Structure and Mechanism in Complex General Acid-base Catalyzed Reactions BY WILLIAM AND JANEM. SAYER P. JENCKS Department of Biochemistry Brandeis University Waltham Massachusetts 02154 U.S.A. Received 28th April 1975 The mechanisms of acid and base catalyzed carbonyl and acyl group reactions are determined largely by the lifetime of the initial addition intermediate. Catalysis through diffusion-controlled trapping by relatively strong acids or bases niust occur if reversion of the intermediate to reactants is faster than proton transfer involving solvent and product formation. If breakdown to reactants is faster than separation of the intermediate and catalyst the intermediate is formed within a solvent cage containing the catalyst through a pre-association mechanism.If the " intermediate " is still more unstable or if there is no barrier for proton transfer a stepwise reaction is impossible and the reaction must be concerted. Thus changes in structure of the reactants can be correlated with changes in the lifetime of the intermediate and the mechanism of catalysis. We would like to know the nature and the magnitude of the rate enhancements that are brought about by general acid and base catalysis of carbonyl and acyl group reactions a field in which R. P. Bell has been a pioneer.l* A few examples are described here in which changes in the structure of reactants and catalysts and the lifetimes of intermediates provide some insight into the mechanism and driving force of these reactions. The mechanisms are divided into two broad classes and several subclasses although there is not always a sharp dividing line between them.I. TRAPPING OF ADDITION INTERMEDIATES A. STABLE INTERMEDIATES When a strong nucleophile attacks a reactive carbonyl compound the initial product is likely to have enough stability to be trapped by proton transfer involving the solvent before it reverts to reactants so that trapping by added catalysts is unnecessary. For example the pK of the oxygen atom in the addition compound that is formed from the attack of trimethylamine on formaldehyde is 9.3 so that the rate constant k for protonation of T* by water is 4 x lo5 s-l (eqn (1); T' based on this pK and diffusion-controlled deprotonation by hydroxide ion in the reverse direction).Since the rate constant kl for amine expulsion to form the unstable formaldehyde molecule is only 3.4 x lo3 s-l (eqn (1)) the addition compound T"will always go on to products and added buffers do not catalyze the rea~tion.~' The more basic addition compounds formed from anionic nucleophiles generally have pK values of the order of 11-13 and will abstract a proton from the solvent 41 GENERAL ACID-BASE CATALYZED REACTIONS 102-103faster; they are correspondingly less likely to exhibit buffer catalysis. Thus cyanide hydroperoxide hydroxide bisulphite and basic thiol anions add to carbonyl compounds without buffer catalysis. 0 0-HO kl t I ks I HX + ,C I1 =HX-C-X-C-(2) k-I I k- I If there is a proton on the attacking nucleophile the intermediate can be trapped by a proton switch mechanism with a rate constant k of ca.106-108 s-l (eqn (2) see later) so that if expulsion of the attacking nucleophile is slower than this no buffer catalysis by trapping will be seen. B. DIFFUSION-CONTROLLED PROTON TRANSFER When a less stable intermediate breaks down to starting materials faster than it is trapped by proton transfer involving solvent there must be catalysis by buffers. When methoxyamine a less basic amine attacks p-chlorobenzaldehyde a less reactive aldehyde expulsion of the attacking amine (k.-l = 3 x lo8s-I) is faster than either proton abstraction from water (k2 = 3 x lo4s-l) or a proton switch (k,= 6 x lo6s-l) so that trapping of the intermediate by encounter with a moderately strong buffer acid increases the observed rate (eqn (3)).5 When the acid strength is too ti MeONH2 + >C=O kl +I cA-lMeONH2-C-OH -2 MeONH2-C-O-kACHA’ ~ k-1~3 x 108s-1 I k-A 1 (3) T’ weak to give a thermodynamically favourable proton transfer trapping will not occur on every encounter and the Bronsted plot becomes an “ Eigen curve ” with a change in slope from limiting values of 0 to -1 with increasing pK of the acid.When the catalyst concentration is increased sufficiently trapping becomes faster than reversion to reactants methoxyamine attack becomes rate determining and the plot of kobsagainst buffer concentration levels off as the rate becomes independent of buffer concentration. Trapping of T by diffusion-controlled reaction with hydroxide ion or proton transfer to a second molecule of attacking amine occurs in the reactions of piperazine and sarcosine with ~yridine-4-aldehyde.