) is obtained for a pure liquid from a solution of three coupled equations; the continuity equation, the Navier-Stokes equation, and the energy transfer equation which leads to Rayleigh-Brillouin scattering (see next Section).The total intensities from Cz(t = 0) and Cy(t= 0) are obtained by substituting the appropriate initial or static conditions into equations (24) and (25). Using Flygare P(R,0) and P(R, 8, 9,O) = P(R,0) &cos 8 -cos 80) 8(cp -qo) in equations (24) and (25) gives the intensities: Iv" = Cz(0) = ANp(K,0) [a2+ &aaa -abb)2] = Iiso + 4Ianis (26) Ih" = Cy(0) = ANF(K,0) (&)(a:a@ -abb)2 = Ivh = Ihh = Ianis (27) Thus, if z-polarized incident radiation is used and observations are made in the xy plane, the ratio of perpendicular to parallel polarized light intensities is This depolarization ratio is a well-known result. Wilson et al.9 have derived a number of depolarization ratios by using similar methods. Equations (26) and (27) are general for any type of molecule by replacing (aaa -Olbb)' with &[(auu-am)2 + (am -acc)2+ (acc-aaa)2].The Ivv and Ih" notation is shown in Figure 1 and the other types of 1~~~~~~~are also evident from the geometry in Figure 1. The Ihh expression in equation (27) is easily derived giving a correlation function equal to the result in equation (25) with the cos 8 sin 8 sin cp dependence being replaced by sin2 8 sin cp cos cp. It is then easy to show that Ihh = Ihv as indicated in equation (27). We now return again to equations (11) and (12) and repeat the analysis described above including the effects of parallel vibrations in the linear molecule. Substituting equations (20) and (21) into equations (11) and (12) and repeating the quenched rotational state analysis described above gives CZ(t)= A exp (-iwot) N(cxO)~P(K, N(aaaO-abbo)2t) + 4A -9 exp (-iwot) --47r x H(K,8,cp, t) + A exp (-iwot) exp (iconj,,njt) c(K,t)Jj exp (-t/~j) N(aaa0 -abbo)2 r Cy(t)= A exp (-hot) P(K,8, cp, t) + A exp (-iwot)47T c(K,t) = exp (-iK*R)G(R,t)d Vs Jv3 E.B. Wilson, jun., J. C. Decius, and P. C. Cross, 'Molecular Vibrations', McGraw-Hill, New York, 1955. Light Scattering in Pure Liquids and Solutions x G(R,8,q,t) d Vsd cos 80 d cos 8 dqo dq x (cos 8 sin 8 sin q) x P(R, 6,cp, t)dV, d cos 80 d cos 6' dqoo dq x (COS8 sin 6 sin cp) x G(R,8,cp, t)d Vsd cos 00 d cos 8 dcpo dq Nn,is the number of molecules in the rz, state. The first two terms in G(t)are identical to the result in equation (24) and the first term in C&) is identical to the result in equation (25). The remaining terms in G(t)and Cy(t)are due to parallel molecular vibrations in the cylindrically symmetric molecule (aaa# (llbb = act).The probability functions and their spatial Fourier transforms (which includes the orientational averaging) use the important approximation that the integral over the scattering volume, Vs,can be extended to an integral over a scattering volume where R -+ 03 to give the spatial Fourier transform. The P(R, t) and P(R, 8,cp, t) terms contain both self and distinct terms. G(R,t) and G(R,8,cp, t) differ from P(R,t) and P(R,8,cp, t), respectively, in that the G functions contain only the self terms. This is because the distinct terms in the G functions involve the molecular vibrations of pairs of different molecules which will have random phases with respect to each other.Thus, the distinct terms, involving the sums over pairs of vibrating molecules, are expected to vanish. Of course, if we are examining a system where distinct terms are in general negligible, the G and P functions will be identical. We have also added the exponential vibrational relaxation process to the correlation functions in equation (29) where rj is the vibrational relaxation time for thejth normal mode of vibration. Returning to equations (26) and (27), we can generalize to include the Raman terms : Ih"(W) = ZVh(W) = Zhh(W) = ZR;$T(U) + zyy(w) , (30) where the superscripts RAY and RAM indicate Rayleigh and Raman scattering, respectively.FIygare In the next Sections we will examine in detail the nature of P(K, t),P(K, 8, q, t), G(K,t), G(K,8,'p, t), p(K, 8, rp, t), and E(K, 8, rp, t) for a variety of condensed p hase-scat t er ing sys t ems . 2 Isotropic Rayleigh and Brillouin Scattering in Dense Gases and Pure Liquids We now examine the spectra arising from the isotropic first term in CZ(t)in equation (29), CZ(t)= A exp (-iwot)N(LXO)~p(K, t) , (31) where p(K, t)is also defined in equation (29) as the Fourier transform of P(R,t), the space-time correlation function. Of course, both the self and distinct cor- relations are contained in P(R,t). The contribution made by this P(R,t) term to the Ivv(w) spectrum in a pure liquid is most easily observed in systems where (aaca: a~b)= 0 and the remaining terms in CZ(t)in equation (29) go to zero.-Mountain10 and Pecorall have discussed the evaluation of P(K, t) in a dense gas or liquid in terms of the density-density space-time autocorrelation function. This is equivalent to evaluating p(K, t) directly from the following three coupled differential equations12 in P(R,t) which is the reverse spatial Fourier transform of P(K, t) needed in equation (29): the continuity equation the Navier-Stokes equation ar 2192 (T)vs24po VT(R,t) -47s + rlB V2I = 0 (33)-+ -VP(R,t) + -at Y Y and the energy-transport equation W(R,t) Cv(y4 -1) aP(R, t) -NAxV~T(R,t) = 0 . (34)pocv ----at at P(R,t) is the probability per unit volume that a scatterer is at R at t and I is the probability current or flux (the number of particles passing through a unit cross-section per unit time).T(R,t) is the temperature at R at time t, us is the velocity of sound in the medium, y = Cp/Cv,where C, and Cv are the heat capacities at constant pressure and volume, respectively, 8 is the thermal expan- sion coefficient, qs and are the shear and bulk viscosities, x is the thermal conductivity, and p is related to the number density, denoted by p = (M/NA)~o, where M is the molecular weight and NA is Avogadro's number. The use of the above linearized equations will be valid in the hydrodynamic realm with small excursions from equilibrium. Only the longitudinal coupling lo R. D. Mountain, Rev.Mod.Phys., 1966,38, 205. l1 R. Pecora, J. Chem. Phys., 1964,40, 1604. l4 K. F. Herzfeld and T. A. Litovitz, 'Absorption and Dispersion of Ultrasonic Waves', Academic Press, New York, 1959. Light Scattering in Pure Liquids and Solutions of the velocity to the density is included. This simplification, where angular correlations between molecules is unimportant, will limit the final results to polarized spectra, Ivv .We also note that density, or probability per unit volume, and temperature are used as the independent variables. Equations (32), (33), and (34) can be solved by using Laplace and Fourier transform methods to give P(K,t) z p(K,Oj -21. p?r -+ XNA(1 -;)) . (35)P pocv P(K, 0) is the static correlation which we will discuss later in this Section x is recognized as the thermal diffusion coefficient and r is the eflective mass diffusion coefficient for sound waves in the medium.Equation (35) is only valid in the limit where vsK & KK~.The first part of the p(K,t) in equation (35) arises from the fluctuations in entropy at constant pressure. The decay of these fluctuations have a time constant of r = I/KK~.The second part of &K, t) in equation (35) arises from fluctuations in pressure at constant entropy which leads to a propagating sound wave with velocity T~ and decay-time constant of rs = i/T'K?13 We now proceed to evaluate the spectra of the scattered light. Substituting equation (35) into equation (31) gives the correlation function for the isotropic scattering of Cz(K,t) = ANaT(K, 0) {(1 -i)exp (-iwot -rcK2t) + -1 [exp [ -i(w0 -v&)t -TKzt] + exp [ -i(w0 + vsK) -rK2t112Y where we have expanded the cos v,Kt term into its complex components.The real Fourier transform of Cz(K,t) [equation (7)] gives the isotropic spectrum as a sum of normalized Lorentzian functions, YKand Yr: -41) 1 KK~-w)= -7T (00 -0)2+ (KK2)21 1 rK --0)= -(37)7i-(a0 * ZlSK -0)2 + (rK2)21 l3 I. L. Fabilinsky, 'Molecular Scattering of Light', Plenum Press, New York, 1968. We also ignore in our discussion here any possible coupling with the internal modes of relaxation in the molecules. See R. D. Mountain, J. Research N.B.S., 1968, 72A, 95, for further details. Flygare The isotropic spectrum predicted in equation (37) is composed of a Rayleigh line centred around the incident radiation frequency, wo, with half-width at half-height of dw = KK~.In addition to the central line at 00,there are two Brillouin lines shifted symnietrically from 00 by fv,K on each side of the central line with half-widths given by dw =rK2.The ratio of integrated intensities for the Rayleigh and Brillouin curves give the value of y = Cp/Cv: This result is the well-known Landau-Placzek ratio. Returning now to a discussion of the Rayleigh-Brillouin spectrum, we can calculate the relative half-widths at half-height of the central Rayleigh [dv = (rK2)/(2r)] and side-band Brillouin [dv = (KK2)/(2T)] spectra from equation (37). We normally find that r 3-K.Thus, the central Rayleigh spectrum is normally a sharper spectrum than the side-band Brillouin spectra.14 Normally the Rayleigh-Brillouin triplet is observed with a Fabry-Perot interferometer.15 The parameters obtained by light scattering according to equation (35) can also be related to the absorption coefficient, y, for sound in the pure liquid, according to where T~ = (l)/(rK2) is the sonic relaxation time. Ultrasonic techniques have been used extensively to measure molecular relaxation processes.12 The tie-in by analysing the Brillouin half-widths has been a more recent development. In principle, according to the simple first-order theory given here, classical sound absorption experiments (y is the absorption coefficient) which measure y/vs2 and 7~, as a function of vs give exactly the same information through equation (39) as the Brillouin shifts, VB, and widths, ~VB,which give the velocity and mass-diffusion coefficient, respectively, as a function of frequency.In order to evaluate the static correlation, H(K,0) in equation (37), we return to our earlier discussion of the first terms in equation (22) leading to the isotropic term in Iv"(w)as finally written in equation (37). Rewriting the Np(K, 0) part of Cz(K,t = 0) from equations (22) and (24) gives l4 The width of the central line can be measured by using optical mixing techniques as shown by J. B. Lastovka and G. B. Benedek, Phys. Rev. Letters, 1966, 17, 1039 who measured K in toluene by this method. l6 See, for instance, G.I. A. Stegeman, W. S. Gornall, V. Velfera, and B. P. Stoicheff, J. Acoust. Soc. Amer., 1971, 49, 979. Light Scattering in Pure Liquids and Solutions NF(K,0) = N s exp (iK-R) P(R,0) d Vs(: = exp { -iK-[rr(O) -rj(O)]} = (C exp { -iK-[ri(O)-C exp { -iK*[ri(O) -rj(O)]}) 9 i=j ri(O)]1) + (i, (40) where N = poVSis the total number of scatterers in the scattering volume, Vs. We have rewritten the bracketed term as a sum of self (i = j) and distinct (i # j) terms. Accordingly, we can write P(R,0) in terms of a self and distinct part. The self part of P(R,0) is clearly a delta function in R and the second (distinct) term can be written in terms of a two-body radial distribution function, g(R), giving [g(R)is dimensionless] P(R,0) = 8(R) + pog(R) .(41) g(R)is the probability of finding a particle at R if there is another particle at the origin. g(R) is normalized to unity at large distances which requires P(R,0) to reduce to PO as R + a.Substituting P(R, 0) in equation (41) into equation (40) and using I exp (-%OR) 8(R)d Vs = 1, we can write P(K,0) + ~S(K)c~(K,O)isot = ~~112~ = ACX~N[S(K) J S(K) = J exp (-iK*R) (8(R)+ po[g(R) -l]} dVs . (42) S(k) is called the structure factor for the liquid16 and the 8(K)term leads to the forward scattered light which will be indistinguishable from the forward travelling incident light. Thus, the S(K) term is the only measurable K-dependent term in the scattered light intensity. The integral in equation (42) can be simplified considerably if we are dealing with optical radiation where the distances probed by the radiation are considerably larger than the distances from R = 0 to the first few molecular diameters or periodic variations in g(R). Under these circumstances, exp (-iK-R) = 1 -iK*R + .. . z 1 and we can write Po exp (-iK*R)[g(R)-1] d Vs z PO J [g(R)-1 3 d V,, which can be evaluated by statistical mechanics17 to give ---1 = pokTPT -1 . (43)PO s [g(R) -l]dVs = pokT (-L:;)T PT = [-(11V)a V,@P]Tis the gas or fluid isothermal compressibility at temperature T and k is Boltzmann’s constant (PThas units of inverse pressure). This final result is independent of K. Of course, if static fluctuations extend out to a dis-tance XO(radiation) or if shorter wavelength radiation were used, the exp (-iK-R) part of the integrand must be included leading to a K-dependence in the final result.In summary, we note that in the low K limit where 1/K is large relative to the mean free path in a gas or where 1/K is large relative to the scatterer-scatterer P. A. Egelstaff, ‘An Introduction to the Liquid State’, Academic Press, New York, 1967. l7 T. L. Hill, ‘Statistical Mechanics’, McGraw-Hill, New York, 1956. Flygare distance in a liquid, S(K) = kTp&T, and the intensity of the isotropic scattered light (which excludes the forward scattered light) is proportional to the compress- ibility of the scattering medium from P(K,O) = pokTP~. (44) Substituting this result into equation (37) gives the complete result. 3 Anisotropic Rayleigh and Raman and Isotropic Raman Scattering in Liquids; Translational and Rotational Diffusion Returning to equation (29) we will examine P(K, 8,q, t)leading to I:$T(o), @K, 8, q,t) leading to Z::F(m), and G(K,t)leading to Iy$p(m).In Figure 2, the Zvv(v)R*Y = ZRky(v) + $ZR,$,Y(v) and Ihv(v)RAY = I:;T(v) scattered spectra of nitrobenzene are shown. The depolarized spectrum in the lower curve is pure anisotropic and the upper curve is a combination of the isotropic and anisotropic spectra. The Z(v)R&T triplet arises from density fluctuations as described in the last Section. Several depolarized or anisotropic Rayleigh and Raman spectra are shown in Figure 3. We note that the half-widths at half-height for both Z(v)!$T and Z(v)5tiF in CS2 and benzene are considerably larger than in nitro- benzene. Typical spectral Iinewidths at 8, = 77/2 (see Figure 1) are of the order of dv = 3 x 109-3 x 1011 Hz for these small molecules.First, we examine e(K, 8,q,t)or G(R,8, q,t)in equations (29) which leads to Zt$y(v) as shown, for instance, in Figure 3. We remember that G(R,8,q,t) is composed entirely of self terms, the distinct terms being zero due to the random phasesofvibration. Thus, G(R,8,q,t)contains only single particle contributions. We will now review the hydrodynamic Debye model for G(R,8,q, t)which describes both the centre of mass (c.m.) position (translational diffusion) and orientation (rotational diffusion) of the particle.The Debye model assumes that many collisions are required to reorientate the molecule. Beginning with the in equation (29), we average over the initial spherical polar 1;"sI exp (iK-R) PZ(COS6) Y(R,8, q,t)dV, d cos 8 dq1:"1;Pz(c0s 8) @K, 8,q,t)d cos 8 dq , (45) where %(R,8,q,t)is the probability per unit volume of finding the molecular c.m. at R with orientation 8 and q at time t. We start with a discussion of one-dimensional translational diffusion of a cylindrically symmetric molecule from a planar delta function in number density in the xy plane. The flux (number of particles per unit area per unit time), J(z), away from this plane of high concentration, is proportional to the gradient of the number density, N(z,t),along the z axis according to Light Scattering in Pure Liquids and Solutions I I 111111111111111111 I11 IIIII -10 -5 0 5 10 15 GH z Figure 2 The IVV(v)RAY and Ihv(V)RAy(& = n/2) spectra of nitrobenzene recorded with a Fabry-Perot interferometer at T = 297 K by A.K. Burnham and S. J. Bertucci with an Ar+ ion laser with A,, = 5145 A (see also A. Szoke, E. Courtens, and A. Ben-Reuven,Chem. Phys. Letters, 1967,1,87). The ZVV(v)RAY spectrum is a combination of the isotropic and anisotropicparts and the IhV(V)RAY spectrum is due only to the anisotropic component. Flygare s= c=s '/T=213K T=343K 50 -30 -10 10 30 50 CZ1 1266 Figure 3 Depolarized Zhv(v) = Z(v)anis Rayleigh and Raman spectra for CS, and benzene. The vertical bars indicate the approximate instrumental widths.Polarized lasers were used as radiation sources and the Rayleigh spectra on the left were taken with a Fabry-Perot interferometer and the Raman spectra on the right were taken with a grating optical spectrograph. All spectra were taken with a 7712 scattering angle. We also note that 1 cm-l = 30 gHz. The data are adapted from S. L. Shapiro and H. P. Broida, Phys. Rev.,1967, 154, 129, F. J. Bartoli and T. A. Litovitz, J. Chem. Phys., 1972,56, 404, and A. K. Burnham and S. J. Bertucci, unpublished work. where Dzz is the laboratory z-axis translational diffusion coefficient. In the absence of external or internal orientating fields, the fluid will be isotropic and Dzs = Dyy = Dzz. However, Daa # Dbb = Dec for a cylindrically symmetric molecule.Using the mass-continuity equation leads to the diffusion equation, The translational diffusion tensor, D(xyz), can be written in terms of the mole- cular-fixed axis (abc) according to D(xyz) = eD(abc)C,where C contains the direction cosines. Using arguments similar to those preceding equation (20), we find Dzz = D + $(Daa-Dbb) Pz(cos 8), where D = Q(Daa+ Dbb + Dee). Substituting this result into equation (47)gives where 8 is the spherical polar angle as before. For an isotropic fluid, translation along the laboratory x,y, and z axes are equivalent and we generalize this expression to give Light Scattering in Pure Liquids and Solutions where 6' is now the angle between the cylindrical molecular axis and the R vector.In the absence of any rotational motion, the Pz(c0s 6') term in equation (48) vanishes over an isotropic distribution of molecules. However, if the mole- CUk is rotating as Well as translating, the Dna -Dbb term above can lead to a coupling between rotation and transIation. Using the above developments, we now write a general equation for 9(R,8,q,T) as needed in equation (45). First, we note that the Laplacian operator, V2, for a cylindrically symmetric molecule can be expressed as a sum of two terms; the first term is the c.m. Cartesian Laplacian and the second term is the internal co-ordinate part written in spherical polar co-ordinates : where ri is the distance from the c.m. to the ith atom in the molecule. We now substitute this Laplacian into equation (48) to write an equation in 3(R, 6, q,t).First, we note that there is no rg dependence in 9(R,0, q,t) which allows us to omit the first term in brackets in V2. We are left with an equation in Daa, @ = D/cr?, the rotational diffusion coefficient (with units of ED+ $was -D~~) 3(~,p2(cOs e)1~2~.~ e, 40, t) In order to separate the rotational and translational co-ordinates in this equation the #(Daa-Dbb) Pz(c0s 6) term must vanish. We will drop this Daa -Dbb term here as Daa z Dbb in near spherical molecules, but we note that its effects can be obtained by using perturbation techniques. Setting Daa -Dm = 0 in equation (49) and spatial Fourier transforming the result gives Writing @(K,O,q, t) in terms of separated variables, g(K,6,q, t) = g(K, t) x 9(8,q), solving the corresponding differential equations, and using the Y(R,0)= 8(R)initial conditions for the self terms, leads finally to the following solutions for G(K,t) and c(K,8,q,t): Flygare c(K,t)= exp (-K2Dt) 47TG(K,8,q, t) = -exp (-[6@ + K2D)t], (51)5 where Tor = 1/(6@) is the single particle reorientation relaxation time.It is also easy to show by the above methods, that E(K,8,cp, t) in equation (29) is also equal to +G(K,8,q, t) in equation (51) for the cylindrically symmetric scatterers considered here. Substituting these results into the appropriate parts of the correlation functions in equation (29) and Fourier transforming gives the Raman spectra in the rotationally quenched limit : DK2 + 60 + 1/73 (w-wo+w~?,,~,)~+ [(l)/(~j)+ DK2 + 6012 Normally only the Raman Stokes transitions are observed [En,< Ennjin nj =on,,,(Enej-Enj)/(fi)]due to the Boltzmann factors in and h3 [equation (29)1.The 9(co)isot Raman spectra have a half-width at half-height of dw = l/~j+ DK2 where 73 is the vibrational relaxation time for thejth normal mode of vibration. Noting that D z (10-c10-5) cm2 s-1 for most liquids and 73 z 10-l2 s for most molecular vibrations in liquids, we can safely write 1/73 >K2Dfor optical radiation and any scattering angle. Thus, a measure of the Raman -Y(W)isot gives, from the half-width at half-height, a direct measurement of the vibrational relaxation times. Several values of 7vib obtained in this way are listed in Table 1.The 9(o)anis Raman spectra have a half-width at half-height of do = 1/73 + DK2 + 60. Typical small molecule values of @ range from 109-1012 s-l and if we are using optical radiation we are safe in writing 60 3-DK2 for any scattering angle. Thus, a measure of the Raman g(W)anis gives from the half-width at half-height, a direct measurement of the rotational diffusion coefficient, @, or the orientational relaxation time, Tor = (1)/(6@). Several values of Tor obtained in this way are listed in Table 1. Of course, IVw(v)RAM= IR&y(v)+ +I:$y(v) and Ih"(u)RAM = I:$Y(v) are measured. IR&f/l(v) can be extracted from IV2)(v)R*~~by subtracting 41#(v)RAM. Keeping in mind our original model of a cylindrically symmetric near-spherical shaped molecule reorientating about its symmetry axis, we note that the values Light Scattering in Pure Liquids and Solutions Table 1 Vibrational relaxation times, 7vi b, and rotational orientation times Tor = 1/(6@)for several molecular liquids from the Raman spectra Vibrational transition 1 MoIecule /cm-l Tvib/10-12S 60 = 'TO~/~O-~~S carbon disulphide CS2 656 10.6a 1.5a acetonitrile CH3CN 2943 3.2b 0.9b methyl iodide CHd 525 2.0a 1.Y 1245 2.Oa 1.4u chloroform CHC13 667 2.oa 1 9 3019 1.1a 1.5" bromoform CHBr3 222 4.1a 5.3a 3019 1.2c 4.4c benzene C6H6 992 4.7d 2.8d hexafluorobenzene C6F6 558 2.2a 6.6a a F.J. Bartoli and T. A. Litovitz, J. Chem.Phys., 1972, 56, 404. b J. E. Griffiths, J. Chem. Phys., 1973, 59, 751. c G. D. Patterson and J. E. Griffiths, J. Chem. Phys., 1975, 63, 2406. d K. T. Gillen and J. E. Griffiths, Chem. Phys. Letters, 1972, 17, 359. of @ and Dcan be written according tof = 67rqr for the frictional force constant18 as D = (kT)/f = (kT)/(67rqr)and @ = (kT)/(8777r3)= (kT)/(6V*q),where k is Boltzmann's constant, q is the shear viscosity of the solution, r is the effective particle radius in the fluid, and V* = 477r3 is the effective molecular volume. We now examine I:$y(v) which arises from the Fourier transform of P(K, 8, q,t) in equation (29). p(K, 8,q, t) is similar to G(K, 8, q,t) considered above where c(K, 8, cp, t) contains only the self terms and P(K, 8, q,t) contains both the self terms and the distinct terms.Thus, we expect the difference between P(K, 8,q,t) and c(K, 8,cp, t) or l",y(v) and It$y(v), respectively, to reveal the distinct effects or the two-particle orientational pair correlations. A direct comparison of It$F(v) and IR$M(v) for CS2 and benzene is shown in Figure 3. It is evident that the half-width at half-height of l~~~(v)is smaller than in It$F(v) in CS2, thus, reflecting the effects of the orientational pair correlations. However, in benzene it appears that the half-widths of IR$y(v) and Z$$y(v) are the same indicating no orientational pair correlation effect. The values of 7R,4y = l/(dw) [where do is the half-width at half-height in lt::(v)] for several molecules are listed in Table 2.Comparing rRkYin Table 2 with Tor in Table 1, for the few molecules which have entries in both tables, shows that Tor 5 rRhYor that the effects of orientational pair correlations cause an effectively longer rotational relaxation time. The orientational pair correlations also affect the integrated intensities of the anisotropic Raman and Kayleigh scattered light. According to equations (28) and (29), the above discussion, and arguments similar to those preceding equation (40), we can write L. D. Landau and E. M. Lifshitz, 'Fluid Mechanics', Addison Wesley Publishing, Reading, Massachusetts, 1959; J. Frenkel, 'Kinetic Theory of Liquids', Dover Publishing, New York, 1955. Flygare Table 2 Rotational orientation times rRkY,which are obtainedfrom the half- width at half-height of the depolarized Rayleigh lines, dw = I/rRky, as shown for instance in Figure 3.The temperatures are near 300 K Molecule TRAY/10-12 s carbon disulphide 1.P acetonitrile 1.7a methyl iodide 2.3a chloroform 2.9b bromoform 10.1c benzene 2.9& hexafluorobenzene 14.0" a S. J. Bertucci and A. K. Burnham, unpublished data, 1976. G. R. Alms, D. R. Bauer, J. I. Brauman, and R. Pecora, J. Chem. Phys., 1973, 59, 5310. G. D. Patterson and J. E. Griffiths, J. Chem. Phys., 1975, 63, 2407. G. R. Alms, D. R. Bauer, J. I. Brauman, and R. Pecora, J. Chem.Phys., 1973, 58, 5570. D. R. Bauer, J. I. Brauman, and R. Pecora, J. Chem. Phys., 1974, 61, 2255. for thejth normal mode in Z::? where N = VSpois the number of scatterers.Substituting from equation (51) and assuming that rotational and translational motion are separable gives The remaining independent sum over i and j where i # j is over all pairs of molecules within the volume element Vs.Thus, if each of the scattering molecules are identical, all terms in one of the sums will be the same and we can write where the last term includes the long-time average of the N -1 identical i terms which are summed over j (j # i). We now use the spherical harmonic 6)i PZ(COSaddition formula to write PZ(COS 6)j in terms of Pz(c0s &g), where 6ij is the angle between the cylindrical symmetry axes of the ij pair of molecules. Using the ergodic hypothesis, we replace the time average with a spatial average.Making these changes, we write 129 Light Scattering in Pure Liquids and Solutions (? 1+ where the last step assumes that (N -1)/N = 1 which requires a large number 8ij))is the sum of average of particles in the scattering volume. CPZ(COS (j values of Pz(c0s 8ij) = B(3 6032 8ij -1) between the N -1 ij-pairs of molecules in the fluid. Substituting the results in equations (56) and (51) into I:,$: and I::? respectively, in equation (53) gives the integrated intensities,19 1 1;;: = -Ag2N(aaao -CXbbo)215 Thus, according to these results, if a pure solution of monomeric cylindrically symmetric molecules would suddenly dimerize with symmetry axes aligned, cPz(c0s 8ij))= 1, and g2 = 2, thereby increasing the intensity of 1:2:.(j If the dimerization would occur with symmetry axes perpendicular, ofCPz(cos8ij))= -4, gz = fr and the intensity Z:;: would decrease. (j Similar arguments can be made of trimers and higher order polymers. A diagram of g2 as a function of density from I:,$: in the isotropic liquid phase in MBBA, a rod-like molecule which forms a liquid crystal phase, is shown in Figure 4. The evidence for increasing alignment with increasing density, as measured from I::: and the resultant g2, is quite convincing. Of course, g2 + a3 as the system approaches the liquid crystal phase. In the limit of Debye diffusion, where many collisions are necessary to cause a molecular reorientation, Keyes and Kivelson2O have shown that the time- dependent part of the correlation function also contains g2 according to [see equation (51) where Tor = 1/(6@)] Is We have ignored a very important problem in extracting the value of g, from the intensities of the scattered light. This problem involves the effects of shielding of the incident radiation field in the scattering medium ;the local field problem.Fortunately, experimental methods have been devised to measure and cancel the effects of the local field (T. D. Gierke and W. H. Flygare,J. Chem.Phys., 1974,61,4083, and A. K. Burnham, G. R. Alms, and W. H. Flygare, J. Chem. Phys., 1975, 62, 3298.) 2o T. Keyes and D. Kivelson, J. Chem. Phys., 1972, 56, 1057. Flygare i Figure 4 Experimental determination of g, as a function of density fionz as shown in equation (57).The system studied is the isotropic liquid p-methoxybenzylidene- n-butyhniline (MBBA) at T = 318 K which is above the transition temperature for the liquid crystal phase. The data are from G. R. Alms, T. D. Gierke, and W. H. Flygare,J. Chem. Phys., 1974,61,4083. $~where 70rg2 = T~ is defined in Table 2 as obtained from the half-width at half-height of the I:$?(CLJ) spectrum. More recently, theory and experiment has been extended to a study of orien- tational pair correlations in a series of CzV type molecules (substituted benzenes) where a definite correlation is found between the magnitude of the orientational pair correlation and the dipole moment of the molecule.21 Another active area of research in interpreting the depolarized light scattered from small molecules involves the interpretation of the dynamics of the diffusion process.This topic has been recently reviewed.22 4 Concentration Fluctuations and Electric Field Effects In this Section a number of principles which are applied to light scattering from solute molecules in solutions are examined. Before discussing scattering from the solute molecules in a dilute solution, we remember that the solvent will certainly scatter light as described in Sections 2 and 3 where we considered pure liquids. a1 S. J. Bertucci, A. K. Burnham, G. R. Alms, and W. H. Flygare, J. Chem. Phys., 1977, 66, 605. *a D. Bauer, J. Brauman, and R. Pecora, Ann. Rev.Phys. Chem., 1976, 27,443. Light Scattering in Pure Liquids and Solutions In this Section we will show that concentration fluctuations of solute molecules give rise to an additional isotropic scattering which allows the measurement of the translational diffusion coefficient of the solute molecule in the solvent.We will also show the effects of an electric field and the measurement of mobilities by light scattering. In the case of a small dilute solute non-vibrating molecule with cylindrical symmetry (aaa# abb = act), the distinct terms in the correlation function will be insignificant and the resulting correlation function and spectrum can be given by arguments similar to those leading to equation (52) for Raman scattering. The results for concentration fluctuations of a solute non-vibrating molecule give the following spectra: A Zh"(U)R*Y = (aaa -abb)2N9(o -W0)anis r DK2 1 L(w -WO)~+ (DK2 + 6@)2_] where A is defined in equations (12) and (13) and all other terms have been defined previously.In the case of a very dilute solute, D is the solute self-diffusion coefficient in the solvent. In a binary mixture of A and B at a higher concentration of solute, the measured diffusion coefficient will be the mutual diffusion coefficient, DAB,given to first order (in an ideal A-B solution) by DAB= DAXB+ DBXA , (60) where XA is the mole fraction of A in the solution. In order to relate the single particle intensities in equation (59) to the pro- perties of a solute in a solution, we assume that the fluctuations in concentration which give rise to the scattered spectra in equation (59) are independent of the density fluctuations giving rise to the Rayleigh-Brillouin spectra.Under these circumstances, Tanf~rd~~ has shown that a2and (aaa-abb) should be replaced with 1 cy2 (1/M + 2B1C + 3B2C2 + . . . %[-$).]'(i) ).(61) for the solute in the solvent where C is the concentration (C/p= M), M is the mass of the scatterer, n is the refractive index of the solution, and no is the refrac- 23 C. Tanford, 'Physical Chemistry of Macromolecules', John Wiley and Sons, New York, 1961. Flygare tive index of the solvent. B1 and B2 are the virial coefficients which give rise to a decrease in the isotropic scattering intensity at higher concentrations. Sub-stituting equation (61) into equation (59) and assuming no orientational correlations gives the complete spectral function for polarized incident radiation (Figure 1) in a dilute solution of N small solute scatterers, 1 1/M + 2B1C + 3BzC2 + .. ] (?[(g).]’ N(aaa -~BB)~Z(W-u0)anis where A?(m -WO)isot and A?(u -W0)anis are given in equation (59). The relative intensities of the isotropic concentration dependent effect in equation (62) can be compared directly with the total Rayleigh-Brillouin intensity for the isotropic density fluctuation effects considered in Section 2. Substituting equation (44) into C(K,0) in equation (36) and comparing with the frequency-integrated form of the isotropic term in equation (62) shows that the relative intensities are given by Solute in Pure liquid dilute solution M2N’no2--[(2)0]2a2NpokqT 47T2 9 where N = PO Vs is the number of pure liquid scatterers and N’ = pbVs is the number of solute scatterers where Vs is the scattering volume.The value of (an/aC)oneeded above can be measured for the particular system in question. We can estimate (an/aC)o by assuming an ideal solution where the refractive index of the solution can be evaluated by the mole fraction weighted sum of the individual pure-fluid refractive indices, n = naXa + nbxb = na + Xb(nb -na) . (63) n is the refractive index of the solution, na and nb are the refractive indices of liquids A and B, and Xa and xb are their respective mole fractions. The mole fraction of a dilute solution of solute B in the solvent A is given by where C is the concentration of the solute B, Mb is the mass of a solute molecule, and pa is the number density of the solvent.Substituting equation (63), differen-tiating with respect to the concentration of B giving an/aC, and substituting this result into the above expression under ‘solute in a dilute solution’, gives Light Scattering in Pure Liquids and Solutions It is evident from this equation that the solute (B) scattering is proportional to the square of the difference in solvent-solute refractive index difference. Consider now the benzene solute concentration fluctuation scattering intensity in a CC14 solvent at T z 300 K. The x2NpokTpT = a2Vspo2kT,~scattering factor for the CC14 solvent at 239 K can be obtained by using a = 10.5 x 10-30 m3, PO = 1.04 x molecules m-3, and ,8~= 10.7 x 10-10 m2 N-1 to give a2p02kT,8~= 3 x lo-".The benzene solute concentration fluctuation scattering factor in a CC14 solvent is obtained from equation (65), where pa(CC14)is given above, Mb = 1.297 x kg, na = 1.4590 and nb = 1.5011, which gives C(nb -na)2/(4rMbpa2)= 6.8 x C. According to these numbers, the total benzene scattering intensity in the cc14 solvent will exceed the solvent scattering at concentrations above C = 10 kg m-3 (for equal scattering volumes), which is a relatively low concentration. We have also chosen a solute solvent (ria -nb) which is quite small. In conclusion, it is evident that concentration fluctuations scattering can be much more intense than the background solvent Rayleigh scattering even at relatively low concentrations.24 We now note that the intensity of the anisotropic Lorentzian [equation (62)] will normally be less than the isotropic term.Of course, in cc14,aaa-abb = 0 and the anisotropic term is zero. However, assuming that aaa-abb = m3 and remembering that PO = C/M, we can write the multiplier of the second Lorentzian in brackets in Ivv(~)RAYin equation (62) as -&0(aaa -cIIbb)2 = $F(C/M)(aaa-= 6.8 x C, which is considerably less than the multiplier of the first Lorentzian in Ivv(v)RAYas shown above. Thus, the Lorentzian with dw = K2D normally dominates the Iv"(w)R*Y spectrum in equation (62). It is also evident that the DK2 half-widths considered here will be smaller than the KK~half-widths considered in Section 2, because D< K.~~ Of course, we have also assumed that there is no coupling between the con- centration fluctuations described here and the density fluctuations described in Section 2.As the solute particles became larger relative to the solvent molecules, the above separation between the concentration fluctuations (and the measure- ment of the translational diffusion) and the other effects described above become more pronounced. Several diffusion coefficients for macromolecules are listed in Table 3. In some of the large molecules listed in Table 3, the size of the scatterer approaches the wavelength of the radiation. Under these conditions, scattering from different parts of the same molecule leads to both static and dynamic (if the molecule is rotating) correlations.The static correlations lead to a K-depen-24 A thorough study of mutual diffusion in binary systems of small molecules has been given by S. J. Bertucci and W. H. Flygare, J. Chem. Phys., 1975, 63, 1 and K. J. Czworniak, H. C. Andersen, and R. Pecora, Chem. Phys., 1975, 11, 451. These efforts show that the translational diffusion in ideal binary systems can be expressed in terms of transferable molecular diameters for the molecules. z5 P. Berge, P. Calmettes, M. Dubnis, and C. Laj, Phys. Rev. Letters, 1970, 24, 89. Flygare dence in the total scattered isotropic intensity leading to a measurement of the radius of gyration.The dynamic correlations lead to spectral characteristics which allow the measurement of the rotational diffusion constant, 0,by examining the spectrum of the isotropically scattered light.26 Table 3 Diflusion constants measured at T = 293 K in water solutions and molecular weights in atomic mass units of several molecules of various sizes Molecule 0/10-11 m2 s-1 M/u* Glycine 93.4 75 Sucrose 45.9 342 Ribonuclease 10.7 13 683 Bovine serum albumin 6.1 67 OOO Oval bunim 7.1 45 OOO Lyzozyme 11.5 14 100 Tropomyosin 2.20 93 OOO Fibrinogen 2.00 330 OOO Myosin 1.10 490 OOO Tobacco mosaic virus 0.29 40 x lo6 Latex spheres (d = 910 A) Calf thymus DNA 0.2 0.52 ca. 5 x 106 * Data from ref. 23, and S.B. Dubin, J. H. Lunacek, and G. B. Benedek, Proc. Nat. Acad. Sci. U.S.A., 1967, 57, 1164. We will now examine the principles involved in electrophoretic light scattering2‘ which involves the observation of laser light scattered from a solution of charged macromolecules which are moving with a drift velocity Vd in the presence of an electric field. We will consider only relatively small molecule scattering here where equation (62) gives an appropriate description of the spectrum of the light scattered in the absence of any perturbing electric fields. Most protein inacromolecules are charged at a given pH and in the presence of an electric field the ions in solution will experience a force in the field causing them to translate with a drift velocity, Vd, given by Yd = pE, where Eis the electric field vector and p is the scalar mobility.In the case of spheres, the mobilities and diffusion coefficients of the ionic molecules are related by p/D = eZ/kT, where e is the electronic charge, 2 is the effective number of charges on the translating ion (including the effects of the electrophoretic counterions), k is Boltzmann’s constant, and Tis the temperature. Return now to equation (46) and consider the flux of molecules along the z axis following the concentration fluctuation in the xy plane as shown. In the presence of an electric field component along the z axis, the electrostatic force on the charged (ionic) molecules in solution will give rise to an additional flux along they axis.The total flux is given by modifying equation (46) according to (Daa = Dbb = Dcc = D) J(z) = -D dP(z, t)/dy + 26 For details see B. Berne and R. Pecora, ‘Dynamic Light Scattering’, John Wiley and Sons, New York, 1976. 27 B. R. Ware and W. H. Flygare, Chem. Phys. Letters, 1971, 12, 81. Light Scattering in PureLiquids and Solutions udP(z, t). Combining this expression with the z axis mass-continuity expression gives the modified one-dimensional Fick’s law in P(z, t) in the presence of the electric field according to The solution to this equation is easily obtained by Fourier transform methods as described before to give P(K,, t) = p(&, 0) exp (-KU2Dt -iK,vdt) , where p(K,, 0) = 1 .O for uncorrelated scatterers. In three dimensions this equa-tion takes the form p(K, t) = exp (-K2Dt -iK*vdt) = exp (-K2Dt -ipK-Et) .(67) Kovd = pK.E = pKEcos a, where a is the angle between the static electric field and the K vector. Remember that the K vector always bisects the angle between the direction of the incident and scattered light (see Figure 1). Sub-stituting equation (67) into the first term of G(t)in equation (29) for self terms only gives the correlation function in the presence of the electric field. Sub-stituting a2 from equation (61) and Fourier transforming gives, finally, 1 IvW(0)isot = -1/M + 2B1C + 3&C2 + . .. It is quite apparent that the only difference between this expression and the isotropic result in equation (62) is the translation in oof pK-E.By using optical mixing techniques the real part of the correlation function for concentration fluctuations has been observed in solutions of bovine serum albumin (BSA) in the presence of an electric field and the data are shown in Figure 5 for several electric fields.A spectrum showing the field-dependentDoppler shift is shown in Figure 6. The experiments illustrated are all at low angles of scattering and the electric field orientation is perpendicular to the direction of the incident light. This configuration leads to the highest resolution in the electric field effect where the shift in frequency divided by the half-width of the line defines the resolution, Rep Re=-Vd*K - VdK COS (8/2) -)lo/& COS (q2)-K2D K2D 47rnD sin (8/2) (69) For very small angles, )copElimR -----Ze)loE 84 ’-2mD8 2~nkT6’ Flygare ZERO FIELD E=11500 Vm-’ E=13500 Vm-l E =15000Vm~~ Figure 5 The observed real part of the correlation function in equation (67) for several electric fieidsfrom reference 27.where n is the solution refractive index and p/D = (eZ)/(kT)is also used in the last step of equation (70). Additional details and a summary of the literature in the field of electrophoretic light scattering have been given recently.28 3n W. H. Flygare, S. L. Hartford, and B. R. Ware, in ‘Molecular Electro-Optics’, ed. C. T. O’Konski, Marcel Dekker, New York, 1977. Light Scattering in Pure Liquids and Solutions 6 = 6.25O I I 25 50 75 100 c v/H z Figure 6 The spectrum of light scattered from concentration fluctuations in calf-thymus DNA where optical mixing techniques reduce the laser frequency reference to zero.0.1 mg DNA was dissolved per ml water with an ionic strength of 0.01 at T = 213 K. The analysis of the shijt in frequency of 50 Hz at 5000 Vrn-l leads to a mobility of p = 3.5 x 10-9 m2V-1s-1. The half-width at half-height leads to a translational difiision coeficient of D = 1.2 x 10-l1m2s-'. The data are from S. L. Hartjord (see also S. L. Hartford and W,H. Flygare, Macromolecules,1975, 8, 80).