~.Similarly the synthesis and cleavage of ureas RNHCONH, shows no buffer catalysis when RNHz is basic but the reaction of 4-anisidine with cyanic acid shows 0 k + RNH + C II RNH2-I1 k-1 N H W. P. JENCKS AND J. M. SAYER general acid and base catalysis with nonlinear Bronsted plots and similar maximum rate constants for strong acids and bases.' With basic amines the intermediate T' is relatively stable (k-l < k, eqn (4)) but as the amine becomes less basic the expulsion of amine from T* becomes faster than proton transfer to form urea ( k- = 3 x 10' s-l k = 10' s-l) and the intermediate can be trapped by added catalysts through k and k,.The reaction with anisidine and the cleavage of 1-phenylcarbamoylimidazole exhibit a change in rate-determining step from proton transfer to C-N bond formation or cleavage as the buffer concentration is increased.8* Addition intermediates formed from the attack of nucleophiles on acyl compounds are usually less stable than those formed from aldehydes and ketones because of electron donation by resonance from the 0 N or S atom that helps to expel the nucleophile and stabilize the acyl compound. The value of k- for expulsion of the aliphatic amino group from T* in the intramolecular aminolysis of S-acetylmer- captoethylamine is 6.6 x 10' s-' and the intermediate is trapped by encounter with acids giving a nonlinear " Eigen-type " Bronsted plot (eqn (5)).lo T' A separate proton transfer step is also required by the dependence on acid concen- tration of the aminolysis rate and the product distribution from the hydrolysis of 2-methyl-A2-thiazoline which proceeds through the same intermediates.lo*' The estimated value of k- for the addition compound T* formed from phenyl acetate and methylamine is 3 x lo9 s-' and the available data are consistent with a mechanism for ester aminolysis in which general acid and base catalysis involves trapping of this intermediate by proton transfer (eqn (6)).12 The addition inter- mediate T' can also be generated upon hydrolysis of the corresponding imido ester. T+ 0 0- ZNH t II COR 'H I-N-C-OR I I T' kf I+ TO Amide - T- Amide The partitioning of this intermediate between ester and amide is independent of the pK of the leaving phenolate ion and shows nonlinear Bronsted plots for general acid catalysis as expected if the partitioning is controlled by proton transfer.I2 The rate constants of aminolysis reactions characteristically exhibit a large dependence on the GENERAL ACID-BASE CATALYZED REACTIONS basicity of the amine (Pnuc= 0.8-1.0) when the rate-determining step is a reaction of T* involving (a) encounter-controlled catalysis by general acids or bases (b) a proton switch to give TO;or (c) direct breakdown by expulsion of a relatively good leaving group via k*.c. PRE-ASSOCIATION OR " SPECTATOR " MECHANISMS When the rate constant for the breakdown of an intermediate becomes larger than that for diffusion apart of the intermediate and catalyst (e.g.kLi and k- H kt 'H I HN -t ;C-0 HN-C-0' R k-I R I -Products H k'l +H I HI BH+.N-C-O'-B.HN.>C-O B.HN-C-0-w-R kLt RI RI respectively in eqn (7)) the pathway of lowest free energy for the breakdown and formation of the intermediate must involve a preliminary association of the reactants and catalyst in an encounter complex; the catalyst may be present as a " spectator " during heavy atom rearrangement l4 When the catalyst does not provide appreciable stabilization of the transition state for formation of the intermediate the Bronsted a or p value for strong acids or bases will be zero but for weak acids or bases the subsequent proton transfer and separation of product and catalyst will become rate determining so that the Bronsted plot exhibits a break similar to that in simple diffusion-controlled proton transfer reactions.Since the pre-association mechanism provides a lower energy faster pathway than the diffusion-controlled mechanism the larger rate constants for strong acid or base catalysts will cause the break in the Bronsted plot to be shifted as shown in the upper line of fig. 1.15*l6 PKBH + FIG.I.