ISSN:0306-0012    

DOI:10.1039/CS9770600109

出版商:RSC    

年代:1977

数据来源: RSC

摘要:

The Binding of Heavy Metals to Proteins By T. L. Blundell and J. A. Jenkins LABORATORY OF MOLECULAR BIOLOGY, DEPARTMENT OF CRYSTALLOGRAPHY, BIRKBECK COLLEGE, MALET STREET, LONDON WClE 7HX 1 Introduction Although heavy metals such as cadmium, mercury, platinum, and lead may have a physiological role, it is their toxicity which has attracted most attention. Thus cadmium from a mine in Japan became a real hazard to the local population giving rise to itai-itai disease (itai means pain); thallium is a rat poison and was used effectively by the Resistance in a factory in Holland against the Nazi management;chronic lead poisoning has been reported not only from occupa- tional exposure but also in children who have eaten flaking paint; mercury caused headline news when mercury-containing waste from a factory manu- facturing PVC and acetaldehyde was dumped in the sea in Minimata Bay and eventually found its way into fish which were the staple diet of the local inhabitants; and similar but less advertised tragedies have occurred with platinum and other toxic metals.Increasing concern over the toxicity of heavy metals in the environment has lead to increased research activity to identify the fate of metal ions in the organ- ism and the dependence of toxicity on dietary deficiencies of metal ions such as zinc and calcium (see reviews by Bremnerl; Vallee and Ullmer2). Proteins are involved in the action of most heavy-metal ions at normal and toxic concentra- tions, although nucleic acid and lipid interactions are also important.For example, proteins in intranuclear inclusion bodies bind lead in chronic poisoning by this metal3 and much of any cadmium and mercury absorbed is sequestered by a soluble kidney protein, metallothionen.4 In vitro experiments have given further information concerning heavy-metal-protein interactions. Thus cadmium can mimic zinc in forming insulin hexamers ;zinc-free carboxypeptidases with lead, mercury, or cadmium at the active site are effective in ester hydrolysis but cannot hydrolyse peptides,5 and mercury binds many plasma proteins and gives rise to dimerization of albumin.6 It is clear that heavy-metal ions may be trans-ported or sequestered in the organism by proteins; they may also bind specifically to certain proteins with a concomitant enhancement, modification, or inhibition of the normal biological activity.I. Bremner, Quart. Rev. Biophys., 1974, 7, 75. a B. L. Vallee and D. D. Ullmer, Ann. Rev. Biochem., 1972, 41, 91. R. A. Goyer, P. May, M. M. Cates, and M. R. Krigman, Lab. Invest., 1970, 22, 245.'P. Pulido, 3. Kagi, and B. L. Vallee, Biochemistry, 1966, 5, 1768. J. E. Coleman and B. L. Vallee, J. Biol. Chem., 1961, 236, 2244. W. L. Hughes and H. M. Dintzis, J. Biol. Chem., 1964, 239, 845. The Binding of’Heavy Metals to Proteins Early studies of the nature of these different heavy-metal-protein interactions emphasized the importance of covalent binding to thiols of cysteine residues. The stability of the metal-sulphur bond allowed a ready identification of the ligand.However, observations such as the binding of cadmium to insulin, and mercury to the active site of carbonic anhydrase and of carboxypeptidase, which cannot involve cysteine, imply that cystine disulphides, histidine imidazole, methionine sulphur, and aspartic and glutamic carboxylic acids must also play an important role. The problem of defining these ligands has been partly operational; they are difficult to identify in a large protein. However, the difficulties are partly con- ceptual. Proteins exist in a complex biochemical environment in living organisms and even in the laboratory are kept in bufiered solutions containing many potential small molecular weight ligands, such as acetate, citrate, tris, phosphate, ammonia, and so on.These ligands can complex the metal ions and so modify their reactivity giving rise to many varied potential protein binding species and a complicated pattern of interaction. The pH and ionic strength can also affect the protein ligand and change its affinity for the metal. This complexity is further confused by the multidentate nature of the protein. Most metal ions are bound through several protein ligands in a specific three-dimensional arrangement. Thus the interaction with the protein tends to be largely entropically driven; the geometrical specificity is increased but energetic considerations of the strength of metal-ligand bonds become relatively less important. At the same time chemi- cal analysis of the metal-protein interaction becomes prohibitively difficult ; the interaction depends critically on the correct protein conformation whereas chemical analysis usually involves denaturation and degradation of the protein.This depressing account of the difficulties in the study of heavy-metal-protein interactions ignores one important source of data. Protein crystallographers have long taken advantage of these heavy-metal ions to label proteins for use in X-ray analysis by the method of multiple isomorphous re~lacement.~-lO In fact the successful use of metal ions in protein structure analyses is a fairly good indication of their toxicity : mercury, platinum, uranyl, thallium, and lead compounds have all found use. Theoretically, if the protein sequence is known, X-ray analysis can lead to a complete description of the metal ion binding site in a way that is not feasible otherwise.In reality this has not been the main interest of the study and most of the useful information has remained unpublished, and that which has been put in print lies scattered in many different journals. In this review we have brought together much of this information, published and unpublished, and present here a detailed analysis of protein-heavy-metal interactions. Although many bound metal ions may not affect the active site of the proteins, these interactions may affect their solubility or supramolecular T. L. Blundell and L. N. Johnson, ‘Protein Crystallography’, Academic Press, London, 1976. K.C. Holmes and D. M. Blow, ‘The Use of X-ray Diffraction in the Study of Protein and Nucleic Acid Structure’, Interscience, New York, 1966. D. C. Phillips, Adv. Res. Difraction Methods, 1966, 2, 75. lo A. C. T. North and D. C. Phillips, Prog. Biophys., 1968, 1. Blundell and Jenkins organization and so lead to toxic effects. Alternatively, binding sites, especially in plasma proteins, may be important in transport of metal ions within the organism. In any case the nature of the ligands binding metal ions may be little different in situations which have physiological effects and those where the effects are neutral. 2 X-RayStudiesof Metals in Protein Crystals Protein crystals exist as two phases: a solid phase of protein molecules packed in an open lattice and a liquid phase occupying the channels and spaces in this lattice.(For a review, see ref.7.) For instance, rhombohedra1 crysals of zinc insulin hexamers contain about 30% solvent, but other crystals of globular proteins can comprise as much as 60% solvent. Thus most of the protein is bathed in aqueous solvent in a way which is probably not unlike that found in biology. In insulin the solvent is a 0.2M citrate buffer at pH 6.3 and the lattice is stable with respect to changes of pH in the range of a few pH units. Other protein crystals require a high salt concentration to prevent dissolution but these may often be desalted after prior cross-linking with a bifunctional agent such as glut araldeh yde , The solvent of crystallization is in equilibrium with the mother liquor sur- rounding the crystal, and the crystals must be kept covered in liquid to avoid evaporation of the solvent and consequent disordering of the lattice.However, this equilibration of the mother liquor and solvent in the lattice provides a straightforward mechanism for introduction of metal ions. Metal ions may be dialysed into the mother liquor and allowed to diffuse into the crystal. Studies using concentrations from O.OlmM to 1M of heavy-metal salt in a buffered solu- tion have shown that the metal ions may diffuse into the crystal in a matter of minutes, although the reaction with the protein may take days or even weeks. Generally a metal ion concentration of 1mM is satisfactory for specific binding.Higher concentrations to give non-specific association of the metal in the lattice channels, but in a few cases can increase the occupancy of a site of interaction. Occasionally different interactions are given by co-crystallization or even controlled reaction between metal ion and protein prior to crystallization. The interaction of metal ions with proteins can be monitored by following changes in the X-ray diffraction pattern. The method can be extremely sensitive and with modern methods of intensity measurement changes due to about 10 electrons in a protein of molecular weight 20000 or more can be identified. However, some caution is required. Occasionally metal ion binding leads to lack of isomorphism. This manifests itself either in changes in dimensions of the crystal unit cell and consequently in the geometry of the diffraction pattern, or in large changes in the diffraction pattern at higher angles only.For the X-ray analysis to be successful the dimensions and symmetry of the diffraction patterns should not differ between the native and heavy atom derivative, and the changes in intensities should be distributed fairly evenly throughout the diffraction pattern. Intensity changes at low angle only usually result from non-specific binding and concentration of the metal ions in the lattice of the crystal. The iljlindi'ng OJ hTeavyildetais to Proteins The positions of the heavy atoms in the cell can be determined without solving the phase problem for the protein crystals, but this is a rather academic point as far as this discussion is concerned as it is also necessary to find the structure of the protein to define the nature of the protein-metal interaction.In fact, high-resolution analyses of over fifty proteins have now been reported,7 and where protein sequences are available the metal co-ordination can be de- scribed. Generally the heavy-metal position will be more precisely defined than that of the protein ligands, and it is often difficult to fix the precise orientation of the liganding side-chain as protein maps, with the exception of two or three carefully refined structures, are not at atomic resolution. For most of the data tabulated here detailed metal co-ordination geometries are not available ; but even identification of the ligands is a very important step forward. 3 The Metal Ions, Potential Ligands, and Factors Mecting the Stability of Metal-Protein Complexes The ions of interest in this discussion include not only ions such as barium, uranyl, and rare earths, which are hard ions and usually bind preferentially to hard ligands such as water, fluoride, carboxylate, or alcohol hydroxyl, but also the B metals such as platinum, silver, gold, cadmium, and mercury, which bind softer ligands such as iodide, bromide, RS, R2S, CN- and \-N more/strongly.The TI+ and Pb2+ ions have non-group valencies and larger radii; unlike the other B metal ions they prefer harder ligands such as carboxylates but may also become oxidized to TP and PbIV, which binds soft ligands more strongly.The potential ligands include not only the amino-acid functional groups of the protein but also the small molecules and ions present in the salt and buffer. Thus soft ligands are histidine imidazole, cysteine thiol, cystine disulphide, bromide, iodide, and cyanide. Terminal glutamate and aspartate carboxylates, hydroxyls of threonine and serine, amides of glutamine and asparagine, water, fluoride, acetate, citrate, and phosphate are harder ligands. The intermediate chloride, ammonia, and amino-groups will also often be present; like the softer ligands, these bind the B metals preferentially. The reaction of the metal with the protein depends critically on the chemistry of the metal salt or compound.Thermodynamically stable complexes such as Pt(CN)42- will not react with the protein through ligand displacement whereas other complexes, such as PtCh2- of the same metal, may do so. Kinetic lability may also affect the course of the reaction. Thus platinum(I1) substituted with a nitrogen ligand is further substituted slowly trans to this ligand, and sulphur atoms give stable complexes because they are not only poor leaving groups but also strong nucleophiles. The charges of the metal ion and ligand are also important factors in deter- mining the nature and extent of the reaction. Thus platinum complexes can exist as anions, neutral compounds, or cations. Clearly a negative ion such as PtC12- will interact less readily with a negatively charged thiol : but a neutral compound Blundell and Jenkins such as PtC12(NH3)2 will more easily bind.In a similar way a negatively charged thiol or carboxylate will be more nucleophilic than the protonated species. Hydrophobicity also plays an important role. MeHg+ will more easily penetrate the hydrophobic core of the protein than the bivalent cation, Hg2+. Data on the relative importance of all these factors are available from X-ray studies (see Chapter 8 of ref. 7). This interplay of the various characteristics of the metal and protein underlie the interaction of metal ions with proteins not only in crystals but also in biology. Thus MeHgCl is a much better inhibitor of membrane-bound enzymes than inorganic mercury,11 while the highly hydrophobic MezHg will be concentrated almost entirely in lipids and the mechanisms of its toxicity will reflect both this and its comparative lack of reactivity with nucleophiles. 4 The Protein Chemistry of Hard Cations Let us first consider the binding of actinide and lanthanide ions.The results with the relevant references are summarized in Table l.13-27The U0z2+ ion binds carboxylate groups of glutamate or aspartate, and occasionally hydroxy side chains of threonine or seriiie as expected of a hard cation. Table 1 shows that ribonuclease S, rubredoxin, and insulin bind uranyl cations at many sites and there is considerable variation in occupancy. Quite often sites of low occupancy are clustered together. The reactivity of the uranyl cation is modified by complex- ing it with a hard ligand such as fluoride in or acetate in UOzAc3-.In lysozyme, elastase, and insulin, the binding of the uranylfluoride anion is close to sites occupied by other uranyl salts, but the binding is less extensive. The carboxylate ions almost certainly co-ordinate by displacing the fluoride ligands. Tervalent lanthanide ions also bind to carboxylate side-chains (Table 1). Samarium in the form of its acetate binds to two glutamate carboxylate groups l1 D. R. Storm and R. P. Gunsalus, Nature, 1974, 250, 778. l2 H. J. Segall and J. M. Wood, Nature, 1974, 248, 458. l3 T. L. Blundell, J. F. Cutfield, G. G. Dodson, E. 5. Dodson, D. C. Hodgkin, D. Mercola, and M. Vijayan, Nature, 1971, 231, 506.l4 E. S Mathews, P. Argos, and M. Levine, Cold Spring Harbour Symposium, Quunt. Biol., 1971, 36,387. l5 F. R. Salemme, S. T. Freer, N. G. H. Xuong, R. A. Alden, and J. Kraut, J. Biol. Chem., 1973,248, 3910. Is E. T. Adman, L. C. Sieker, and L. H. Jensen, J. Biol. Chem., 1973, 248, 3987. l7 J. R. Herriott, L. C. Sieker, L. H. Jensen, and W. Lovenberg, J. Mol. Biol., 1970, 50, 391. H. C. Watson, D. M. Shotton, J. M. Cox, and H. Muirhead, Nature, 1970, 225, 806. lS M. S. Geisow, personal communication, 1975. 2o C. C. F. Blake, Adv. Protein Chem., 1968, 23, 59. a1 A. Tulinsky, personal communication, 1974. G. N. Reeke, personal communication, 1974. 23 T. L Blundell, unpublished results. 24 P. M. Colman, J. N. Jansonius, and B. W. Matthews, J.Mol. Biol., 1972, 70, 701. 25 W. N. Lipscomb, G. N. Reeke, J A. Hartsuck, F. A. Quiocho, and P. H. Bethge, Phil. Trans. Roy. SOC.,1970, B257, 177. 2E G. M. Edelman, B. A. Cunningham, G. N. Reeke, J. W. Becker, M. L. Waxdal, and J. L. Wang, Proc. Nat. Acad Sci. U.S.A., 1972, 69, 2580. 27 W. G. J. Hol, PhD Thesis, Rijksuniversitet, Groningen, 1971. .* E Table 1 The ligands of uranyl, lanthanide, plumbous and thallous cations and their complexes in proteins. Further details are 9 given in reference 7 and in the references given in the table. Highly occupied sites are indicated by an asterisk 8 Conc. of Site % Protein (Ref.) reagent Bufler/salta PH number Binding site 4 Insulin (13) 1 mM O.05M-acetate 6 I* 0.01M-ZnAc:! Cluster of sites 4:I Cytochrome 65 (14) 100 mM 4M-AS 7.5 1* ASP-66; Glu-48 0.1 M-tris 2* Glu-78 3 Asn-1 1 ? 4 ASP-83, LYS-86; Glu-38, I?.9 Glu-30 56 Ferricytochrome c2 10 x 3M-AS 5.8 1* (Rhodospirillum rubrum) (1 5) protein 2* conc.34 Bacterial ferredoxin (16) 100 mM 3.3M-AS 7.5 12 sites 0.7M-tris/ HCI Rubredoxin (17) N 100 mM 3.5M-AS 4 6 sites Tosyl elastase (18) 5 miM 1 -2M-NazS04 5.8 1" 0.01M-NaAc Glu-43, Glu-37 z GIu-44Glu-64Thr-63Lys-9761~-37, Lys-112 Clusters of sites, somc common with Yb3+ and Sm3+Most highly occupied close to Asp-47. Cluster of 4 sites Glu-70, Glu-80, Try-82,Val-67, Leu-73 (Mainchain CO) UOz(N03)2 Prealbumin (19) 10 mM 3.1M-AS 5.0 6 major Between 2 Glu one on each sites of two subunits U02(NOs)2 Lysozyme (hen egg white) (20) 0.85M-NaC1 4.7 5 sites Asp and Glu carboxylates Thr-89 UOz(NO3)2 a-Chymotrypsin (21) 3.6 Major site.Glu-21, Asp-1 53 uo2(No3)2 Concanavalin A (22) 0.1 mM 1" ASP-80, ASP-83 2 Not close to any side-chains K3UOzFs Insulin (1 3) 1 mM O.05M-citrate 6.3 1 2 sites close together 0.01 M-ZnAcz 2 Glu-B13, Glu-B13' (similar to UOZAc2 but lower occupancy) K3UOzFs Tosyl elastase (1 8) 30 mM 1.2M-NazS04 5.0 1 Glu-70, Glu-80, Tyr-82, 0.01 M-NaAc Val-67, Leu-73 (mainchain). (same as uoz(No3)zbut lower occupancy) KsUOzFj Lysozyme (hen egg white) (20) 0.OM-NaCl 4.7 1 Asp and Glu groups; (same 2 as two largest UOz(N03)z sites) smAc3 Insulin (23) 1 mM 0.05M-NaAc GdAc3 0.01 M-ZnAcz 6.2 1 Glu-B13, Glu-B13' (same as DyC13 UOzAcz major site) Sm(N03)3 Bacterial ferrodoxin (1 6) 40 mM 3.3M-AS 7.5 N 10 Many sites clustered and bPrC13 0.7M-tris/HC1 some common with 2 1 UOz(N03)z gLac13 Thermolysin (24) Ca2+ double site REuC13 Tr islace t at e 5.5 2 ASP-57 4 3 ASP-200 3 (continued overleaf) 2 5' Table 1-continued Conc.of Site Reagent Protein (Ref.) reagent Buflerlsalt PH number Binding site Srn(N03)~ Concanavalin A (22) 10 mM 1 Glu-87, Asp-1 36, Asn-1 3 1, 2 Gln-132, Gly-152, Asp-80, 3 Asn-82, Asp-83 PbAcz Insulin (apo protein) (13) 10 mM O.OSM-acetate 6.3 1* B13-Glu, B13-Glu 2 His-B 10 3 His-B10 4 N terminus B1, A17-Glu PbAcz Insulin (13) 100 mM O.05M-acetate 6.3 4 sites As at 10 mM but site 4 is higher in occupation PbC12 Carboxypeptidase (25) 3 mM 0.01 M-citrate 7.5 1* Glu-270 0.02M-tris 2" Citrate, not protein Pb(N03)z Concanavalin A (26) 0.3M-NaN03 6.8 1* Gln-87, Asp-136, Asp-80, 0.01 M-Na 2* LYS-82, ASP-83 maleate TlF Subtilisin novo (27) O.OSM-glycine/ 9.1 1 ASP-197 NaOH 55% acetone TlAc Insulin (23) 10 mM 0.05M-NaAc 6.3 B13 Glu, B13 Glu (same as 0.01 M-ZnAcz uranyl but lower occupancy) a) AS = ammonium sulphate; Ac = acetate.Blundell and Jenkins in insulin. Similar substitution is given by gadolinium and dysprosium. In flavo- doxin, samarium binds close to an aspartate, and in lysozyme, europium and gadolinium bind between the active site glutamate and aspartate groups. The difference in specificity of the lanthanides is mainly due to the difference of size, and heavier and smaller lanthanides may often bind where others will not.This has been neatly illustrated in the work on thermolysin where lanthanides bind at calcium sites (see section 8). In proteins T~T and Pb2+ also bind in a similar way to the hard cations (Table 1). In insulin the major site for both Pb2+ and T1+ is at two glutamates in the same way as Sm3+ or U022+ ions. In carboxypeptidase, Pb2+ binds to glutamate but in concanavalin it binds to two sites both of which involve aspartic acids. The uptake of T1+ by cells will reflect its similarity to IS+though toxicity may depend on its unique redox characteristic. The similarity of the reactivity of Pb2+ and other hard cations is reflected in the 50% inhibition of the noradrenaline sensitive adenylcyclase of Purkinje cells in the rat cere- bellium28 by both Pb2+ (at 3 pM) and La3+ (at 2 pM).This has been suggested as the primary site of lead neurotoxicity. In this case Ca2+ ions do not prevent the inhibition but Pb2f and Ca2+ can compete, as suggested by the finding that voluntary ingestion of lead is linked to subclinical Ca2+ deficiency in Rhesus m0nkeys.~9 The intermediate character of the Pb2+ ions compared to uranyl and lanthanide ions is shown by the fact that Pb2+ binds to a small extent with the imidazoles which bind zinc in zinc-free insulin and to a terminal amino-group when the lead is at high concentration. In conclusion, it appears that lanthanides are more selective than uranyl ions. In contrast to uranyl, which often gives multisite binding, samarium (and lead) may give one major site.In insulin the extent and rate of lead binding was found to be very dependent on the concentration of ions and the tem- perat~re.~3 The binding of uranyl and lanthanide ions is sometimes prevented by formation of insoluble phosphates and hydroxides. Chelating agents such as citrate will also bind the metal ions and inhibit the binding; for instance uranyl acetate and samarium acetate show little reaction with insulin in citrate buffer although they bind strongly in acetate or tris buffer. However, they bind readily to proteins in the presence of large concentrations of nitrogen ligands. 5 Protein Chemistry of Soft Ions :Covalent Interactions B-metals such as platinum, mercury, and gold tend to form covalent complexes with ligands such as thiol, imidazole, and thioethers rather than to harder ligands such as carboxylate or hydroxy-groups.Their chemistry has been widely reviewed.7~209~0-~~ 28 J. A. Nathanson, R. Freedman, and B. J. Hoffer, Nature, 1975, 261, 330. as J. L. Jacobion and C. I. Snowden, Nature, 1975, 262, 51. 30 C. S. G. Phillips and R. J. P. Williams, ‘Inorganic Chemistry’, Vol. 11, Oxford University Press, 1966. 31 R. E. Dickerson, D. Eisenberg, J. Varnum, and M. L. Kopka, J. Mol. Biol., 1969, 45, 77. 32 A. 5. Thomson, R. J. P. Williams, and S. Reslova, Structure and Bonding,1972, 11, 1. 33 G. A. Petsko, DPhil. Thesis, Oxford University, 1973.34 P. J. Sadler, Structure and Bonding, 1976, 29, 171. 147 The Binding of Heavy Metals to Proteins A. Mercuric Compounds.-In 1954 Perutz and his colleagues35 exploited the different reactions of the thiols in the a-and p-chains of haemoglobin towards p-chloromercuribenzoate (PCMB). Binding at the a-chain was achieved by first blocking the more reactive P-chain thiols with iodoacetamide and then crystallizing in the presence of mercuric acetate. Binding was also observed with dimercuriacetate (DMA) and 1,4-diacetoxymercuri-2,3-dimethoxybutane (Baker’s mercurial), which contain two mercury atoms and were found crys- tallographically to have their mercury atoms separated by 1.7 and 4.9 A, respectively. Rossmann and his co-workers in their study of lactate dehydro- genase found that two thiol groups are reactive with PCMB, HgCl2, and DMA (see Table 2) but only one was reactive with the more bulky Baker’s mercurial.Different mercurials all bind to the same cysteine thiol in calcium-binding pro- tein. These data are summarized in Table 2.36-56 Many of these reagents have a covalent mercury-carbon bond which is not easily broken. The chloride, acetate, or nitrate ligands are not bound strongly, and the mercury cation is particularly reactive towards the negatively-charged and polarizable -S groups. The cysteines will be less reactive at lower pH when the thiol is protonated. Ammonia will complex the mercury but will not dis- place carbon substituents and is unlikely to change its charge.An excess of chloride decreases the reaction by complexing with the mercury and giving it a net negative charge; this has been observed with calf rennin in 2M-NaCI The -SMe of methionine is less nucleophilic than the thiol group, -S-, and anyway has no negative charge. This probably explains the fact that it rarely binds mercury reagents. One example is the binding of HgIa- to rubredoxin. 36 D. W. Green, V. M. Ingram, and M. F. Perutz, Proc. Roy. SOC.,1954, A225, 287. 36 A. Liljas, PhD. Thesis, Acta Universitatis Upsaliensis Weilands Tryckeri, Uppsala, 1971. R. H. Kretsinger and C. Nockolds, J. Biol. Chem., 1973, 248, 3313. 38 M. M. Bluhm, G. Bodo, H. M. Dintzis, and J. C. Kendrew, Proc. Roy. SOC.,1958, A246, 369. 39 W.E. Love, personal communication, 1971. I0 E. A. Padlan and W. E. Love, Nature, 1968, 220, 376. 41 K. Hardman, personal communication, 1975. C. C. F. Blake and 1. D. A. Swan, Nature New Biol., 1971, 232, 12. I3 R. Huber, 0. Epp, and H. Formanek, J. Mol. Biol., 1970, 52, 349. 44 M. J. Adams, A. McPherson, M. G. Rossman, R. W. Schevitz, and A. J. Wonnacott, J. Mol. Biol., 1970, 51, 31. 45 J. Drenth, J. N. Jansonius, R. Koekoek, and €3. G. Wolthers, Adv. Proteiri Chem., 1971, 25, 79. A. F. Cullis, H. Muirhead, M. F. Perut?, M. G. Rossman, and A. C. T. North, Proc. Roy.SOC.,1961, A265, 15. Peking Insulin Research Group, Scientia Sinica, 1973, 16, 136. S. M. Cutfield, private communication, 1974. G. E. Schultz, M. Elzinga, F. Marx, and R.H. Schirmer, Nature, 1974, 250, 120. R. Huber, 0. Epp, W. Steigemann, and H. Formanek, European J. Biochem., 1971, 19,42. s1 E. Hill, D. Tsernoglou, L. Webb, and L. J. Banaszak, J. Mol. Biol., 1972, 72, 577. s2 P. B. Sigler, B. A. Jeffery, B. W. Matthews, and D. M. Blow, J. Mol. Biol., 1966, 15, 175. 63 C. I. Branden and E. Zeppezauer, private communication, 1974. s4 R. E. Dickerson, T. Takano, D. Eisenberg, 0. B. Kallai, L. Samson, A. Cooper, and E. Margoliash, J. Biol. Chern., 1971, 246, 15 11. 6s C. S. Wright, R. A. Alden, and J. Kraut, Nature, 1969, 221, 233. 6o T. Takano, R. Swanson, 0.B. Kallai, and R. E. Dickerson, Cold Spring Harbour Syinposium, Quant. Biol., 197I, 34, 397. 148 Table 2 The ligands of niercurbals in proteins. The chemical formulae for these mercurials are further detailed in reference 7 C'onc.of Site Reagenta Protein (Ref.) reagent Bufer/salta PH number Binding site HgAcz Carbonic anhydrase (36) 2.3 M-AS 8.5 1 His-93, His-95, His-1 17, (apoenzyme) (zinc site) HgACz Calcium binding protein (37) 0.8 mM 4M-phosphate 6.8 1 cys-18 HgBrz Hg (succin- Thermolysin (24) 5 mM 5% DMS 4 1 His-231 imide)z 0.01 M-CaAca 0.01 M-trislacetate HgC12; LiCl Thermolysin (24) 1 mM 5% DMS 6 1 His-23 1 20 mM 0.01 M-CaAcz 0.01 M-tris acetate HgClz Carboxypeptidase A (25) 0.8 mM 0.2M-LiCl 7.5 1* His-69, Glu-72, His-1 96, 0.02M-tris zinc site 2* His-1 20 3* His-29, Lys-84 4 His-303 Hg(NH3)z2 Myoglobin (38) Equimolar 3 M-AS 6.5 1 His GH-1 close to AG+ site + (HgO in AS) with GH-4 (same as Zn2+)protein LYSA-14 HgAc:! Glycera haemoglobin (39, 40) 6.8 1 Cys-B30 -b HgAcz Concanavalin (41) 2.1 M-phosphate 6.0 1 His-1 27 2 Met-129, His-127 % .1.-3 Asp-1 18 E4* ASP-80, ASX-83, Tv-100, % His-205 25 Lys-135 fi 6 Tyr-12, His-205 g (continued overleaf) Table 2-continued Conc. of Site Reagent Protein (Ref.) reagent Bufer/salta pH number Binding site HgClz Hen egg white lysozyme 0.85M-NaC1 4.7 1 major Arg-14, His-15, Asn-93, chloride (20) Lys-96, Arg-128 (same as PdCh2-, PtCk2-) HgAcz Prealbumin (1 9) 0.5 mM 3.1M-AS 5.0 4 sites Cys-10 (one per monomer) HgAcz Human lysozyme (42) 50 mM 3M-NaCI 4.5 1* 0.02M-NaAc 2 HgAca Chironomus haemoglobin 3.75M-phosphate 7 1* His-G2, Asn-G7 (43) 2 His-G 19 Hg(CN), Lamprey haemoglobin (39) 0.5 mM 2M-phosphate 6.8 1* Three sites close to SH of 10 pM-NaCN 2 Cys-141 position, i.e.side-3 chain occupies different positions. (HgCN42- does not bind) HgClz Lactate dehydrogenase (44) 10 x conc. 1 CY@H) of protein 2 CY s(SH) HgClz Papain (HgCI Blocked-SH) mM 1* His-159 (45) 2 Asn-194 HgCIz Haemoglobin (46) 1.9M-AS 7.0 1* Cys-l040!(SH) 2* Cys-93 P(SH) Hg(Ac)z Glycera haemoglobin (40) 2 x protein 1 CYS-30 (B39) conc. prior to crystal- lization EtHgCl Insulin (47, 48) Sat. O.05M-acetate 6.3 1 His-B5 0.01 M-ZnAcz EtHgCl Calcium binding protein (37) 0.8 mM 4M-phosphate 6.8 1 Cys-18-SH MeHgCl MeHgNOs Lactate dehydrogenase (44) MeHgNOz Adenyl kinase (49) 0.05 mM 1* cys-25 2 CYS- 1 87 3 His-36 MeHgCl Concanavalin (41) 2.1 M-phosphate 6.0 1* Met-129, His-127 2* His-205 3* Tyr-100 4* Asp-1 18 5 Tyr- 100 6 His-180, Gln-87, Trp-88, Trp-182 PCMBS Papain (45) 10 mM Methanol 9.3 1 His-8 1 Water 2 His-1 59 PCMBS Myoglobin (38) Equimolar 3M-AS 6.5 1 His-G14 (Hg binding) with Asn-H8, Lys-FG2, Ser-F7 protein (Sulphonate binding) PCMBS Lysozyme (hen egg white) (20) 0.85M-NaC1 4.7 1 Sulphonate of PCMBS binds Arg-68 PCMB Papain (45) 1.5 mM Methanol 9.3 1* His-81 (Same as PCMBS) PCMA Water 2 His-159 PCMB Bovine pancreas basic Sat.2.25M-phosphate 10 1* C-terminus trypsin inhibitor (50) 2 Lys-41, Tyr-10 3 N-terminus PCMB Calcium binding protein (37) 0.8 mM 4M-phosphate 6.8 1* Cys-18 (SH) PCMA PCMA Myoglobin (38) Equimolar 3M-AS 6.5 1* His-GH-1 with Asn-GH-4 protein Lys-A- 14 R.PCMB Lactate dehydrogenase (44) 10 x protein 1* CYS(SH) conc. 2 Cys(SH) PCMB Haemoglobin (46) 1* Cys-93 P(SH) (continued overleaf) Table 2-continued v1 h, Conc. of Site Reagent Protein (Ref.) reagent Buflerlsalt PH number Binding site HMSA Bovine pancreas basic 4 mM 2.25M-phosphate 10 1 Asn-24, Gln-3 1, Lys-15 G3 trypsin inhibitor (50) 2 N-terminus 3 N-terminus % Q4 Asn-24, Gln-3 1, Lys-15 3 5 C-terminus G MSSS Bovine pancreas basic 6 mM 2.25M-phosphate 10 1* N-terminus %trypsin inhi bit or (50) 2* Tyr-21, Arg-19 3 3 Asn-24, Gln-31, Lys-15 ci 4 N-terminus 3 5 LYS-46 3 HMTS Bovine pancreas basic 6 mM 2.25M-phosphate 10 1 N-terminus s$’trypsin inhi bit or (50) HMTS Lysozyme (20) 0.85M-NaC1 4.7 Sulphonate binding to Arg-68 Baker’s Haemoglobin (46) 1.9M-AS 7 1 CYS-93P(SH) Dimercurial Baker’s Lactate dehydrogenase (44) 10 x protein 1 Dimercurial conc.PHMB Lactate dehydrogenase (44) 4 x protein conc. PHMBS Maiate dehydrogenase (51) 0.1 mM 2.8M-AS 5 3 sites 0.1 M-NaAc PMA a-Chymotrypsin (52) 3.5M-phosphate 4.2 1, 2 N-terminus PhHgAc 24% dioxan s-s cys-1-27 Thiomersal Calcium binding protein (37) 0.8 mM 4M-phosphate 6.8 1 CyS-l8(SH) PhHgAc PhHgNOz DMA Therrnolysin (24) 0.001M 5%DMS 7.5 1 His-231 0.01 M-CaAca 0.01 M-tris/acetate Thiomersal Liver alcohol dehydrogenase 10-2 rnM tris/HCI 8.4 1* cys-240 (53) 0.05M 1* cys-9 DMA Lactate dehydrogenase (24) 10 x protein 2 CysiSH) conc. DMA Haemoglobin (46) p-chain SH Mersal yl Ferricyt ochrome 4.6M-phosphate 6.2 His-33 c (horse) (54) Mersalyl Cytochrome bj (14) 0.3 mM 3M-phosphate 7.5 Glu-48 Tyr-27(0) Arg-84 Mersalyl Subtilisin BPN’ (55) 0.9 niM 2.1M-AS 5.9 1 His-64 0.05M-NaAc 2 N-terminus 3 His-64 Mersaly1 Chironomus haemoglobin 3.75M-phosphate 7 1* His-G2, Asn-67, His-G19.(43) 2* Same as HgAc2 Mersaly1 Concanavalin A (22) 0.1 mM 1* His-127, Met-129 2 His- 127 K2HgT4 Ferricytochrome c (tuna) 0.3 mM 4M-AS 6 1 Gln-16 b (56) Cys-17 thioether bridge to haem g2 Gly-37, Asn-60 Rubredoxin (17) High conc. 3.3M-AS 4 1* Gly-43, Met-1 8 3 (continued overleaf) $ 3 Table 2-continued Conc. of' Site Reagent Protein (ReJ) reagent Buflerlsalt a PH number Binding site KzHgI4 Papain (45) 5 mM Methanol water 1 N-terminus.(same as His-1 59) 2 IrCls3- and PtCl6'- KHgI3 Myoglobin (35) Same as 3.0M-AS 6.5 1* Next to haem in protein hydrophobic pocket conc. 2 Lys-FG2, Gln-F6, Asn-118, Gln-114 Lysozyme (20) 0.85M-NaC1 4.7 1 major Arg-13, Arg-13' 2 minor (2 sites 5.6 A apart) (same as IrCW, AuCIg-, PdI42-) Chironomus haemoglobin 3.75M-phosphate 7 1 haem (43) 2* His-G19 Lamprey haemoglobin (39) 0.5 mM 2M-phosphate 6.8 1* His-73 10p.M-NaCN 2 cys- 14 1 3 cys-141 Glycera haemoglobin (40) 1.5 mM 3.2M-AS 6.8 3 sites Cys-B30 (major site) phosphate (a) AS E ammonium sulphate; Ac = acetate; Ph 3 phenyl; Me = methyl; Et = ethyl; PCMB = p-chloromercuribenzoate; PCMBS = p-chloro-mercuribenzene sulphonate; PHMB = p-hydroxymercuribenzoate; PHMBS = p-hydroxymercuribenzene sulphonate; HMSA = 3-or 5-hydroxy-mercurisalicyclic acid ; MSSS = 3-hydroxymercuri-5-sulphosalicyclicacid ; PCMA = p-chloromercurianiline; Mersalyl = salyrganic acid ; DMA = dimercuriacetate; Baker's Dimercurial = 1,4-diacetoxymercuri-2,3-dimethoxybutane;HMTS 3 2-hydroxymercuritoluol-4-sulphonicacid.The chemical formiilae for these mercurials are further detailed in reference 7. Blundell and Jenkins Sulphur atoms are not only possible ligands for mercurials. In fact, histidines very often bind mercury reagents. Imidazole becomes a very good ligand above pH 6-7 when it loses its proton. For instance, Table 2 shows that thermolysin, which has no methionine or cysteine, binds DMA, mercurisuccinimide, and HgCl2 through histidines.The mercury-containing compound, mersalyl, has been called a histidine specific reagent as it binds to a histidine in cytochrome c and subtilisin BPN.31 However, neither of these proteins has thiol groups. In calcium-binding protein, mersalyl binds to a thiol group. The specificity, if any, of mersalyl is most likely due to its large size. The specificity of mercurials seems to be very dependent on their size, shape, and substituent groups. .We have seen that in lactate dehydrogenase, the bulky Baker’s mercurial binds only one thiol whereas smaller reagents bind two. The binding of mercurials to the immunoglobulin fragment Fab New was also studied by varying the nature of the reagent until ones were found which bound specific- ally.In alcohol dehydrogenase, which contains 14 thiols, most mercurials denatured the protein, but one, thiomersal, gave a more specific reaction which did not lead to denat~ration.5~ The kind of interaction which may give rise to binding of mercurials to pro- teins was evident from the early studies of myoglobin and lysozyme. In myo- globin, PCMBS (see Table 2) binds to a histidine. Its negative charge interacts with the positive charge of a lysine on a neighbouring molecule in the lattice. PCMBS does not bind at this site; this has no negative charge. In fact this kind of interaction may easily give rise to binding of PCMBS in a way which does not involve the mercury atom at all.In lysozyme, PCMBS binds only through the sulphonate ! In the case of carbonic anhydrase mercury-containing sulphon- amides bind as a bivalent ligand with the mercury interacting with the imidazole of His-63 and the sulphonamide group binding to the active site zinc (see Figure l).36 This results in strong inhibition. Figure 1 The interaction of a sulphonamide inhibitor with carbonic anhydrase (Reproduced by permission from A. Liljas, PhD Thesis, University of Uppsala, 1971) The Binding of Heavy Metals to Proteins Anionic mercury complexes such as HgC1g2 -, HgBr42-, or Hg(SCN)g2 -also bind proteins. Despite the fact that the anions have a negative charge the mer- cury can still become bound by the negatively charged cysteine, as in lamprey haemoglobin.This is probably due to the dissociation of the complex. Thus Hg142- dissociates to give rise to HgIs-, HgI2, and T-in solution, and the reac- tion may be through an uncharged species. On the other hand the anionic forms can interact electrostatically as in lysozyme or in the minor sites of ferrocyto- chrome and myoglobin or through hydrophobic interactions as in myoglobin or chironomus haemoglobin. This is discussed in Sections 6 and 7. The hydro- phobicity of these and organomercurial reagents may be reflected in a tendency to interact specifically with membrane components. Thus some membrane adenyl cyclases are specifically inhibited by MeHgCl at low concentrations.ll B. Mercuration of Disu1phides.-Steinberg and Sperling57 have shown that mercury atoms may be inserted into disulphide bonds of cystines.The mercura- tion of the cystine may be carried out either by reduction of the disulphide followed by reaction with mercuric ions58 or in one stage by using the reducing mercurous ions : Hgs2++ RSSR = 2(RSHg)f = RSHgSR + Hg2+ The S-Hg-S system formed in this way is linear and is about 3 longer than the disulphide bond. Despite this, substitution into one disulphide of ribo- nuclease58 or papain59 appears to have little effect on the conformation or biological activity, and insertion of mercury atoms into immunoglobulin frag- ments has little effect on antibody binding.60 However, mercurated ribo-nuclease ~rystallized,~8 and the mercurated immunoglobulin fragment, Fab New,6* crystallized non-isomorphously.Nevertheless, a mercurated Bence-Jones dimer61 and mercurated a-lactalbumin62 gave a smaller change in structure of the protein in the crystal. In insulin a selective mercuration of one disulphide, A6-All, is given. C. Silver Ions.-Table 3 shows that silver nitrate reacts either with cysteine as in haemoglobin or more often with histidine as in myoglobin, trypsin,63 and carboxypeptidase. The ions react in a similar way to mercuric ions such as Hg(NH3)z2+ and probably also form an ammonia complex, Ag(NH3)4+, when ammonium sulphate is present. The fact that Ag+ is less polarizing and not as 67 I. Z. Steinberg and R. Sperling, in ‘Conformation of Biopolymers’, ed.G. N. Ramachandran, Academic Press, New York, 1967, p. 215. R. Sperling, Y. Burstein, and 1. Z. Steinberg, Biochemistry, 1969, 8, 3810. 6B R. Arnon and E. Shapira, J. Biol. Chem., 1969, 244, 1033. 6o L. A. Steiner and P. M. Blumberg, Biochemistry, 1971, 10, 4725. 61 K. R. Ely, R. L. Girling, M. Schiffer, D. E. Cunningham, and A. E. Edmundson, Biochem. J., 1973,12,4233. 6a R. Aschaffenberg, personal communication, 1976. O3 J. L. Chambers, G. G. Christoph, M. Krieger, L. Hay, and R. M. Stroud, Biochem, Biophys. Res. Comm.,1914,59,70. Table 3 The binding sites of silver in proteins Conc. of Site Reagent Protein (Ref.) reagent Buferlsalt pH number Binding site AgN03 Thermolysin (24) 5 mM 5% DMS 5.5 1 His-231 0.01 M-CaAc2 2 His-88 0.01 M-tris/Ac AgNOa Carboxypeptidase A (25) 5 mM 0.2M-NaAc 8.0 1 His-166, Ser-158 2 His-1 20 3 His-29, Lys-84 4 His-303 AgN03 Myoglo bin (3 8) equimolar 3M-AS 6.5 1 His-B-5 with His-GH-1 protein AgN03 Haemoglobin (46) 1.9M-AS 7.0 1* CYS-1044SH) 2* Cy~-93p(SH) AgN03 Trypsin (63) 12 mM MgS04 7.5 1 His-57, Asp-102 The Binding of Heavy Metals to Proteins reactive as Hg2+ may explain why in acetate buffer Ag+ ions bind glugacon in a similar way to the mercuric chloride but with less disordering.64 D.Tetrachloroplatinate(u), Tetrachloroaurate(m), and Analogous Reagents.-Many data are available for the reactive tetrachloroplatinate ion, PtC142-. The reaction conditions and binding sites of this and related heavy atom reagents are summarized in Table 4.Platinum, palladium, and gold are fairly soft, forming stable covalent com- plexes with soft ligands such as chloride, bromide, iodide, ammonia, imidazole, and sulphur ligands. The stereochemistry of their complexes depends critically on the number of ‘d’ electrons. Thus ds ions of PdII, PtII, and AuIII are pre- dominantly square-planar. This includes PtCI&, Pt(NH3)42+, Pt(CN)42 -, and AuC14-. Occasionally these ions accept one further ligand to give a square pyramid or two ligands to give octahedral co-ordination, but the fifth and sixth ligands are much more weakly bound. On the other hand PtIV has a d6 electron configuration and forms stable octahedral complexes such as PfC1G2- with six covalently bound and equivalent ligands.In order to understand their protein chemistry we must consider the factors affecting the thermodynamic and kinetic stability of the complexes, not only the potential protein ligands but also the effect of salting-out agent, buffer, and pH on the reaction. Sigler and Blow74 have drawn attention to the fact that NH3 from (NH4)2S04 may displace chloride from the square-planar PtCh2-, and alter the reaction with proteins. They transferred a-chymotrypsin crystals into phosphate and found this led to faster and more reproducible reaction with .~~PtCh2-. Wyckoff et ~ 1found further evidence for the displacement of chloride by ammonia in their study of ribonucleases; only freshly prepared PtCh2- reacted. The platinum complexes found in the presence of ammonium sulphate and phosphate are illustrated in Figure 2.33 The formation of the phosphate complex may be minimized if a large excess of chloride ions is present.Acetate may also form complexes in time if there is an excess of acetate and no chloride. These complexes vary in the nature of the ligands and in their charge; they will therefore react in very different ways with a protein.33 The charged groups PtCh2-, PtC13(P04)4-, PtNH3C12(P04)2-, and Pt(NH3)42+ do not penetrate into an hydrophobic protein core. The anionic groups do not react with anionic re- agents such as RS-but are attacked more readily by neutral nucleophiles such as R-SH, R-imidazole, or R-NH2. The cationic group Pt(NH3)42+ is rather inert due to the weak trans effect of the ammonia ligands, and it is most likely to form electrostatic complexes with anionic groups such as carboxylate. The neutral Pt(NH3)2C12 molecule can penetrate into hydrophobic areas. It will require a stronger nucleophile but will be reactive to anionic nucleophiles such as R-S-.K. Sasaki, S. Dockerill, D. Adamiak, I. J. Tickle, and T. L. Blundell, Nature, 1975, 257, 751. 74 P. B. Sigler and D. M. Blow, J. Mol. Biol., 1965, 12, 17. Table 4 Protein ligands of platinum, palladium, and gold complexes Conc. of Site Reagent Protein (Ref.) reagent Buferlsalt PH number Binding site K2PtC14 Concanavalin A (41, 65) 0.5 mM 2.1 M-phosphate 6.0 1* Met-129, His-127 2 Met-1 29 Chironomus haemoglobin 3.75M-phosphate 7.0 1* Met-H17 (43) 2* His-G2, C-terminus 3 His-G 19 Pt(N02)2 Chironomus haemoglobin 3.75M-phosphate 7.0 1 His-G2 (NH3)2 (43) 2* His-G 19 Pt(en)Cl2 Ribonuclease S (66) 2 mM 3.2M-AS 8 1 His-1 19 Pt(en)Clz Ribonuclease S (66) 2 mM 3.2M-AS 5.5 1* Met-29 Pt(en)Clz Lactate dehydrogenase (44) 2.5 mM 1 CYS(SH) 2* Cys(SH) K2PdC14 Lysozyme (20) 0.85M-NaC1 4.7 1 Arg-14, His-15, Asn-93, K2PdBr4 Lys-96, Arg-128 K2PdI4 Lysozyme (20) 0.85M-NaC1 4.7 1 Arg-13, Arg-13' (same as IrCltj3-, Hgh2-, AuCh-) K2PtC14 Concanavalin (22) 1 mM 1* His-127, Met-129 2 Met-42 K2PtC14 Ferricytochrome c (horse) 4.6M-phosphate 6.2 1* (54) 2* Met-65 Close together bY 3 His-33 5 Tuna ferrocytochrome 0.1 mM 95% AS 6 1 Met-65 8-c (67)Cytochrome C550 (68) 1.3 mM 1* % a-Chymotrypsin (69) 3.5M-phosphate 4.2 1, 2* N-terminus and S-S of 3* 24% dioxan cys- 1-1 27 5' 3, 4* Met-192 (continued overleaf) Table 4-contiiiued Conc.of Site ' Reageut Protein (Ref.) reagent Buferlsalt PH number Binding site K2PtC1.4 Subtilisin BPN' (55) 0.65 mM 2.1 M-AS 5.9 1* 5 Met-50 % O.OO5M-acetate 2 His-64 G K2PtCh Subtilisin novo (27) 1 Met-50 % 2 Trp-241 His-238 !$9Trp-106 3 Ala-1 (N-terminus) $-b QK2PtC14 Thermolysin (24) 6 mM 5%DMS 5.8 1 His-250 2;. ..r0.01-CaAc2 0 0.01 M-tris/ 2 His-21 6 %Y acetate ..* K2PtC14 Carboxypeptidase A (25) 0.2M-LiCl 7.5 1 C~S-2.161 (-S-S-) E 0.02M-tris 2" Met-103 3* N-terminus : Ala-1 4 His-303 KzPtC16 a-Chymotrypsin (69) 3.5M-phosphate 4.2 1,2* Terminal amino-group and S-S of Cys-1-127 (same as PtCV-) KzPtCls Papain (45) 5 mM Methanoljwater 9.3 1 N-terminus 2 His-1 59 (same as Hgk2-, IrCl$-) KzPtCls Csncanavalin A (65) 3 mM 2.1 M-phosphate 6.0 1* Met-129 (same as PtCV-) 2 Met-129 3 Met-42 KzPtCls Lysozyme (20) 0.85M-NaC1 4.7 1 Arg-14, Hisl5, Asn-93, Lys-96, Arg-128 (same as HgC12 and PdC14) Thermolysin (24) Sat.5% DMS 5.6 1 major His-250 soh. 0.01 M-CaAc2 0.01 M-trislacetate 5 minor Todination of tyrosine High Potential Iron Protein : HiPIP (70) 0.5 mM 3.2M-AS 6.5 1* Met-49 2 3 High Potential Iron Protein: HiPIP (70) 10 mM 3.2M-AS 6.5 1* Met.-49 2 K2P t (Nod4 Adenyl kinase (49) 2 mM 1* His-36 2 KAuC14 Carbonic anhydrase (36) 1 Zn, Thr 197, X139 2 HzO on Zn, His 128 3 Arg 25, carbonyl of His 35 KAuC14 Ferricytochrome c2 10-100 x 3.2M-AS 5.8 1* His-42 (Rhodospirillum rubrum) protein (15) conc. 2 ASP-3 KAuC14 Lysozyme chloride (20) 0.85M-NaC1 4.7 1 Arg-12, Arg-13’ (same as Hgh2-) NaAuC14 Lactate dehydrogenase (44) 1 mM 1 CYS(SH)2 CYS(SH) CYs(SH) hrHAuC14 Glycera haemoglobin (40) 1.5 mM 2.6M-AS 6.8 1 CYS-30 (B39) 2 O.06M-phosphate His-72 c i$ KAuC14 Myoglobin (38) Equimolar 3M-AS 6.5 1 His-B5 (same as Ag) z with His-GH1 %protein 3a-Chymotrypsin (71) 1 Met- 1 92, Cys-191-220 % s-(same as PtC142) Y (continued overleaf) 2 !$ Table &continued 8Conc.of Site Reagent Protein (Ref.) reagent Bu-er Jsalt PH number Binding site % QHAuC14 Adenyl kinase (49) 1 His-36 8 MMTGA Carbonic anhydrase (36) 2.3M-AS 8.5 1 Zn, Thr-197, X139 L" + K2Pt(CN)4 aK2R(CN)4 Ferrocytochrome c (tuna) 6 mM 95% AS 6 1 No near neighbour f 2(72) Same as Hgh2- rr Lys-53, Ala-4, Lys-7, "cl2 Ser-100, Val-3 a rrGlu-44, Gln-70, Lys-72, 3.Lys-73 h 3 Lys-99, Lys-99', Ser-103, 4 Ser-103' Glu-21, LYS-7, LYS-25 5 Ile-269 (main chain) Liver alcohol 1 mM O.OSM-tris/HCI 8.4 1 ASP-223, LYS-228, dehydrogenase (73) Arg-47, Arg-369 Adenyl kinase (49) 2 mM 1* 2 Major site near 3 His-36 4 KAu(CN)2 Carbonic anhydrase (36) 20 mM 2.3M-AS 8.5 1* H2O on Zn, His-128 KAu(CN)2 Lamprey haemoglobin (39) 1 mM 3.6M-AS 6.8 1 Lys-106, Ser-107 20pM-NaCN Glu-92 2 Ser-107, Val-8 3 cys- 141 Not in presence of phosphate 'b K. D.Hardman and C. F.Ainsworth, Biochemistry, 1972, 11,4910. H. W. Wyckoff, K. D. Hardman, N. M. Allewell, T. Inagami, D. Tsernoglou, L. N. Johnson, and F. M. Richards, J. Biol. Chem., 1967,242,3984."N. Tanaka, T. Yamane, T. Tsukihara, T. Ashida, and M. Kakudo, J. Biochem., 1975, 77, 147. R.Timkovich and R.E. Dickerson, J. Mol. Biol., 1973, 79, 39. 'eP.B. Sigler, D. M. Blow, B. W. Matthews, and R. Henderson, J. Mol. Biol., 1968, 35, 143. 'O C. W. Carter, J. Kraut, S. T. Freer, and R. A. Alden, J. Biol. Chem., 1974, 249, 6339. 71 A. Tulinsky, personal communication, 1974. 7t T. Takano, R.Swanson, 0. B. Kallai, and R. E. Dickerson, Cold Spring Harbour Symposium Quant. Biol., 1974, 34, 397. 79 C. I. Branden, H. Eklund, B. Nordstrom, T.Boieve, G. Soderlund, E. Zeppezauer, 1. Ohlsson, and A. Akeson, Proc.. Nut. Acud. Sci. U.S.A., 1973,70,2439. c. The Binding of Heavy Metals to Proteins I trans [Pt(NH,)C1,(P0,)13-(4 Figure 2 The complexes of platinum which may exist when PtC142-reacts with ammonium sulphate or phosphate (Reproduced by permission from G. Petsko, DPhil Thesis, Oxford University, 1973) These observations provide a rationale for the observed reactions of protein crystals when soaked in PtC142-.33 At acid pH it reacts with methionines, cystine disulphides, N-termini, and histidine imidazole, which all form stable complexes. These are good nucleophiles which can displace chloride from platinum com- plexes. Cysteine -SH groups are less nucleophilic.PtC142- does not react with the cysteines of erythrocruorin or triosephosphate isomerase at about pH 7 in phosphate buffer, or the cysteines of malate dehydrogenase at pH 5.0 in the presence of ammonium sulphate. However, prealbumin and triosephosphate isomerase at about pH 7 in ammonium sulphate react with PtC12- through their cysteine groups, and these reactions occur in one to two days. In these cases the nucleophile is probably -S-and the reaction may occur with PtC12(NH&, which will have formed within 24 h. The reaction of methionine and ionized cysteine appear to be faster than histidine; and so time of reaction may provide a further variable controlling the specificity. From the discussion above it can be seen that tetrachloroplatinate is not a very specific reagent but reaction conditions can enhance binding at certain protein ligands relative to others.Thus in ribonucleases PtC142- binds to a methionine at pH 5.5, but a further site at a histidine is partially occupied at higher pH. PtC142- binds different sites in triosephosphate isomerase depending on whether ammonium sulphate or phosphate are present. In most cases a Blundell and Jenkins square-planar platinum complex results. It is possible that a square-pyrimidal complex of platinum(I1) is formed but unlikely that these complexes are oxidized to octahedral platinum(1v) complexes as suggested by Dickerson et aL31 Wyckoff et aZ.66 have shown that sometimes cis-PtCIz(ethy1enediamine) prevents substitution by two protein ligands trans to each other, a cross-linking reaction which leads to disorder of protein crystals. The rate of the reaction with the protein is slowed down by using the other square-planar anions such as Pt(N0~)4~-.AuC14- reacted with sperm whale myoglobin, but only after 6-9 months at pH 6.5 in ammonium sulphate. The reaction which occurred at two histidines may have occurred through an intermediate amine complex such as AuC13(NH3), AuCh(NH3)2+, etc. Tn aqueous solutions AuC14- is hydrolysed to Au(OH)4- in about one hour34 and AuC14- may also be reduced to free AuI by the oxidation of methionine.34 This complicates interpretation of the reactions. Pt(CN)42- does not allow nucleophilic substitution; in ribonuclease S the substitution is quite different from that of PtCl2-. The binding of stable anions like Pt(CN)42- is described in Section 6. E.Osmium and Iridium Reagents.-The binding sites of osmium and iridium are listed in Table 5. TrCls3- can bind proteins through imidazole or amino- groups as in papain where the binding sites are the same as those given by HgI42- and PfC162-. The anions may also bind basic groups as found in lysozyme and subtilisin novo. In ferrocytochrome CZ, Ir(NH3)& appears also to bind to basic groups, predominantly lysines, but the electron density could possibly be due to I-binding rather than the metal ions. 6 Electrostatic Binding of Heavy Atom Anions to Proteins Proteins contain a number of positively charged groups, including the terminal a-amino and lysine €-amino functions and the guanidinium group of arginine, which may form ion pairs with heavy atom anions.Histidine may also bind anions especially at lower pHs where it is positively charged. We have seen that ions like PtC12- or HgTP tend to be bound covalently by soft ligands such as cysteine, methionine, and histidine by displacement of the halide ligands. However, very often these ions are bound electrostatically. Thus in lysozyme and in the minor site of myoglobin, Hg142- dissociates to HgI3- and binds ionically to the proteins. The myoglobin site is shown in Figure 3 and involves a lysine, two glutamines, and an asparagine (Table 2). The lysozyme site involves two arginine guanidinium groups. The same site can be taken up by PtC142-, AuC14-, IrCl63-, OsCl63-, and Pdh2-.The binding as an ionic species gives rise to lack of specificity of these reagents. If the Pt(CN)42- ion is used the ligands are less likely to be displaced by protein ligands, and so an ionic binding becomes the most likely mode of inter- action with the protein. Thus Table 4 shows that Pt(CN)42- binds at several sites involving lysines in ferricytochrome. In carbonic anhydrase and liver alcohol dehydrogenase Pt(CN)42- binds at positively charged sites in the active site. Table 5 The ligands of osmium and iridium reagents in proteins Conc. of Site Reagent Protein (Ref.) reagent Buferlsalt pH number Os(NH3)& Ferricytochrome c2 10-100 x 3.2M-AS 5.8 1* (Rhodospirillum rubrum) (15) protein conc.2* 3 NasIrCls Subtilisin novo (27) 1 2 3 4 5 NaaIrCls Papain (45) 5 mM Methanol 9.3 1* water 2 0.85M-NaCl 4.7 1 Binding site Glu-37, Lys 112 Lys-56, Met 55 Lys-lo9 LYS- 136 Lys-27, Asn-118 Asn-25 Trp-241, His Trpl06 Gln-103, Asn-240 N-terminus His-1 59 Arg-13, Arg-13' I-or anion binding (same as Hg142-and PtCltj' -) (same as Hgh2-, Pd h2-AuCld) Blundell and Jenkins El5 Leu ser 4 H8-$%YAsn H FC3 His Heme propionic acid (C 1 Figure 3 The binding of Hg13-to myoglobin (Reproduced by permission from J. Mol. Biol., 1968, 31, 305) The heavy-atom anions Pt(CN)42- and Au(CN)2- often act as inhibitors. In carbonic anhydrase they bind close to the zinc atom.In liver alcohol dehydro- genase they bind in strict competition with the coenzyme, NAD. The auricyanide ion binds at two sites. One site is normally occupied by the phosphate groups of the coenzyme and the other by the adenosine part. The tetracyanoplatinate ion binds only to the phosphate ~ite.~3173 binds at In ribonuclease, P~(CN)I~- quite different sites from PtCh2-. Au(CN)2- will also tend to bind at anionic sites, but this is two-co-ordinate and in the presence of soft ligands such as cysteine may give tetrahedral complexes. Similar anion-binding sites may also be occupied by iodide and other halide ions, as occurs in carbonic anhydra~e.~~ 76 J. E. Norne, T. E. Bull, R. Einarsson, B. Lindman, and H. Zeppezauer, Chemica Scripta, 1973, 3 142.167 The Binding of Heavy Metals to Proteins The fact that halide ions bind in similar ways implies that halide in the buffer or salting-out agent could interfere with binding of Au(CN)2- or Pt(CN)q2-. Phosphate may also bind in a similar anion pocket. In chironomus haemo- globin43 Au(CN)2- does not bind in the presence of phosphate, but can be bound in other buffers such as acetate. 7 Binding by van der Waals Interactions Although the early experiments with Hgh2- on sperm whale myoglobin were designed to bind a mercury to a rnethi~nine,~~it was later found that the ion was bound as Hgh- in a hydrophobic pocket close to the haem group76 as shown in Figure 3. This is not so surprising as the iodine ligands are very soft and would give rather good van der Waals interactions which would stabilize the binding.This description of the binding is consistent with the finding that AuI4- and 13-also bind at the same site but HgBrs- does not. Even a single xenon atom can be bound77 when myoglobin is equilibrated with xenon at 2.5 atm, and this interaction cannot be through either ionic or covalent links but must be due to London interactions and induced dipole moments which make up van der Waals interactions. The protein groups are slightly distorted by inclusion of these large groups of atoms, and it is therefore not surprising that the binding depends critically on the nature of the globin. Seal myoglobin binds HgIa- in the hydrophobic pocket whereas haemoglobin does not. 8 Metal Ion Replacement in Metalloproteins A number of metalloproteins have been studied by X-ray analysis.Among these are included several zinc proteins (carbonic anhydrase, carboxypeptidase, thermolysin, and insulin) and calcium proteins (staphylococcal nuclease and thermolysin) where the metals are weakly bound and can be replaced by heavy metals. In some of these, the metal can be replaced by soaking the crystals in a solution of the heavy-metal salt. Thus in nuclease, calcium is directly exchanged for bar- ium by soaking in a solution of barium chloride.78 Similarly, in subtilisin novo, thallous ffuoride replaces a sodium ion although at a slightly different ~ite.2~In others a more drastic procedure is required as the zinc or calcium is more firmly bound.However, initial attempts at removing zinc cofactors in solution were unsuccessful. For instance, addition of 1 ,lo-phenanthroline to carbonic anhydrase gave a zinc-free enzyme which did not crystallize,79 and insulin crystallized in a different space group in the absence of zinc.80 More successwas achieved by soaking the crystals themselves in a solution of a suitable chelating agent. Thus dialysis of carbonic anhydrase crystals against 2,3-l6 R. H. Kretsinger, H. C Watson, and J. C. Kendrew, J. Mol. Biol, 1968, 31, 305. I1 B. P. Schoenborn, H. C. Watson, and J. C. Kendrew, Nature, 1965, 207, 28. 18 A. Arnone, C. J. Bier, F. A. Cotton, V. W. Day, E. E. Hazen, D. C. Richardson, J. S. Richardson, and A.Yonath, J. Biol. Chem., 1971,246,2301. qB B. Tilander, B. Strandberg, and K. Friborg, J. Mol. Biol., 1965,12,740. tio B. W. Low and J. E. Berger, Acfu Cryst., 1961, 14, 82. f 68 BIundeLL and Jenkins dimercaptopropanol in an hydrogen atmosphere produced crystals of the zinc- free enzyme.79 Use of 5-hydroxyquinoline-8-sulphateand ethylenediaminetetra- acetic acid (edta) with carboxypeptidasesl and rhombohedra1 2-Zn insulins2, respectively, gave zinc-free crystals. It appears that the crystal packing stabilizes the zinc-free structure and prevents the conformational charge or disaggregation which occurs in solution. Electron-density maps of insulin later showed that the side-chains in the regions vacated by the zinc atoms were rather disordered and were not reordered on addition of heavy atoms with the exception of cadmium.Imidazole groups of histidines are frequently involved in coordination of replaceable zinc atoms: there are three histidine ligands to zinc in carbonic anhydrase and in insulin, and two histidines and glutamate in carboxypeptidase and thermolysin. Zinc is most easily replaced by transition or B-metal ions such as Fez+, Co”, or Cd2+. Dialysis of the crystalline apoenzyme against 0.0003M mercuric acetate for ten days leads to substitution of mercury at the zinc site in carbonic anhydra~e.7~ Dialysis against 0.005M mercuric chloride gives a similar mercury substitution for carboxypeptidase.81 Cadmium also binds at the zinc site of insulin and the same derivative can be prepared by co-crystallization.*2 Plumbous ions (Pb2+) give low substitution at the zinc sites in insulin, whereas in zinc-free carboxypeptidase two sites 4 A from the zinc site were given.These results are not unexpected. Although lead is a B-metal, the non-group valency of a plumbous ion means that it is larger than mercuric and less polarizing. It tends to bind electronegative groups such as carboxylate rather than imidazole and will not easily replace zinc. Calcium tends to bind oxygen ligands, such as carboxylate side-chains, and is best replaced by other alkaline earth ions such as Sr2+, Ba2+ or tervalent lanthanide ions. Thus in solution the lanthanide, neodymium, will replace calcium in trypsin and trypsinogen83 and in a-amylase.84 Colman et al.24 have demonstrated that three of the four calcium ions of thermolysin may be replaced by either lanthanide ions or by strontium or barium.The crystals were first equilibrated with calcium- free tris-acetate buffer at pH 5.5 for one hour and then a solution of metal ions in the same buffer was added. The four calcium sites in thermolysin can be charac- terised in the following way: Ca 1, the inner double site, Glu-190, Glu-177, H20, Carbonyl 187 Ca 2, the outer double site, Glu-190, Glu-177, Carbonyl 183, H2O Ca 3, the single site at Asp-57 Ca 4, the single site at Asp-200. The lanthanides bind at sites 1,3, and 4, and calcium is concurrentIy ejected from site 2 also. This presumably is the result of the higher charge of the lanthanides; W.N. Lipscomb, J. C. Coppola, J. A. Hartsuck, M. L. Ludwig, H. Muirhead, J. Searl, and T. A. Steitz, J. Mol. Biol., 1966, 19, 423. *a M. J. Adams, E. Collier, G. Dodson, D. C. Hodgkin, and S. Ramaseshan, Abstracts of the Seventh Meeting of the International Union of Crystallography, 1966, A 165. 53 D. W. Darnall and E. R. Bimbaum, J. Biol. Chem., 1970,245, 6684. G. E. Srnocka, E. R. Birnbaum, and D. Darnell, Biochemistry, 1971,10,4556. 169 The Binding of Heavy Metals to Proteins barium and strontium replace all four calciums. The heavy atom usually has a larger radius than the atom it replaces. In therniolysin a more similar substitu- tion is given when the metal ion has a smaller radius than calcium, for example, by the smaller lanthanides such as Lu3+.85 Replacement of calcium by barium involves the introduction of a larger metal ion, and it is not surprising to discover that the barium atom in nuclease is 0.75A from the calcium ion position.78 Similarly the mercury positions in carbonic anhydrase36 and carboxypeptidase25 differ by 0.7 and 0.25 A, respect-ively, from the zinc positions.These small displacements clearly contribute towards the differences of catalytic activity found in the heavy-metal enzymes. Thus esterase activity of Hg-carboxypeptidase is slightly increased in comparison to the zinc enzyme, whereas the lead enzyme has only 5004 esterase activity. But neither Hg nor Pb carboxypeptidases have peptidase activity.5 9 Biological Implications The implications of the heavy-atom binding sites found in proteins are rather different for the two classes of metals.The hard cations UOz2+, Pb2+, T1+, Ba2+, and the lanthanides can be seen to be bound only by proteins acting as multidentate ligands. These ligands are generally oxygen ligands except for Pb2+ when nitrogen ligands are sometimes found. Carboxylates are most prom- inently found as ligands both in the proteins and in buffers such as citrate which interfere with these reactions. (Thus it is possible that negative staining density is affected by the concentration of carboxylates when Pb2+ or U0z2+are used.) The toxicity may be expected to depend on the nature of the complex. Thus Pb2f is less toxic as the citrate or aspartate complex to Aspergillus niger.86 The diuretic effect of uranyl is reduced by citrates7 in rats.In the case of lead and thallium this picture is complicated by the possibility that PbIV and TP species will be formed which would tend to bind to softer ligands. However, even in this case it is clear that initially the uptake and distribution of the compounds will depend on their binding to hard ligands such as carboxylates. The compounds of mercury, cadmium, silver, gold, palladium, and platinum show very different reactivity. However, the range of possible ligands is clearly larger than has been suggested on the basis of reaction with isolated amino-acids. This is again, presumably, due to the protein acting as a multidentate ligand.Thus Hg2+ is frequently found to bind to imidazole. The metals show a strong dependence of reactivity on the nature of the other ligands. Thus methylmercury compounds can bind to ligands in hydrophobic pockets not accessible to Hg2+. These compounds can also enter membranes. In protein crystallography iodide complexes of mercury have found use for the same reason. A spectacular instance of the specific requirement for ligands is afforded by the great difference in the toxicity between the cis and trans isomers of Pt(NH3)zClz. Although this probably involves reaction with nucleotides rather than proteins it underlines the need to *6 B W. Matthews and L. H. Weaver, Biochemistry, 1974, 13, 1719. 86 I. V. Zlochevsbaia and I. L. Rabotnova, Mikrobiologiya, 1968, 37, 691.V. Nigrovic and E. J. Caftuny, Nature, 1974, 247, 381. Blundell and Jenkins consider biological ligands as multidentate. Protein crystallography also offers evidence of the importance of buffering ions, as many reactions of particularly PtII and AuIII are affected by the presence of NH or amines in the soaking solution. Conversely some complexes such as Pt(CN)q2- and Au(CN)2- are stable and not easily attacked. These bind as anions rather than as electrophiles and may mimic the binding of other anions such as phosphate. Finally, many cases have now been investigated where a heavy metal has replaced a lighter one in a protein. A generalization from this behaviour is that the exact position and stereochemistry are rarely identical.This may explain the observed differences of specificity and activity. The detailed description of heavy-metal protein interactions presented here shows that toxicity depends critically not only on the chemistry of the metal ion but also on the nature of the ligands in the reagent and in the medium. The metabolism of the metal ion leading to chemical modification will alter the poten- tial interactions in the organism and change the toxicity of the reagent. General- izations concerning the ‘toxicity’ of any heavy atom clearly require the kind of detailed discussion considered proper for other metabolites. We would like to thank all those who supplied unpublished data concerning heavy metal binding sites in proteins studied by X-ray analysis. We are also grateful to Dr P.Sadler and Dr A. Thompson for reading and making helpful comments on the manuscript.