-Bronsted plot illustrating that the break in the curve for a base-catalyzed pre-association mechanism is at a higher pK than that for a diffusioncoiitrolled proton transfer mechanism. W. P. JENCKS AND J. M. SAYER This mechanism has been suggested for general base catalysis of the attack of 2-methylthiosemicarbazide on p-chlorobenzaldehyde which shows a break at least 1.4 pK units above the expected pK of 3.1 for T* and gives a calculated Bronsted plot in good agreement with experiment based on a value of kL = 5 x loll s-l.Such a shift can provide presumptive evidence for a preassociation mechanism; the absence of a dependence of the rate on solvent viscosity l7 would provide further evidence. IT. CATALYSIS WITH TRANSITION STATE STABILIZATION A. HYDROGEN BONDING Structure-reactivity considerations predict that general acid catalysis of the attack of a nucleophile on the carbonyl group will involve more proton transfer (larger Bronsted a) for a more weakly basic nucleophile.l** l9 In accord with this the Bronsted a value of 1.0 for the formation and breakdown of hydrogen peroxide- p-chlorobenzaldehyde addition compounds is much larger than the a values for aldehyde and ketone hydration.2o Although the interpretation of Bronsted co- efficients has been questioned,21 it is an experimental fact that this a value of 1.0 means that polar substituents on the catalyst have the same effect on the stability of the transition state as they do on the equilibrium ionization reaction so that we can say that so far as substituent effects are concerned the catalyst in the transition state resembles the conjugate base of the acid. The fact that acid-catalyzed breakdown is 6 to 18 times faster for the p-methoxybenzaldehyde than for the p-chlorobenzaldehyde adduct means that significant C-0 bond cleavage has occurred in the transition state since even complete protonation should be favoured by only about two-fold in the p-methoxy compound.These data suggest that the rate-determining step is the formation and cleavage of the C-0 bond and that the essentially complete proton transfer in the transition state is stabilized by hydrogen bonding to the catalyst (eqn (8)). The rate constant for the solvated proton deviates downward by approx- imately 50-fold from the Bronsted line for carboxylic phosphoric and arsenic acids. This makes the detection of general acid catalysis possible in spite of the a value of 1.0 and suggests that bifunctional hydrogen bonding to the two acidic protons in the transition state may account for the relatively high activity of the latter catalysts.B. CONCERTED CATALYSIS The strongest evidence for concerted acid-base catalysis is for reversible additions of ROH to unsaturated centres with Bronsted p values (for ROH addition) and a values (for ROH expulsion) near 0.5. In the general acid catalyzed breakdown of an alcohol-phthalimidium addition compound (eqn (9)) the value of a increases steadily R R k GENERAL ACID-BASE CATALYZED REACTIONS from 0.49 to 0.74 as the leaving alcohol becomes more basic in the series from tri- fluoroethanol to As the catalyzing acid becomes stronger there is a decrease and then a reversal in the dependence of the rate on the pK of the leaving alcohol (Qlg changes from -0.23 for acetic acid to +0.24 for the proton).These results are inconsistent with a stepwise mechanism of catalysis and also cannot be explained by a hydrogen-bonding mechanism in which the proton rests in one well of a double potential well hydrogen bond during C-0 bond formation and break- down.22*23 If the alcohol leaves as the anion with hydrogen bonding to the catalyst there must be negative charge on the oxygen atom in the transition state (1). This is inconsistent with the faster acid catalyzed expulsion of ethanol than of trifluoroethanol. I 2 If the alcohol is first protonated and leaves with hydrogen bonding to the buffer base there must be positive charge on oxygen in the transition state (2). This is inconsistent with the faster expulsion of trifluoroethanol than of ethanol catalyzed by the weaker acids.The linear Bronsted plots and the gradual change in a and in sensitivity to leaving group pK support a concerted mechanism rather than a sudden shift from one to the other of the hydrogen-bonding mechanisms 1 and 2. The data may be described by reaction coordinate contour diagrams such as that shown in fig. 2gfor the 'acetic acid-trifluoroethanol reaction. Changes in the position +I 1 A-HO-C-AH O-C-RI RI FIG.2.-Possible reactioncoordinate contour diagram for the breakdown of a trifluoroethanol- phthalimidium ion addition compound catalyzed by acetic acid. The horizontal axis represents proton transfer and the vertical axis represents cleavage and formation of the C-0 bond. of the transition state with changing structure of the reactants and catalysts provide a rationalization for the observed changes in a and leaving-group effects.These changes correspond to an interaction coefficient l/c = 0.07 in the Cordes equation that interrelates these parameters. 