ISSN:0306-0012    

DOI:10.1039/CS9770600139

出版商:RSC    

年代:1977

数据来源: RSC

摘要:

The Chemistry and Binding Properties of Aluminium Phosphates By J. H. Morris, P. G. Perkins, A. E. A. Rose, and W. E. Smith DEPARTMENT OF PURE AND APPLIED CHEMISTRY, UNIVERSITY OF STRATHCLYDE, GLASGOW, G1 1XL This review covers the chemistry of some aluminium phosphates and their application as systems which will bind together particulate or fibre materials to form structurally cohesive products. No comprehensive review of aluminium phosphates has appeared since the work of Kingery.l 1 Classification of Binders A binder may be defined as1 ‘a bonding material formed by reaction which is capable of imparting strength and elasticity to a body of aggregate in which it is present in minor proportion, while retaining in part the properties of the aggregate’. Attempts have been made to systematize and classify the large variety of bindersl-3 on the basis of the type of ‘bond formation’, and a comprehensive classification which is in common usage has been suggested by Sychev (Figure 1).2 Mineral binders are divided broadly into two classes, adhesives and cements.Adhesives are considered to be those binders, such as water glass and metal acid phosphates, which are suitable for immediate use. The second class is by far the larger and so has been subdivided into those which harden by physico- chemical processes, and those which harden by formation of a new insoluble or sparingly soluble phase as the result of a chemical reaction. The first sub-group includes hardeners such as highly concentrated, finely divided suspensions of clays, oxides, and carbides; here water is the binder. Evaporation of water from the system causes capillary forces to act in the pores.As a result, surface forces pull the particles together and a coagulation structure is formed. The second sub-group can be further divided into those cements which react with water, i.e., the hydraulic cements, and those involving acid-base interaction. The latter include metal oxide and hydroxide reactions with phosphoric acid, cements made from oxides and water glass, e.g. ZnO + NazSiO3, and silicate coatings, e.g., Ca(OH)2 + NazSiO3. With regard to phosphate binding, Sychev2 differentiates between bonds formed from phosphate binders (adhesives) and bonds involving phosphates which are formed as a result of chemical reaction (cements), in which the phosphoric acid is the material which is added as the binder.J. F. Wygant, in ‘Ceramic Fabrication Processes’, ed. W. D. Kingery, Chapman and Hall, London, 1958, p. 171. M. M. Sychev,Zhur. priklad. Khim., 1970,43,528; 1970,43,758, ’1. Teoreanu, Rev. Roumaine Chim., 1972,17, 121 1. The Chemistry and Binding Properties of Aluminium Phosphates ~ .I mI-. 174 Morris, Perkins, Rose, and Smith 2 General Binding Principles The requirements for cementitious bonding have been studied,l-4 and with particular regard to aluminium phosphate are as follows: Two essential properties of binders are cohesion and adhesion. Cohesion of a binding material requires that a continuous structure be formed in situ from a large number of ‘nuclei’.Crystalline materials always fulfill this requirement, although they generally do not show good bonding characteristics. For, although strength can arise from crystal interlocking, such crystalline structures tend to be rigid, and the lack of residual forces in a well-ordered crystal leads to poor adhesive properties. Amorphous structures may also possess three-dimensional continuity and, in contrast with crystalline phases: (i) they are more flexible and can accommodate internal stresses and strains more readily, (ii) the rate of their formation is not so critical, (iii) random structures incorporate residual forces which promote adhesion. Kingery5 has related the formation of disordered structures to the cationic size in a number of phosphate systems of potential binder interest.Large metal cations can form structures in which they have a high co-ordination number. The higher the co-ordination number, the greater the tendency towards ordered regular structures ; conversely, the smaller the cation, the greater the tendency to form structures of low co-ordination number, thereby increasing the degree of randomness in polymers. Thus aluminium, with a relatively small cationic radius, increases the bonding power of a phosphoric acid binder, whereas with the large thorium cation, strength is correspondingly decreased. This is shown in the Table. Table The relation of cationic size to the strength of the phosphoric acid bond (Kingery5) Cation Basicity Ionic radizis6/nm Phosphoric acid bond Be2+ amphoteric 0.035 strength increased AP + amphoteric 0.051 strength increased Fe3+ weak base 0.064 strength increased Th4+ weak base 0.102 strength decreased It has been noted that, in order for the binding phase to be cohesive, it must be capable of forming continuous structures throughout the bonded material.There are two essential requirements for this to happen. (i) Prior to the formation of the binding phase, the binder is in the form of a well-dispersed system which allows for mobility of ions. Phosphate binders are always added as aqueous solutions partly for this reason. (ii) The binder is required to have functional groups if continuous network structures are to be formed by condensation- M.M. Sychev, Zhur. priklad. Khim., 1971,44, 1740. 6 W. D. Kingery, J. Amer. Ceram. SOC.,1950,33,242. ‘Handbook of Chemistry and Physics’, ed. R. C. Weast, Chemical Rubber Company, Ohio, 197213, 53rd. edn., p. F-177. 175 The Chemistry and Binding Properties of Aluminium Phosphates polymerization reactions; e.g. phosphate binders contain hydroxy-groups and undergo the following condensation-polymerization reaction on dehydration : 0 0 0 0 II II II II HO-P-OH t-HO-P-OH -+-HO-P-0-P-OH + H20 I I 1 I 0 0 0 0I I I I The binding phase must also possess adhesive properties which are either inherent in the binder or are formed during the binding process.Adhesion results from the development of molecular attractive forces acting between the binding phase and the surface of the aggregate or filler. Mechanical adhesion due to surface roughness is also possible, and this can be increased by external pressure. The first requirement for adhesion is close contact between the inter- acting surfaces. The presence of an efficient wetting liquid increases the contact surface and allows Brownian motion to contribute to contact establishment.4 Filler materials with ionic lattices are readily wetted by polar liquids such as water or aqueous solutions. In the case of phosphate reaction cements, the function of the phosphoric acid is three-fold: first to provide the phosphate linkages, secondly to act as a dispersing medium to promote cohesion, and thirdly to serve as a wetting liquid to increase adhesion.A further requirement for adhesion is that the binding phase incorporates residual polar force fields.5 Not only are acid phosphate binders capable of forming amorphous glass phases, the polar forces in which promote adhesion, but they also contain hydrogen bonds, formed through their functional groups (1). These probably serve to increase adhesion by bonding to oxygen atoms on the surface of a filler. I I 0 0 I In reaction cements the rate of the chemical reaction in which the binding phase is formed is important. Fast reactions can produce a breakdown in cohesion or adhesion, resulting in a weak cement. A good example of this is the hydration of the tricalcium aluminate phase in portland cements. Unless a retarder is added to slow its reaction with water, a ‘flash set’ occurs which produces a weak structure.Morris, Perkins, Rose, and Smith With acid phosphate reaction cements, a close relationship has been shown to exist between the rate of reaction and the strength of the bond.59’ The rate of reaction of oxides with phosphoric acid depends on the basicity of the oxide. Because the reaction between ‘amphoteric’ aluminium oxide and phosphoric acid is slow, the phosphate bonds are gradually developed. Basic zinc oxide, on the other hand, reacts so violently with phosphoric acid that the structure is effectively destroyed and no cohesion results. The reaction between metal oxides and phosphoric acid can be retarded in some cases by calcining the oxide to give a less reactive surface or by partially neutralizing the acid or both.Preliminary firing of the oxides of zinc and magnesium to temperatures of 1373-1473 K is sufficient to slow down the reaction to a rate which will allow adhesive bonding to occur. For especially basic oxides, such as calcium oxide, a better technique is to use partially neutra- lized phosphoric acid. It is significant, from the results of Kinger~,~ that aluminium phosphate possesses the highest modulus of rupture of all the phosphates he studied. This is probably due to a fortuitous combination of factors, in that the binding phase which is formed is amorphous and the rate of formation of the product from uncalcined alumina is low.3 Classification of Aluminium Phosphates Phosphates of aluminium exist in many different forms with varying stoicheio- metry, structure, and properties. A number exist in the natural mineral state and feature in early phosphate studies.8~~ These minerals are widely distributed but most deposits are small and not commercially viable. However, the material is actively mined in North Africa and in the U.S.A., where aluminium phosphates are found in association with apatites in the Florida phosphate fields,l0 In Florida, wavelite, A13(PO&(OH)3,5H20, crandallite (also known as pseudo- wavelite), CaAl3(PO&(OH)5,H20, and millisite, (Na,K)CaAl6(PO4)4(0H)g, 3&0, have been identified. Aluminium phosphate is a valuable mineral source of both alumina and phosphate; the latter is obtained by nitric or sulphuric acid leaching to produce mixed and superphospha te-type fertilizers, respectively.lo Studies of the A1203-P205-H20 system have been carried out by several groups under a range of conditions.11-15 Although a hydrate with an A1203: P205 molar ratio of 1 :2 is known, it is apparent that the chemistry of aluminium S. L. Golynko-Vol’fson and L. G. Sudkas, Zhur. priklad. Khim., 1965,38, 1466. * C. Ramelsberg, Pogg. Ann., 1845,64,251. G. C. Wittstein, Pogg. Ann., 1856, 97, 158. lo H. M. Stevens, in ‘Phosphorus and its Compounds’, ed. J. R. Van Wazer, Interscience, London, 1961, Vol. 11, p. 1056. IIJ. C. Brosheer, F.A. Lenfesty, and J. F. Anderson, jun., J. Amer. Chem. SOC.,1954, 76, 5951. l2 R.F.Jameson and J. E. Salmon, J. Chem. Sac., 1954,4013. l3 S. A. Sigov and G. Ya. Sadykova, Uzbek. khim. Zhur., 1961, 2, 7 (Chem. Abs.. 1965, 62, 467 1). li H.Guerin and R. Martin, Compt. rend., 1952,234, 1799. l5 V. N. Sveshnikova, Zhur. neorg. Khim., 1960,5,477. The Chemistry and Binding Properties of Aluminium Phosphates phosphates is based on compounds with Al203:P205 molar ratios of 1 :1 (dP04,XHzO) and 1 :3 [AI(H2P04)3]. Only these will be discussed in detail in this review. Aluminium Phosphates of Ratio &03:P205 = 1:l.-Phosphates with an A1203:P205molar ratio of 1:1are the most common and, to date, have been the most widely studied. This stems from (i) their being naturally occurring, (ii) their structural similarity to silica, and (iii) their commercial value.Hydrates. Two methods have been reported for the preparation of the hydrates of aluminium phosphate.16J7 Preliminary preparation of a solution of alumina in phosphoric acid, with an A1203:P205 molar ratio of 1:2.7, and subsequent refluxing, gave a precipitate of AlP04,2H20 the form of which was concentration dependent;I6concentrated solutions produced metavariscite, and dilute solutions variscite. A number of metastable hydrates of AIP04, with between one and two molecules of water per atom of aluminium, were formed successively in small amounts before transforming to variscite or a mixture of variscite and meta- variscite, depending on the initial concentration of the solution.Variscite and metavariscite have also been obtained17 from a solution of sodium dihydrogen phosphate and aluminium chloride. Variscite was obtained after digestion of the solution at 333 K for 7 days and metavariscite after digestion at 363 K for 27 days. The preparation of the two compounds has been carried out also by the present Reviewers.18 Thermogravimetry16 has shown that variscite and metavariscitels both have two moles of water per gm atom of aluminium and that both are capable of undergoing rehydration if dehydration is incomplete. The latter compound begins to lose weight at 358 K and weight loss is complete at 483 K. Furthermore, differential thermal analysis18 of the materials shows an irreversible endothermic peak between 357 and 446 K, reaching a maximum at 428 K.1.r. measurements have revealed that both hydrates have the structure AIP04,2HzO rather than AI(OH)2,H2P04. Close similarity of the i.r. and X-ray powder diffraction patterns of the synthetic variscite and metavariscite with those of their naturally occurring counterparts was observed, The interplanar distances measured also agree with later work.18 The dihydrate formulation of variscite and metavariscite has been confirmed by lH n.m.r.19 and by X-ray crystal determination of the structure of meta- variscite.20 Here Po4 tetrahedra share vertices with four A104(&0)2 octahedra, and the two water molecules are cis to each other. The surface and porosity properties of the Xerogel non-stoicheiometric l6 F.D’Yvoire, Bull. SOC.chim. France, 1962, 1762. l7 E. 2.Arlidge, E. C. Farmer, B. D. Mitchell, and W. A. Mitchell, J. Appl. Chem., 1963, 13, 17. J. H. Morris, P. G. Perkins, A. Rose, and W. E. Smith, J. Appl. Chem. Biotechnology, 1976, 26, 385; A. E. A. Rose, PhD Thesis, University of Strathclyde. l8 C. Doremieux-Morin, M. Krahe, and F. D’Yvoire, Bull. SOC. chim. France, 1973,409. 2o R. Kniep and D. Mootz, Acra Cryst., 1973, B29,2292. Morris, Perkins, Rose, and Smith Alp04 have also been investigated,21 and its exchange properties with Fe3+ have been studied using Mossbauer spectroscopy.22 Polymorphic forms of the di- hydrate have also been investigated by i.r. spectro~copy.~~ Anhydrous Forms.Anhydrous aluminium phosphate has attracted the interest of many investigators since the observation that Alp04 is isostructural with silica; Alp04 exists in quartz (as berlinite), tridymite, and cristobalite forms. Each of these forms can be prepared by several methods. The berlinite or quartz form of Alp04 is obtainedZ4 by heating precipitated, amorphous, alumin- ium phosphate to 823 K with lithium fluoride. Phosphotridymite and phos- phocristobalite have been prepared by similar methods in the presence of sodium, potassium, or lithium chlorides25 or sodium or potassium fluoride or carbonate26 as promoters of crystallization. Both the hydrates, variscite and metavariscite, are transformed into phos- photridymite on heating (variscite at 673 K and metavariscite at 723 K) and to phosphocristobalite at higher temperatures.16 The quartz, tridymite, and cristobalite forms of aluminium phosphate result when alumina reacts directly with phosphoric a~id.~'*~* The temperature-stability relationships of the aluminium phosphate poly- morphs, studied by X-ray diffraction, thermal analysis, and thermal expansion, revealed29 a close parallel of the system with earlier results for silica, and com- parable results have been found by differential thermal analysis (Scheme l).30 Alp04 1088 & 4K 1298 & 5 K berlinite f--------,tridymite-form -cristobalite-form 446 K 3662 x K 403 K 483 2 x K p +--+ a p f----------,a1+--a2 P-a Si02 1140 K 1743 K quartz +-----+ tridymite -+ cristobalite 446 K 390 K 436 K 493-543 K p-a p+-+Cll1-a2 p++a Scheme 1 V.M. Chertov, R. S. Tyutyunnik, and I. E. Niemark, Adsorbtsiya Adsorbenti, 1974, 2, 109 (Chem. Abs., 1974, 81, I1 1721).*: A. S. Plachinda, V. M. Chertov, I. P. Suzdalev, E. F. Makarov, R.S. Tyutyunnik, and I. E. Niemark, Teor. i eksp. Khim., 1974,10,549 (Chem. Abs., 1974,81, 159417). ZR S. V. Gevorkyan, L. N. Egorova, and A. S. Povarennykh, Geol. Zhur. (Russ.edit.), 1974, 34, 27 (Chem. Abs., 1974, 81, 127842). J. Papailhau, Compt. rend., 1956, 242, 1191 (Chem. Abs., 1956, 50, 12717). 25 M. Orliac and J. Papailhau, Compt. rend., 1963, 256, 202 (Chem. Abs., 1963, 58, 7444). 26 I. Papailhau, Compt. rend., 1955,240,2336 (Chem. Abs., 1955,49, 15408). 27 M. Tsuhako, I.Motooka, and M. Kobayashi, Nippon Kagku Zasshi, 1971,92, 318 (Chem. Abs., 1971, 75, 58098). 28 A. S. Yutina, Z. D. Zhukova, and S. V. Lysak, Izvest. Akad. Nauk S.S.S.R., neorg. Materialy, 1966,3,2020. 29 W. R. Beck, J. Amer. Ceram. SOC.,1949,32,157. 3o 0.W. Floerke and H. Lachenmayr, Ber. deut. keram. Ges., 1962,39, 55 (Chem. Abs., 1962, 56, 12387). The Chemistry and Binding Properties of Aluminium Phosphates However, these must be regarded with caution, since the tridymite form of silica is probably not a true polymorph. Conversions from one form into another are much more rapid than for the corresponding transformations of silica and, with the exception of the berlinite inversion, take place at lower temperatures, The temperature inversion of phosphocristobalite depends on its thermal history31 and, although the cristobalite inversion temperature showed similar variation, the presence of a double endothermic peak during the low-high inversion of phosphocristobalite may indicate that the mechanism is different from that involved in the low-high cristobalite inversion.The enthalpy of the berlinite-phosphotridymite transition, calculated32 from the pressure-temperature curve up to loo0 atm, is 1096 J mol-1, although this value is based on an observed transition temperature of 978 f 7 K, which is about 110 K lower than that found by Beck29 and Fl~erke.~O The a+? berlinite phase transition occurs33 at 854 K, with a change in the heat capacity of 14.6 x J g-l K-l. The enthalpy of the transition is 11.05 J 8-1 and the gain in volume is 8.8 x nm per unit cell.As in quartz, the transformation involved dilation in the c-direction of the unit cell. The latent heat of transformation and the increase in the length of the c-axis and cell volume are much greater for Alp04 than for Si02. Disputably, the parallel between Alp04 and Si02 continues into the melt and A1203,P205 glasses can be made.34-36 The addition of small amounts of silica to aluminium phosphate inhibits the formation of well-ordered crystal structures.30 Silica also lowers the inversion temperature of phosphocristobalite,37 owing to the formation of molten surface layers of Si02-P~O5, whereas all phosphates have a retarding effect on the conversion of quartz to cristobalite, with aluminium phosphate the least effect i~e.3~ There are 15 compounds related by crystaIline structures39 to silica poly- morphs. They have been classified40 into two categories (i) ‘half-breed’ deri- vatives, (ii) ‘stuffed’ derivatives.Alp04 occurs among the ‘half-breed’ derivatives, in which half of the silicon positions are occupied by aluminium atoms and half by phosphorus atoms. In addition to an AlP pair of atoms being isoelectronic with two Si atoms, the combined radii are also comparable6 (AP+, 0.051 nm; 37 F. A. Hummel, J. Amer. Ceram. Sac., 1949, 32, 320. 32 E. C. Shafer and R. Roy, Z. phys. Chem. (Frankfurt), 1957, 11, 30 (Chem. Abs., 1957, 51 I 1023). 33 M. Trocaaz, C. Berger, M. Richard, and L.Eyraud, Bull. SOC.chim. France, 1967, 4256 (Chem. Abs., 1968, 68, 73 102). 34 F. Drexler and W. Schutz, Glustech. Ber., 1951,24, 172. P. Beyersdorfer, Silikut. Tech., 1962, 13, 346 (Chem. Abs., 1963,58, 11075). 36 A. Dietzel and H. J. Poegel, Nuturwiss., 1953, 40, 604. 37 0. W. Floerke, Monatsh., 1972, 103, 81 (Chem. Abs., 1972, 76, 130966). 38 T. Chvatal, Sprechsaal Keram., Glas. Email. Silikate, 1972, 105, 537 (Chem. Abs., 1972, 77, 155 788). ng M. J. Buerger, J. Chem. Phys., 1947, 15, 1. 3o M. J. Buerger, Amer. Min., 1948, 33, 751. 180 Morris, Perkins, Rose, and Smith Si4+, 0.042 nm; P5+, 0.035 nm) and it is required that A1 and P alternate in the str~cture.~l n-Bonding between aluminium and oxygen may also occur, although this is likely to be less effective than that involving pho~phorus.~~ A consequence of similar radii is that corresponding phases of Alp04 and SiOz have cell dimensions which are similar or simple multiples.The cell con- stants are as follows. Berlinite (trigonal, space group P3121: z = 3);33934-48 a0 = 0.494 nm (0.4934.497s nm); co = 1.095 nm (1.084-1.097 nm) (cf. quartz?S ao = 0.4903 nm; co = 4 x 1.078 nm). P-Cristobalite form of AIP04 at room temperature (orthorhombic or pseudotetragonal, space group C2221);~~9~O ao = bo = 0.7099 nm; co = 0.7006 nm (c-P-cristobalite a0 = 0.7003 nm; co = 0.6950 nm). At 523 K, the P-cristobalite form is cubic with ao = 0.711nm.51 Related Studies. The formation of insoluble iron and aluminium phosphates helps to cause soluble phosphate fertilizers to become unavailable to plants and crops.The nature of these phosphates, their availability to plants, and other precipitation properties of aluminium phosphate, have received attention. Digestion52 of precipitated aluminium or iron phosphate in solutions containing alkali and alkaline earth cations leads to a wide range of crystalline species, many of which correspond to natural minerals, e.g., varisci te, strengite, tarana- kite, leucophosphate, and minyulite. The removal of soluble orthophosphates from aqueous solutions using activated alumina53 may be useful for controlling phosphate in rivers and lakes. Early work suggested that aluminium is strongly complexed by ortho-phosphoric acid,S4,55 and the dissociation constants of assumed aluminium complexes with orthophosphoric acid have been reported.56 When ca.1.0-1.5 mol A1203 per mol P205 is dissolved in phosphoric acid, an extremely vis- cous solution is formed which can be dried, yielding an amorphous ~olid.~~?~~ 4’ R. Brill and A. P. De Bretteville, jun., Acra Cryst., 1955,8, 567. 4i J. R. Van Wazer, ‘Phosphorus and its Compounds’, Interscience, London, 1961, Vol. 1, p. 553. 43 B. Sharan and B. N. Dutta, Acfa Crysf., 1964, 17,82. 44 B. N. Dutta, Indian J. Pure Appl. Phys., 1964,2, 362 (Chem. Abs., 1965,62,4720). 45 D. Schwarzenbach, Naturwiss., 1965, 52, 343. R. W. G. Wyckoff, ‘Crystal Structures’, Interscience, London, 1965,Vol. 3, p. 28. 47 H. F. Huttenlocher, Z. Krist., 1935, 90A, 508.48 H. Strunz, Z. Krist., 1941, 103, 228; A. N. Winchell and H. Winchell, ‘The Microscopic Characters of Artificial Inorganic Solid Substances’, Academic Press, New York and London, 1964. 49 R. Brill and A. P. De Bretteville, jun., Amer. Min., 1948, 33, 750. 50 R. C. L. Mooney, Acra Cryst., 1956,9,728. 51 G. Tromel and B. Winkhouse, Fortschr. Min., 1949,28,82 (Chem. Abs., 1951,459980). 52 J. F. Haseman, J. R. Lehr, and J. P. Smith, Proc. Soil Sci. SOC.Amer., 1951, 15,76. 63 R. D. Neufield and G. Thodos, Environ. Sci. Technol., 1963, 3, 661 (Chem. Abs., 1969, 71, 51 717). j4 L. Dede, 2.anorg. Chem., 1923,125,28. j5 S. R. Carter and H. F. Clews, J. Chem. SOC., 1924, 125, 1880. N. Bjerrum and C. R. Dahn, Z. Phys. Chern., 193 1,627.j’ H. H. Greger, Brick and Clay Record, 1950,117,63. 58 H. H. Greger, U.S.P.2460344; B. P. 597 169. 181 The Chemistry and Binding Properties of Aluminium Phosphates These solutions probably59 consist of aggregation polymers of aluminium and phosphate ions in randomized three-dimensional networks; their stability and degree of polymerization are strongly dependent on the pH of the rnediums42 The impurity complexes of PCh3-and A13+ in a KBr lattice have been studied by i.r. spectroscopy.60 The mass spectra of Alp04 vapour in thermodynamic equilibrium with the condensed phase at various ionization energies show that thermal dissociation occurs into a complex mixture of oxides of phosphorus.61 The Chemistry of Aluminium Dihydrogen Phosphate.-The acid phosphate of aluminium is frequently represented in terms of the 1: 3 stoicheiometry of its oxides as Ala03,3P205,6H20.The chemistry of Al(H2P04)3 is complex, and many differing and often contradictory results are reported in the literature. Its chemical properties appear to vary according to the method of its preparation. The first preparations of AI(H2P04)3 were reported in 187862 and 1888.63 A solution of alumina was heated in concentrated phosphoric acid at 373 K, and partial evaporation of the solution produced AI(H2P04)3 as a crystalline solid. Little work was done on the chemistry of Al(H2P04)3 until 1957, when D’y~0h-e~~~~~studied the thermal and X-ray diffractive properties of powders of Al(H~P04)3 and showed that at least two species of AI(H2P04)3 (labelled A and B) and probably another (labelled D) existed, all of which produce X-ray line diffraction patterns different from that of the so-called classical (or C) form prepared by the original meth0d.~~~G3 Clays may also be used as alumina sources for preparing aluminium phos- phates,18s66*67 although the composition of the final product depends on the A1: P ratio in the reaction mixture and aluminium extraction is rarely complete, even with excess phosphoric acid (I : > 6).Because the basis of phosphate binding is the formation of polymers as a result of dehydration, the thermal properties of Al(H2P04)3 have been studied by a number of inve~tigators.l8~~5~6*-72Correlations of thermogravimetric analysis, !I C.F. Callis, J. R. Van Wazer, and P. G. Arvan, Chem. Rev., 1954,54,777. Ya. P. Tsyackenko and V. M. Zaporozhets, Fiz. Tverd. Tela. (Leningrad), 1974, 16, 2444 (Chem. Abs., 1974, 81, 129282). 61 G. A. Semenov, K. A. Frantseva, E. Nikolaev, L. L. Schetnikova, M. G. Tretnikova, V. M. Ust’yansev, L. B. Khoroshavin, and D. S. Rutman, Ogneupory, 1974, 52 (Chem. Abs., 1974,81, 128512). 62 E. Erlenmeyer, Annalen, 1878,194,196. 63 P. Hautefeuille and J. Margottet, Compt. rend., 1888, 106, 136. 64 A. BoulIe and F. D’Yvoire, Compt. rend., 1957,245,531. O5 F. D’Yvoire, Bull. SOC. chim. France, 1961,2277. c6 A. S. Yuting, Z. D. Zhukova, and S. V. Lysak, Sb. Naucli. Tr. Ukr. Nauch-Isslell, Insr. Ogreuporov, 1970, No. 13, 173 (Chem. Abs., 1973,78,165057).6i F. D’Yvoire, Bull. Soc. chim. France, 1962, 1243. 68 A. A. Chistyakova, V. A. Sivkina, V. I. Sadkov, A. P. Kashkovskaya, and L. G. Povysheva, Izvest. Akad. Nauk S.S.S.R., neorg. Materialy, 1969,5,536. G9 G. D. Salmanov and G. N. Aleksandrova, Zzvest. Akad. Nauk S.S.S.R.,neorg. Materialy, 1969,5, 148. io R. N. Rickles, J. Appl. Chem., 1965,15,74. ” E. Eti and W. D. Hall, Amer. Ceram. SOC.Bull., 1971,50,604. M. Shiota, T. Kato, and H. Numato, Reports Res. Lab., Asahi Glass Co. Ltd., 1970, 20, 93. Morris, Perkins, Rose, and Smith differential thermal analysis, and X-ray diffraction studies reveal distinct stages of dehydration, represented by Scheme 2. The first product is essentially amor- phous, and no evidence for the crystalline second product (acid tripolyphosphate) was obtained by some workers, who found direct crystallization of the aluminium metaphosphate Al(P03)3 (B) from the amorphous phase.498-563 K 2Al(HzPO& -A1203,3Pz05,3HzO +3H20 f 563-778 K A1203,3Pz05,3HzO-2AIHzP30io +H2O f 778-898 K 2AlHzP3010-2Al(P03)3 +2H20 f Scheme 2 More recently, it has been reported73 that the dehydration of a solution of aluminium phosphate of ratio 1:3.7 proceeds via the formation of crystalline Al(H2P04)3. Subsequently, a mixture of the diphosphate AhH12(P207)9 and the triphosphate H2AlP3010,2H20 is formed in the 293-573 K range. At 598 K, crystalline Al(P03)s is formed, which has a phase transition at 723-773 K. Other workers have reported the formation of AlP04,A12(HzPz07)3, H2AlP3010, and Al(P03)s during the dehydration of H3Al(P04)2,2HzO, Al(H2P04)3, and Alz(HP04)3.74*75 Al(H2P04)3 (B) is also formed from the decomposition of the disubstituted acid phosphate of aluminium, A1H3(P04)2,3H20.76 Thus: -6H20 2AIH3(P04)2,3H20-AI(H2P04)3 (B) +Alp04 (cristobalite) 423473 K Traces of AI(H2P04)3 (A) and of another phosphate, which may be Al(H2P04)3 (D), are also present.AI(H2P04)3 has also been observed to be formed from an aluminium phosphate solution in which the molar ratio is A1203:P205 = 1:2.2.77178 A disubstituted phosphate of aluminium was formed which thermally decomposed to equimolar quantities of AI(H2P04)3 and AlP04. It appears that thermal transformations of AI(HzP04)3 (B) exhibit a pattern similar to those of the classical form, Al(H2P04)~ (C), although opinions differ as to the mode of decomposition.Initial dehydration produces an amorphous pha~e~6-7~ aat 73 M. I. Kuz’rnenkov, V. V. Pechkovski, and I. T. Buraya, Zhur. neorg. Khim., 1973, 18, 958 (Chern. Abs., 1973, 78, 15440). 74 E. G. Levitas and L. B. Romanovskji, Ukrain. Khim. Zhur. (Run. edn.), 1972, 38, 866 (Chem. Ah., 1973, 78, 10952). 75 V. F. Tikavyi, K. N. Lapko, A. N. Lobanok, A. N. Chivenkov, and A. A. Sokol’chik, Vesti Akad. Navuk Belarus. S.S.R., Ser. khim. Navuk, 1972, 5, 74 (Chem. Abs., 1973, 78, :‘F. 11 061).D’Yvoire, Bull. SOC. chim. France, 1961,2283. i7 I. L. Rashkovan, L. N. Kuz’minskaya, and V. A. Kopeikin, Izvest. Akad.Nauk S.S.S.R., neorg. Materialy, 1966, 2, 541. V. A. Kopeikin, A. I. Kudryashova, L. N. Kuz’rninskaya, 1. L., Rashkovan, and 1. A. Tananaev, Zzvest. Akad. Nauk S.S.S.R., neorg. Materialy, 1967, 3, 737. ’~4 F. D’Yvoire, Compt. rend., 1958, 247,297. The Chemistry and Binding Properties of Aluminium Phosphates temperature about 30 K below that of the corresponding phase transition of Al(H2P04)3 (C), probably owing to differences in crystal size between the two forms, and some doubt exists as to whether A12P3010 is formed as an intermediate in the transformation of Al(H2P04)3 (B).76-80 The formation of amorphous phases is considered to be importantlp5 for the cementing properties of a binder, because of their ability to adhere and cohere, and their ability to absorb strain.Al(H2P04)3 forms two amorphous phases: the first, partially dehydrated, is formed between 503 and 573 K and the second, an anhydrous phosphate, at temperatures greater than 1273 K. Between 523 and 573 K, binders containing AI(H2P04)3 [or disubstituted aluminium phos- phates which produce Al(H2P04)3], acquire ~trength~~~~l~~~ and moisture resistance. The latter property arises because the amorphous phase which is formed is a vitreous substance insoluble in water,77 whereas its precursor phase, AI(H2P04)3, is extremely hygroscopic. The formation of the amorphous phase is responsible for the cold-setting properties of aluminium phosphate binder^.^ However, not much is known about the properties of the low-temperature amorphous phase, although it has been described as ‘possessing a randomized Its binding action is due to a ‘polymer shell of chains and rings of aluminium phosphate and alumina tetrahedra’.83 Although its presence is readily identified by X-ray powder diffraction, it is somewhat metastable and slowly crystallize~,77~~0 yielding AlHzP3010,(2-3)H20.~~ Its stoicheiometry is probably A1203,3P205,3Hz05~71~7g and, although its structure is unknown, a pyrophosphate formulation, A12(H2P207)3, has been proposed.71 9 739 The high-temperature amorphous phase formed at temperatures greater than 1273 K is a metastable glass of unknown composition and stru~ture.~~~~~ The thermal transitions in AI(H2P04)3 (B) are summarized in Figure 2.The optical indices of refraction of both amorphous phases are 513-573 K, 7 = 1.512; 1373-1573 K, 7 = 1.510.68 Aluminium tripolyphosphate, AlH2P3010, has been reported77 to have the chain structure (2).The low-temperature form of aluminium trimetaphosphate, Al(P03)3 (B), has a chain structure (3) in which the Po4 tetrahedra are linked together to form infinite (P03-)n chains. The high-temperature form, aluminium tetrametaphosphate, AI(P03)3 (A), has the ring structure (4).84 Evidence for cyclic character is the presence of bands at 1311 and 1028 cm-1 in the i.r. spectrum.85 Both these bands are present in all metaphosphates containing the A. A. Chistyakova, V. A. Sivkina, V. I. Sadkov, A. P. Khashkovskaya, and L. G. Povysheva,Izvest. Akad.Nauk S.S.S.R.,neorg. Materialy, 1969,5,1573. H. Bechtel and G. Ploss, Ber. deut. keram. Ges., 1963,40, 399. ’% G. I. Duderov and Yu. P. Gonchavrova, ‘Progress in Electroceramics’, Izd. Vniiem, Moscow, Vol. 7. 8a B. N. Bogomolov and V. M. Sergeeva, Ogneupory, 1964,29,520. H. Remy, ‘Treatise on Inorganic Chemistry’, Izd. Inostr., Moscow, 1963. 85 V. M. Medvedeva, A. A. Medvedev, and I. V. Tananaev, Izvesf. Akad. Nauk S.S.S.R., neorg. Materialy, 1965,1,211. Morris, Perkins, Rose, and Smith m Gn I 9 n The Chemistryand BindingPropertiesof AluminiumPhosphates phosphorus-oxygen ring, but are absent in the spectra of the polyphosphates.86-89 0 0 0 0 II II It II IIAIH, -0 -P -0 -P-o---P -o-P -0-I I I I I 0-0-0-0 0 0 0 0- 1ro II I -o-P-o-P=oI.!III 0 0 I 1 Io=P-o-P-o-1-bl II At higher temperatures (1073-1273 K), Al(P03)3 (B) is non-reversibly converted into Al(P03)3 (A),6538972977although the latter has been observed at lower temperatures76~s0~90apparently where traces of phosphoric acid have catalysed its formation. Relatively few studies have been made on the physical and chemical properties of aluminium metapho~phate.~~~~~ Potentiometrictitration of base against AI(HzP04)3between pH 2 to 12reveals three end-pointsl8993 similar to H3P04, and the similarity of the behaviour of AI(H2P04)3 with that of H3P04 is suggested to be due to the liberation of H3P04 before the first end-point. The first end-point at pH 4.5 is strong, the next at pH 8.5 is less strong, and that at pH 11 is weak but observation can be facilitated by adding a precipitating agent.When AgN03 is added at pH 2.8, a precipitate of amorphous Alp04 is formed,ls producing free H3P04, which Rfi E. Steger, Z. anorg. Chem., 1958,294,146.*' D. E. C. Corbridge and E. Y. Lowe, J. Chem. SOC.,1954,493. 88 D. E. C. Corbridge, J. Appl. Chem., 1956,6,456. 89 U. V. Klyucharov and L. I. Skoblo, Doklady Akad. Nauk S.S.S.R.,1964, 154,634 (Chem.Ah., 1964, 60, 11 584).A. A. Chistyakova, V. A. Sivkina, A. P. Khashovskaya, and V. I. Sadkov, Zzvest. Akad. Nauk S.S.S.R.,neorg. Materialy, 1969,5,1738. 91 R. Arstanova, A. B. Bekturov, V. V. Tikhonov, and V. K. Esik, Izvest. Akad. Nauk S.S.S.R., Ser. khim., 1972, 22, 1 (Chewi.Abs., 1972, 76, 148317). 92 M. Tsuhako, I. Motooka, and M. Kobayashi, Bull. Chew. SOC.Japan, 1971, 46, 2343 (Chem. Ah., 1973, 79, 11 1246). g3 V. F. Tikavyi, K. N. Lapko, I. A. Zakharov, E. P. Antonyuk, and L. I. Dorozkhina, Izvest. Akad. Nauk S.S.S.R., neorg. Materialy, 1971, 7, 1629. 186 Morris, Perkins, Rose, and Smith is then partially neutralized in two steps by base. Finally, Alp04 redissolves in NaOH, giving NaAlO2 and Na3P04 at the third end-point.l8 The effectiveness of Al(H3P04)2 as a binding agent may be due to chemical interaction with fillers such as silica. A study of the cross-breaking strength of a series of samples containing the phosphate with a varying proportion of silica as filler (Figure 3) is consistent with the existence of a silica-phosphate inter- I I 1 I I200 -1 2 3 4 5 6 hilica aluminium phosphate Figure 3 Plot of the relatim between silica content and cross-breaking strength of test boards using an aluminium phosphate binder action.Studies of the interaction by analysis, thermal, and X-ray methods, rule out a bulk chemical reaction between the two components. Further studiesls show that both aluminium and phosphate ions are absorbed on a quartz silica surface, but that aluminium is successively replaced by phosphate as more aluminium phosphate is added. When precipitated amorphous silica is employed, it is found that only aluminium ions are absorbed on its surface. The mechanism of this absorption process appears to be complex, and at least three stages can be postulated: (i) primary absorption of AP+, H+, and Pod3- ions on the surface; (ii) Hf ions are then displaced by PCh3-and A13+ until all active sites are occupied; (iii) replacement of A13+ by POa3- up to an ill-defined limit.The aluminium displaced possibly ends up as hydroxylated The Chemistry and Binding Properties of Aluminium Phosphates polymers, such as those described by HSU.~~ The interaction on the interfaces between silica and aluminium phosphate is probably due to hydrogen bonding, as has also been suggested95 for the interaction of phosphoric acid and talc. The bulk binding properties of a crystalline compound may be influenced by the crystal habit.96 In the Al(H2P04)3 series both needle and plate forms of crystal, which are crystallographically identical, can be formed. The needle and plate crystals, grown under a variety of conditions, have the same chemical compo- sition, are both of C-form, and appear to be merely different crystal habits.Crystal growth was observed under polarized light on a Reichert hot-stage microscope. Viscous solutions favour the formation of needle crystals and the rate of growth is important. The plates are converted into a shapeless mass on heating to 573 K, whereas the needle crystals retain their shape. Since the material is amorphous after this treatment, the retention of form by the needles is clearly important. The effect on the strength characteristics of test boards containing the different crystal habits showed that those containing needIe crystals had an approximate five-fold increase in cross-breaking strength over those containing platelets.Aluminium Phosphates of Ratio A1203:P205 = 1:2.-Aluminium phosphate solutions with a molar ratio range of Ah03:P205 = 1 :2-4 form the basis of commercial thermostable binders and coatings, the exact molar ratio depending on the iise.*O The literature data concerning these solutions are extensive, although contradictory. It has been proposed that aluminium phosphate solutions contain chiefly the anionic triphosphatoaluminate complex [AI(HP04)3I3-, but other auth0rs~~9~~ have suggested that the viscous aluminium phosphate system contains polymeric molecules with a branched three-dimen- sional structure. The curve of the dependence of the reduced viscosity on con- centration for aluminium phosphate solutionsg7 has a shape characteristic of polyelectrolytes;the change in the conductivity also corresponds to that for polyelectrolytes. No evidence for condensed polyphosphate ions with P-0-P bonds was found using chromatographic technique~,~4 and the high viscosities of such solutions are suggested to be due to the formation of a polymer structure with a complex system of hydrogen bonds.The composition and thermal transformations of the precipitates which form from these solutions have been widely st~died.~~,~~ From solutions of molar ratios Of A1203 :P205 = 1:2.74, D’Yvoire isolated the disubstituted phosphate, AIH3(P04)2,3H20 at room temperature. This compound dehydrated between 373 and 423 K to form equimolar quantities of Alp04 (cristobalite) and AI(H2P04)3 (B), with trace amounts of both Al(H2P04)3 (A) and another phos- phate, probably AI(H2P04)3 (D).Dehydration under varying conditions of 94 P. H. Hsu, in ‘Trace Inorganics in Water’, ed. R. F. Gould, American Chemical Society Washington, 1968, Advances in Chemistry Series No. 73, p. 115. 95 R. Robinson and E. R. Segnit, Austral. Ceram. SOC.,1967,3,9. y6 J. H. Morris, P. G. Perkins, A. E. A. Rose, and W. E. Smith, to be published. 97 V. N. Sveshnikova and S. N. Zaitseva, Zhur. neorg. Khint., 1964,9,1232. 188 Morris, Perkins, Rose, and Smith humidity produced differing crystal forms : heating AlH3(P04)2,3HzO under vacuum (low humidity) produced appreciable amounts of Al(H2P04)3 (D) as well as Al(H2P04)~ (B) and Alp04 (cristobalite); under extremely moist con- ditions, AlH3(PO&,H20 was first formed before rapidly decomposing to a mixture of the A-, B-, and C-forms of Al(H2P04)3 and Alp04 (quartz).Evaporation of the binder solution, A1203 :P205 (molar ratios = 1:2.7-4) at 388 K produces A1H3(P04)2,H20.76 This disubstituted phosphate has been reported to lose two moles of water between 408 and 473 K to form a mixture of the B-and C-forms of AI(H2P04)3 and the quartz- and cristobalite-forms of AlP04. It is interesting to note that, if solutions having the molar composition of A1203 = 1:2.7-3 are refluxed for some days, hydrates of the trisubstituted phosphate, Alp&, are formed.16 If, however, the reflux is stopped after some minutes, an amorphous compound of variable stoicheiometry forms.76 This undergoes initial dehydration at temperatures between 293-323 K and 388-423 K, the range depending on the value of x,as follows: 2YH20 2AIH3(x- 1)(P04)r,yHz0 -(3 -x)AlP04 (cristobalite) + (X -1) Al(H2P04)3 The form c.f Al(H2P04)~ which is obtained depends on the value of x: for x = 2.07, form C is obtained; for the majority of other values of x (i.e.,x < 1.5), the B form of Al(H2P04)3 is produced, with traces of another phosphate which may be the D form of Al(HsP04)3. The results of D’Yv0i1-e~~ also appear to suggest that, if a freshly prepared binder solution is cured, it will give rise to a crystalline phase composition at 473 K different from one which has been allowed to stand for some time, as shown in Scheme 3.freshly prepared binder solution -+ AlH3(P04)2,H20 -+ Al(H2P04)3 (B)+ Al(H2P04)3 (C) + Alp04 (quartz) + Alp04 (cristobalite) aged binder solution -+ AlH3(P04)2,3HzO .1 Al(H2P04)3 (B) + Alp04 (cristobalite) Scheme 3 D’Yvoire’s76~79 results have been quoted in some detail here for they serve, not only to show the complexity of the chemistry of these solutions and of alum- inium phosphate chemistry in general, but also to show how apparent con-tradictions between the work of various researchers can arise. Thermal transformations of individual phases are not always identical with changes undergone when the aluminium phosphate solutions used in technology are heated.However, despite the wide variation in the results of the thermal transformations of the aluminium phosphates obtained from binder solutions having A1203:P205 molar ratios of between 1:2 and 1:4, it is apparent that formation of the disubstituted phosphate of aluminium, A1H3(PO4)2,3HzO7 occurs. Moreover, subsequent dehydration produces equimolar amounts of trisubstituted Alp04 and monosubstituted A(HzP04)3.75 189 The Chemistry and Binding Properties of Aluminium Phosphates The effect of phosphoric acid on the nature of the thermal transformations on a 1:2 aluminium phosphate have been studied98 and found to lead to for- mation of AI(H2P04)3, which subsequently promotes the formation of pyro- phosphates at lower temperatures.The presence of H3P04 catalyses the de- composition of H3[AI(P04)2] into primary and secondary phosphates. A summary of the low-temperature thermal transformations of aluminium phosphate binder solutions is given in Figure 4. Commercial Aluminium Phosphate Binders.-The physical and mechanical properties of compositions containing aluminium phosphate binders result from a combination of factors, such as the method of preparation, the amount and concentration of the phosphoric acid, the amount of binder present, the degree of compression, the temperature of curing, the type of aggregate or filler, and the presence of a catalyst. Essentially, there are two methods of achieving the aluminium phosphate binding action.One method is to add phosphoric acid to an aluminous material, usually either alumina or aluminium hydroxide, which is contained in the com- position to be cemented; binding is then obtained by heating directly. The second methcd is to preform the aluminium phosphate, which is then added as a solution to the composition to be cemented and binding obtained by heating. If the first method is used, there is generally a continuous increase in strength up to high temperature~.~~-l01 Concentrated phosphoric acid (88 %), however, was foundlo2 to impart a maximum crushing strength to alumina compacts at 873 K. Above 873 K, strength deteriorated rapidly with temperature. With aluminium phosphate as the binding agent, however, higher strengths are achieved at lower temperat~res.~~~~J03 A maximum in shear,lo4 transverse-breaking,5 and crushing strengths105 is reached at about 873 K.The maximum in trans- verse-breaking strengths occurred at 1473-1 673 K in an aluminium phosphate- alumina composition.5 The weaker compacts produced at lower temperatures by the phosphoric acid method may be due to polymerization reactions involving free phosphoric acid not being completed until a temperature greater than 673 K is reached.lo3 Phosphoric acid added to corundum produced a bonded system that was not stable until the material had been treated to over 723 K.6g Un-polymerized, free phosphoric acid is extremely hygroscopic and causes deterior- ation of the binding action through rehydration, and insufficient reaction occurs ’* A.N. Chivenkov, K. N. Lapko, A. K. Lobanok, A. N. Chivenkov, and A. A.. Sokl’chik Vesti Akad. Navuk Bclarrrs. S.S.R. Ser. Khint. Navuk, 1973, 124 (Chent. Abs., 1973, 79, 5 1 084). gB P. A. Gilham-Dayton, Trans. Brit. Ceram. Soc., 1963,62, 895. looV. A. Bron, A. I. Uzberg, T. P. Karzhavina, and G. S. Krotova, Ogneupory, 1972, 1, 37 (Chem. Ah., 1972,76,103245). lol A. V. Amenkov and A. A. Pirogov, Ogneupory, 1960, 25, 527. lo2 K. Fisher, Proc. Brit. Ceram. SOC., 1969, 12, 51. O3 G. N. Duderov, Ogneupory, 1964,29,460. F. Ya. Boradai and T. M. Evdokimova, Zzvest. Akad. Nauk S.S.S.R., neorg. Materialy, 1969,5, 1406. lo5 H. Bechtel and G. Ploss, Ber. deut. keram.Ges., 1960,37,362. A B c ridymiti ristobalite J J J J J J 413-443 K" J Jbinder AIH,(PO4),,31 12O K'hcsolution 373 4'3 i3CCb J J J J II \ J J J J 373 KY6 4AIH,(PO,),,l~,O > J J J JAIL03:PLOI= I :2.7-4 J J Figure 4 Low-temperature thermal transformations of aluminium phosphate binder solutions The Chemistry and Binding Properties of Alumitiium Phosphates if the phosphoric acid solution is too dil~te.l~~**~~ The optimum concentration generally appears to be 60-70 %.69Jo3910*-112 The filler and its proportion in the compact affects both strength and thermal expansion. Using alumina, with phosphoric acid as the binder, only 0.2 % expansion at 573 K and 0.6% at 1073 K was 0b~erved.l~~ An aluminium phos- phate binder showed 0.8 % at 1273 K,102 although with vitreous silica the magni- tude depended on the amount of silica present and on the temperature.104 A considerable change in the coefficient of thermal expansion could be obtained by addition of alumina or titanium dioxide to the silica.This modification is useful because compositions can be tailored to match the thermal expansions of the material to be cemented and thus reduce thermal stresses. For the preparation of suitable aluminium phosphate binder solutions, the optimum molar ratio of A1203:P~05is in the range 1 :3 to 1:3.5.103J09J14J15 If the ratio is greater than 1:4,strength decreases.lo3 This is probably due to the presence of free phosphoric acid not involved in the reaction.A commercially available aluminium phosphate binder,l16 a process for the manufacture of condensed aluminium phosphates,ll7 and the use of mortars with aluminium phosphate binders have been descri bed.1l8 Aluminium phosphate gives strong binding with a variety of fillers, e.g. a~~mina5,102,103,10~,107,~~~,~~~,~20and silica.99,104.121-126 In addition, other refractory materials such as ~irconia,103~109~~10~127silicon carbide,flOJ2* and mulliteg9 can be employed. Io6 Unpublished data. lo' A. Petzold and M. Rohrs, 'Concrete for High Temperatures', Maclaren, London, 1970, p. 185. lea L. B. Khoroshavin, P. N. D'yachkov, B. V. Ponomarev, L. Ya. Pivnik, and V. K. Bogatikova, Ogneupory, 1968, 33, 40. log T. I. Rodina, Steklo i Keram., 1969,26,24.'Iu H. D. Sheets, J. J. Bulloff, and W. H. Duckworth, Brick and Clay Record, 1958, 133, 55. G. D. Salmonov, V. F. Gulyaeva, and G. N. Aleksandrova, Zharostoikie Betony, Nauchn.- Issled. Inst. Betony i Zhelezobetona, 1964, 72 (Chem. Abs., 1965, 62,2609). Yu. G. Duderov, Ogneuporv, 1972, 5,46 (Chem. Abs., 1972, 77, 51 897). ]I3 W. H. Gitzen, L. D. Hart, and G. MacZura, Bull. Amer. Ceram. SOC.,1956,35217. llJ0.M. Margulis and A. B. Kamenetskii, Ogneupory, 1964,29,329. 115 N. B. Semirkhanova, D. A. Kuznetsov, A. 1. Malakhov, and E. P. Antonyuk, Trudy Mosk. Khim.-Tekhnol. Inst., 1967,54,251 (Chem. AbA., 1968,68,52805). *I6 M. Maretti, Refrattari, 1972,4,251. A. Hloch, N. Medic, and R. Kohlhaus, U.S.P. 3650683 (Ceram. Abs., 1972,55, 175h).118 A. B. Kamenetskii, Ogneupory, 1973, 34 (Chem. Abs., 1973,78,15084).J. C. Bidard, U.S.P. 3538202(Ceram. Ah., 1971,54,79f). lf0 E. Kupzog, M. Koltermann, and P. Bartha, Ber. deut. keram. Ces., 1967, 44, 445 (Chem. Abs., 1967,67, 119864). Yu. G. Duderov and P. P. Budnikov, Iivest. Akad. Nauk S.S.S.R.,neorg. Materialy, 1970, 6,928. Iz2 P. P. Budnikov and Yu. G. Duderov, Izvest. Akad. Nauk S.S.S.R.,neorg. Materialy, 1966, 2,187. A. Cser, Epitoanyag, 1957,9,293. Iz' K. H. Bauer and M. Weiler, Swiss P. 521 180 (Chem. Abs., 1972,77,78966). 12j J. H. S. Skoning, U.S.P. 3057740 (Chem. Abs., 1962,57, 16168). lZ6A. Yamashita, Zmono, 1969,41,415 (Chern. Abs., 1969,71,63495). Iz7 E. Tauber, H. J. Pepplinkhouse, D. N. Crook, and R. J. O'Brian, Australas.Inst. Mining Met., Proc., 1972,244,47 (Chem. Abs., 1973,78,51541). N. I. Krasotkina, V. V Levchuk, and N. I. Voronin, Ogneupory, 1969,34,59. Morris, Perkins, Rose, and Smith The characteristics of aluminium phosphate binding may be altered by the addition of various catalysts. Ammonium fluoride accelerates the setting process,110 whilst the following substances have been used as inhibitors : Rodine 78 (a complex amine),llO 5-sulphosalicylic acid, acetylacetone, and dextrin.'l Magnesium oxide can also be used to control the hardening process.lZ6 It is possible to reduce the curing temperature of the aluminium phosphate binder to room temperature by the addition of small amounts of curing catalysts such as aluminium powder129 or magnesium oxide masked with either sodium hexa- metaphosphate or sodium tetrametaphosphate.130 Addition of hydrated zirconia also has the effect of lowering the temperature of hardening and increasing water resistance.131 The consistency of the viscous aqueous aluminium phosphate solution may be improved by addition of powdered iron or an iron 0xidel3~ or c]ay.107,120 Aluminium phosphate binding, produced directly or indirectly, has been used in a variety of applications in the field of refractorie~.100~111,118.133-138It has also found application as a binder for ~oatings,~~~-144 thin films,l45 and furnace linings.146 One of the most important advantages of aluminium phosphate binding in high-temperature furnace linings is its high resistance to attack by sIag.147-149 With suitable fillers or foaming agents, aluminium phosphate can be used to produce light-weight articles where a high strength/density ratio is lZ9 T.Nishigori and K. Matsuno, Imono, 1967,39, 192 (Chem. Abs., 1968,68,89392). lB0 T. Akaiwa, Jap. P. 14257 (Chem.Abs., 1972,76,103 343). lT1 I. V. Tananaev, V. A. Kopeikin, E. V. Maksimchuk, I. A. Rozenberg, 0. N. Dementeva, and V. I. Platun, U.S.S.R.P. 2201 17 (Chem.Abs., 1968,69,109515). 13p W. H. Kreidle, B. P. 996297, Austrian Appl., 1960 (Cnem. Abs., 1965, 63, 6700); B. P. 1003745; Ger. Appl., 1961 ;Addn. to B. P. 996297 (Chem.Abs., 1965,63,14520). l?B L. A. Tseitlin and A. P. Gubatenko, Ogneupory, 1968, 33, 25 (Chem. Abs., 1968, 69, 12599). T. Chvatal, Ger.Offen. 2126521; Austrian Appl., 1970 (Chem. Abs., 1972, 76, 131056). 135 R.E. Fisher, U.S. P. 3 6493 13 (Chem.Abs., 1972,77,24 159). IR6 Ukranian Scientific Research Institute of Refractory Materials (by L. A. Tseitlin and A. P. Gubatenko), U.S.S.R. P. 196594 (Chem.Abs., 1968,68,89624). IRi Farbwerke Hoechst AG, B. P. 1262 162 (Brit. Ceram. Abs., 1972,71,65A, 1067). 13* A. K. Radaman and V. E. Mukhin, U.S.S.R. P. 240525 (Chem.Abs., 1969,71,41928). IB9K. Momose, K. Jibiki, I. Yamamoto, and K. Okabayashi, Imono, 1969, 41, 479 (Chem. Abs., 1969,71,82705). I4O N. N. Silina, R. 1. Sagitulina, and N. I. Moskvitin, Polim. Mashinostr., 1972, 6, 157 (Chem.Ah., 1973,78,88012). 1(1 E. W. Koeing and G. R. Cope, U.S. P. 3 352 (Chem.Abs., 1968,68,52889). lg2 R.J. Fuchs, C. W. Lutz, and L. E. Cohen, U.S. P. 3547670 (Ceram.Abs., 1971,54,99c). lg8 N. Silinaand R. I. Sagitullina,U.S.S.R.P. 245633 (Chem.Abs., 1969,71,128 162). 144 A. S. Frolov, M. G. Trofimov, and E. M. Verenkova, Vysokotemp. Pokrytiya, Trudy Seminara, Leningrad, 1965,153 (Chem.Abs., 1967,67,119 880). A. 1. Avgustinik, V. P. Alekseev, and G. 1. Zhuravlev, Zhur. priklad. Khim. (Leningrad), 1969,42,434(Chem.Abs., 1969,71,7652). ld6A. J. Owen, R. Visser, and J. Van Laar, Ger. Offen. 1801 875, Brit. Appl., 1967 (Chem. Abs., 1969,71,24470). 14' W. Doornenbal, J. T. Van Konijinenburg, J. Van Laar, R. Visser, and A. Waasdorp, Cent. Nat. Rech. Met., Brussels, Met. Rep., 1970,25,11 (Chem.Abs., 1971,74,145767). 148 R.Visser, Tonind.-Ztg. Keram.Rundsch., 1972,96, 182 (Chem.Abs., 1972,77,129994). J. A. Halpin, Ber. deut. keram. Ges., 1972,49,219 (Brit. Ceram. Abs., 1973,72,6A, 119), The Chemistry and Binding Properties of Aluminium Phosphates required.112,122,150-153Dense materials from boron and aluminium nitride with a continuous skeleton derived from aluminium dihydrogen phosphate have been successfully used as slide beari11gs.1~~ Aluminium phosphate binding is also used in catalytic bodies,155 diamond abrasives,156 electrically conducting porcelain corn position^,^^^ castable refractorie~,15~ foamed ceramics for building and insulating materials,160 high-temperature nozzles,161 lacquers for electrical sheets,162 and vacuum163 and dental ~ements.~~J~~ Halogenated aluminium phosphates in solution can be used to coat glass fibres in order to increase their ~trengthl~~J~~ or they can be crystallized to form resin-reinforcing fibre~.l~~ The long-term action of high temperature on the mechanical properties and microstructure of glass fibre laminate in an aluminium phosphate binder has been found to result in some degradation of the protective film coating on the fibres, and then corrosion of the surface of the fibres.168 Other patents have been summarized,l* and work on mixed ahminium- chromium phosphates has been rep~rted.l~~J~~-l~~ M.Marinov, D. Petkov, G. Grigorov, and T. Petkova, Stroit. Materialy Silikat. Prom. 1968,10,12 (Chem. Abs., 1969,71,128249). l‘,l M. Marinov, D. Petkov, G. Grigorov, and T.Petkova, Stroit. Materialy Silikat. Prom., 1969,10,11 (Chem. Abs., 1969,71,63 791). P. Artelt and W. Holloh, Ger. Offen. 2045 497 (Clzem.Abs., 1972,76, 157 707). 153 Steinwerke ‘Fuerfest’ Karl Albert G.m.b.h., Ger. P. 1191 734 (Chem. Abs., 1965,63, 1572). 15’ L. I. Prikhod’ko and L. I. Sobolevskii, Porosh. Met., 1968,8,22. Isi V. P. Saltonova and A. I. Malakhov, U.S.S.R. P. 170 91 5 (Chem. Abs., 1965,63, 14 1 lo). D. I. Mendeleev Chemical-Technological Institute, Moscow (by V. L. Balkevich, I. A. Surkova, L. B. Borokova, B. Yu. Kopp, and 0.D. Sadkovskaya), U.S.S.R. P. 204181 (Chem. Abs., 1968,68,116955). lii Morgan Crucible Co. Ltd., B. P. 959398 (Chem. Ah., 1964,61,2805). 15* M. G. Sivchikov and G. V. Yurchenko, Leglc. Prom., Nauk.-Virobn.Zb., 1968,2, 18 (Chem.Abs., 1969,71,53060). T.Taniguchi, Seramikkusu, 1966,1,66 (Chem.Abs., 1968,68,71800). 160 A. G. Eubanks and R. E. Hunkeler, U.S. P. 3 382082 (Chem. Abs., 1968,69,12658). 161 D. Iordanov, D. Georgiev, D. Ivanov, G. Dirnitrov, G. Ganev, and P. Petrov, Rudodobiv Met. (Sofia) Met., 1966,21,28 (Chem. Abs., 1967,67,25007). 162 C. A. Aakerblom, Ger. Offen. 2 161 162 (Chem.Ah., 1972,77,154024).N. A. Meshcheryakov, L. A. Kabanova, and B. I. Boiks, Prib. i Tekh. Eksp., 1967,4,219. 16‘ A. D. Wilson, Nat. Bur. Stands. (US.)Spec. Publ., 1972, NO.35485. Ie5 J. E. Cassidy, Ger. Offen. 2046934; Brit. Appl., 1969 (Chem. Abs., 1971,75,78267). 166 J. D. Birchall and J. E. Cassidy, Ger. Offen. 2161436; Brit. Appl. 59040 (Chent. AbJ.1972,77,116 002). 16i J. D. Birchall and J. E. Cassidy, Ger. Offen. 2046932; Brit. Appl., 1969 (Chem. Abs., 1971, 75, 141 524). lea B. A. Kiselev, V. N. Bruevich, V. A. Kudishina, N. A. Rozdina, I. A. Deev, Yu. V. Zherdev, A. I. Mikhalskii, and A. Ya Korolev, Izvest. Akad. Nauk S.S.S.R.,neorg. Matcrialy, 1973,9,692 (Chem. Abs., 1973,79,43 523). 169 T. Chvatal, Trans. Internat. Ceramic Congr., loth, 1966, 377 (Chem. Abs., 1969, 70, 80527). T. Chvatal, Bol. Soc. Espan. Ceram., 1968,7, 165 (Chem. Abs., 1968,69,89432).T. Chvatal, Austrian P. 260770 (Chem. Abs., 1968,68,98335).T. Chvatal, Ger. P. 1281 916 (Chem.Abs., 1969,70,50 150). Ii3 T. Chvatal, Austrian P. 231 337 (Chem. Abs., 1964, 60, 7776; Austrian P. 259441; Addn. to Austrian P. 231 337 (Chem. Abs., 1968,68,62367); Austrian P.2 59442; Addn. to Austrian P. 231 337 (Chem.Abs., 1968,68,62371). 174 V. P. Volkova, V. A. Kopeikin, V. I. L’Vov, 1. L. Rashkovan, and V. V. Yarkina, U.S.S.R. P. 356257 (Chem. Ah., 1973,78,101540). lii A. V. Bromberg, A. G. Kasatkina, V. A. Kopeikin, A. 1. Kuzminskaya, I. L. Rashkovan, and I. V. Tananaev, Izvest. Akad. Nauk S.S.S.R.,neorg. Materialy, 1969, 5, 805 (Cheni. Abs., 1969,71,6312). lie G. N. Aleksandrova, Trudy Vost. Inst. Ogneuporov, 1968, 10, 167 (Chem. Abs., 1971, 7§, 67116). 194

ISSN:0306-0012    

DOI:10.1039/CS9770600173

出版商:RSC    

年代:1977

数据来源: RSC

摘要:

The Organic Chemistry of Superoxide By E. Lee-Ruff DEPARTMENT OF CHEMISTRY, YORK UNIVERSITY, TORONTO, ONTARIO 1 Introduction The importance of oxygen in sustaining life fornis directly or indirectly is unquestionable and has been the subject of intensive studies by biologists and chemists over the past two centuries. Although reactions of oxygen with organic compounds are in general highly exothermic, such reactions are usually slow at physiological temperatures. The fact that the electronic spins of oxygen (triplet) and most stable organic substrates (singlet) are incongruent is probably the most important reason why oxygen does not behave as a random oxidizing agent capable of destroying complex organic material. A number of mechanisms and intermediates for the interaction of molecular oxygen with specific biological substrates have been suggested.l These include singlet oxygen, metal-oxygen complexes, and various reduced states of molecular oxygen either free or bound to an enzyme.Recent evidence indicates that the superoxide anion is present in all aerobic organisms and is believed to be involved in several enzyme oxidation processes.2 However, its presence does not neces- sarily imply that it is the reactive form, and it may be either the metabolite present in the terminal electron transfer to molecular oxygen during the oxidation- reduction process or a precursor to other reactive species directly involved in oxygen addition. The involvement of 02-in biological oxidations can be investigated by studying its reactions with simple substrates with functionalities similar to those encountered in the more complex living system.Until recently the chemistry of superoxide has been limited to physicochemical studies, detection in enzymes and related systems, and upper-atmosphere studies in the gas phase. With the discovery of new superoxide reagents and the preparation of stable solutions of 02-in aprotic organic solvents, the chemistry of superoxide can be extended to include a wide range of simple organic substrates which may serve as models for the biological counterparts. This review summarizes these recent chemical investigations and relates the findings to certain biological oxidation mechanisms. 2 Occurrence and Preparation of Superoxide Superoxide can result from reduction of molecular oxygen [equation (l)] and from oxidation of peroxide [equation (2)].The reduction potential of the 02/02-G. A. Hamilton, in ‘Chemical Models and Mechanisms for Oxygenases in Molecular Mechanisms of Oxygen Activation’, ed. 0. Hayaishi, Academic Press, New York, 1975, p. 405. I. Fridovich, Adv. Enzymol., 1974, 41, 35. The Organic Chemistry of Superoxide 022--+ 02-+ e couple is highly dependent on the nature of the medium and ranges from -0.2 to -0.5 V [vs. normal hydrogen electrode (n.h.e.)].3-5 The electron affinity of oxygen, determined as +0.42 to +0.44 eV,617 indicates that one- electron reduction of oxygen is an exothermic process in the gas phase. The reduction potential of the 02-/022-couple and electron affinity of 02-are -1.8 and + 1 V, respectively;*~8 however, attempted electrolytic production of 02-from peroxide yields molecular oxygen.9 The generation of 02-from peroxide can be accomplished by non-electrolytic oxidative methods (see below).A number of electroreducing methods for the generation of 02-in both aprotic solvents and water have been des~ribed.~-~~~J~ Many of these procedures call for the use of tetra-alkylammonium perchlorate as the supporting electrolyte; in an atmosphere of oxygen, tetra-alkylammonium superoxides are formed in appreciable concentrations. Other methods for generating 02-in aqueous solutions include reduction of oxygenated solutions by hydrated electrons or by hydrogen atoms generated during photolysis in the far ultraviolet region,ll-1* and radiolysisl0~l5 or ultrasonication of water.16 The short lifetime of 02-(k2 decay = 107 1 mol-l s-l) in aqueous solutions preclude studies of substrates oxidized relatively slowly by 02-.In principle, any substance with an oxidation potential greater than + 0.5 V (n.h.e.) should be capable of reducing molecular oxygen to 02-. Thus a number of alkali metals and transition-metal ions react in this way. The combustion of potassium to yield ‘oxides of potassium’ was described by Gay Lussac in 181 1, and is presently used in the commercial preparation of K02.17J8 Both sodium and potassium superoxides can be prepared by direct oxygenation of the metals in diglyme with catalytic quantities of xanthone.19 In the case of lithium, only peroxide is obtained under these conditions. Superoxides have also been found J.Divisek and B. Kastening, J. Electroanalyt. Chem., 1975, 65, 603. D. T.Sawyer and J. L. Roberts, jun., J. Electroanalyt. Chem., 1966, 12, 90. M. E.Peover and B. S. White, Chem. Comm., 1965, 183. R. S. Celotta, R. A. Bennett, J. L. Hall, M. W. Siegel, and J. Levine, Phys. Rev., 1972, A6, 631. W. T.Zemke, G. Das, and A. C. Wahl, Chem. Phys. Letters, 1972, 14, 310. * D. R. Bates and H. S. W. Massey, Trans. Roy. SOC.,1943, A239, 269. J. Chevalet, F. Rouelle, L. Gierst, and J. P. Lambert, J. Electroanalyt. Chem., 1972, 39, 201. lo D. Behar, G. Czapski, J. Rabani, L. M. Dorfman, and H.A. Schwarz,J. Phys. Chem., 1970, 74, 3209. l1 J. H. Baxendale, Radiation Res., 1962, 17, 312. l2 G. E. Adams, J. W. Boag, and B. D. Michael, Proc. Roy. SOC.,1965, A289, 321. l3 E.Hayon and J. McGarvey, J. Phys. Chem., 1967,71, 1472. l4 J. M. McCord and I. Fridovich, Photochem. and Photobiol., 1973, 17, 115. l6 D. Klug, J. Rabani, and I. Fridovich, J. Biol. Chem., 1972, 247, 4839. l8 B, Lippitt, J. M. McCord, and I. Fridovich, J. Biol. Chem., 1972, 247, 4688. l7 A. W. Petrocelli and D. L. Kraus, J. Chem. Educ., 1963,40, 146. L. Andrews, J. Mol. Spectroscopy, 1976, 61, 337. lBN. S. Sokolov and G. A. Matsulevich, Neorg. Perekisnye. Soedinenii. Doklady Vses. SoveAhch., 1973, 95 (Chem. Abs., 1975, 83, 212027). 196 Lee-Rufl on a number of transition-metal surfaces, e.g.Pt, Au, Co, Rh, W, and Mo, when these are exposed to oxygen.20 Silver dispersed on silica gel is oxidized by oxygen to Ag02.21 A number of superoxo-complexes of transition metals have been prepared from direct oxygenation of lower oxidation state metal complexes such as cobaIt(11),22*34 iron(11),~3 iridi~m,~4 rhodium(n),25 and chromium(n) although the superoxide nature in the iridium and rhodium complexes is questionable.26 In many of these systems superoxide is present as a tightly bound ligand with little dissociation to free superoxide. These systems have been studied as possible models for biological oxygen carriers, oxygenases and oxidases.Some p-super- oxobiscobalt(m) complexes have been prepared by oxidation of the corresponding ,x-peroxobiscobalt 2 :1 add~cts.~~~~~The superoxide nature of dioxygen in these complexes has been inferred from Rarna~~,~~ and electron spinu.v.-~isible,~O resonance (e.s.r.) spectro~copy,~~ X-ray crystallographic analysis,32 and molecular orbital calculations of the 0-0 bond distance.33 For example the 0-0 bond distance in p-superoxobis(penta-aminocobalt) was determined as 1.31 A.32933 This compares with the 0-0 bond distance of 1.20 8, for molecular oxygen, 1.28 8, for alkali-metal superoxides, and 1.35-1.45 8, for peroxides.34 E.s.r.studies have shown that ferrocene forms an Ferxl superoxide on porous Vycor glass rods.35 The ability to reduce oxygen is not restricted to metals and metallic ions. Certain organic anions, radicals, and neutral species have been found to undergo efficient one-electron transfer to 02.These include flavins (mono- and di-nucleo- tides),36,3' reduced dyes such as methylene blue,38 fluorescein,39 saponified chlorophyll,40 and various aromatics (tryptophan,39 perylene,41 and catechols42 zu S.L. McLellan, Diss. Abs. (B), 1975, 36, 744. 21 N. Shimizu, K. Shimokoshi, and Y. Yasumori, Bull. Chem. SOC.Japan, 1973, 46,2929. E2 G. McLendon and A. E. Martell, Coordination Chem. Rev., 1976, 19, 1. ZR J. P. Collman, R. R. Gagne, and C. A. Reed, J. Amer. Chem. SOC.,1976,96,2629. 24 L. Vaska, Science, 1963, 140, 809. 25 L. Vaska, L. S. Chen, and W. V. Miller, J. Amer. Chem. SOC.,1971, 93, 6671. 26 R. Kellerman, P. J. Hutta, and K.Klier, J. Amer. Chem. SOC.,1974, 96, 5946. 27 M. Balnor, J. Biol. Chem., 1970, 245, 6125. 28 A. G. Sykes and J. A. Weil, Progr. Znorg. Chem., 1970, 13, 1. *@ T. Strekas and T. G. Spiro, Inorg. Chem., 1975, 14, 1421. aoV. M. Miskowski, J. L. Robbins, I. M. Treitel, and H. B. Gray, Znorg. Chem., 1975, 14, 2318. M. Mori, J. A. Weil, and J. K. Kinnaird, J. Phys. Chem., 1967, 71, 103. 32 W. P. Schaefer and R. E. Marsh, Acra Cryst., 1966, 21, 735. 3R I. Hyla-Kryspin, L. Natkaniec, and B. Jezowska-Trzebiatowska, Chem. Phy5. Letters, 1975, 35, 311. 34 M. M. T. Khan and A. E. Martell, 'Homogenous Catalysis by Metal Complexes. Activation of Small Inorganic Molecules', Vol. 1, Academic Press, New York, 1974. 3s T. H. Vanderspurt, J. Turkevich, M.Che, and E. Buchler, J. Catalysis, 1974, 32, 127. 36 C. Beauchamp and I. Fridovich, Analyr. Biochem., 1971, 44, 276. 37 V. Massey, S. Strickland, S. G. Mayhew, L. G. Howell, P. C. Engel, R. G. Mathews, M. Schuman, and P. A. Sullivan, Biochem. Biophys. Res. Comm., 1969, 36, 891. 38 J. M. McCord and I. Fridovich, J. Biol. Chem., 1972, 245, 1374. nn C. Balny and P. Douzon, Biochem. Biophys. Res. Comm., 1974, 56, 386. 1o L. S. Jahnke and A. W. Frenkel, Biochem. Biophys. Res. Comm., 1975, 66, 144. Y. Kodratoff, C. Naccache, and B. Imelik, J. Chim. phys., 1968, 65, 562. A2 H. P. Misra and I. Fridovich, J. Biol. Chcrn., 1972, 247, 3170. The Organic Chemistry of Superoxide and their semiquinones43). In some cases36~39~40 the reduction of 02 is promoted by irradiation with U.V.or visible light. In an earlier reportg4 on the oxidation of perinaphthenide (1) to perinaphthyl radical (2) by molecular oxygen, the reduced product was not identified but it was proposed to be peroxide. It is likely that this oxidation, as well as other autoxidations of carbanions with oxygen, proceeds with concomitant production of 02-. Formation of 02-from oxidation by peroxide is less common although a number of such preparations have been reported. Hydrogen peroxide can be oxidized by ceric(~v)~~ ions. Certain metal peroxides can be and peri~date~~ oxidized by ozone to a mixture of superoxides and ozonides [equation (3)] at low temperatures.47 M(0z2-)%+ 03 -+ M(Oz-)% + M(03-) M = Li, Mg, Ba, or Sr (3) A number of superoxides have been prepared by ligand substitution with 02-.Thus rare earth superoxides can be synthesized from their perchlorate salts by direct substitution with sodium superoxide [equation (4)].48 Tetramethylammonium superoxide can be conveniently made by heterogenous reaction of either its fluoride or hydroxide salt with potassium superoxide and subsequent extraction with liquid ammonia.49 Other ammonium or tetra-alkylammonium superoxides can be prepared by electrolysis in an aprotic medium with a suitable ammonium salt as the supporting electrolyte.3-5.9910 The greater solubility of tetra-alkylammonium superoxides in organic solvents make these more suitable reagents than the metal salts in the study of the organic chemistry of superoxide. 4.3 H.P. Misra and I. Fridovich, J. Biol. Chem., 1972, 247, 188. 44 D. H. Reid, Chem. and Ind.. 1956, 1504. 45 E. Saito and B. H. J. Bielski, J. Amer. Chem. SOC.,1961, 83, 4467. 46 P. F. Knowles, J. F. Gibson, F. M. Pick, and R. C. Bray, Biochem. J., 1969, 111, 53. 4’(z) I. I. Volnov, S. A. Tokareva, V. N. Belevskii, E. I. Latysheva, V. I. Klimanov, and G.P. Pilipenko, Neorg. Perekisyne Soedinenii, Doklady Vses. Soveshch., 1973,110; (b) S. A. Tokareva, G.P. Pilipenko, V. I. Klimanov, and I. I. Volnov, Tezisy Doklady Vses. Soveshch. Khim. Neorg. Perekisnye Soedinenii, 1973, 14. V. N. Belevski, E. Ramashov, A. E. Kharakoz, T. B. Durnyakova, and S. V. Bleshinski, Izvest. Akad. Nauk Kirg. S.S.R., 1975, 57. *9 A. D.McElroy and J. J. Hashman, Inorg. Chem., 1964, 3, 1798. Lee-Ruf Both iron(@ and cobalt(Iz1) superoxo-complexes have been obtained by nucleophilic displacement by 02-. Aquocobalamin reacts with 02-to give superoxocobalamin,50 and perchlorate ion is displaced from iron protoporphyrin dimethyl ester perchlorate. 51 A number of oxyanions can expel 02-under different conditions. y-Irradiation of chromate ion in alkaline solution,52 X-radiation of solid potassium chlorate,53 and decomposition of potassium perox~chromate~~ have been reported to generate superoxide. 3 Detection of Superoxide A number of physical methods have been developed for the detection of 02-, including c~nductimetry,~~ and u.v., i.r., Raman, and e.s.r. spectroscopy.Transient 02-in aqueous solution exhibits a Amax at 240 nm (E = 1060),56 although an absorption at 443 nm has been reported in ~yridine~~ and attributed to a pyridine superoxide complex. The i.r. spectrum of 02-shows no characteristic sharp band@ as would be expected for homopolar molecules. On the other hand, the Raman spectrum of KO2 shows a dv at 1145 cm-l attributed to the 0-0 stretch,5* which is in close agreement with results obtained from electron scatter- ing experiments with oxygen (dv = 1089 ~m-l).~~ The i.r. absorption at 1100 cm-1 of potassium superoxopentacyanocobaltate(rrr)and p-superoxopentamino- cobalt(m) has been assigned to the 0-0 stretch of superoxide rather than that of peroxide, which would exhibit an absorption at 800 cm-1.60 The free radical nature of 02-permits its detection by e.s.r.spectroscopy. Pyridine solutions of 02-gives an e.s.r. signal at g = 2.020,57 just slightly above the free electron values. This compares with values ofg,, = 2.175 andg, = 2.002 for solid KO2 and NaO2. E.s.r. detection of 02-in enzyme systems indicates somewhat lowerg values (g,, = 2.08 andg, = 2.00);61 however, further evidence that the paramagnetic species was 02-was obtained by the use of 170-enriched oxygen and observation of the resultant hyperfine splitting.62 Under conditions where 02-cannot be detected by direct e.s.r. measurements a method of spin trapping has been devel0ped.6~ 5,5-Dimethyl-l-pyrroline-l-oxide(3) is used to trap 02-and produce the stable paramagnetic N-oxide radical (4).J. Ellis, J. M. Pratt, and M. Green, J.C.S. Chem. Comm., 1973, 781. s1 H. A. 0. Hill, D. R. Turner, and G. Pellizkr, Biochem. Biophys. Res. Comrrt., 1974, 56, 739. 52 J. Kalerinski, Bull. Acad. polon. Sci., Ser. Sci. chim., 1966, 14, 137. 53 T. E. Hasty, W. B. Ard, jun., and W. G. Moulton, Phys. Rev., 1959, 116, 1459. 54 E. K. Hodgson and I. Fridovich, Biochemistry, 1974, 13,3811. 55 S. Ander, Strahlentherapie, 1967, 132, 135. 56 G. Czapski and L. M. Dorfman, J. Phys. Chem., 1964, 68, 1 169. 57 W. Slough, Chem. Comrn., 1965, 184. 63 E. G. Brame, jun., S. Cohen, J. L. Margrove, and V. W. Meloche, J. Inorg. Nuclear Chem., 1957, 4, 90. 59 M. J. W. Boness and G. J. Schulz, Phys. Rev., 1970, A2, 2182. aoT.Shibahara and M.Mori, ‘Proceedings of the 16th International Conference on Co-ordination Chemistry, 1974’. (Chem. Abs., 1976, 85, 54 057). 61 P. F. Knowles, J. F. Gibson, F. M. Pick, and R. C. Bray, Biochem. J., 1969, 111, 53. R. C. Bray, F. M. Pick, and D. Samuel, European J. Biochem., 1970, 15, 352. fi3 J. R. Harbour and J. R.Bolton, Riochem. Biophys. Res. Comm., 1975, 64, 803. The Organic Chemistry of Superoxide I0-Chemical tests for 02-rely on its oxidative and reducing properties. A widely used method [equation (5)] is the reduction of tetranitromethane to nitroform anion, which absorbs in the U.V. at 350 nm (E = 15 Although the test is 02-+ C(N02)4 -+ C(N02)Z + NO2 + 02 (5) efficient with bimolecular rate constants approaching diffusion control values, the reduction is not specific to 02-and other reducing species such as hydrogen atom, hydroxyl, and solvated electrons also reduce tetranitromethane.Another commonly used assay for 02-is reduction of nitrotetrazolium blue (5) to its diformazan (6) (A,,, = 560, E = 30 x lo3), which can be monitored 7*2 by visible spe~troscopy.~5 However, it is conceivable that in this case other reducing radicals may also give a positive test. 61 B. H. J. Bielski and A. 0.Allen, J. Phys. Chem., 1967, 71, 4544, 65 R. W. Miller and C. T. Kerr, J. Biol. Chem,, 1966, 241, 5597. 200 Lee-Rufl A less efficient method for 02-detection involves reduction of ferricytochrome derivatives to ferrocytochromes, which can be monitored by visible spectroscopy (A,,, = 550 nm).66-68 Again, this method is not specific for 02-and furthermore, the bimolecular rate constant for ferrocytochrome reduction is only lo61 mol-l s-1.67 Epinephrine (adrenalin) (7) is oxidized to adrenochrome (8) by 02-, and HO IMe \ NHMe adrenochrome production (h = 480 nm, E = 4020) can be monitored as a method of assay.G8 Lengfelder and co-workers have found that this oxidation involves several reactive intermediates, not all of which lead to adrenochrome69 and also that adrenochrome, once formed, can be reduced to a semiquinone.Caution, therefore, has to be exercised in interpreting results of this test on a quantitative basis. Another oxidative method for detection of 02-involves observation of chemi- luminescence from luminol oxidation with 02-.70However, the background luminescence must be evaluated concurrently, since the same phenomenon with ground-state triplet oxygen has been observed in alkaline solutions of luminol.71 Seyb and Klemberg72 have described a volumetric analysis of oxygen evolution upon quenching of alkali-metal superoxides with acetic acid [equation (6)].202-+ 2H+ + H2Oz + 02 (6) In water this dismutation is quite efficient (k N lo7 1 mol-l s-l) and follows second-order kinetics in 02-.73 It is catalysed (k > lo91 mol-l s-I) by a number of specific metalloenzymes, called superoxide dismutases,2 which can be used in combination with any of the above-mentioned chemical tests to provide unam- biguous evidence for the presence of 02-.However, there has been some recent controversy over the specificity of these enzymes which is discussed in Section 5. E6 A. Azzi, C. Montecucco, and C. Richter, Biochem. Biophys. Res. Comm., 1975, 65, 597. 67 G. M. Simic, I. A. Taub, J. Tocci, and P. A. Hurwitz, Biochem. Biophys. Rex. Comm., 1975, 62, 161. 68 J. M. McCord and I. Fridovich, J. Biol. Chem., 1969, 244, 6049. 6QW. Bors, M. Saran, C. Michel, E. Lengfelder, C. Fuchs, and R. Spottl, Internat. J. Radiative Biol., 1975, 28, 353. 70 J. R. Totter, E. C. de Dugros, and C. Riveiro, J. Biol. Chem., 1960, 235, 1839. 71 E. H. White, J. Chem. Educ., 1957, 34, 275. 72 E. Seyb and J. Klemberg, An&. Chem., 1951, 23, 115. 73 G. Czapski and B. H. J. Bielski, J.Phys. Chem., 1963, 67, 2180. The Organic Chemistry ofSuperoxide 4 Reactivity of Superoxide A. 02-as a Reducing Agent.-The relatively low reduction potential of oxygen is consistent with the ease with which 02-can act as a reducing agent, and reactions involving electron transfer from 02-to metallic ions in their higher oxidation states as well as organic substrates have been mentioned as assays for 02-. The reduction of cytochrome c (FeI’I) by 02-deserves some comment since the cytoclwomes are found in cells of all aerobic organisms and play an important role in the electron-transport chain and oxidase activity. The reaction is described by equation (7), which appears to be the reverse of the reaction for electron cyt c (Fe3+) + 02--+ cyt c (Fe2+) + 02 (7) transport from ferrocytochrome c to molecular oxygen and its conversion into water and oxidized organic substrate.The oxidation potential for the Fe2+/Fe3+ couple is reported to be -0.771 V.74 If the reduction potential for the 02/02-couple is taken as -0.29 V, EOfor reaction (8) would be + 1.06 V. Ray0 and Fe3++ 02--+ Fe2++ 02 (8) Hay0n7~ measured an EO of +0.27 V for cytochrome c reduction by 02-, indicating that ferrocytochrome c is a stronger reducing agent than uncomplexed ferrous ions. It is therefore conceivable that in a different medium, the oxidation potential of the Fe2+/Fe3+ cytochrome couple may become more positive thus shifting the equilibrium to the left in equation (7). Such reversal has been ob- served76 in the formation of 02-by oxygenation of haemoglobin [equation (9)].Hb(Fe2+) + 02 + Hb(Fe2+)02-+ Hb(Fe3+) + 02-(9) The ability of edta-complexed iron(@ to catalyse the 02-dismutation reaction129 (see Section 9,in which the mechanism involves alternate reduction and reoxidation of Fe3+, points to the ease with which iron can either oxidize 02-or reduce 02. Valentine and Q~inn~~ recently reported the reduction of a MnXI1 tetraphenylporphyrin (TPP) salt to MnIITPP, and showed that the reverse oxygenation of MnIITPP to MnlIITPP+ does not lead to superoxide formation. Transfer of an electron by 02-to ortho- orpara-quinones leads to semiquinone radicals [equations (10) and (1 1)].75J8-84 The reaction is reversibles2 and the balance of equilibrium depends on the one-electron reduction potential of the 74 ‘Handbook of Chemistry and Physics’, 44th Edn.Chem. Rubber Co., 1962-63, p. 1744. 75 P. S. Rao and E. Hayon, J. Phys. Chem., 1975,79, 397. 76 K. H. Winterhalter, Chimia, 1976, 30, 9. 77 J. S. Valentine and A. E. Quinn, Inorg. Chem., 1976, 15, 1997. 78 E. Lee-Ruff, A. B. P. Lever, and J. Rigaudy, Canad. J. Chem., 1976, 54, 1837. 79 R. Poupko and I. Rosenthal, J. Phys. Chem., 1973, 77, 1722. so M. Simic and E. Hayon, Biochcm. Biophys. Res. Comm., 1973,50, 364. R. L. Willson, Chem. Comm., 1970, 1005. 82 K. B. Pate1 and R. L. Willson, J.C.S. Faraday I, 1973, 69, 816. 83 I. B. Afanasev, S. V. Prigoda, T. Y. Maltseva, and G. I. Samokhvalov, Internat. J. Chem. Kinetics, 1976, 6, 643.84 Y. A. Ilan, D. Meisel, and G. Czapski, IsruelJ. Chem., 1974, 12, 891. 202 Lee-Ruf $02-fO2 + 02-fO2 quinone. Willson and Pate182 obtained one-electron reduction potentials of a number of quinones by determining equilibrium concentrations. Whereas 02-transfers an electron to ubiquinone, 1,6naphthoquinone, p-benzoquinone, and methylated p-benzoquinones, semiquinones of vitamin K, 2,3-dimethylnaph-thoquinone, and anthroquinone-2,6-disulphonate will reduce dioxygen to superoxide. Since semiquinones are involved in electron transport in cells, dioxygen reduction by these radicals may be one possible source of 02-. In addition to quinones, a number of other organic compounds will oxidize 02-. These include tetranitromethane and nitrotetrazolium blue, which have already been mentioned as characteristic tests for 02-.Studies of the chemistry of 02-in the gas phase have revealed that ozone, nitrogen dioxide, and oxygen itself can act as oxidizing agents, as shown in equations (12)-(14).85-88 These processes represent the major decomposition pathways for 02-decay in the ionospheric regions. Studies of Fenton’s reagent led Haber and Weisssg to propose reaction (15), in which an extremely reactive oxidant, the hydroxyl radical, is generated. This reaction seems to occur readily in buffered aqueous systems and is believed to be responsible for a number of aromatic hydroxylations occurring in living organisms 85 D. A. Parkes, J.C.S. Faraday I, 1972, 68, 2103. F. C. Fehsenfeld and E.E. Furguson, Pfanet Space Sci., 1968, 16, 701. ni E. E. Furguson, Accounts. Chem. Res., 1970, 3, 402. J. D. Payzant and P. Kebarle, J. Chem. Phys., 1972, 56, 3482. F. Haber and J. Weiss, Proc. Roy. SOC.,1934, A147,332. The Organic Chemistry of Superoxide (see Section 4B). Organic peroxides and hydroperoxides are reduced by 02-according to equations (16) and (17).90vQ1 In the reduction of hydroperoxides, ROOR + 02--+ RO-+ ROO+ 02-(1 6) ROOH + 02--+ ROO+ OH-(17) alkoxyl radical formation appears to be dominant to hydroxyl radical formation,Q1 as would be expected from electronegativity considerations. B. Superoxideasan Oxidizing Agent.-The reduction potential of the 02-/0z2-couple has been determined as -2.02 V (vs. standard calomel electrode) in dimethyl sulphoxide;g however, the potential for the transformations written in the direction shown in equations (18) and (19) are + 1.51 and + 0.94 V, HO2 + H+ + e -+ HZOZ (18) 02-+ 2H+ + e -+ HZOZ (19) respectively (n.h.e.).g2s93 It is somewhat surprising that the latter potential is lower as the pKa of 02-is 4.81° and protonation of 02-would be expected to be exothermic.In any case, 02-would be expected to react as a powerful oxidant. Hydrogen abstraction from diphenols to produce semiquinones [equation (20) ] has been reported for a number of catechols,78~94-95 hydroq~inones,~8~~~ and ene-1,Zdiols such as ascorbic acid.97 It is not known whether the process is con-certed or involves a sequence of H and H+ abstractions.However, it appears unlikely that the concerted mechanism would operate in the case of the p-hydroquinones owing to the large inter-hydrogen distance. Two possible schemes can be envisaged for two-step elimination : hydrogen-atom abstraction by 02-followed by proton transfer to the resultant HOO-[equation (21)] or + 0,-," HOO-(21)+ a'*+a'* OH OH 0-the reverse order [equation (22)]. If the pK value for HO2 of 4.81° is reliable, the equilibrium constant for the first step of equation (21) would be 10-4 (taking A. LeBerre and Y.Berguer, Bull. SOC.chim. France, 1966, 2363. 91J. W. Peters and C. S. Foote, J. Amer. Chem. SOC., 1976, 98, 873. st D. H. Busch, H. Shull, and R. T. Conley, 'Chemistry, Allyn and Bacon, 1973, p. 447. D3 P.M. Wood, F.E.B.S. Letters, 1974, 44, 22. 94 R. W. Miller and U. Rapp, J. Biol. Chem., 1973, 248, 6084. 85 Y. Moro-Oka and C. S. Foote, J. Amer. Chem. SOC., 1976, 96, 1510. D6 P. S. Rao and E. Hayon, J. Phys. Chem., 1973, 77, 2274. 97 M. Nishikimi, Biochem. Biophys. Res. Comm., 1975, 63,463. Lee-Ruff pK = 9 for phenols). It has also been found that phenol will not transfer a proton to 02-in the gas phase.98 On the other hand, the decomposition of 02-in water (pK = 15) indicates the intermediacy of hydroperoxyl (HO2) radicals, which subsequently disproportionate according to equation (23),1° so that HO2 + HO2 -+ H202 + 02 (23) although the acid-base equilibrium between 02-and catechol may favour the reactant side, subsequent oxidation by hydroperoxyl radicals may be efficient enough for the overall transformation to occur by this mechanism.No visible evolution of oxygen accompanied catechol oxidation, as would be expected if perhydroxyl radical were present (in sufficient quantities for bimolecular decay). In some cases subsequent oxidation of the semiquinones results in the production of quinones, as in the case of adrenalin,69 p-hydroquinone,78 and ascorbic acid.97 The production of quinones could also arise from bimolecular redox reaction between two semiquinones according to equation (24).69 In the case of catechol Q*-+ Q'--,Q + Q2-(24) semiquinones, oxidative ring-opening takes place to yield muconic acid deri- vatives [equation (25)]78199 or o-hydroxymuconic semialdehy des [equation (26)].95 This mode of oxidative decomposition is similar to the oxidation of catechols by pyrocatechase and metapyrocatechase.lO09 lol ' 9* I.Dzidic, D. I. Carrol, R. N. Stillwell, and E. C. Homing, J. Amer. Chem. SOC.,1974, 96, 5258. ¶* M. Tezuka, Y. Ohkatsu, and T. Osa, Bull. Chem. SOC.Japan, 1975, 48, 1471. loo 0. Hayaishi and K. Hashimoto, J. Biochem. Japan, 1950, 37, 371. lU1 S. Dagley and D. A. Stopher, Biochem. J., 1959, 73, 16. The Organic Chemistry of Superoxide Vitamin E (a-tocopherol) acts as an antioxidant in the inhibition of peroxi- dation of lipids, a chain reaction believed to be initiated by 02-and singlet oxygen.102 Nishikimilo3 investigated the interaction of superoxide and a chro- manol derivative (9),a model for tocopherol, and reported the formation of quinone (11) via the ketonized chromanone (10).Quinone (11) or a similar derivative from a-tocopherol could subsequently act as a free radical scavenger and inhibit the peroxidation chain reaction. (9) With simple primary and tertiary alcohols, 02-behaves in much the same fashion as its dismutation in aqueous solution [equation (27) 1.90 With secondary alcohols, auto-oxidation of the metal alkoxide with oxygen generated in situ 2 ROH + 202--+ 2RO-+ H202 + 02 (27) is observed, with ketones being the end products [equation (28)]. Aldehydes are susceptible to oxidation by 02-to give the corresponding carboxylic acids in yields of 59-72%, according to the stoicheiometry given in lo* D.D. Tyler, F.E.B.S. Lerrers, 1975, 51, 180. Io3 M. Nishikimi and L. J. Machlin, Arch. Biochem. Biophys., 1975, 170, 684. Lee-Ru$ equation (29).104 Reaction conditions are mild so this constitutes an efficient 2Na02 + RCHO -+ RCO2Na + 02 + NaOH (29) method for alkaline oxidation of aldehydes. Ketones are inert to oxidation by 03-, although enolizable ketones undergo self condensation in the presence of 02-.105 A novel oxidative ring-opening transformation of tetracyclone (12) to give a ketohemiketal (13) has been reported by Rosenthal.106 In a related studylo5 + 02--Ph Ph OH 0 0 dibenzal acetone (14) reacted with 02-to yield cinnamic acid and benzaldehyde, which resisted further oxidation. Although the mechanism of these trans- 0 0 formations is not clear it would appear that the initial step may involve a Michael addition of 02-to the enone.Oxidative cleavage of a-diketones by 02-has been reported by Le Berre and Berguer.lo4 Phenanthraquinone (1 5) and 1,2-naph- thaquinone (17) are transformed into carboxylic acids (16) and (18) + (1 9), respectively. Benzil is converted by 02-into a mixture of benzoic and benzilic acids. San-Filippo and co-workers made similar observations in their study of 02-oxidations of a-keto-, a-hydroxy-, and a-halogeno-ketones, esters, and carboxylic acids;lo7 in all cases carboxylic acids are produced. A number of superoxide-mediated systems that are capable of hydroxylating aromatic compounds have been discovered. Horseradish peroxidase with dihydroxyfumarate and molecular oxygen can hydroxylate p-coumaric, p-hydroxybenzoic, and salicylic acids.108 These reactions are inhibited by catalase lopA.LeBerre and Y. Berguer, Bull. SOC.chim. France,1966, 2368. lo5 E. Lee-Ruff, unpublished results. log I. Rosenthal and A. Frimer, Tetrahedron Letters, 1975, 3731. lo’ J. S. San-Filippo, C. I. Chern, and J. S. Valentine, J. Ow. Chem., 1976, 41, 1077. loRB. Halliwell and S. Ahluwalia, Biochem. /., 1976, 153, 513. The Organic Chemistry ofSuperoxide ( 6) 0 0 and superoxide dismutase, indicating that both hydrogen peroxide and 02-are necessary for hydroxylation. This in turn suggests that the hydroxyl radical produced by the Haber-Weiss reaction is the reactive species in these processes.Another 02-mediated oxidation involving hydroxyl radical is the oxidation of methional (20) to ethylene. Methional sulphoxides are also detected in the MeSCH2CHKHO -,CH2CH2 + CO (20) reaction mixture.log Peroxidation of lipids is initiated by 02-generating systems.ll0 Competitive inhibition tests have shown that both superoxide dismutase and catalase will inhibit such oxidations, again implicating hydroxyl radical as the initiator in the chain reaction-lll However, trapping experiments112 have shown that hydroxyl radicals are not directly involved in peroxidations, and moreover, in the absence of hydrogen peroxide, 02-adsorbed on silica gel will react with olefins to produce alkylperoxyl radicals.ll3 Oxidative cleavage of olefins and formation of epoxides with 02-have been reported in the case of benzalfluorene (21) and cyclohexanone (22).11* Oxidation of benzilic carbon by 02-with production of hydroperoxide has looW.Bors, E. Lengfelder, M. Saran, C. Fuchs, and C. Michel, Biochem. Biophys. Res. Comm., 1976,70,81. 1lOP. B. McCay, K. L. Fong, M. King, E. Lai, C. Weddle, L. Poyer, and K. R. Hornbrook, Lipids, 1974, 1, 157. ll1 R. Zimmerman, L. Flahe, U. Weser, and H. J. Hartman, F.E.B.S. Letters, 1973, 29,117. 112 D. D. Tyler, F.E.B.S. Letters, 1975, 51, 180. lla 0. I. Lyubimova, A. G. Kotov, and L. Y. Karpova, Kinetika i Kataliz, 1972, 13, 1603 (Chem. Abs., 1973,78, 71 040). 114 R. Dietz, A. E. J. Forno, B. E. Larcombe, and M. E. Peover, J. C/zem.Suc.(B),1970,816.Lee-Ruf 0 02-__II, been observed in the transformation cumene into cumene hydroperoxide.99 Hydroperoxides may also intervene in the conversion of fluorene into fluorenone and diphenylmethane into benzophenone. Efficient dehydrogenation of 1,9-dihydroanthracene to anthracene has been rep0rted.9~ It is conceivable that in lipid autoxidations mediated by 02--producing systems, formation of hydro- peroxides as initiators of the chain reaction may be the result of allylic hydrogen abstraction by 02-in a typical free radical process. C. Superoxide as a Base, Nucleophile, and Ligand.-The pK of the hydroperoxyl radical (conjugate acid of 023 has been reported as 4.8.1° This implies that under physiological conditions superoxide exists in the non-protonated form and may be expected to exhibit weakly basic and nucleophilic properties.Decay of the hydroperoxyl radical follows second-order kinetics (kN 107 1 mol-1 s-I), giving oxygen and hydrogen peroxide.lO Superoxide will abstract a proton in the gas phase from a number of oxygen and carbon acids including benzoic acid, malononitrile, p-nitrophenol, trinitrotoluene, picric acid, and acetic acid.98 On the basis of the solution pK reported for 02-, the acidity of HO2 should be comparable with that for acetic acid. In the gas phase, however, acetic acid is a stronger acid than HO2. Such reversal in trends of acidities on going from the condensed to the gas phase have been observed for other acids.ll5 Superoxide acts as a nucleophile in displacement reactions on alkyl halides, tosylates, acyl chlorides, esters, and anhydrides.One of the earliest such reports116 was of reactive chlorinated derivatives such as benzoyl and triphenylmenthyl chloride giving the corresponding peroxides in the presence of water. Trimethyl- chlorosilane (23) reacts with potassium superoxide to yield disilyl ether (24). Me3SiC1 + KO2 + MesSiOSiMe3 (23) (24) 116 D. K. Bohme, E. Lee-Ruff, 1972,94, 5153.and L. B. Young, J. Amer. Chem. SOC.,M. Schmidt and H. Bipp, 2. anorg. Chem., 1960,303, 190, The Organic Chemistry of Superoxide Less efficient reactions with 02-were observed for saturated alkyl halides and sulphates,lo4 and in certain cases alcohols are obtained, presumably from the further reduction of the peroxides initially formed. The inefficiency of these reactions is probably due to the limited solubility of NaO2 and KO2 in organic solvents since in all cases suspensions of NaO2 were used.However, effective nucleophilic substitutions were observed in electrogenerated superoxide solutions containing saturated alkyl halides, resulting in initial alkylperoxyl radical formation (Scheme l).l14J17 Once formed, the peroxyl radical is reduced either by hydrogen-donating reagents giving hydroperoxides or by 02-to give peroxide anion which can undergo nucleophilic substitution with another alkyl halide to yield peroxides. Peroxides and hydroperoxides are reduced by 02-to give alcohols.g0 RX + 0,-+ROO. 4-X' ROOR ,RX ROO-ROOH do2-RO-5tOW RQH Scheme 1 With the use of crown ethers to solubilize superoxide in organic solvents,ll* a number of preparative methods for the syntheses of alkyl peroxides,llg acyl peroxides,l20 and alcohols121J22 have been developed using 02-as an oxygen nucleophile [equations (30)-(32) 1.2RX + 02--+ ROOR (X = C1, Br, or I) (30) 00 II II 2R-C + 02-+ RCOOCR \ c1 2RX + 02-2RO-+ 302 + 2X-(32)3 Alkyl peroxides and alcohols are formed with net inversion of stereochemistry of the alkyl group,122 indicating a bimolecular mechanism for 02-substitution in these systems. Cyclic peroxides can be obtained in fair yields from substrates possessing two reactive leaving groups. Thus dimesylate (25) and di-iodides (27) 11' M.V. Merritt and D. T. Sawyer, J. Org. Chem., 1970, 35, 2157. ]I8 J. S. Valentine and A. B. Curtis, J. Amer. Chem. Soc., 1975, 97, 224. R. A. Johnson and E. G. Nidy, J. Org. Chem., 1975,40, 1680. lzfl R. A. Johnson, Tetrahedron Letters, 1976, 331. lB1 J. S. San-Filippo, C. I. Chern, and J. S. Valentine, J. Org. Chem., 1975, 40, 1678. laaE. J. Corey, K. C. Nicolaou, M. Shibasak, Y. Macluda, and C. S. Shiner, Tetrahedron Letters, 1975, 3 183. Lee-Ruff OMS. OMS LPh + 202-(35) Ia + 202-t (37) WI + 202-(39) and (29) react with 02-to give cyclic peroxides (26), (28) and (30), respectively.I229 123 Alkyl and aryl carboxylic acid esters and phosphate esters are sensitive to superoxide hydrolysis 125 whereas amides and nitriles are inert.The mechanism of these hydrolyses is summarized by Scheme 2. It was shown by the use of a chiral group (R1)that 99% retention occurred, indicating that 0 0 0 // // 0,-//RC + 02-* RC -+ RC \ \ \ OR1 02. 02-0 0 00 // // 0,-II II RC + 02 + RC t-RCOOCR \ \ 02-0-Scheme 2 lZ3E. Lee-Ruff and J. Rigaudy, unpublished results. Iz4 J. S. San-Filippo, L.J. Romano, C. I. Chern, and J. S. Valentine, J. Org. Chem., 1976, 41, 586. lZ5F. Magno and G. Bontempelli, J. ElectrQpnplyt, Ghem., 1976,68, 337. 21 1 The Organic Chemistry of Superoxide acyl oxygen bond cleavage had taken place. Also acyl peroxides and acyl peroxy-esters are reduced under these conditions to carboxylic acids. Superoxide can act as a ligand in transition-metal complexes, as is shown by the nature of dioxygen in oxygenated complexes of cobalt.126 Theoretical molecular orbital calculations and other physical data indicate the superoxide character of dioxygen.However, the majority of these complexes are formed by addition of ground-state dioxygen and only a few have been made by direct ligand exchange or attachment with 02-. One such case reported is the synthesis of superoxocobalamin by addition of 02-to the cobalt(n1) aquocobalamin complex (vitamin BI~).~O The same oxygenated complex can be obtained by direct oxygenation of the reduced cobalt(n) aquocobalamin complex. An example of ligand exchange by 02-has been reported by Hill and co-workers51 in which iron(@ protoporphyrin dimethyl ester perchlorate, in the presence of 02-, gives iron@) protoporphyrin dimethyl ester superoxide.Liberation of 02-by ligand exchange of chloride ion in oxyhaemoglobin has been reported [equation (33)].127 Since chloride ion is a naturally occurring nucleophile it is conceivable that HbO2 + C1--+ HbCl + 02-(33) superoxide production in haemoglobin oxidation may arise in part from such a mechanism. 4 Superoxide Dismutase and Singlet Oxygen It is not the purpose of this review to dwell on the details of superoxide dismutases since numerous review articles already exist. However, some mention as to the mechanism of catalysis should be made in light of the chemical properties of 02-discussed above. These enzymes catalyse reaction (34).The active site is a 02-+ 02-+ 2H+ + H202 + 02 (34) histidine-complexed Cu2+ and the mechanism involves alternate reduction and reoxidation of Cu2+ during successive interactions with 02-[equations (35) and (3Q1.128 A number of simple copper complexes and uncomplexed copper salts E-CU2+ + 02--+ E-Cu+ + 02 (35) E-Cu+ + 02-+ 2H+ + E--Cu2+ + H202 (36) catalyse this reaction with rate enhancements of the same order of magnitude as superoxide dismutase. However, complexing of copper with NH3, C1-, Sod-, and certain amino-acids leads to decreased catalytic activity attributed to changes 126 A. Dedieu, M. M. Rohmer, and A. Veillard, J. Amer. Chem. SOC.,1976, 98, 5789, and references therein. 12’ W. J. Wallace, J. C. Maxwell, and W.S. Caughey, Biochem. Biophys. Res. Cnmm., 1974, 57, 1 104. 128 D. Klug-Roth, 1. Fridovich, and J. Rabani, J. Amer. Chem. SOC.,1973, 95, 2786. 212 Lee-Ru$ in the ionization potential of c0pper.~28-~3~Metals other than copper, e.g. iron in the iron(n)-edta complex, are equally effective in dismuting 02-by a redox cycle involving two equivalents of 02-similar to the superoxide dismutase mechanism. 33 Oxygen can exist in excited states that are exceedingly reactive, and thermo- dynamic data suggest that singlet oxygen is produced in the dismutation of 02-.13* Since superoxide dismutase can inhibit chemiluminescence associated with 02--producing systems,135J36 and the chemiluminescence attributed to singlet oxygen dimeric emi~sion,l3~ it has been proposed that singlet oxygen can be catalytically quenched by superoxide di~mutase.1~6 Other observations that solid potassium superoxide generates singlet oxygen138 and that decomposition of peroxochromate, a known singlet oxygen s0urce,~39 results in luminol chemiluminescence which is inhibited by superoxide dismutase, led to suspicions that this enzyme is not selective to 02-catalysis.However, subsequent investi- gations of the peroxochromate system clearly showed that along with singlet oxygen, 02-is also prod~ced~~J~0 resulting in the earlier misinterpretation of the superoxide dismutase quenching results. Furthermore, direct measurements of the quenching effects of superoxide dismutase in inhibiting singlet oxygen reaction gave negative results, indicating that those enzymes do not act as catalysts in singlet oxygen decay.lgl However, one cannot rule out the possibility of singlet oxygen production in the 02-dismutation although the chemiluminescence attributed to singlet oxygen dimer emission, which is quenched by superoxide dismutase, may be due to other species produced from 02-.A CO2 dimer has been suggested as the light-emitting species in these systems.142 Attempts to trap singlet oxygen with selective scavengers in the 02-dismutation reaction were unsuccessful.143 Recently it has been shown that 02-is an effective photo- chemical quencher of singlet oxygen with rate constants of quenching approaching lZ9 R. Brigelius, R. Spottl, W. Bors, E. Lengfelder, M.Saran, and U. Weser, F.E.B.S. Letters, 1974, 47, 72. J. Rabani, D. Klug-Roth, and J. Lilie, J. Phys. Chem., 1973, 77, 1169. 131 D. Klug-Roth and 5. Rabani, J. Phys. Chem., 1976, 80, 588. K. E. Joester, G. Jung, U. Weber, and U. Weser, F.E.B.S. Letters, 1972, 25, 25. 133 B. Halliwell, F.E.B.S. Letters, 1974, 56, 34. 134 W. M. Latimer, ‘Oxidation Potentials’, Prentice-Hall, New York, 2nd Edn., 1952. lR5 R. M. Arnson, Arch. Biochem. Biophys., 1970, 136, 352. 136 A. Finazzi-Agro, C. Giovagnoli, P. Desole, L. Celabrese, G. Rotilio, and B. Mondovi, F.E.B.S. Letters, 1972, 21, 183. 13’ D. R. Kearns, Chem. Rev., 1971,71, 395. 138 A. U. Khan, Science, 1970, 168,476. 13@J. W. Peters, J. N. Pitts, jun., I. Rosenthal, and H. Fuhr, J. Amer.Chem. SOC.,1972, 94, 4348. 140J. W. Peters, P. J. Bekowies, A. M. Winer, and J. N. Pitts, jun., J. Amer. Chem. SOC.,1975, 97, 3299. 141A. P. Schaap, A. L. Thayer, G. R. Faler, K. Goda, and T. Kimura. J. Amer. Chem. SOC., 1974,96,4025. 142 J. Stauff, U. Sander, and W. Jaeschke, ‘International Conference on Chemiluminescence Biochemistry’, ed. M. Corimer, Plenum Press, 1972, p. 131. 14R R. Nilsson and D. R. Kearns, J.Phys. C/ient., 1974, 78, 1681. 213 The Organic Chemistry of Superoxide diffusion control values (k = 2 x lo91mol-l s-1);144J45 thus any singlet oxygen generated would have to be scavenged for detection at least as efficiently, as it is quenched by 02-, and failure to detect singlet oxygen is probably due to the 02-quenching process.5 Summary Superoxide has been shown to be present in all oxygen-metabolizing organisms and its toxicity has been known for the last quarter century. However, there is a dearth of information on its reactivity with simple organic compounds. Only recently has the organic chemistry of superoxide been reopened for study. Some general trends of reactivity can be summarized. Superoxide undergoes efficient oxidation and reduction with organic and inorganic substrates having reduction or oxidation potentials exceeding that of 02-. It is a weakly reactive radical and base. It can undergo nucleophilic reactions with a number of substrates possessing reactive leaving groups and can act as a ligand in a number of metal complexes. Under certain conditions it can lead to production of highly reactive radicals such as hydroxyl which are capable of indiscriminate oxidation of most organic compounds. Its stability in aqueous solution appears to be minimal, and its decomposition to hydrogen peroxide and oxygen is catalysed by a family of enzymes called superoxide dismutases which are found in all aerobic cells.Its general oxidation and reduction properties may be responsible for its important function in biological oxidation and oxygenation processes. It is clear that before any studies of the interactions of 02-with proteins, lipids, polysacharide, nucleic acids, and other biochemicals are undertaken in order to elucidate biochemical oxidation mechanisms involving 02-and its toxicity, further studies on simpler molecules with mono-and multi-functional groups have to be carried out.Unfortunately this is an area in which serious inquiries have only just begun and it is to be hoped that the basic mechanisms of interaction of 02-with thiols, thioethers, disulphides, phosphates, monosocharides, and polyunsaturated fatty acids will be investigated. More quantitative data are required for the evaluation of relative free radical and nucleophilic reactivities of 02-in order to assess its importance and its interactions with more complex systems. However, due restraint has to be exercised in correlating data from reactions of free superoxide with those involving (superoxo) metal complex reactions and more complex enzymes. The effect of metal in the latter plays an important role in such pheno- mena as binding of the substrate and altering the nature of dioxygen. As a result such systems may be quite different from free superoxide, as found in crown- ether-complexed solutions of its alkali-metal salts. 144 H. J. Guiraud and C. S. Foote, J. Amer. Chem. SOC.,1976, 98, 1984. I. Rosenthal, Israel J. Chem., 1975, 13, 86.