22 The analogous general acid catalyzed hydrolysis W. P. JENCKS AND J. M. SAYER of benzaldehyde phenyl acetals exhibits an even larger sensitivity of changes in transition state structure to changing structure of the catalyst and leaving with l/c z 0.2. Evidently the energy gradients in the contour diagrams that permit these large shifts are unusually shallow. These shifts and the normal or low catalytic constants for the proton in these reactions appear inconsistent with an S-shaped reaction coordinate with only vertical motion for C-0 cleavage and the proton in a stable potential well in the transition state.The general acid catalyzed dehydration of carbinolamines to imines (eqn (10)) is a similar reaction in which the value of a increases from 0.62 to 0.73 as the basicity of the amine decreases in the series hydrazine (pK 8.3) to thiosemicarbazide (pK 1.9).25 This change is as expected from the increase in the energy of the lower part of the reaction coordinate diagram and corresponds to an interaction coefficient l/c2 = 0.02. The addition of methoxyamine to p-chlorobenzaldehyde is catalyzed by the proton through a class I1 (concerted or hydrogen bonded) mechanism (eqn (1 1)) concurrently with the previously mentioned class I stepwise mechani~m.~ As the pK of the attacking amine is decreased the dipolar addition intermediate becomes less stable the stepwise mechanism is correspondingly less favourable and only the class I1 mechanism is observed for the reaction with 2-methylthiosemicarbazide.Electron-donating substituents on the aldehyde also favour amine expulsion and destabilize the dipolar intermediate relative to the aldehyde and to the transition state of eqn (1 1). Accordingly the addition of semicarbazide to p-nitrobenzaldehyde proceeds through concurrent class I and class I1 mechanisms whereas only the class I1 mechanism is observed for p-methoxybenzaldehyde.26 Thus "harder " reactions that require more unstable intermediates are more likely to proceed through class I1 mechanisms. c.DIFFUSION-CONTROL LED CATA L Y Z ED RE ACT I0 N S When structural changes make an intermediate progressively less stable a point may be reached at which the reaction occurs with a strong catalyst at a diffusion- controlled rate but with a weak catalyst at a slower rate through a class I1 mechanism. This will result in a break in the Bronsted plot and other structure-reactivity correla- tions. For example strongly basic thiol anions add to acetaldehyde with rate- determining nucleophilic attack (kl eqn (12)) no buffer catalysis and essentially no dependence of the rate on nucleophile basicity but weakly basic anions exhibit a I k2 I RS-+ ;c=o RSCOH.A-RSCOH 4-A-I k-2 1 sharp downward deviation in the structure-reactivity correlation and a value of Pnuc = l.0.27928 This is because the rate-determining step becomes the separation of hydroxide ion from the product (k2,eqn (12) A-= OH-); in the reverse direction the diffusion-controlled encounter of hydroxide ion with the addition compound is rate determining (k20.9 x 1O1O M-1 s-l) .When the thiol is weakly basic the = intermediate T-is unstable so that thiol anion expulsion is faster than protonation by water; the estimated rate constants for methyl mercaptoacetate (pK = 7.8) are 48 GENERAL ACID-BASE CATALYZED REACTIONS k- = 5 x lo8 s-' and K,k2 = 2.5 x lo8s-'." For still less basic thiols k- is estimated to be in the range 1010-1012.4 s-' and a general acid catalyst must be present in the reacting complex for the reasons given in I-C.Accordingly these reactions exhibit general acid catalysis with a Bronsted a value of 0.2 and a small dependence of the rate on thiol anion basicity (Pnuc = 0.15 for two thiophenols). This is consistent with either hydrogen bonding or a concerted mechanism of catalysis. The position of the break in the Bronsted plot for this and other reactions of this kind does not correspond to the pK of the intermediate. The breaks in the nonlinear Bronsted plots for general base catalysis of the hydrazinolysis of acetylimidazole and for general acid and general base catalysis of the methoxyaminolysis of acetyl- triazole 29 are more than 2 pK units away from the estimated pK values of the addition intermediates. This is consistent with the mechanism of eqn (13) for these reactions 0 0-0 11 kl + I kd[Catj kc n RNH 4-,C-X RNH,-C-X L[T'.Cot] ,CNHR t Cat (I3) k-I I k-d in which strongly basic or acidic catalysts react with the intermediate T' at every encounter (k,) and weak catalysts cause a slower concerted breakdown of Tk (kc); the rate constants for strong acid and base catalysts in the acetyltriazole reaction are equal corresponding to a common value of kd.Mechanisms for acid catalyzed addition of a nucleophile N to a carbonyl group are summarized schematically in fig. 3. The preferred mechanism depends largely on the lifetimes of the intermediates. It is obvious (although not always recognized) that if an " intermediate " has too short a lifetime to exist (<10-'3-10-'4 s) a reaction cannot be stepwise and must be concerted.What is not yet known is whether concerted catalysis is possible when the intermediate does exist. The concurrent Type I and Type I1 reactions of methoxyamine and semicarbazide with benzaldehydes demonstrate that the intermediate T' has a significant lifetime (ca. s) with a barrier for C-N cleavage and that concurrent reaction pathways I I FIG.3.-Mechanisms for general acid catalysis of the addition of a nucleophile N to a carbonyl group. The mechanisms are drawn in the form of a reaction coordinate diagram but contour lines are omitted. The dashed lines represent free energy barriers for association and diffusion processes. * For HA = H20 based on pKa = 12.4 for To,from a measured pKa of 12.4for HOEtSCH20H (R. Kallen personal communication) and structure-reactivity correlation^,^^ and an assumed value of k-2 = 10'' M-' s-'.Note that if the breakdown is concerted the K; step is not required. W. P. JENCKS AND J. M. SAYER 49 inside and outside the central box of fig. 3 are possible. If there is no barrier for proton transfer in the T'*HA complex this intermediate does not exist and the reaction must be concerted. However we do not know whether there is such a barrier nor whether coiicerted catalysis is possible when the existence of this barrier provides a potential well for the complex. The mechanism is likely to be stepwise when the proton transfer occurs through one or more water molecules as it does in many simple proton transfer reactions 30 whereas a concerted mechanism is favoured when there is direct proton transfer between the reactant and catalyst and an initially unfavourable proton transfer suddenly becomes strongly favourable in the course of the reaction.R. P. Bell Acid-Base Curalysis (Oxford Univ. Press London 1941). R. P. Bell The Proton in Chemistry (Cornell Univ. Press Ithaca N.Y. 2nd ed. 1973). T. D. Stewart and H. P. Kung J. Amer. Chem. SOC., 1933 55 4813. J. Hine and F. C. Kokesh J. Anrer. Chenr. SOC. 1970 92 4383. S. Rosenberg S. M. Silver J. M. Sayer and W. P. Jencks J. Anier. Chern. SOC.,1974,96 7956. H. Diebler and R. N. F. Thorneley J. Amer. Chenz. SOC.,1973 95 996. R. N. F. Thorneley and H. Diebler J. Anrcr. Clrern. Soc. 1974. 96 1072. A. Williams and W. P. Jencks J. C. S. Perkin II 1974 1753 1760.A. F. Hegarty C. N. Hegarty and F. L. Scott J. C. S.,Perkin I/ 1974 1258. lo R. E. Barnett and W. P. Jencks J. Amer. Cliem. SOC., 1969 91 2358. R. B. Martin and R. I. Hedrick J. Anzer. Chetn. SOC.,1962 84 106; R. B. Martin R. I. Hcdrick and A. Parcell J. Org. Cliern. 1964 29 3197. A. C. Sattcrthwait and W. P. Jencks J. Atner. Cficm. SOC. 1974 96 7018 7031. l3 W. P. Jencks and K. Salvesen J. Amer. Cfietn. SOC. 1971 93 1419. L. D. Kershner and R. L. Schowen J. Amer. Clienr. SOC.,1971 93 2014. l5 M. I. Page and W. P. Jencks J. Amer. Chem. SOC. 1972 94 8828. J. kl. Sayer and W. P. Jencks J. Amer. Chenr. SOC.,1973 95 5637. l7 C. Cerjan and R. E. Barnett J. Phys. CJietn. 1972 76 1192. E. H. Cordes and W. P. Jencks J. Amer. Cfienr. Soc. 1962 84 4319.l9 W. P. Jencks Chem. Rev. 1972 72 705. 2o E. Sander and W. P. Jencks J. Anrer. Chern. SOC., 1968 90 4377. 21 F. G. Bordwell and W. J. Boyle Jr. J. Anier. Chetn. SOC.,1972 94 3907; A. J. Kresge Curiud. J. Clienr. 1974 52 1897. 22 N. Gravitz and W. P. Jencks J. Amer. CJiern. SOC. 1974 96 507. 23 J. Hine J. Amer. Cliern. Soc. 1972 94 5766. '' B. Capon personal communication. 2.i J. M. Sayer M. Peskin and W. P. Jencks J. Anrer. Cfreni. SOC. 1973 95 4277. 26 J. M. Sayer B. Pinsky A. Schonbrunn and W. Washstien J. Amer. Clrem. SOC.,1974,96 7998. 27 G. E. Licnhard and W. P. Jencks J. Anier. Chenr. SOC. 1966 88 3982. 2R R. E. Barnett and W. P. Jcncks J. Ainer. Chenr. SOC., 1969 91 6758. 2y J. P. Fox and W. P. Jencks J. Anrer. Chem. Soc. 1974 96 1436. 30 D. Roscnthal and E. Grunwald. J. Anrer. Chenr. SOC.,1972 94. 5956 and references therein.