ISSN:0306-0012    

DOI:10.1039/CS9770600195

出版商:RSC    

年代:1977

数据来源: RSC

摘要:

Immobilized Enzymes By C. J. Suckling DEPARTMENT OF PURE AND APPLIED CHEMISTRY, UNIVERSITY OF STRATHCLYDE, 295 CATHEDRAL STREET, GLASGOW G1 1XL 1 Introduction Chemists working closely with natural products or biological processes have been aware for a long time of the advantages that the use of enzymes as catalysts would have, Enzymes demand attention chiefly because of the efficiency, selectivity, and mildness with which they catalyse reactions but their use has been restricted because, being high molecular weight proteins, their stability is limited. One possible way out of this dilemma is to use the chemical principles of enzymic catalysis that have been established by physical-organic chemical studies to design stable, synthetic catalysts. It is hoped that such systems will rival enzymes in their efficiency and this approach has been called ‘biomimetic chemistry’.l An alternative is to invent some way of enhancing the thermal, mechanical, and chemical stability of an enzyme so that it becomes a recoverable catalyst.The chemistry of immobilized enzymes has grown from this germinal idea. An immobilized enzyme molecule is prevented from diffusing freely through the reaction medium by being attached physically or chemically to a support material: the reaction system thus consists of two phases, the bulk solution, and the immobilized enzyme with its support environment. Commonly the support materials are synthetic polymers and the first indications that such materials might be chemically compatible with biological macromolecules came from Merrifield’s pioneering studies on solid-phase peptide synthesis.2 Shortly afterwards, there developed applications of his techniques to immobilize chemical reagents3 In both fields, the range of polymers used was gradually extended from hydrophobic polystyrenes to hydrophilic organic polymers.The development of much of the polymer chemistry required for enzyme immobili- zation has run parallel with studies of enzyme purification by the technique of affinity chromat~graphy.~ In addition, enzymes have been immobilized upon inorganic and non-polymeric supports and the wide range of immobilization techniques available opens up a multitude of potential applications. The chief areas of current application of immobilized enzymes are in clinical :R.Breslow, Chem. SOC.Rev., 1972, 1, 553. R. Merrifield and G. R. Marshall, in ‘Biochemical Aspects of Reactions on Solid Sup-ports’, ed. G. R. Stark, Academic Press, New York, 1971, p. 111; L. 5. Marnett, D. G. Neckers, and A. P. Schaap, in ‘Techniques in Chemistry’, ed. J. B. Jones, C. J. Sih, and D. Perlman, Interscience, New York, 1976, Vol. 10, p. 995. C. C. Leznoff, Chem. SOC.Rev., 1974,3,65. H. Guilford, Chem. SOC.Rev., 1973,2,249. Immobilized Enzymes chemistry and in the food processing industry. Clinical chemists need to analyse for metabolites accurately, rapidly, and repeatedly; soluble enzymes are already widely used for this purpose. The industrial conversions of milk into cheese and of corn syrup into fructose are also catalysed by enzymes.If immobilized enzymes were used for these tasks, there would be a considerable saving of materials and money. The potential of applications such as these has stimulated much basic research into the modification of enzyme properties by immobiliza- tion and into the design and operation of both analytical and large-scale im- mobilized enzyme reactor systems. Much has already been written in detail on these topi~s.~-8 Rather than compose new variations upon often heard themes, this review will try to show how the chemical properties of immobilized enzymes fit them for their application in general terms. As is often the case in those areas of chemistry related to biology, nature evolved the chemistry of interest before the chemist invented it.Immobilized enzymes are a good example of this. Many metabolic enzymes are not diffusing freely in solution but are compartmentalized, agglomerated, or bound to membranes. These three natural immobilization techniques are exemplified firstly by the enzymes of the citric acid cycle and of oxidative phosphorylation, which are contained within subcellular particles known as mitochondria, secondly by the fatty-acid-synthesizing enzyme complex, in which a high metabolic efficiency is achieved by a number of enzymes acting in concert, and thirdly by several enzymes that hydroxylate foreign compounds. The last group of enzymes is bound to lipid membranes in liver and it is very difficult to free the enzymes from the lipid without destroying their catalytic activity.The membrane thus provides a safe working environment for the enzyme. Chemists have endeavoured to incorporate each of these features into synthetically immobilized enzyme systems and the following discussion illustrates how far they have succeeded. 2 Techniquesof Immobilization The applications mentioned in the introduction require enzymes immobilized in quite different forms. For instance an analyst might need an enzyme bound to a flexible membrane that he can attach to an electrode; in contrast, an enzyme immobilized upon a rigid support might suit an industrial reactor. In any case, it is obviously essential that the catalytic activity of an enzyme is not seriously impaired by the immobilization process.In this respect, it could be argued that a physical means of immobilization such as adsorption upon a suitable material would be preferable to chemical methods in which the enzyme becomes covalently bonded to its support. However, the regions of enzymes at which catalysis takes place, the so-called active sites, are rarely found on the surface 0. R. Zaborsky, ‘Immobilized Enzymes’, CRC Press, Cleveland, 1973. ‘Handbook of Enzyme Biotechnology’, ed. A. Wiseman, Wiley, New York, 1975.’K. Mosbach, in ‘Techniques in Chemistry’, ed. J. B. Jones, C. J. Sih, and D. Perlman, Interscience, New York, 1976, Vol. 10, p. 969. ‘Immobilized Enzyme Principles’, Applied Biochemistry and Bioengineering, Vol.1, ed. L. B. Wingard, E. Katchalski-Katzir, and L. Goldstein, Academic Press, New York,1976. 216 Suckling of the protein molecule; they are usually clefts in the molecule’s interior. This means that a bulky reagent such as a polymer can be coupled to the surface of an enzyme molecule without disrupting the catalytic machinery. Provided that extremes of pH and ionic strength, temperatures greater than 40 “C,and small, highly electrophilic reagents are avoided in the immobilization process, both physical and chemical techniques are generally applicable. A. Physical Immobilization Methods.-One of the simplest techniques for immobilizing an enzyme is to adsorb it upon an inert, insoluble matrk9 Many metal oxides, including glasses, will function as matrices for adsorption, prin- cipally by hydrogen-bonding between oxygen functions of the oxide surface and polar groups on the enzyme’s periphery.Similarly, ion-exchange resins will bind enzymes by ionic association with charged groups on the surface of the enzyme. Recently, hydrophobic association has been used as an immobilization tech- nique.10 With these procedures it is common to find that some desorption and leakage of the enzyme into the bulk medium occurs and consequently more tenacious immobilization techniques are required. If a gel is prepared in a solution containing an enzyme, the enzyme becomes trapped within the forming gel matrix. Since the molecular weights of enzymes exceed 15 OOO, it is easy to prepare a gel with a pore size too small to allow the enzyme to escape.This has been achieved with carbohydrate and polyacrylamide gels, both of which are essentially hydrophilic.9 If a hydrophobic container for the enzyme is required, the enzyme can be encapsulated within droplets of poly- merizing monomers and the resulting microcapsules containing the enzyme can be filtered from the polymerization suspension.ll Less permanent methods that are on the fringe of immobilization are liquid emulsions12 and aqueous- organic two-phase systems.*3 Finally, small but highly efficient immobilized enzyme reactors can be prepared by enclosing enzymes within semi-permeable devices ;bundles of hollow fibres have been especially ~uccessful.~~ In all of the physical methods of immobilization that enclose an enzyme within a small pored wall the size of the pores controls the transport of molecules to and from the enzyme’s active site.This effect may be detectable either as a change in the kinetic properties of an enzyme or, more dramatically, as an alteration in the substrate specificity of the enzyme. Since the chemist can determine the pore size, the opportunity exists to control the reactivity of the enzyme. B. Chemical Immobilization Methods.-There is usually an abundance of R. Goldman, L. Goldstein, and E. Katchalski, in ‘Biochemical Aspects of Reactions on Solid Supports’, ed. G. R. Stark, Academic Press, New York, 1971, pp. 4-5. lo K. C. Caldwell, R. Axen, M. Bergwall, and J. Porath, Biotechnol. Bioeng., 1976, 18, 1573,1589.‘Microencapsulation’, ed. J. E. Vandegaer, Plenum Press, New York, 1974. l3 S. W. May and N. N. Li, in ‘Enzyme Engineering’, ed. E. K. Pye and L. B. Wingard, Plenum Press, New York, 1974, Vol. 2, p. 77. l3 A. Pollak and G. M. Whitesides, J. Amer. Chem. SOC.,1976,98,289. l4 P. R. Rony, J. Amer. Chem. SOC.,1972,94,8247. Immobilized Enzymes nucleophilic groups (amino-acid side-chains) on the surface of enzyme molecules. Using these groups as reactants, three distinct types of immobilization are possible.15 Firstly, an enzyme can be covalently bonded through a peripheral reactive group to a suitably functionalized polymer (Scheme 1). Polystyrenes are usually Polystyrene Partially hydrolysed nylon CNH3+co, -jii NC NHCOCHNEnz I H R Nylon $;BNH v OEt Metal oxide or glass Carbohydrate Carbohydrate Cl HNEnz Scheme 1 Reagents: i, EtOCH,Cl; ii, Enzyme-NH,; iii, RCHO-CN(CH,),NC; iv, RCHO-Enzyme- NH,; v, Et,O+BF,-; vi, H,N(CH,),NH,; vii, OCH(CH,),CHO; viii, Si(OR),Y (Y = CN, SH, efc.);ix, CNBr; x, cyanuric chloride activated by preparing chloromethylated derivatives following Merrifield.2 Metal oxides can be treated with silylating agents bearing reactive groups with which the enzyme can bond.16 Polyamides, after partial hydrolysis, can be reconstituted incorporating the enzyme into the polymer through amide b0nds.l' This method is suitable for large-scale uses of nylon, but on a smaller laboratory l5 R.D. Falb, ref. 12, p.67. l6 G. Baum and M. Lynne, Process. Biochem., 1975, 10, No. 3, p. 14. l7 L. Goldstein, A. Freeman, and M. Sokolovsky, ref. 12, p. 97. Suckling scale the activation of amides by conversion into an imino-ether has been used to prepare an automatic analyser for glucose.l* Recently, radiolysis of acrylic and alkene polymers has been used to graft enzymes to a supportlg and such immobilized systems are characterized by high resistance to microbial attack. Carbohydrates are usually activated with cyanogen bromide or cyanuric chloride.2*121For laboratory work, a number of commercial polymers with pendant reactive groups are available (agaroses from Pharmacia with amino or carboxylate functions, ‘Enzacryls’ from Koch Light with amino, hydrazino, thiol, or thiolactone functions).The second group of chemical techniques involves the incorporation of the enzyme by covalent bonding into a growing polymer chain. To obtain water- insoluble products it is usual to employ copolymers such as acrylamide or acrylic acid with maleic anhydride in which a degree of cross-linking occurs.22 Although stable preparations usually result, this method runs the risk of wasting a quantity of enzyme by confining it within recesses of the polymer that are inaccessible to substrate molecules. Thirdly, some enzymes can be immobilized by cross-linking the enzyme molecules to each other with polyfunctional reagents such as bis-diazonium salts, dialdehydes, or cyanuric chl0ride.~3 This technique is not generally appli- cable because the small cross-linking reagents may penetrate the enzyme’s active site and react there, causing inactivation of the enzyme.However, bifunctional reagents are more widely useful in stabilizing physically adsorbed enzymes by cross-linking.Z4 One of the chief drawbacks of all of these chemical methods for large-scale applications is the cost of the support material. Recently it has been suggested that the reversible chemical immobilization of an enzyme might offer a solution to this problem. Once an enzyme has lost its catalytic activity through prolonged use, it could be removed from the valuable support which is then recovered.25 Enzymes that have free, non-catalytic thiol groups have been successfully used to demonstrate this principle.The thiols can be oxidized to form disulphide bridges with pendant thiols on the support. When it is desired to strip the enzyme from the support, the disulphide bond can be cleaved with a suitable reagent such as another thiol. Just as physical immobilization techniques impose a barrier to substrates reaching the active site of the enzyme, so chemical methods modify the im- mediate environment of the enzyme molecule at the same time as restricting D. L. Morris, J. Campbell, and W. E. Hornby, Biochem. J., 1975, 147, 593. l9 J. L. Garnett, R. S. Kenyon, and M. 5. Liddy, J.C.S. Chem. Comm., 1974, 735; H. Maeda, H. Suzuki, A. Yamauchi, and A. Sakamane, Biotechnol. Bioeng., 1975, 17, 119. H. H. Weetall and C. C. Betar, Biotechnol.Bioeng., 1975, 17, 295. 21 G. Kay and M. D. Lilly, Biochim. Bioph-vs. Acta, 1970, 198,276. 22 D. Jaworek, ref. 12, p. 105. 23 R. Goldman, L. Goldstein, and E. Katchalski, in ‘Biochemical Aspects of Reactions on Solid Supports,’ ed. G. R. Stark, Academic Press, New York, 1971, p. 22. zg M. Charles, R. W. Coughlin, E. K. Paruchuci, B. R. Allan, and F. X.Hasselberger, Biotechnol. Bioeng., 1975, 17, 203. 25 J. Carlsson, R. Axen, and T. Unge, European J. Biochem., 1975,59, 567 219 Immobilized Enzymes access to the enzyme. Through both of these mechanisms the chemist gains a measure of control over the enzyme’s reactivity and this is a major stimulus for research into the modification of properties that occurs when an enzyme is immobilized.3 The Effect of Immobilization upon the Properties of the Enzyme A. Stability.-To achieve a successful application of an immobilized enzyme it is of great importance to know under what conditions the preparation can be stored and used. The immobilization techniques outlined above cause minimal changes in the conformation of the enzyme molecule and consequently any improvements in the apparent robustness of the enzyme will be functions of the protection that its new environment offers. These are, however, only broad generalizations and each case must be treated on its merits. Storage stability is usually at least as good as for soluble enzymes under the same conditions of pH, ionic strength, and temperature.26 At best, greater than 60% of initial activity can be retained over several years’ storage.It appears that physically immobilized enzymes fare better than chemically treated ones and there are a few examples of stable preparations that have used a combina- tion of physical and chemical techniques. This is most effective when a pre- adsorbed enzyme can be clipped into its active conformation by a cross-linking agent.24~27 A wider range of operating conditions can be expected to be available with immobilized enzymes compared with their soluble reIatives, including some unusual solvents for enzymes. Polar immobilization matrices can protect enzymes against denaturants such as urea and guanidinium ions and it has also been possible to conduct reactions in 76% ethanol-water solutions with the enzyme trypsin, a peptidase, attached to gIass.28 Most matrices offer some protec- tion against microbial degradation of the bonded enzyme.However, the thermal stability of enzymes is rarely enhanced on immobilization and it is consequently not reasonable to expect an enzyme-catalysed reaction to be accelerated by heat. Enzymes in general have a fairly sharply defined pH for optimal catalytic activity. Immobilization sometimes broadens the optimal range and shifts it to a different pH,29 but the contrary has also been observed.30 B. Kinetic Behaviour and Catalytic Activity.-Apart from stability, the potential user of an immobilized enzyme will wish to know what modifications to the catalytic activity and specificity of the enzyme are likely consequences of immobilization. The investigation of these factors requires a study of the kinetic properties of the immobilized enzyme under the conditions of intended use.J. J. Marshall and M. K. Rabinowitz, Biotechnol. Bioeng., 1976, 18, 3, 9.*’ 0.R. Zaborsky, ref. 12, p. 115. 28 H. H. Weetall and W. P. Wann, Biotechnol. Bioeng., 1976, 18, 105. 29 L. Goldstein, M. Pecht, S. Blumberg, D. Atlas, and Y.Levin, Biochemistry, 1970, 9, 2322. 3u L. Goldstein and E. Katchalski, Z. analyt. Chem., 1968, 243, 375. 220 Suckling Enzyme kinetics31 centre upon the Michaelis-Menten equation which relates the rate of an enzyme-catalysed reaction to the substrate concentration, [S], the maximum velocity of the reaction when the enzyme is saturated with substrate, V&x, and the Michaelis constant, Km: VmaxRate = 1 + Km/[S] The Michaelis constant is a measure of the affinity of the enzyme for its substrate; a small Km indicates that the substrate is tightly bound by the enzyme.Km and are characteristic properties of an enzyme and both may be altered by the microenvironment of the enzyme’s supporting polymer. Catalytic activity and specificity are usually measured in terms of relative rates of reactions with different substrates. Many enzymes will tolerate foreign substrates and will transform them into products, usually at rates less than for the natural substrates. Absolute specificity is encountered only with respect to stereoisomeric compounds. Since immobilization can alter the rates at which substrates can react, chiefly by modifying Km, both the activity and the specificity of an immobilized enzyme may well differ markedly from those of the soluble enzyme.Sufficient is known about the physicochemical factors that affect the Michaelis constant for it to be possible to design an immobilized enzyme, an analytical membrane for instance, with the specificity that the chemist desires. Once immobilized, an enzyme can be considered to be in a separate phase from the bulk solvent: by making this factorization it is possible to develop a theoretical explanation for the changes in kinetic properties that accompany immobilization.32 Consider an enzyme immobilized within an anionic matrix, such as produced by carboxylate ions in a maleic anhydride copolymer.Protons in the reaction medium will tend to drift towards the matrix to annul its negative charge. As a result, the enzyme on the matrix will experience a pH lower than the surrounding medium and, in order that the ionizable groups on the enzyme’s active site retain their correct ionization state for catalysis, the pH of the bulk phase must be raised to compensate for the ‘protonation’ of the matrix and the enzyme. This phenomenon is observed as a shift of the pH optimum to higher pH for an enzyme in an anionic environment. A polycationic resin will, of course, shift the pH optimum to lower pH. The pHs prevailing in each domain can be measured and from them an electrostatic potential of 56-150mV between the two phases can be calculated, a value in good agreement with theoretical calculations.33 Related effects can be illustrated quantitatively by measuring the Michaelis constant for the immobilized enzyme.Intuitively it is obvious that Km will be reduced if the microenvironment due to the support and the substrate are oppositely charged, and vice versa. For example, ATP-creatinine phospho- 31 H. R. Mahler and E. H. Cordes, ‘Biological Chemistry’, 2nd Edn., Harper and Row, New York, 1971, p. 275. 32 L. Goldstein, Y. Levin, and E. Katchalski, Biochemistry, 1964,3,1913. 33 R.Goldman, L. Goldstein, and E. Katchalski, ref. 23,p. 34. Immobilized Enzymes transferase catalyses the phosphorylation of creatinine by the polyanionic molecule adenosine triphosphate (ATP; Scheme 2).34 The non-immobilized 0 0 0 0 II II It -0-P-0-P-OCH N I I HO 0-110 HO OH HO OH Adenosine triphosphate ------+ 0-I I MeN -CH,CO,-Crealine H,N+ NH, HPN+ NC/ \\,/""' I Iiq H NHI I (C11*)3 (CH33I I PhCONH --C -CO,Et PhCONHC -COP-Scheme 2 enzyme has Km = 6.5 x mol 1-1 but the enzyme immobilized on carboxy- methyl cellulose, an anionic polymer, had the less strongly bound value of 7 x rnol 1-l.In contrast, the peptidase ficin hydrolysed benzoylarginine ethyl ester, a cationic substrate, with Km = 2 x mol 1-1 in the soluble state but Km = 2 x lop3mol 1-1 when immobilized on carboxymethyl cellulose. To emphasize that there is a difference between the kinetic constants for free and immobilized enzyme, the Michaelis constant for the latter is written Kin'.Since the immobilization of an enzyme upon a charged support introduces an electric field effect into the enzyme's kinetic behaviour, it has been suggested that the application of an external electric field could generate useful properties. A theory of these effects has been propounded.35 A further distortion of kinetic parameters is caused by diffusion limitations in immobilized preparations. An insoluble enzyme particle in solution is immediately surrounded by a layer of solvent that is unstirred. As the substrate is consumed a4 W. E. Hornby, M.D. Lilly, and E. M. Crook, Biochem. J., 1968,107,669. 36 Y.Set0 and S.T. Hsieh, Biotechnol.Bioeng., 1976, 18,813. 222 Suckling within the immobilization particle during reaction, a substrate concentration gradient will be set up leaving the immediate environment of the enzyme depleted in substrate with respect to the bulk solution. Consequently more substrate will be required in order to saturate the enzyme’s active sites and hence to attain maximal rates of reaction. In other words, Km’ for the immobilized enzyme will be larger than Km for the soluble enzyme. The obvious expedients for circum- venting this difficulty have proved successful. Smaller immobilized particles allow K,’ to approach Km more nearly,23 as does increasing the stirring rate of the reactor. In the limit, the best that can be done is to immobilize the enzyme upon a soluble support such as a short-chain polysaccharide in aqueous solution.Thus chymotrypsin, a peptidase, has Km = 3.3 x 10-3 mol 1-1 for the non- natural substrate acetyltyrosine ethyl ester; the enzyme immobilized upon a water-insoluble Sephadex showed Kn,‘ = 3 x 10-2 moll-1 whereas an analogous water-soluble derivative had Knl’ = 3 x 10-3 mol l-1.36 Extending this notion, the more soluble the enzyme support in the reaction medium, the more reactive the preparation is likely to be, but at the cost of the loss of the convenience of an insoluble preparation in recovery of the enzyme. A general theory of both diffusion and electrostatic effects on immobilized enzymes has been put forward by Lilly.34 4 Applications of Immobilized Enzymes A characteristic that distinguishes two broad classes of applications of im- mobilized enzymes is the contribution that further technological research and development has made to the realization of the application.On the one hand, for mechanistic studies of enzyme catalysis, for research into membrane proper- ties, and for laboratory synthesis, a bottle of immobilized enzyme, a beaker, and a stirrer will usually be adequate hardware. On the other hand, industrial, analytical, and medical applications require much research into the whole system that surrounds the immobilized enzyme. It is perhaps as much a reflection of the technological problems to be solved as an indication of the intense current interest in applications of immobilized enzymes that publications of industrial, analytical, or medical significance appear more frequently than those dealing with the more academic aspects of the field.A. Biochemical and Bio-organic Research.-Immobilizat ion of enzymes opens up new methods for the study of chemical interactions within proteins: inter- actions between sections of a polypeptide chain give rise to the biologically active forms, the tertiary structure, or folded protein chain, and to the quaternary structure (Le. the agglomeration of completely folded chains or subunits). By immobilization upon an inert support, a protein chain can be prevented from interacting with other proteins in solution and a single subunit can then be studied in complete isolation. A postulate of protein chemistry is that a given sequence of amino-acids will 38 R.Axen, P.A. Myrin, and J. C. Janson, Biopolymers, 1970, 9,401 ImmobilizedEnzymes spontaneously fold into a specific tertiary structure which, for an enzyme, will be the catalytically active structure. It has been possible to verify this postulate for simple, single subunit enzymes but the complications caused by subunit agglomeration have made studies of multicomponent enzymes difficult. Lactate dehydrogenase, an enzyme that has four subunits, has been immobilized on glass and it was found that after mild denaturation of the enzyme, causing the tertiary and quaternary structure to be disrupted, activity was recoverable on removal of the denaturing agent.37 This result confirms that careful immobiliza- tion does not interfere with the folding of polypeptide chains, at least when glass is the support.38 Aldolase is another tetrameric enzyme and it catalyses the reversible cleavage of fructose 1,6-diphosphate into dihydroxyacetone phosphate and glycer-aldehyde 3-phosphate.By means of Sepharose activated to only a small degree, it was possible to immobilize the whole enzyme globule intact and subsequently to split off most of the subunits, leaving immobilized only those few single polypeptide chains that were covalently bonded to the support. The polymer held these immobilized subunits sufficiently far away from each other to prevent interaction between them, and the properties of the individual subunits could then be studied.Interestingly the single subunits showed some enzymic activity and, as with lactate dehydrogenase, reconstitution of the tetrameric enzyme restored almost full catalytic activity.39 In the Introduction it was pointed out that many functional units in biology consist of groups of enzymes embedded in membranes, one example being the mitochondria1 enzymes of the tricarboxylic acid cycle. To understand the be-haviour of such enzyme systems it is valuable to investigate the ways in which smaller groups of enzymes function when acting in concert. Synthetically im- mobilized enzymes offer an opportunity to study such systems, and Mosbach has immobilized malate dehydrogenase, citrate synthase, and lactate dehydro- genase together and studied their behaviour as a model for oxalacetate produc- tion in mitochondria.40 His immobilized trio was eight times more active than the three soluble enzymes interacting randomly in solution (Scheme 3).B. Synthetic Organic Chemistry.-Enzymes have potential as catalysts for synthetic operations but, biosynthetic studies apart, little has yet been a~hieved.~l In natural product chemistry it may be expected that an immobilized enzyme will accept many substrates structurally related to its natural substrates. With steroids especially, enzymes from micro-organisms have wide specificities and it has been possible to use columns of immobilized fungal cells having 11-/3- hydroxylase and 1 ,Zdehydrogenase activity in syntheses of cortisol and pred- nisolone (Scheme 4).42 37 H.E. Swaisgood and 1. C. Cho, Riochim. Biophys. Acta, 1972, 258,675. :H H. R. Horton and H. E. Swaisgood, ref. 12, p. 169. 39 W. W.-C. Chan, Biochem. BiophyA. Res. Comm., 1970, 41, 1198. P. A. Srere, B. Mattiason, and K. Mosbach, Proc. Nat. Atad. Sci., U.S.A., 1973,70, 2534. C. J. Suckling and K. E. Suckling, Chem. SOC.Rev., 1974,3,387. Az K. Mosbach and P. M. Larsson, Biotechnol. Bioeng., 1970, 12, 19. 224 Suckling CH,COSCo A CHSCO$ICHZCOZH malate CHzC02H 1 dehydrogenax -1 I H-C-OH CO2H NAD+77NADHZHO lactate dehydrogenase I COLH Scheme 3 00 II aH II 1 I/?-hydroxylase z-0 Cortisol dehydrogenase1 Prednisolone Scheme 4 ImmobilizedEnzymes Many enzymes of synthetic potential, such as alcohol dehydrogenase, require coenzymes in addition to substrate in order to function.It is often difficult to regenerate the consumed coenzyme with chemical reagents and immobilized enzymes thus become obvious choices for the task. The problem of regenerating the redox coenzyme nicotinamide adenine dinucleotide has been solved for small-scale work but remains a difficulty when larger-scale operations are contemplated. However, the synthesis of adenosine triphosphate from adenosine monophopshate and chemically prepared acetyl phosphate by means of a two- enzyme coupled system itnmobilized on Sephadex or polyacrylamide has been successfully developed to the verge of being a practical pr0cess.~3 C.Analytical Applications.-Enzymes have for many years been used in the routine assay of metabolites in biological fluids but each analysis necessarily consumes a quantity of soluble enzyme. An obvious improvement is to employ a re-usable immobilized enzyme preparation. Compounds such as glucose, lactic acid, urea, and ethanol are particularly amenable to this approach because suitable specific enzymes are readily available (Scheme 5). The concentrations of HH -2NH,+H,NCON€I, uteasc + CO, Scheme 5 43 C. R. Gardner, C. K. Colton, R. S. Langer, B. K. Hamilton, M. C. Archer, and G. M. Whitesides, ref. 12, p. 209; G. M. Whitesides, A. Chmurny, P. Garrett, A. Lamotte, and C. K. Colton, ibid., p. 217. 226 Suckling metabolites present in the sample are determined by measurement of the change in concentration of either a co-reactant, such as NADH in dehydrogenase- catalysed reactions,44 or a product such as amonium ions from the hydrolysis of urea by urease.45 Spectrophotometry and potentiometry are the respective analytical techniques.However, for those enzymes that consume oxygen as a reactant, the oxygen concentration is best followed by means of an oxygen electrode46 and it is a short step to combine the enzyme with the electrode membrane to produce the so-called enzyme electrode.45~46 An enzyme electrode consists of a glass electrode surrounded by a membrane that contains the immobilized enzyme. It is sometimes sufficient simply to parcel the enzyme up between the glass and a semipermeable membrane such zu cellophane, but more usually the enzyme is covalently bound to the membrane. With such a system, three modes of analytical selectivity are available. Firstly, there is the enzyme’s intrinsic selectivity, secondly, the glass electrode can be chosen to respond only to certain ions,47 and thirdly, the choice of the membrane (anionic, cationic, hydrophobic, etc.) offers selectivity. Many applications of enzyme electrodes are built into automated systems, e.g.glucose oxidase with an oxygen electrode.46 Similarly, amino-acid oxidases, which catalyse the oxi- dative deamination of amino-acids to keto-acids (Scheme 5) can be used in conjunction with a cation-selective electrode to permit rapid and continuous assay for amino-a~ids.~~ A more detailed discussion of these techniques, in- cluding multiemyme assays can be found in a review by Guilbalt.49 Automated amino-acid analysis is a well established technique and im- mobilkd enzymes make it possible to extend automation to the peptide degradation processes.In peptide analysis, the chief advantage of enzyme- catalysed degradation is that amino-acid side-chains such as the alcohol functions of serine and threonine and the indole ring of tryptophan, which are decomposed by the normal acidic hydrolysis conditions, are quite stable to enzymes. Studies of immobilized derivatives of the broad-spectrum peptidase, pronase and of other more selective peptidases show that enzymic degradation has the potential for development into fully automatic anal~sers.~~ Research has also begun into the use of immobilized phosphatases for the hydrolysis and sequencing of nucleic acids.51 It should not be thought that the scope of analysis using immobilized enzymes is limited to organic compounds. Hydrogen peroxide is a product of some of the oxidation reactions in Scheme 5 and it can readily be determined using im- mobilized ~eroxidase.~~ Further, some enzymes have obligatory requirements 44 M.K. Weibel, ref. 12, p. 385. 4b G. G. Guilbalt and J. G. Montalvo, J. Amer. Chem. SOC.,1969,91,2164. 46 S. J. Updike and G. P. Hicks, Science, 1967,158,270; Nature, 1967,214,986. 47 R.A. Durst, Amer. Scientist, 1971,59,353. 48 G. G. Guilbalt and E. Hrabankova, Anal.Chim.Ada, 1971,56,285. 48 G. G. Guilbalt, Enzymology, 1975,1,293. G. P. Royer and G. M. Green, Biochem. Biophys. Res. Comm., 1971,44,426. 51 R. A. Zingaro and M. Uziel, Biochim. Biuphys. Acra, 1970,213, 371. ba B. F.Rocks, Proc. Soc. Analyt. Chem., 1973, 10, 164. 227 Immobilized Enzymes for certain metal ions. Tyrosinase, which catalyses phenol oxidation, requires copper(xx) ions and these ions can be removed from the immobilized enzyme which becomes inactive. If the enzyme is then immersed in a solution containing an unknown concentration of copper, the extent of reactivation of the enzyme gives a measure of the copper concentration. Nanogram quantities are reported to be detectable.58 An assay for zinc can be set up similarly using immobilized alkaline ph~sphatase.~~ A recent development that is causing much interest in the biomedical field is the use of immobilized enzymes in immunoassay techniques as an adjunct to radioimm~noassay.5~An antibody to an enzyme can be rapidly purified by means of an immobilized enzyme: if the system is then reversed and the purified antibody itself is immobilized, a material that will absorb the target enzyme with high selectivity is produced and this facilitates the determination of the concentration of the enzyme in a sample.In this survey of analytical methods, little has been said about the type of immobilization technique best suited to the application. Much analytical work requires enzyme membranes and, accordingly, either covalent attachment to or entrainment within a polymer is suitable. On the other hand, a continuous analyser involving degradative enzymes would need enzymes adsorbed or bonded to materials suitable for use in columns such as glasses or polymer beads.The variety of approaches being pursued is well illustrated by contributions to the 1973 Symposium on Enzyme Engineering.56 D. Medical Applications.-The idea of using enzymes as drugs is not new; protein- and carbohydrate-hydrolysingenzymes, for instance, are widely used as digestive aids, especially in Japan. However, the use of immobilized enzymes in medicine has wider therapeutic implications which can be regarded chiefly as either the removal of toxic substances or the synthesis of metabolites that are in deficiency.j7*5E It has been suggested, for example, that urease could be used to control urea levels and that phenylalanine ammonia lyase could help alleviate the problem of phenylketonuria.The attraction of using an immobilized enzyme in place of a soluble one in medicine is four-fold. Firstly, hydrolysis of an enzyme drug by digestive processes can be avoided; secondly, the elimination of the enzyme by filtration through the kidneys is prevented; thirdly, immunological rejection of the enzyme drug can be controlled; and finally, the drug can be localized into its required site of action. These considerations apply to the use of immobilized enzymes in invasive therapy, and clinicians are rightly slow to adopt new invasive methods O3 J.V. Stone and A. Townshend, J.C.S. Chem. Comm., 1972, 502. 64 A. Townshend and A. Vaughan, Talanta, 1970,17,289. 65 D. L. Eshenbaugh and E. James, in ‘Immobilized Biochemicals and Mnity Chromato- graphy’, ed. R. B. Dunlap, Plenum Press, New York, 1974, p. 61. 56 ‘Enzyme Engineering’, ed. E. K. Pye and L. B. Wingard, Plenum Press, New York, Vol. 2, 1974. 67 G. Brown, ref. 12, p. 433. c8 T. M. S. Chang, Enzymology, 1975,1,245. 228 Suckling until both their efficiency and safety have been rigorously established. There are also practical difficulties to surmount. The effectiveness of any treatment using an immobilized enzyme will depend upon the ability of the required substrates and cofactors to reach the enzyme and upon the active lifetime of the enzyme.Solutions to these problems have still to be found but the following example illustrates what may become possible in the future. The control of urea levels in blood has been studied by Chang using im- mobilized preparations of urease.59 Both the implantation of immobilized enzyme particles and the construction of extracorporeal shunt devices have been investigated. Urease is especially suitable for pioneering studies because it catalyses a simple hydrolysis reaction without coenzyme requirements. It has been estimated that an artificial kidney as small as 10 cm long x 2 cm diameter is feasible using these techniques. Apart from the treatment of patients, immobilized enzymes have been shown to be useful in clearing air in rooms from virus particles.In order to reproduce in the host and cause disease, viruses depend upon the translation of their nucleic acids by the infected host cells. If enzymes that specifically hydrolyse the viral nucleic acids can be immobilized, the virus will be killed upon contact with the enzyme. Porous glass and ceramic materials have been used as supports for such enzymes and the materials have been shown to disinfect airborne herpes simplex, influenza, and cocksackie viruses rapidly.60 E. Industrial Applications.-The idea of using enzymes in large-scale processes goes back to the early twentieth century when Rohm introduced hydrolytic enzymes into laundering.61 Today, a wide range of processes falls within the scope of industrial applications of enzymes, but, whereas applications to research, analysis, and medicine require only small amounts of enzyme that can easily be obtained from small batches of a natural source, a large-tonnage application would necessitate a substantial investment in the production of the enzyme itself.In addition there are toxicological problems associated with the large quantities of biological waste that would be produced62 and there is also the high cost of enzyme immobilization itself, with respect to both the support material and the immobilization reagent. Together these factors make it im- probable that any of the large-tonnage petrochemical plants, which are highly efficient, will be superseded by plants based upon enzyme technology in the immediate future. However, in those industries where natural products are synthesized either for food or for pharmaceutical purposes, the application of immobilized enzymes is now having a major impact.(i) Reactor Design.Once the immobilization and isolation problems are solved, there remains the technological problem of the form that the catalyst should 59 T. M. S. Chang, Sci. J., 1967,3,62; T.M.S. Chang, L. J. Johnson, and 0.J. Ransome, Canad. J. Physiol. Pharntacol., 1967,45,705. 6o J. Enright, J. Gainer, and D. J. Kirwan, Environ. Sci.Technol., 1975,9,586. 61 0. Rohm, Ger. P. 283 923. 62 L. D.Scheel, D. E. Richards, V. B. Perone, and W. B. Tolos, ref. 12, p. 351. ImmobilizedEnzymes take and the type of reactor that is appropriate.63-66 Two chief factors must be considered.Firstly, the enzyme must be readily removable from the reaction mixture and, secondly, the kinetics of the immobilized enzyme system must be well established so that the optimum contact time for the reactants can be defined. As we shall see, these factors interact. Ready removal of the enzyme can be simply accomplished by packing a column with the particles bearing the enzyme and passing a solution of the reactants through at a rate such that only products are eluted from the column. To determine the correct flow rate, the kinetic parameters for the immobilized enzyme system must be known. It has been deduced theoretically for a column reactor that the conversion into product depends upon the height of and the flow rate through the column, provided that the enzyme is saturated by an excess of substrate. The saturation concentration of substrate depends upon K,’, and it has been shown that the theoretical principles are born out well by experiments with immobilized hydrolytic enzyme~.67-~~ The flow rate will further be influenced by the material upon which the enzyme is immobilized.For this reason, rigid supports such as metal oxides are often preferred to organic polymers, provided that stable bonding between the enzyme and the metal oxide can be obtained. Apart from columns, fluidized-bed, packed-bed, and stirred-tank reactors have been investigated for use with enzyme~,~O the choice usually depending on the reaction in question.In stirred reactors, K,’ varies with the degree of agita- tion, indicating the importance of substrate diffusion effects as was noted earlier. Many enzyme-catalysed reactions are slow compared with conventional chemical processes and in these cases batch or plug flow processes are preferable. Organic supports have not been notably successful in such reactors because a sufficiently high degree of agitation to overcome diffusion limitations cannot be attained. Consequently there has been a widespread interest in the use of oxides of ferro- magnetic metals as supports; the reactants can then be easily separated from the support by restraining the latter in a magnetic field. So far, these ideas have not been developed beyond initial research experiment^.^^^^^ (ii) Processes in Operation. The Japanese are the undoubted leaders in this field and the production of optically pure amino-acids by immobilized enzyme- catalysed hydrolysis of their acylated derivatives has been an established process in Japan for many years.73 The reported capacity of these processes is substantial, 63 H. H.Weetall, ref. 55, p. 191. 64 W. H. Pitcher, jun. and N. B. Havewala, Enzymology, 1975, 1, 83. O5 M. D. Lilly and P. Dunnill, Process. Biochem., 1971, 6, No. 8, p. 29. Re A. M. Filbert and W. H. Pitcher, Process. Biochem., 1976, 11, No. 7, p. 3. 67 R. Goldman, L. Goldstein, and E. Katchalski, ref. 23, p. 39. 68 A. BarEli and E. Katchalski, J. Biol. Chem., 1963, 238, 1690. 69 M.D. Lilly, W. E. Hornby, and E. M. Crook, BioLhem.J., 1966,100,718. 7o M. D. Lilly, G. Kay, A. K. Sharp, and R. J. H. Wilson, Biochem. J., 1968, 107,5p. M. Charles, R. W. Coughlin, B. R. Allen, E. K. Paruchi, and F. X. Hasselberger, ref. 55, p. 213. 72 G. Gellf and J. Boudrant, Biochim. Biophys. Ata, 1974, 334, 467. 73 T. Tosa, T. Mori, N. Fuse, and I. Chibata, Biotechnol. Bioeng., 1967,9, 603. 230 Suckling greater than 700 kg day-1.74 Indeed amino-acid production from cheap1 feed- stocks is a continuing area of interest for immobilized-enzyme technologists. For example, a large number of amino-acids can be prepared by enzyme- catalysed transaminations in the presence of the coenzyme pyridoxal phosphate (Scheme 6). Tryptophan and tyrosine have been manufactured by means of an Ar i IMeCOC0,H + NH4++ ArH + f H,O/““YC0,-HsN+ Scheme 6 Reagent:i, tyrosine phenol lyase-pyridoxal phosphate aryl lyase enzyme immobilized on columns from pyruvic acid, ammonia, and either indole or phenol as starting materiak75 The process also works for hydroxy-substituted indoles and phenols to afford the pharmaceutically im- portant compounds L-DOPAand serotonin.Under optimum conditions, 90% of aromatic substrate is converted into product. The neutral support used, Sepharose, scarcely alters the kinetic properties of the enzyme although the pH optimum is slightly higher for the immobilized system. The chief expense in such a system is the coenzyme, which here was used as a component of the reaction buffer solution.As we have seen, the problem of coenzyme recovery and regeneration is a severe current limitation of immobilized-enzyme technology, especially where the redox coenzyme, NAD, or the phosphorylating coenzyme, ATP, is involved. The problems of ATP regeneration are being tackled successfully using coupled enzyme systems (see Section 4B)and much effort is currently being expended to find a similar system for NAD.76977Without a solution to this problem, the large-scale use of enzymes catalysing such reactions as the oxidation and reduc- tion of aldehydes, ketones, and alcohols cannot become practicable although the enzymes are readily available. An interesting immobilized adaptation of established enzyme-catalysed reactions in the pharmaceutical industry is the production of 6-aminopenicillanic 74 T.Sato, T. Mori, T. Tosa, and I. Chibata, Arch. Biochem. Biophys., 1971,147,788. 7b S. Fukui and S. Ikeda, Process. Biochem., 1975,10, No. 6, p. 3. 76 R. W. Coughlin, M.Aisiwa, and M. Charles, Biotechnol. Bioeng., 1975, 17, 209; R. W. Coughlin, M. Aiziwa, M. Charles, and B. F. Alexander, ibid., p. 515. 77 J. R. Wykes, P. Dunnill, and M.D. Lilly, Biotechnol. Bioeng., 1975,17,51; R. P. Chambers, J. R. Ford, J. H. Allender, and W. Cohen, ref. 12, p. 195. 231 Immobilized Enzymes acid, an important intermediate in the production of semi-synthetic penicillins (Scheme 7).78The process uses either carbohydrate or maleic anhydride-methyl vinyl ether polymers and yields of over 90 % are claimed in a continuous process.Similar results have been described for the related antibiotics the cephalo- sp~rins.~g Scheme 7 The food processing industry is a further major area of application of im- mobilized-enzyme technology; crude enzymes have been used in beer, bread, and cheese production for many years.80s81 A case that hit the chemical head- lines recently was the establishment of a plant to isomerize the cheap but tasteless sugar glucose into its toothsome isomer fructose. The catalyst consists of compounded particles of the whole bacterium that contains the enzyme glucose isomerase, and a column reactor is used. In this way, the expensive isolation and immobilization steps are replaced by the cheaper and simpler operations of growing and harvesting the micro-organism and compressing it into catalyst pellets.82 The process is attractive because its feedstock, corn syrup, is currently cheap.For those interested in the development of glucose isomerase technology from enzyme isolation to reactor design, an interesting case study has been published.83 Cheese making has traditionally used rennet, an enzyme preparation from the stomach mucosa of unweaned calves, to coagulate the milk, but a rise in cheese production has brought about a world shortage of rennet.B4 If an immobilized enzyme that catalyses the same reactions can be obtained from a cheap microbial source and immobilized, the establishment of an economic process seems probable.85 Development work has shown that milk can be coagulated con- tinuously in a fluidized-bed reactor86 and it has been estimated that immobilized enzymes could offer up to an eighty-fold saving in the cost of rennins.87 Dairy processing is likely to prove to be a rapid growth area of enzyme technology.78 M. A. Cawthorne, Ger. P. 2 356 63011974 (Chem. Abs., 1975, $2, 15 244). 79 T. Fuji, K. Matsumoto, and T. Watanake, Process. Biochem., 1976, 11, No. 8, p. 21. 80 'Immobilized Enzymes in Food and Microbial Processes', ed. A. C. Olson and C. L. Cooney, Plenum Press, New York, 1973. 81 H. H. Weetall, Process. Biochem., 1975, 10, No. 6, p. 3. a* J. L. Meers, Chem. in Britain, 1976, 115. 83 B. K. Hamilton, C. K. Colton, and C. L. Cooney, ref. 80, p.85. 64 J. L. Sardinas, Process. Biochem., 1976. 11, No. 4, p. 10. 85 T. F. Richardson and N. F. Olson, ref. 80, p. 19. 86 M. Cheryan, P. J. van Wyk, N. F. Olson, and T. F. Richardson, Biorechnol. Bioeng., 1975, 17,585. 87 B. Wolnak, ref. 12, p. 369. Suckling Chemists are also pursuing the long-term potential application of immobilized enzymes to the production of food from metabolic wastes.*S 5 Conclusions The market for enzymes used for all purposes is growing continuously and in the U.S.A. in 1975reached a cash value of &25 n1illion.87 Very little of this turnover is accounted for by large-scale processes and most industrialists concerned with enzyme technology recognize the market potential of immobilized enzymes. However, they contrast the advantages of mild reaction conditions and selec- tivity with the drawback of the cost of immobilization. Costs may be cut, as in the glucose isomerase case, by using immobilized whole cells of bacteria and this technique may well prove to have wider applications to reactions in heavy organic chemical industry.Pye,89 in a thought-provoking article, is not confident about future develop- ments in industry, partly because of the coenzyme problem that has been discussed. In a more optimistic vein, Mosbach7 looks forward to the solution of this problem through coupled multienzyme processes and anticipates an era of ‘biochemical synthesis’ analogous to the era of natural product synthesis that has absorbed so much of the efforts of organic chemists over the years.Un- doubtedly, new applications of immobilized enzymes will be developed for analytical and medical purposes. However, major developments in the capital intensive areas of chemistry depend much upon forecasts of the economic situation twenty years from now, in particular in assessing what processes will become too costly to run. Even in special reports on biochemical engineering, little attention has been paid to the contribution that enzymes could make to energy produ~tion.~O While oil is available, immobilized-enzyme technology will not compete with established processes but its use to bridge a future resources gap is a possibility worthy of serious consideration. D. L. Marshall, ref. 55, p. 345. 8sE.K. Pye, ref. 80,p. 1. Do A.N. Emery, ‘Biochemical Engineering Survey’, Science. Research Council, 1976; E, M. Crook, Trends in Biochemical Sciences, 1976,1, N.195.

ISSN:0306-0012    

DOI:10.1039/CS9770600215

出版商:RSC    

年代:1977

数据来源: RSC

摘要:

The Polymerization and Copolymerization of Butadiene By D. H. Richards EXPLOSIVES RESEARCH AND DEVELOPMENT ESTABLISHMENT, MINISTRY OF DEFENCE, WALTHAM ABBEY, ESSEX EN9 1BP 1 Introduction The polymerization of butadiene, whether uniquely to form homopolymers or with other monomers to form copolymers, is a process of great industrial importance. Production of butadiene-containing elastomers in the Western World in 1975 exceeded 3.6 million tons, and constituted about 50% of all rubber and 70% of all synthetic rubber consumed. The properties of polybutadiene (PBD) depend on the way in which the mono- mer is incorporated into the chain. Butadiene can be polymerized by either 1,2- or 1,4-addition. If polymerization takes place to give the 1,2- or vinyl form exclusively, then two types of stereoregular sequences can arise; one in which the asymmetric carbon atoms have the same configuration [isotactic placement, (1) ] or one in which the configuration alternates [syndiotactic placement, (2)].Stereospecific polymerization in the 1,4-mode produces a cis-PBD (3) or a trans-PBD (4). The four distinct stereoregular homopolymers have been prepared I I ICH=CH, &H=CH% ~H=CH, CH=CH, H CH=CHaI I I 11\/" H\? H\/"C=C c=c c=c/\ 1-\ /-CH, CH,-CH, CH,-CH, kH2-H CH,-CH, H \/ \ /" CH,-CH,/"="\ Hf=7/ CH, -H H The Polymerization and Copolymerization of Butadiene and shown to have very different properties. The isotactic and syndiotactic PBD’s are rigid, crystalline and virtually insoluble materials, whereas cis-PBD is a soft, readily soluble elastomer with a glass transition temperature (Tg)* around -100°C.It has a high retractive force and crystallizes at extensions greater than 200% (cf. the isoprene analogue, hevea rubber). In contrast, trans- PBD is hard, difficultly soluble and crystallizes without elongation (cf.the isoprene analogue, gutta percha). PBD’s of mixed structure (atactic PBD’s) may be conveniently regarded as copolymers with different isomers being taken as individual component mono- mers. In this way the distribution of components along the chain can more easily be considered, and PBD’s may be envisaged with structures ranging from alternating copolymers to block copolymers. Indeed the ‘blockiness’ of PBD’s produced by various catalyst systems is currently a subject of considerable debate. The best elastomeric properties are exhibited by high cis-PBD and, initially, methods of preparing this material were sought.It was later found that the sole important structural requirement for a good cross-linked or vulcanized rubber was that it should have a low vinyl content. cis-trans Isomerization occurs under vulcanization conditions,l and the final properties of the product after correct compounding and vulcanization are insensitive to the initial cis-trans ratio. 2 Structural Analysis Present day methods of determining the structures of PBD’s are entirely based on spectroscopic techniques. The earliest and most used of these is infra-red (i.r.) spectroscopy, but nuclear magnetic resonance (n.m.r.) spectroscopy can now give a more accurate estimate of the molecular architecture of those polymers with low vinyl content.A. IIR Spectroscopy.-The trans and vinyl content of PBD’s may be readily determined from the bands at 970 cm-1 and 910 cm-1 respectively. The cis component, however, absorbs over a broad band near 715 cm-1 which varies unpredictably in wavelength with the sample and can overlap other peaks. This problem has been critically reviewed2 and a method has been recommended which involves integrating the absorbance between 830 cm-1 and 635 cm.-l A simpler method was later proposed3 which is satisfactory in most cases. *Tg is an important characteristic of a polymer system.Although not thermodynamic in origin, it indicates the onset of segmental motion of polymer chains and marks the temperature above which the physical properties of a material transform from those characteristic of glass to those of rubber. J. I. Cunneen, G. M. C. Higgins, and W. F. Watson, J. Polymer Sci., 1959, 41, 1. 2 R. S. Silas, J. Yates, and V. Thornton, Annlyt. Chem., 1959, 31, 529.* J. Haslam, H. A. Willis, and D. C. M. Squirrell, ‘Identification and Analysis of Plastics’, Iliffe Books, 1972. Richards B. Raman Spectroscopy.-Cornell and Koenig4 have shown that the Raman double bond stretching vibrations can be used to describe PBD’s. All three C=C vibrations are Raman active, and the stretching frequencies have been determined as 1650 cm-1 (cis), 1664 cm-1 (truns), 1655 cm-1 (vinyl) and 1639 cm-1 (isolated vinyl).Since the vibrations are all in the same spectral region, the authors assume the extinction coefficients to be in a constant ratio to each other, and they determine the structural composition from the peak areas. C. hiSpectroscopy.-The proportions of the three isomers may be determined by 1H n.m.r. spectroscopy, and both this technique and 13C n.m.r. spectroscopy may be employed to determine the isomer distribution along the chain. The success of the latter analysis relies upon the identification of resonances as being due to specific structural sequences. For example, in a cis-trans (c-t) PBD in which resonances are observed at frequencies characteristic of a triad structure, the intensities of the tcc, tct, ctt and ctc bands relative to the ccc and ttt bands should give a direct measure of the ‘blockiness’ of the sample.These resonances must therefore be accurately assigned and, unfortunately, controversy still exists in this area. Early low field 1Hn.m.r. studies were able to determine the relative proportions of vinyl and 1,4-linkages in PBD but could not distinguish between the cis and truns configurations5 (Table 1, column 3). More recently, Morton and co- workers6 working at 300 MHz and Hatada et nZ.7 working at 100 MHz showed that an analysis of cis and trans structure could be obtained in high 1,4-PBD’s. At 300 MHz both the olefinic and aliphatic resonances consist of a pair of broad peaks (Table 1, column 4) and, when the aliphatic peaks were irradiated, the olefinic resonances were each resolved into three peaks corresponding to triads of cis and trans units.However, when the olefinic peaks were irradiated, one group found no change in the aliphatic resonance,6 whilst the other obtained a triplet centred at 2.06 a7which was assigned to the three possible doublets (Table 1, column 5). This discrepancy may, however, be more apparent than real since the measurements were carried out in different solvents and at different temperatures. As the proportion of vinyl units increases, interference occurs in the olefinic region and it is recommended that the cis-trans content be determined in the aliphatic region. Even here analysis becomes difficult at more than 60% vinyl structure because of overlap with the methine proton of the vinyl group.The I3C n.m.r. spectra of PBD’s containing all possible linkages are extremely S. W. Cornell and J. L. Koenig, Rubber Chem. Technol., 1970, 43, 322. H. Y. Chen, Analyt. Chem., 1962, 34, 1134. 1793. 6E. R. Santee, jun., L. 0. Malotsky, and M. Morton, Rubber Chem. Technol., 1973, 46, 1156; E. R. Santee,jun., R. Chang, and M. Morton, J. Polymer Sci. B, 1973, 11, 449; E. R. Santee, jun., V. D. Hochel, and M. Morton, J. Polymer Sci.B, 1973, 11, 453. K. Hatada, Y.Tanaka, Y. Terawaki, and H. Okuda, Polymer J. Japan, 1973,5,327; K. Hatada, Y.Terawaki, H. Okuda, Y. Tanaka, and H. Sato, J. Polymer Sci. B, 1974, 12, 305. The Polymerization and Copolymerization of Butadiene Table 1 lH N.m.r.absorptions of polybutadiene Structure Hydrogen Low High High Field Location Field Field Spin Decoupled 5.307 (tct)b 5.324 (CCC)~ 5.3a 5.356 (ttt)b 1,&CHeCH=CH-CH2-P a lp 2.0a 2.06 (ct)” a r fa 5.3a 1,2-CH-CH-I CH=CH2 1.3a UP =deuterochloroform solvent, tetramethylsiloxane reference, T =ambient, from ref. 5. b = o-dichlorobenzene solvent, hexamethyldisiloxane reference, T = 110 “C, from ref. 6. e = deuterochloroform solvent, tetramethylsiloxane reference, T = 35 “C, 45 “C, from ref. 7. complicated and it is necessary to consider the various stereospecific types separately. The spectra of 1,4-PBD’s reveal relatively little detailed information even at 67.88 MHz; the aliphatic carbons give two well separated peaks at about 28 and 33 p.p.m.relative to tetramethylsilane (TMS) due to cis and trans linkages respectively,* and even at high fields show no further splitting due to sequence distrib~tion.~Only slightly more information is obtained from the olefinic region; four peaks are observed at 130.10, 130.25, 130.75 and 130.85 p.p.m. relative to TMS8 which are assigned to tc*, cc*, tt* and ct* diad resonances (the asterisk denotes the unit containing the observed carbon). Elgert et al., examining vinyl PBD’s at 67.88 MHz,loa interpreted the olefinic methylene peak at 113.4 p.p.m. and the methine peak at 142.5 p.p.m. in terms of pentad sequences arising from the different stereochemical arrangements of the vinyl units.They subsequently studied cis-vinyl PBDlOb and found these absor- bances complicated by 1,2-~onfigurational and compositional 1,2-1,4 sequence effects, but the cis region at about 129 p.p.m. was intepreted in terms of 1,2-1,4 triad sequences. Similar interpretations have been given by other authors,ll 8 F. Conti, A. SegrB, P. Phi, and L. Porri, Polymer, 1974, 15, 5. 9 K. F. Elgert, B. Stutzel, P. Frenzel, H-J. Cantow, and R. Streck, Makromol. Chem., 1973, 170, 357. 10 (a) K. F. Elgert, G. Quack, and B. Stutzel, Makromol. Chem., 1974, 175, 1955; (b) idem., Polymer 1974, 15, 613. 11 J. Furukawa, E. Kobayashi, N. Katsuki, and T. Kawagoe, Makromol. Chem., 1974, 175, 237. Richardr although Conti et a1.,12a analysing the aliphatic absorbances of equibinary cis-vinyl, 15% cis-85% vinyl, and 6 % trans-94% vinyl PBD’s gave some assign- ments which differ. Elgert et a1.12b also analysed mixed structured PBD’s prepared anionically and assigned the 1,4-olefinic carbon resonances in terms of triad sequences. A similar interpretation was made by Clague et al.13 who also demonstrated that further useful information could be obtained from hydrogenated material, as this simplified the spectrum.3 Emulsion Polymerization Butadiene is not easily polymerized to high molecular weights by conventional free-radical initiators in solution or in bulk, although the polymeric material has the desired high 14 structure (N 80%). The products are mixtures of a low molecular weight soluble fraction (sol) and a cross-linked gel, the proportions of which depend on the reaction conditions.This behaviour is caused by a low velocity of propagation (N 8 I mol-1 s-914 competing with a very fast termination step (N los 1 mol-ls-l)l5 and a rapid cross-linking reaction (N8 x 1mol-l s-l),16 and results in an excessive degree of branching under high conversion conditions. For these reasons, the homogeneous free radical polymerization of butadiene is now used commercially, principally to prepare low molecular weight ( < 5000) functionally terminated liquid polymers; even in these systems the conversion is kept very low (N40 %) and with only a minimal amount of solvent added. The slow growth rates and short lifetimes of butadienyl radicals mean that the rate of termination must be reduced to obtain a high molecular weight product, and this is best effected by confining the propagating radicals in the micelles of an emulsion.These emulsion processes are of great industrial importance, particu- larly for the copolymerization of butadiene with styrene or with acrylonitrile to form styrene-butadiene rubber (SBR) or nitrile rubber. An excellent review of the subject has appeared recently,17 and the salient features of the process as applied to butadiene will be outlined here. An aqueous dispersion of the monomer is made using an emulsifying agent or soap, which consists of droplets of about 1 pm in diameter stabilized by a monolayer of soap, and soap micelles of about 75 8, diameter which contain occluded monomer.The monomer is primarily stored in the droplets, but is in I* (a)F. Conti, M. Delfini, A. L. Segrk, D. Phi, and L. Porri, Polymer, 1974,15,816; (b) K. F. Elgert, G. Quack, and B. Stutzel, Polymer, 1975, 16, 154. l3A. D. H. Clague, J. A. M. von Broekhaven, and L. P. Blaauw, Macromolecules, 1974, 7, 348. l’ M. Morton, P. P. Salatiello, and H. Landfield, J. Polymer Sci., 1952, 8, 215. l6 W. Cooper and G. Vaughan, ‘Progress in Polymer Science’, ed. A. D. Jenkins, Vol. 1, Pergamon Press, 1967, 93. R. A. Hayes, J. Polymer Sci.,1954, 13, 583. l7 D. C. Blackley, ‘Emulsion Polymerisation’, Applied Science, London, 1975. The Polymerization and CopoIymerization of Butadiene equilibrium with that located in the micelles via the small amount in true solution.Addition of a water soluble radical initiator, usually a persulphate or a redox system, allows radicals to be generated in the aqueous phase which migrate to the soap micelles wherein they initiate polymerization. The radicals are predomi- nantly absorbed by the micelles rather than the droplets because of the very much greater surface area they present as a result of their large number (N 1014 cm-3) and small size. Propagation continues, fed by the monomer from the droplets so that they decrease in size as the micelles swell to form monomer-polymer particles. The growth of these particles increases the surface area to a point at which all the free emulsifier is adsorbed, and hence no new particles are formed-end of Stage I of this process.This point is usually reached rather early in the reaction (about 10-15 % conversion at normal soap loadings), and thereafter the constant particle concentration results in a constant rate of polymerization until the monomer droplet phase is consumed-end of Stage IT. This occurs at about 52% conversion with butadiene at 50 "Cusing a standard persulphate recipe.18 The third and final stage involves the polymerization of the residual monomer in the monomer-polymer particles, but emulsion polymerizations involving dienes are normally halted at about 60-70 % conversion in order to avoid excessive branch- ing and cross-linking. Therefore the bulk of the polymer is formed under the steady-state conditions prevailing under Stage 11.After Harkins19 had demonstrated that polymerization was initiated in the micelles and that propagation continued at these sites after their transformation into particles, Smith and EwartZ0 developed an elegantly simple equation to express the kinetics of the process. The theory required the conditions that (i) the monomer was only slightly soluble in water, (ii) the polymer was soluble in monomer, and (iii) the radicals were generated in the aqueous phase. If the assumption were then made that the termination of a radical in a particle took place very rapidly on the introduction of another compared with the interval between the entrance of successive radicals, then at any given time one half of the particles would contain a single radical and the other half would contain none.Therefore, under the steady-state conditions applying at Stage 11, the rate of polymerization is given by equation (l), where kp is the rate constant of pro-pagation in bulk, [MI is the monomer concentration in the particle, and N is the number of particles. In typical emulsion systems the particle concentration is approximately 1014cm-3 and the rate of radical generation is about 1013cm-3 s-1, so that propagating radicals have lifetimes of the order of 10 s. Under these conditions polybutadienyl radicals can grow to high molecular weight despite l8E. J. Meehan, J. Amer. Chem. SOC.,1949, 71, 628. W. D. Harkins, J. Amer. Chem. SOC.,1947, 69, 1427; J.Polymer Sci., 1950, 5, 217. ao W. V. Smith and R. H. Ewart, J. Chem. Phys., 1948,16, 592. Richards their slow propagation rate. Critical reviews of the theories relating to emulsion polymerization have been recently published.21$22 Polymerizations involving butadiene almost exclusively use anionic emulsifiers such as the alkali metal salts of fatty acids between CIZ arid c18used either alone or in conjunction with rosin acid soaps. They are effective between pH 9 and 11, and may be coagulated in acid media. Emulsifier concentrations are of the order of 5 parts per 100 monomer. An electrolyte, typically KCl, is often dissolved in the aqueous phase where it can reduce the critical micelle concentration and, by increasing the particle size through partial agglomeration, lower the viscosity of the final latex.The Smith-Ewart theory predicts that at Stage I1 the relationship indicated in equation (2) should hold (Eand I are the emulsifier and initiator concentrations), -dM/dt K [E]0*6[I]0*4 and results from SBR polymerizations indicated the emulsifier and initiator exponents to be 0.7 and 0.4-0.5 re~pectively,~3in reasonable agreement with theory. This relationship shows that if insufficient emulsifier is present the number of propagating particles is small, so that they grow too large and agglomeration and flocculation results. The initiators used in butadiene polymerization fall into two categories, persulphate-mercaptan combinations and redox systems. Potassium persulphate alone initiates many monomer systems; thermal decomposition at about 50 "C produces two sulphate radical ions [equation (3)].These radicals, however, initiate butadiene only inefficiently in the absence of a mercaptan, and has argued that the radical anion is highly polar and cannot readily penetrate into the non-polar latex particle. He postulates that the use of a mercaptan promoter, typically a C12 molecule, allows the less polar mercaptyl radicals to be formed by reaction (4)which can enter the particles more easily. Care has to be taken with the use of mercaptans, however, as they are also molecular weight modifiers (see later). SBR and nitrile rubbers prepared at 55 "C using the persulphate catalyst system are known as hot polymerized rubbers.The later development of redox initiators allowed the polymerization temperature to be lowered, typically to 21 A. E. Alexander and D. H. Napper, 'Progress in Polymer Science', ed. A. D. Jenkins, Vol. 3, Pergamon Press, 1971, 145. 22 J. Ugelstad and F. K. Hansen, Rubber Chem. Technol., 1976, 49, 536. 23 I. M. Kolthoff and A. I. Medalia, J. Polymer Sci., 1950, 5, 391. E. W. Duckj J. A. Waterman, and G. E. Latteij, J. Appl. Chem., 1962, 12, 469. The Polymerization and Copolymerization of Butadiene 5 "C, and the rubbers resulting, known as cold polymerized rubbers, had gener- ally superior physical and mechanical properties. The most satisfactory type of redox initiator for butadiene consists of a monomer soluble hydroperoxide (p-menthane, pinane or cumene) and a water soluble reducing system.The latter component consists of a ferrous salt complexed with a pyrophosphate or EDTA, and a reducing agent such as glucose or a sulphoxylate. Alkoxide radicals are generated by reduction of the hydroperoxide by free ferrous ions, equation (9, ROOH + Fe2+-+ RO. + OH-+ Fe3+ (5) and the resulting ferric species are converted back to the ferrous state by the reducing agent. Ferric ions can react directly with hydroperoxide to produce peroxy radicals (ROO -)but these are incapable of initiating polymerization. For this reason, air is kept out of emulsion systems as oxygen inhibits polymerization by converting the alkoxy radical into the inactive peroxy radical, Oxygen scavengers such as sodium dithionate are also often added.The molecular weights of emulsion polymers are controlled by the addition of chain transfer agents, otherwise known as modifiers. Mercaptans have most frequently been employed in this role [equation (6)] and their efficiencies have RSH + Mn* + MnH + RS* (6) been shown to be governed by two factors (i) their rates of diffusion from the monomer droplets to the growing particles, and (ii) their rates of chain transfer with propagating radicals. The Cla mercaptans give the optimum performances in this regard and are now generally used. A detailed review of the molecular weight control of emulsion polymers has recently been published.25 Butadiene based emulsion systems are prone to cross-linking and gelling, and polymerization must be terminated at about 65 % conversion to minimize this effect.Compounds used in this role are called short stops, and they deactivate the reagents in the aqueous phase and terminate the growing radicals in the particles. Examples of such compounds are hydroquinone, sodium polysulphide, and phen ylhydrazine. It has been shown that the microstructure of the butadiene component of a polymer is little affected by changes in initiator, soap or modifier.26 Changing the polymerization temperature principally affects the cis-trans ratio, the vinyl content remaining at about 18 %; at 5 "C the trans content is 70% and drops to 63 % at 50 oC.26 The butadiene component in nitrile rubbers possesses a higher trans content than the homopolymer or SBR; thus nitrile rubber prepared at 28 "C and con- taining 28 % acrylonitrile has the butadiene structure 12% cis,78 % trans and 10 % vinyl.Both the cold SBR and the cold nitrile rubbers have narrower molecular weight distributions than the hot variety, and this is attributable to the activation 2b C. A Uraneck, Rubber Chem. Technol., 1976, 49, 536. J. L. Binder, Ind. and Eng. Chem., 1954, 26, 1727. Richards energy of transfer (70 kJ mol-1) being considerably larger than that of propagation (39 kJ mol-l). The emulsion technique as applied to butadiene is used chiefly to produce SBR and nitrile rubbers, and in these circumstances the copolymerization equation [equation (7)] applies, where rl = kll/k12 and r2 = k22/k21.The monomer concentrations refer to those in the particles, which may be different from those in the storage oil droplets. However, Lewis et aZ.,z7 measuring the reactivity ratios (r) of bulk styrene (Ml)-butadiene (M2) systems at 60"C obtained rl = 0.78 and r2 = 1.39, close to those obtained in emulsion systems at 50 oC28(ri = 0.64, r2 = 1.38) so that the differences in monomer concentration are probably not large. The latter values give r1r2 = 1.08 which is near the unity required for ideal copolymerization. This ideal situation arises when the relative reactivities of both monomers are the same toward either of the two possible growing ends, and implies that the relative molar concentration of monomers in the copolymer and in the emulsion mixture are the same at all monomer ratios.The consequent constancy of com- position of product with conversion is quite closely obeyed by SBR,particularly at low polymerization temperatures28 These copolymerization conditions do not apply for nitrile rubbers. Although, like SBR, the reactivity ratios obtained for acrylonitrile (Ml)-butadiene (Mz) systems in bulk and in emulsion agree well,29 the absolute figures of ri = 0.28, r2 = 0.02at 5 "Cand rl = 0.42,r2 = 0.04at 5 "C are very different, with r132 1. This relationship means that there is only one ratio of monomers at a given tem- perature where the polymer composition is constant with conversion. This so called azeotropic mixture is calculated from equation (8), and has a value of 36% [M1]/[M2] = (r2 -l)/(rl -1) (8) acrylonitrile at 5 "C and 42%acrylonitrile at 50 "C.4 Anionic Polymerization Although the polymerization of dienes by alkali metals was first revealed in 1910-11 by patents issued to Matthews and Strange30 and to Harris,31 the first systematic study of these and similar systems initiated by alkyl lithium was carried out by Ziegler and co-workers in the mid-thirties.32 They distinguished between the initiation step and the subsequent propagation step and established F. M. Lewis, C. Walling, W. Cummings, E. R. Briggs, and W. Weiniseh, J. Amer. Chem. SOC.,1948, 70, 1527. 88 R. D. Gilbert and H. L. Williams, J. Amer. Chem. SOC.,1952, 74, 4144. 29 W. V. Smith, J. Amer. Chem. SOC.,1948, 70, 2177.F. E. Matthews and E. H. Strange, B.P. 24 790/1910. 31 C. H. Harries. U.S.P.1058 056/1913; Annalen, 1911, 383, 184. sz K. Ziegler, F. Dersch, and H. Wolltham, Annalen, 1934, 511, 13, 45, 64. The Polymerization and Copolymerization of Butadiene that, whereas initiation with alkyl lithium produced a species which could propagate at one end [equation (9)], direct alkali metal initiation yielded a diadduct capable of propagation at both ends [equation (lo)]. Although the nM RLi 3. M -+ RMLi --+ RMn+lLi (9) nM 2Li + M -+ LiMLi +LiMn+lLi (10) process was not described as anionic, the concept that these systems were ones which need not possess a termination step, now called 'living' systems, was quite clearly perceived. Some general deductions may be made from Ziegler's scheme.Unlike free radical systems, in the absence of chain transfer processes or impurities the polymer molecular weight will increase with conversion. Furthermore, since in alkyl lithium initiation one polymer chain is generated for each initiator molecule consumed (C), the number average molecular weight (Mn)is given by Mn = M/C, where M = weight of monomer consumed. With initiation fast relative to pro-pagation this relationship approximates to Mn = M/CO,where C is the initial catalyst concentration. Similarly, for alkali metal initiation Mn = 2M/C ( N" 2M/Co for fast initiation). Finally, the molecular weight distribution in both cases should be Poisson with fast initiation (Mw/Mn -+ 1 at high molecular weights).More recently, alkali metal adducts of polycyclic hydrocarbons, prepared by direct reaction of the components in polar solvents, were found to act as anionic initiators by an electron transfer mechanism.33 Typically, sodium reacts with naphthalene in tetrahydrofuran (THF) to form the radical anion (5) in which the electron occupies the lowest vacant .Ir-orbital of the naphthalene. These species (Na+ N') initiate by electron transfer to monomer, and the monomer radical anion thus formed dimerizes to give the dimer dianion which propagates further [equation (12)l. In one sense these catalysts may be regarded as solubilized forms Na+N+ -M+N+Na+M'+Na+M;- dim Na++ (n-2M) Na+ M:- Na+ (12) of alkali metal with the consequent advantage of very much faster rates of initiation.Unfortunately, however, they are stable only in good cation solvating solvents which, as will be seen, adversely affect PBD structure. 33 M. Szwarc. M. Levy, and R. Milkovich, J. Amer. Chem. SOC.,1956, 78, 2656. Richards A. Alkali-metal Initiation.-PBD prepared by alkali metal initiation has a structure determined by the nature of the metal rather than the presence or otherwise of inert hydrocarbon solvents. High 1,6PBD ( N 90%) is produced with lithium, about 40% of which is cis.34,35 These heterogeneous reactions are slow, with initiation and propagation proceeding simultaneously, so that broad molecular weight distributions result. Initiation by alkali metals other than lithium, or even their alloys with lithium, produces PBD’s with greatly increased vinyl content (N 6073.3435 Introduction of a polar solvent such as THF accelerates the rate of reaction, but its presence has a dramatic effect on the PBD structure; all cis structure is eliminated and up to 90% vinyl conformation is formed. The same behaviour is observed with electron-transfer catalysts.36 It is therefore clear that the desired high 1,4-PBD’s can only be prepared anionically using lithium as the counter ion, and in the absence of a polar solvent. These conditions may be realised by using alkyl lithium initiators and, as these systems are homogeneous and consequently more reproducible than those involving the metal, they have been the subject of much study.B.Alkyl Lithium Polymerization.-All alkyl lithiums except the methyl derivative are soluble in hydrocarbon solvent in which they exist in associated forms; ethyl lithium and butyl lithium are hexameric whereas the branched and larger chain species exist as tetramers.37 This aggregation complicates the analysis for the polymerization process because propagation results in polymeric lithium species which form mixed aggregates with residual lithium initiator.38 Further, the inadvertent presence of other lithium compounds such as halides or alkoxides can influence the rate of initiation by participating in the aggregati0n.3~ Kuntz and Greber,40 using isobutyl lithiurn in heptane as initiator, showed that PBD possessed a structure unaffected by changes in initiator or monomer concentration or by degree of conversion; it was also relatively insensitive to polymerization temperature over a wide range.Similarly, the alkyl lithium initiator used does not affect the microstructure of the polymer, although it does have a marked effect on the kinetics and hence the molecular weight and disper- sity of the product. Change of hydrocarbon solvent also principally alters the rate of polymerization rather than the molecular architecture of the polymer; catalysts in aromatic solvents initiate more rapidly than in aliphatic media, but PBD’s from these systems have only slightly higher vinyl content.41 s4A.W. Mayer, R. R. Hampton, and J. A. Davison, Rubber Chem. Technol., 1953, 26, 522. a5 F. C. Foster and J.L. Bender, ‘Advances in Chemistry’, Series No. 19, A.C.S., 1957, 26. 36 A. Rembaum, F. R. Ells, R. C. Morrow, and A. V. Tobolsky, J. Polymer Sci.,1962,61, 155. a7 T. L. Brown, Adv. Organometallic Chem., 1965, 3, 365. F. SchuC and S. Bywater, Polymer, 1973, 14, 594. 39 S. Bywater, ‘Progress in Polymer Science’, ed. A. D. Jenkins, Vol. 4, Pergamon Press, 1975, 27. I. Kuntz and A. Greber, J. Polymer Sci.,1960, 42, 299. 41 H. L. Hsieh, J. Polymer Sci.A, 1965, 3, 181. The Polymerization and Copolymerization of Butadiene Even small amounts of polar solvents have a profound effect on the structure of the PBD's formed. Figure 1 shows how the vinyl content of PBD increases with the amount of polar component added to a hydrocarbon solvent; the rate of increase is clearly dependent on its solvating power.42 Other authors15 have studied the effect of THF on PBD structure over the complete range of solvent compositions and shown that the initial rapid increase in vinyl structure soon levels off, only increasing from 61 % in 10% v/v solution to 79 % in pure THF.fx Diglyme 80 Tetramethyl ethylene diamine c1 .s 60 XE3 cqcr Tetrahydro furan 40 o? 20 Diet hy lether A, -! i.//c I I I 5: I 10: I 15: 1 Molar ratio Additive: Butyl-lithium Figure 1 Eflect of polar additives on the structure of polybutadiene.Reproduced by permission from J. Znst. Rubber Ind., (A), 1968, 6, 3407.1 The increase in the rate of polymerization, and particularly the rate of initia- tion, observed on addition of polar solvents is caused by preferential solvation of the active ends, thereby enhancing the dissociation of monomeric ion pairs as well as increasing their concentration by reducing the association number of the alkyl lithium.For example, n-butyl lithium which is hexameric in hexane, is tetrameric in diethylether and probably dimeric in THF.43 C. Kinetics of Polymerization.-Hydrocarbon Solvents. The kinetics of poly- merization is complicated by the fact that initiation and propagation take place E. W. Duck and M. Locke, J. Inst. Rubber Znd. (A), 1968, 6, 3407. P. West and R. Waack, J. Amer. C'hem. SOC.,1967, 89, 4395. 246 Richards concurrently, although the rate of initiation may be enhanced by suitable choice of alkyl lithium. H~ieh~~ has shown that the order of reactivity towards dienes is Bus > Pri > But > Bui > Bun, and that the faster initiator gives the more monodisperse product .45 The rate of initiation is always proportional to the monomer c~ncentration,~~ whereas its dependence on catalyst is determined by the solvent; the order of initiation of isoprene by n-butyl-lithium is fractional in benzene,45 but first order in hexane.46 The latter relationship cannot be explained as initiation by mono- meric initiator in equilibrium with hexameric aggregates, and direct addition of the monomer to the aggregate to form a mixed association [equation (13)] has (RLi)6 + M -+ [(RLi)sRMLi] been suggested.Since there is some evidence that these are more reactive than the original hexamer,47 autocatalysis of the initiation process can occur.Alternatively, the aggregation of initiator may be suppressed in the presence of polymeric ion pairs in favour of cross association,48 and the inadvertent presence of other lithium salts such as alkoxide has also been proposec, as a cause of breakdown of the initiator complexes.49 The propagation reaction is also first order in monomer, but in aliphatic solvents is fractional with respect to the propagating species. More recent esti- mates of this latter value centre around a quarter,50 and again a correlation can be sought between this and the association number. Unfortunately, the two estimates attempted have given two different numbers; Morton et aL50 obtain a value of two, whereas Worsfold and Bywater5I find a figure of four with no evidence of dissociation over a wide range of dilution.A simple relation between reaction order and degree of association would only exist if the monomeric species were the sole active species. Hsieh and Glaze52 have speculated that butadiene might complex with the aggregates before dissociation and addition occurs. Polar Solvents. Addition of polar substances to hydrocarbon solutions of alkyl lithium increases the polar character of the species present, as evinced by a sharp increase in electrical conductivity,53 and, although even in pure THF alkyl lithium is still principally dimerically associated, the formation of monomeric ion pairs is enhanced.44 H. L. Hsieh, J. Polymer Sci. A, 1965, 3, 153, 163, 173. 46 H. L. Hsieh and 0. F. McKinney, J. Polymer Sci. B, 1966, 4, 843. J. E. L. Roovers and S. Bywater, Macromolecules, 1968, 1, 328. p7 S. Bywater and D. J. Worsfold, J. Organometallic Chem., 1967, 10, 1. 48 I. A. Alexander and S. Bywater, J. Polymer Sci., A, 1968, 6, 3407. 4B A. F. Johnson and D. J. Worsfold, J. Polymer Sci., A, 1965, 3, 449. M. Morton, L. J. Fetters, R. A. Pett, and J. F. Meier, Macromolecules, 1970, 3, 327. b1 D. J. Worsfold and S. Bywater, Macromolecules, 1972, 5, 393. I'H. L. Hsieh and W. H. Glaze, Rubber Chem. Technol., 1970, 43, 22. ssA. Kh. Bagdasar'yan, V. M. Frolov, E. I. Tinyakova, and A. B. A. Dolgoplosk, Proc. Acad. Sci.U.S.S.R.,1965, 162, 582. 247 The Polymerization and Copolymerization of Butadiene The kinetics of the initiation process, as distinct from the overall polymerization kinetics, have not been studied in detail. In THF, initiation is very fast, and once propagation takes place the high solvating power of the solvent breaks down any polybutadienyl lithium aggregates to monomeric ion pairs so that the rate of propagation is fist order in catalyst and monomer. The solvation of polystyryl-alkali metal salts in THF and other polar media has been studied by Szwarc and his colleague^,^^ who have established that a dynamic equilibrium exists between contact ion pairs, solvent separated ion pairs, and free ions. Although in THF the relative concentration of free ions is very small (N 1%), its propagation rate is about 800times faster than that of the ion pairs so that polymerization is principally effected by this species.Rapid equilibration ensures that all chains participate equally in the polymerization process so that the polydispersity is low. No similar study has been made of the polydienyl lithium systems, but it may be expected that since there is less extensive charge delocalization the ionization would be lower, although the contribution of free ions to the polymerization process should still be significant. Recently, a study of cumyl potassium initiation in THF confirmed that propagation was almost entirely due to the free butadienyl anion.55 D. Mechanism of Anionic Polymerization.-Two basic observations have to be explained in any mechanistic picture developed for the alkali metal (alkyl) polymerization of butadiene, (i) the high 1,4-PBD’s produced in bulk or inert solvent by lithium and its alkyls, and (ii) the high 1,2-PBD’s prepared using other alkali metals under these conditions, and by all alkali metals and their derivatives when polar solvents are employed.2 3 2 3 CH-CH wCH2 \4 CH, Li + N.m.r. studies of the active ends in hydrocarbon media show them to exist exclusively as the 1,6adduct (6) with a cis-trans ratio of about 3: 1.56 Addition of THF results in an upfield shift in the signal from the methine hydrogen on C-2, suggesting some contribution from the 7-ally1 structure (7). Although in THF the active centre is still primarily located on C-4, delocalization may be sufficient to allow attack by monomer at C-2 and cause vinylic addition.The virtually exclusive vinyl structure in PBD’s prepared with lithium naphthalene cannot be explained in this way unless the regenerated naphthalene helps to delocalize the 64 M. Szwarc, ‘Carbanions, Living Polymers and Electron Transfer Processes’, Interscience, 1968. tj6 A. Siove, P. Sigwalt and M. Fontanille, Polymer, 1975, 16, 605. w S. Brownstein, S. Bywater, and D. J. Worsfold, Macromolecules, 1973, 6, 715. Richards charge. By~ater~~ has warned against looking for too close a parallelism between n.m.r. observations and the structure of the product, pointing out that in hydrocarbon solvents the former record the configuration in the aggregates whereas propagation is probabIy conducted by the unassociated species.Glaze et aZ.57examined the 1,6addition product of t-butyl lithium and butadiene, neopentyl lithium, in ether solvents and confirmed that charge delocalization occurred. cis-trans Isomerization was slow in ether at 30 “C but considerably faster in THF. There has been virtually no spectroscopic examination of polydiene anionic ends with other alkali metal counter ions, and inferences must again be drawn from Szwarc’s results using polystyrene anions.54 Here solvation of the counter ions falls off rapidly as their radii increase, and the active ends involving the larger cations are essentially contact ion pairs, particularly in hydrocarbon solvents.Thus with ‘living’ PBD’s a terminal configuration analogous to (7) might be favoured as the cations could be stabilized by back donation from C-2 and C-4, and in part explain the predominance of vinyl structure in these PBDs.39 E. Termination.-After propagation has ceased, the active centres may be terminated by the addition of suitable reagents. Proton donors such as water or alcohol yield terminally non-functional PBD’s33 [Scheme 11. Alternatively, -MH 4LiOR - -M-Li+ MCOO-Li’ -MCH2CH20-Li+ H+ Hf - MCOOH MCH,CH,OH -MR + LiX Reagents: i, ROH; ii, CO,; iii, CH,-CH,; ivy RX. ‘0’ Scheme 1 carbon dioxide gives a carboxylate end and ethylene oxide gives an alcoholate gr0up.5~ Primary alkyl halide addition yields alkyl terminated PBD,60but p-hydrogen elimination reactions have been observed with second- ary and tertiary halides.61 Halogen terminated materials may be obtained by direct reaction with excess halogen or alkyldihalide, although significant li7 W.H. Glaze, J. E. Hancock, J. Chaudhuri, M. L. Moore, and D. P. Duncan, J. Organo-metallic Chem., 1973, 51, 13. E.* H. Brody, D. H. Richards, and M. Szwarc, Chem. and Ind. (London), 1958, 1473. 58 D. H. Richards and M. Szwarc, Trans. Faraday SOC.,1959, 55, 1644. 6o A. Davis, D. H. Richards, and N. F. Scilly, Makromol. Chem., 1972, 152, 121. 61 A. Davis, D. H. Richards, and N. F. Scilly, Makromol. Chem., 1972, 152, 133. The Polymerization and Copolymerization of Butadiene amounts of Wurtz coupled product result due to the rapidity of the reaction.62 This may be minimized by converting the ‘living’ polymer into its Grignard analogue before terminating.62 The latter technique may also be employed to avoid ketone forming side reactions in the carbon dioxide terminating process.Termination reactions are particularly important in the anionic preparation of telechelic liquid PBD’s. These are low molecular weight (M3000) PBD’s with terminal functional groups which can be cross-linked in a controlled manner by reaction with polyfunctional reagents such as isocyanates. Such systems have been comprehensively reviewed.63 F. Solution SBR’s.-The development of emulsion SBR’s spurred further investigation into the preparation of SBR’s by anionic means. Products from these processes are called solution SBR’s.Because a high 1,4-conformation is required of the butadiene component, polymerization has to be carried out in hydrocarbon solvents. It was found, unexpectedly, that butadiene was preferentially polymerized and was virtually consumed before styrene took significant part in the reactio11,6~ despite the faster rate of homopolymerization of styrene than butadiene. This phenomenon was interpreted as being caused by preferential solvation of the active ends by butadiene thereby preventing styrene access until the diene concentration was greatly reduced. Once accessibility had been achieved the polymerization process was accelerated. A different explanation was given by O’Driscoll and Kuntz65 who developed a copolymerization equation based on the observation that the bulk of the active sites were present as non-propagating dimers.The two theories need not be mutually exclusive, however; the latter may be a formalization of the qualitative picture given by the former. Whatever the detailed explanation, this phenomenon precludes the manu- facture of SBR’s in this way. This difficulty may be overcome in two ways; butadiene may be added to the system incrementally,66 or a randomizing agent may be introduced.67 The latter agent almost certainly acts as a preferential solvating agent, and polar solvents fall into this category although they adversely affect the microstructure. One of the more effective additives is lithium t-b~toxide.~~ Solution SBR’s prepared for tyre purposes contain 20-25 ”/, styrene and possess about 90% 1,6structure (354% cis).They are superior to their emulsion counterparts in that the molecular weight distribution is controllable, and they contain no contaminants such as fatty or rosin additives. They are becoming increasingly important commercially. Nitrile rubbers cannot be prepared anionic- 61F. J. Burgess and D. H. Richards, Polymer, 1976, 17, 1020. 63D. M. French, Rubber Chem. Technol., 1969, 42, 71. gp A. F. Johnson and D. J. Worsfold, Makromol. Chem., 1965, 85, 273. 65 K. F. O’Driscoll and I. Kuntz, J. Polymer Sci.,1962, 61, 19. g6 D. J. Worsfold and S. Bywater, Canud. J. Chem., 1960, 38, 1891.67 U.S.P 3 094 512/1963. Richards ally as the nitrile group is susceptible to anionic attack and the stability of the acrylonitrile anion is too great to initiate butadiene polymerization. G. Butadiene Block Copolymers (Thermoplastic Elastomers).-Block copolymers have been defined as ‘polymers composed of molecules in which two or more polymeric segments of different chemical composition are attached end to end’.68 Modern anionic processes have made their synthesis much more readily attain- able; in particular they have led to the discovery and development of thermo- plastic elastomers of the ABA type, where A represents a polymeric segment of high Tg (plastic component) and B represents a central polymeric segment of low Tg (elastomeric component).Because the entropy of mixing of polymers is low, the components are generally mutually incompatible and form a two phase system in which the plastic micelles are embedded in a rubber matrix. At ambient temperatures, since the elastomeric chain ends are locked to the plastic micelles which act in place of chemical cross-links, the material exhibits elastomeric properties resembling those of a vulcanized rubber. As the temper- ature is raised through the Tg of the plastic component, however, the material exhibits thermoplastic properties and eventually melts. This means that, unlike normal rubber, a device made from these materials may be remoulded as required. The synthesis and properties of thermoplastic elastomers have been reviewed by Fetters.G9 The fist such materials developed, and currently the most important of this class were those derived from styrene and butadiene. Hereafter designated SBS, they are prepared anionically and consist of polystyrene segments of 10-15 OOO molecular weight flanking polybutadiene segments of 50-100 OOO molecular weight.Four distinct ways have been devised to prepare them. Monofunctional Initiation. Styrene, butadiene and then styrene are added sequen- tially to an alkyl (usually sec-butyl) lithium initiator in hydrocarbon solvent. The time scale of initial styrene polymerization at 30 “C is about an hour, after which the butadiene block is formed in five to six hours. THF is added at this stage to speed the final styrene polymerization, and the system is then terminated with a proton donor and precipitated from methanol.The main disadvantage of this procedure is that inadvertent introduction of extremely small quantities of impurities at any stage results in the formation of S or SB blocks. The latter contaminant has a deleterious effect on the properties of the SBS produced.69 DifunctionaZ Initiation. Dianionic PBD is initially formed and styrene is then added to grow both terminal blocks simultaneously. The process has only two stages, but the common difunctional initiators are soluble only in polar solvents and O8 M. L. Huggins, P. Corradini, V. Desraux, 0.Kratley, and H. Mark, Polymer Letters, 1968, 6, 257. OS L. J. Fetters, J. Elastoplastics, 1972, 4, 34.25 1 The Polymerization and CopolJmerization of Butadiene consequently produce undesirable high 1,2-PBD's. Some seeding techniques have been attempted, and 1,4-dilithio-l ,1,4,4,-tetraphenyl butane in hydrocarbon solvent containing 15 % anisole has been used as initiator.70 However, the lack of a suitable cheap difunctional initiator has prevented the development of this otherwise attractive synthetic route into a commercial process. Monofunctional Initiation and Coupling. An S block is prepared on which buta- diene is subsequently polymerized to half the chain length required. A coupling reaction is then carried out with a dihalide (RX2) as indicated in equation (14). RX,Sn-Li+ + m -2 B -+ Sn--Bm/z-Li+ --+ Sn-Bm/2-R-Bm/2-Sn + 2LiX (14) SB contamination can easily occur by inaccuracy in the titration or by a halogen-metal exchange side reaction.The latter can be minimized by adding THF after polymerization and raising the temperature to 40-50 "C before titrating. Phosgene has been found to be a very efficient linking agent.71 Direct Copolymerization. The preferential polymerization of butadiene in mixtures of butadiene and styrene may be used to prepare SBS polymers. A monofunctional initiator is used, initially with styrene, and then with a mixture of both monomers. The main disadvantage of this technique is that, during the polymerization of the mixture, styrene gets increasingly incorporated as the butadiene decreases, and a tapered segment of random composition separates the two essentially homo- polymer blocks.This lessens the incompatibility of the two polymer components and the resulting phase blending gives materials with reduced tensile properties. 5 Polymerization by Alfin Catalysts Alfin catalysts, discovered by Morton in 1947,72 consist of a mixture of a sodium alkoxide, sodium olefin derivative and sodium chloride, the name being derived from the combination of the syllables in italic. The most active catalyst of this class is allylsodium, sodium isopropoxide and sodium chloride. The technological aspects of these systems have been recently reviewed.73 Polymerization is very rapid, resulting in a very high molecular weight PBD which possesses 75-80 % trans and < 5 % cis structure. These molecular weights (N 7 x lo6)mean that the materials are difficult to process, but more recently it has been found that 1,4-dienes, particularly 1,4-dihydrobenzene or 1,4-dihydro- naphthalene, are efficient molecular weight modifiers,73 and [14C] experiments have shown that one molecule of modifier is attached to each chain.Morton investigated the system in some detail, and arrived at the following conclusions : (i) very few organo sodium derivatives exhibited catalytic activity 70 L. J. Fetters and M. Morton, Macromolecules, 1969, 2, 463. 71 M. Morton, R. F. Kammerick, and L. J. Fetters, Macromolecules, 1971, 4, 11. A. A. Morton, E. E. Mogat, and R. L. Letsinger, J. Amer. Chem. SOC.,1947, 69, 650. 73 R. Newburg, H. Greenberg, and T. Sato, Rubber Chem.Technol., 1970, 43,333. Richard and, of these, allyl sodium was the most efficient, (ii) sodium isopropoxide was the most active of the alkoxides, (iii) maximum catalytic activity was obtained at equimolar allyl sodium and sodium isopropoxide, (iv) sodium chloride was a necessary component of the system, the activity of which was maximized when generated in situ, (v) the intrinsic viscosity did not increase with conversion, and (vi) the reaction was first order with respect to monomer, and each chain possessed an allyl group. The equimolar relationship between allyl sodium and sodium isopropoxide led Morton to propose the cyclic co-ordination complex (8) adsorbed on to sodium chloride as the polymerization catalyst. He contrasted the rapid polymerization to give high trans-PBD with the slower polymerization by organo sodium catalysts to yield high 1,2-PBD, and thus suspected different mechanisms were in operation.Further, the lack of dependence of molecular weight on conversion led Morton to postulate a free radical mechanism in which the radicals were fixed on a lattice complex of sodium chloride and allyl sodium. The similarity in structures between Alfin catalysed and emulsion polymerized PBD's seemed to support this view. Uelzmann,74 however, proposed an anionic mechanism in which the iso-propoxide adsorbed on to the sodium chloride lattice stabilized the growing anionic chain as illustrated for allyl sodium in (8). This hypothesis is now favoured as it explains the chain modifying capacity of the 1,4-dienes as being due to their action as hydride ion transfer agents.The constancy of molecular weight with conversion can then be ascribed to the action of a chain transfer process. The role of the salt is clearly of importance. It has been shown that those which are catalytically active have alattice constant between 2.81 and 3.29 k75a Indeed, Bykhovskii and Min~ker7~* have proposed that crystal defects which involve localized electrons (e.g. F centres) are sites of catalytic activity, but this is difficult to correlate with the critical role played by the organo sodium component. It is now accepted that, although this heterogeneous process is complex, the balance of evidence indicates that polymqization occurs by a form of anionic insertion mechanism.74 H. Uelzmann, Rubber Chem. Technol., 1959, 32, 597. 75 (a) L. Reich and A. Schindler, 'Polymerisation by Organometallic Compounds', Inter- science, 1966, 406; (b) Bykhovskii and Minsker, ibid., p. 420. The Polymerization and Copolymerization of Butadiene 6 Polymerization by Transition Metal Catalysts The discovery of the Ziegler-Natta (Z-N) catalytic system~~~,~~ for the stereo- specific polymerization of cc-olefins and of dienes initiated a search for other effective transition metal catalysts. Z-N catalysts were defined as a combination of metal alkyls of Groups I to I11 with transition metal salts of Groups IV to VIII. In practice, however, the most efficient combinations were those in which an aluminium alkyl derivative were interacted with titanium, vanadium or cobalt salts.So far as butadiene is concerned, successful catalysts of the Z-N type have been found as well as those which fall outside the Z-N patents. The distinction between Z-N and other catalyst systems has therefore become blurred and, apart from legalistic considerations, it seems unnecessary to compartmentalize them in this way. Hereinafter they will all be considered as transition metal catalysts. Cooper and Vaughan15 and, more recently, Ledwith and Sherrington78 have discussed the mechanisms by which polymerizations catalysed by various transition metal systems occur, and this aspect will only be touched on in this review. The reactions between aluminium alkyls and transition metal compounds are very complex and have not been elucidated in many cases.They result in catalysts, some of which are soluble in hydrocarbon solvents, and those based on cobalt are very important for synthesizing high cis-PBD. Although it is difficult to generalize about systems involving aluminium alkyls, it appears that in many cases the initial reaction is the alkylation of the metal salt to form an unstable species which decomposes to liberate an alkyl radical and leave the metal in its reduced form [equation (15)]. These species may be further alkylated, and AIR3 + Tic14 -+ AlRzCl + RTiC13 -+ R* + Tic13 (15) polymerization is envisaged as taking place through co-ordination of the mono- mer at the missing site in the octahedral configuration of the transition metal caused by the missing ligand.A further broad generalization may be made about catalytic systems involving aluminium alkyls. PBD’s of the following structure are produced with the transition metal compounds listed : high cis TiBr4, TiI4, P-TiCls, cobalt compounds; high trans TiC14, Tic13 (cc, y, a), vanadium halides; high vinyl, chromium and molybdenum compounds. A. Titanium Compounds.-Addition of equimolar amounts of AIR3 or AlR2X to a solution of a TiIV compound in hydrocarbon solvent yields a precipitate of the Ti111 species79 (the chloride in its 18 form). This may be further reduced to TilI by AlR3, but not by AIRzX.*O The aluminium compounds are bound on to the 76 K.Ziegler, E. Holzkamp, H. Breil, and H. Martin, Angew. Chem., 1955, 67, 541. 77 G. Natta, J. Polymer Sci.,1955, 16, 143. 78 A. Ledwith and D. C. Sherrington, ‘Reactivities, Mechanism and Structure in Polymer Chemistry’, eds. A. D. Jenkins and A. Ledwith, J. Wiley, 1974, Chap. 12. P. H. Moyer, J. Polymer Sci. A, 1965, 3, 209. C. Beermann and H. Bestian, Angew. Chem., 1959,71, 618. Richards TiII precipitate and play an essential role in its activity, and they may be added subsequently if the Ti111 compound has been prepared without their involvement. It is the nature of the TiIII salt that determines the structure of the PBD formed rather than that of the reducing agent. cis-l,4-PBD’s. AlRrTiI4 and analogous systems containing iodine can produce 95% cis-PBDk81 Systems such as AlR3-TiC1212 and AIRrI2-TiC14 are typical, and have the advantage over the tetraiodide that the titanium compounds are more soluble. These three formulations behave similarly because in each case any chloride is eliminated into the solvent, leaving the precipitate as TiI3.They have the following features in common: (i) catalytic activity starts at Al:Ti = 1 and increases with this ratio to a maximum variously quoted at ratios of 1.5 to 382 and 5,79 (ii) catalytic activity decreases as the polymerization proceeds, (iii) molecular weight increases with conversion and varies inversely with catalyst concentration, (iv) the ciscontent increases with decreasing catalyst concentration, reaching 95% at 10-3 molar in benzene; it is also maximized at Al:Ti ratios corresponding to maximum catalytic activity.Electron donors such as di-isopropyl ether have been used to stabilize the ~atalyst,~3and other recipes where aluminium alkyl has been replaced by alumin- ium hydride derivatives also yield high c~s-PBD’s.~~ The feature common to all these systems is, however, that the titanium is present as the tri-iodide. trans-1,4-PBD7s.The most efficient titanium catalysts for preparing the highly crystalline trans-PBD are those containing a-TiCls, which is produced by reaction of hydrogen on Tic14 at high temperature^.^^ Even in these systems, however, significant amounts of amorphous material are formed which can be ether extracted. Vinyl-PBD’s. Although titanium compounds have been used to prepare highly tactic 1,2-PBD’s, the halides have to be excluded as they are 1,4-directing.The systems are generally homogeneous. A1Et3-Ti(O-Bun)4 gives a high isotactic vinyl product .86~87 Interestingly, Ti(NEtz)4 yields high 1,2-PBD’s with AIEt3, AlHClzOEtz or AlHChNMe2; the last two produce syndiotactic material, whereas use of AIEtZCI or AlEtCl2 eliminates any vinylic addition and yields high tran~-PBD.~~ Belg. P. 551 851/1957.P. H. Moyer and M. H. Lehr, J. Polymer Sci. A, 1965, 3, 217. 88 J. F. Henderson, J. Polymer Sci. C,1963, 4, 233. 134 W. Marconi, A. Mazzei, A. Araldi, and M. de Maldb, J. Polymer Sci. A, 1965, 3, 735. P. Tepenitsyna, M. I. Farberov, A. M. Kut’in, :and G. S. Levskaya, Vyskomol.Soedineniya, 1959, 1, 1148. 86 T. A. Zakaharov and Yu I. Ermanov, J. Polymer Sci. A, 1971,9, 3129. 87 G. Natta, L. Porri, and A. Carbonaro, Makromol. Chem., 1964, 77, 126. 88 A. Mazzei, D. Cucinella, W. Marconi, and M. de Maldh, Chimica e Industria, 1965, 45, 528. The Polymerization and Coplymerization of Butadiene B. Vanadium Compounds.-trans-l,4-PBD’s.Vanadium halides in conjunction with AIR3 or AlRzX yield catalysts which are specific for trans-PBD’s (>95%).s9 p or y Tic13 may be added to vc13 before reaction with AIR3 to give a catalyst with increased efficiency. Cooper,9o experimenting with metal alkyls other than aluminium, has shown that the vinyl content of PBD increases at the expense of trans in the series Pb < Cd < Mg < Li < Na.Homogeneous catalysts can be prepared from V(acac)s (acac = acetylacetonate) or vcl3 3THF and AIEt2C1;91 at very high Al: V ratios they yield highly crystalline pure trans-PBD’s, although their molecular weights are lower than those produced by heterogeneous catalysts. C. Chromium and Molybdenum Compounds.- Vinyl-PBD’s. Salts of both these metals with AIR3 form catalysts which give high 1,Z-PBD’s. Chromium based catalysts yield product, however, in which the tactic component is <50%, whereas those based on molybdenum are more specific. MoOZ(OR)Z, MoO(acac)z, and Mo(acac)3 give a 95% vinyl product in which there is 75% syndiotactic material.92 wAllyl complexes of chromium [Cr(v-allyl)s and Cr(wcroty1)s ] polymerize butadiene without co-catalyst to give essentially pure 1,2-PBD.93 However, if other transition metal halides are used as co-catalysts the structures of the resulting polymers are dramatically changed;94 thus NiClz or Ti14 gives PBD possessing >90 % cis structure, whereas NiBrz gives 95 % trans-PBD.D. Cobalt Compounds.-These compounds, with AlRzCl or AIRCh, form soluble catalysts which produce high cis-PBD’s although the former co-catalyst requires water or HCI as an activator. With AIR3 syndiotactic 1,ZPBD is formed. cis-1,4-PBD’s. The structure of the polymer formed with these catalysts is independent of the cobalt salt and is insensitive to the A1 :Co ratio above 30 :1. This is explained as an initial exchange occurring between the chloride of the aluminium species and the cobalt salt to yield CoClz which is then alkylated to RCoCl.This decomposes to an alkyl radical and cobalt@ chloride which is stabilized by complexation with the aluminium compound and which then becomes the active catalytic species.94 Lists of Co-A1 catalytic systems have been made elsewhere,l5 and so will not be reproduced here. Apart from the properties mentioned above, all these cobalt catalysts have the following features in common: (i) the molecular weight of the product increases with time, and is related directly to the monomer concentration 8e G. J. Amerongen, ‘Elastomer Stereospecific Polymerisation’, Adv. Chem., Series 52, 1966, 136. So W. Cooper, Rubber Plastics Age, 1963, 44,44. G.Natta, L. Porri, and A. Carbonaro, Rend. Accad. Maz. Lincei, 1961, 31, 189. ea G. Natta, Nucleus, 1963, 21 1. e3 B. A. Dolgoplosk, S. I. Beilin, Yu. V. Kershak, K. L. Makovetsky, and E. I. Tinyakova,J. Polymer Sci. C, 1973, 11, 2569. ep C. E. H. Bawn, Rubber Plastics Age, 1965, 46, 510. Richards and inversely to the cobalt concentration, (ii) the rate of reaction is first order with monomer and cobalt concentrations, (iii) the polymerization may be carried out over a wide temperature range (-30 “C to + 30 “C) without affecting stereospecificity. Very high molecular weights can be obtained with these systems, and regu- lators, such as hydrogen, olefins, allenes or cyclooctadiene, have been used to reduce the chain length and yield an easily workable product.95 The catalyst CoC12-AlC13 is an entirely inorganic catalyst which forms high cis-PBD.It has been shown to have the structure Co(AlC14)~, but it is more commonly used with thiophene which increases its activity.96 Vinyl-PBD’s. Cobalt salts with A& form soluble catalysts in general, which The amorphous region does not exceed produce highly syndiotactic 1 ,Z-PBD’S.~~ 10% and the crystalline residue possesses exclusively 1,Z-units. Again, a variety of cobalt compounds may be used in these systems, but their activities are lower than those of their analogues for cis-PBD and the molecular weights of the product are significantly less. E. Nickel Compounds.-cis-and truns-l,4-PBD’s.Normal a-bonded nickel salts combined with AlEtzCl catalyse butadiene polymerization to high cis-PBD (N 85 %), but not as efficiently as the cobalt comple~es.~8 The main importance of nickel compounds as catalysts lies in the activity of the wallyl nickel halides.These soluble compounds polymerize butadiene slowly without co-catalyst at a rate dependent on the square root of their concentration. Since they exist as dimers in hydrocarbon solution, it is clearly the dissociated form which is the active ~pecies.9~ The addition of Lewis acids such as Tic14 or AIC13 greatly increases their activity, and the kinetics becomes first order in catalyst.100 The new catalytic complexes are almost certainly monomeric, and the initiating species is envisaged as (9),where butadiene is complexed in the S-cis form Ir 1 CH, CH, O6 C.Longrave, R. Castelli, and M. Ferraris, Chimica e Industria, 1962, 44, 725. O6 H. Scott. R. E. Frost. R. F. Bell. and D. E. O’Reilly, J. Polymer Sci. A, 1964, 2, 3233. D7 E. Susa,J. Polymer Sci. C,1964,’4, 387. O8 W. M. Saltzman and L. J. Kuzma, Rubber Chem. Technol., 1973, 46, 1055. D. E. O’Reilly, C. P. Poble, jun., F. Belt, and H. Scott, J. Polymer Sci.A, 1964,2, 3257. looT.Yashimoto, K. Kamatsu, R. Sakata, K. Yamamoto, Y. Takeuchi, A. Onishi, and K. Ueda, Makromol. Chem., 1970, 139, 61. The Polymerization and Copolymerization of Butadienc through both its double bonds.lo1 The increased ionic character of (9) accounts for its increased activity.1O2 The nature of the halide in the catalyst system determines the structure of the PBD, whether the halide originates from the nickel salt or the co-catalyst.Chloride ions induce a high cis product whereas iodide ions direct the poly- merization toward high frans-PBD. Bromide ligands give an intermediate structure with the cis form generally favoured, but in the majority of these systems the vinyl content is low (N 3 %). The reactivity of the nickel catalysts increases in the order crotyl < methylallyl < allyl, and the configuration of the complex has been shown to be syn, independent of the microstructure of the PDB formed.103J04 F.Remaining Group VIII Elements.-Certain members of the second series of Group VIII polymerize butadiene in polar solvents such as water or ethanol.Although salts of ruthenium105 and palladiuml06 are catalytically active, rhodium compounds are outstanding in this regard and have consequently been most studied. Aqueous rhodium nitrate polymerizes butadiene directly without benefit of additives, although other rhodium salts require the presence of a sulphate or sulphonate emulsifier, the structure of which is critical for the polymerization. Thus sodium lauryl sulphate is the sole active sulphate emulsifier, and sodium alkyl benzene sulphonates are active whereas sodium alkyl sulphonates are not.lo7 It therefore appears that the emulsifier participates in the polymerization process, although the product is invariably high trans-PBD (>96 %). Free radical inhibi- tors have no effect, whereas 1,3-cyclohexadiene (molar ratio to Rh 20 : 1) increases the rate about twenty fold.lo8 In contrast, 1,5-cyclooctadiene,lo7 pyridine or EDTAlO8 are powerful inhibitors.I.CHI IRh Rh ! CH ‘CI ’ CH CHI I I CH,R CH,R CH,R (10) (1 1) lol M. Gippin, Ind. and Eng. Chem. (Product Res. and Development), 1965, 4, 160. lo8J. P. Durand, F. Dawans, and Ph. Teyssit, J. Polymer Sci. A, 1970, 8, 979. losR. Warin, Ph. Teyssi6, P. Bourdandurg, and F. Dawans, J. Polymer Sci. B, 1973, 11,177. lo*V. 1. Klepikova, G.P. Kondratenkov, V. A. Kormer, M. I. Lobach, and L. A. Churlyaeva,J. Polymer Sci. B, 1973, 11, 193. lobA. J. Canale and W. G. Hewett, J. Polymer Sci. B, 1964, 2, 1041. lo6A. J. Canale, W. G. Hewett, T. M. Shyre, and E. A. Youngman, Chem.and Znd.(London), 1962, 1054. lo’ M. Morton and B. Das, J. Polymer Sci. C, 1969, 27, 1. lo*Ph. Teyssit and R. Danby, J. Polymer Sci. By 1964, 2, 413. Richards The polymerization mechanism probably involves co-ordination of the growing chain to form a rr-ally1 rhodium complex of low valency as the active species, possibly in the dimeric form (lO).lOg The action of emulsifiers could then be to stabilize the monomeric species (1 1) in the micelle. Inhibitors probably act through competition with butadiene for the co-ordination sites on the rhodium ion. G. Uranium Compounds.-There have been recent disclosures that uranium compounds polymerize butadiene homogeneously in hydrocarbon solvents to yield very high (99 %) cis-PBD. Tris(n-a1lyl)uranium halide produces this material in high yield, the activity being independent of the nature of the halide.110 The catalytic activity may be improved by adding a Lewis acid, with RAIC12 being preferred owing to its solubility and availability.Alternatively, equally impressive results have been obtained using U(OR)4 and EtAlClz (or AIEt3 and AIBr3).ll1 The very high structural regularity allows fast and very extensive strain induced crystallization to occur, and results in a considerable improvement in processability and properties of the rubber.111~112 These systems have not been examined kinetically as yet, but their products show considerable commercial promise. H. Quibinary Po1ybutadienes.-Interest has recently developed in synthesizing PBD's possessing alternating structure involving two out of the three structural isomers.Thus cis-vinyl equibinary structure was claimed by Furukawa et al. for materials produced from the catalyst systems MOC~~(OR)~-AIE~~(AI : Mo =-6)ll and Co(acac)3-AIEt3-H~0,113 but 13C n.m.r. studies later showed that the structural components were essentially random.lOJl TeyssiC and co-workers produced equibinary cis-trans-PBD using a 2,6,10- dodecatriene-l,12-diylnickel-trifluoraceticacid system as catalyst,lOZ but again its structural regularity was questioned by other workers.8 TeyssiC,114 however, was able to show that the regularity depended on temperature and the nature of the solvent ;the most regular alternating product being prepared in methylene chloride.7 Conclusions Butadiene can be polymerized in a variety of ways, but the structure of the polymer produced is very dependent on the technique employed. An indication of some of the structural variations which may be effected by the methods outlined lo@W. Cooper, Ind. and Eng. Chem. (Product Res. and Development), 1970, 9, 457. 1l0 A. de Chirico, P. C. Lanzani, E. Raggi, and M. Bruzzone, Makromol. Chem., 1974, 175, 2021. M. Bruzzone, A. Mazzei, and G. Guiliani, Rubber Chem. Technol., 1974, 47, 1175. ll2 G. Lugli, A. Mazzei, and S. Poggio, Makromol. Chem., 1974, 175, 2029. 113 J. Furukawa, K. Haga, E. Kobayashi, J. Iseda, T. Joshimoto, and K. Sakamoto, PolymerJ. Japan, 1971, 2, 371. 114 M. Julemont, E. Walckiers, R. Warin, and Ph. TeyssiC, Makromol.Chem., 1974,175, 1673. The Polymerization and Copolymerization of Butadiene herein is given in Table 2. In general the high 1,4-PBD's, favoured for their superior elastomeric properties, are manufactured commercially using alkyl lithium-initiated or transition metal-catalysed systems. The latter are based on titanium, cobalt or nickel. Table 2 Ranges of polybutadiene structures obtained by various polymerization techniques Structure of Polymer/ % Method of Catalyst Polymerization System cis-1,4 trans-1,4 1,2 Free Radical Emulsion 20-3 60-74 22-16 Anionic RLi 34-38 56-50 10-12 Hydrocarbon solvent Co-ordinated Anionic Alfin 10-2 70-75 25-20 Ti Halide 80-94 15-1 9-3 AIR3 or AlRzX V Halide - >95 1-5 AIR3 or AlRzX Transition Mo Salts 4 2-1 92-96 Metal AIR3 Co Salts 93-98 3-1 "1 AlRzCl or AlRCl2 rr-ally1 NiCl 81-94 12-4 4-2 A& or TiX4

ISSN:0306-0012    

DOI:10.1039/CS9770600235

出版商:RSC    

年代:1977

数据来源: RSC