The aldol reaction is a powerful means of forming carbon–carbon bonds in organic chemistry.[1][2][3] Discovered independently by Charles-Adolphe Wurtz[4][5][6] and Alexander Porfyrevich Borodin in 1872,[7] the reaction combines two carbonyl compounds (the original experiments used aldehydes) to form a new β-hydroxy carbonyl compound. These products are known as aldols, from the aldehyde + alcohol, a structural motif seen in many of the products. Aldol structural units are found in many important molecules, whether naturally occurring or synthetic.[8][9][10] For example, the aldol reaction has been used in the large-scale production of the commodity chemical pentaerythritol[11] and the synthesis of the heart disease drug Lipitor (atorvastatin).[12][13]
The aldol reaction is powerful because it unites two relatively simple molecules into a more complex one. Increased complexity arises because up to two new stereogenic centers (on the α- and β-carbon of the aldol adduct, marked with asterisks in the scheme below) are formed. Modern methodology is capable of not only allowing aldol reactions to proceed in high yield but also controlling both the relative and absolute stereochemical configuration of these stereocenters. This ability to selectively synthesize a particular stereoisomer is significant because different stereoisomers can have very different chemical or biological properties.
For example, stereogenic aldol units are especially common in polyketides, a class of molecules found in biological organisms. In nature, polyketides are synthesized by enzymes that effect iterative Claisen condensations. The 1,3-dicarbonyl products of these reactions can then be variously derivatized to produce a wide variety of interesting structures. Often, such derivitization involves the reduction of one of the carbonyl groups, producing the aldol subunit. Some of these structures have potent biological properties: the potent immunosuppressant FK506, the anti-tumor agent discodermolide, or the antifungal agent amphotericin B, for example. Although the synthesis of many such compounds was once considered nearly impossible, aldol methodology has allowed their efficient synthesis in many cases.[14]
A typical modern aldol addition reaction, shown above, might involve the nucleophilic addition of a ketone enolate to an aldehyde. Once formed, the aldol product can sometimes lose a molecule of water to form an α,β-unsaturated carbonyl compound. This is called aldol condensation. A variety of nucleophiles may be employed in the aldol reaction, including the enols, enolates, and enol ethers of ketones, aldehydes, and many other carbonyl compounds. The electrophilic partner is usually an aldehyde or ketone (many variations, such as the Mannich reaction, exist). When the nucleophile and electrophile are different, the reaction is called a crossed aldol reaction; on the converse, when the nucleophile and electrophile are the same, the reaction is called aldol dimerization.
Contents |
The aldol reaction may proceed via two fundamentally different mechanisms. Carbonyl compounds, such as aldehydes and ketones, can be converted to enols or enol ethers. These compounds, being nucleophilic at the α-carbon, can attack especially reactive protonated carbonyls such as protonated aldehydes. This is the "enol mechanism." Carbonyl compounds, being carbon acids, can also be deprotonated to form enolates, which are much more nucleophilic than enols or enol ethers and can attack electrophiles directly. The usual electrophile is an aldehyde, since ketones are much less reactive. This is the "enolate mechanism."
If the conditions are particularly harsh (e.g., NaOMe, MeOH, reflux), condensation may occur, but this can usually be avoided with mild reagents and low temperatures (e.g., LDA (a strong base), THF, −78 °C). Although the aldol addition usually proceeds to near completion under irreversible conditions, the isolated aldol adducts are sensitive to base-induced retro-aldol cleavage to return starting materials. In contrast, retro-aldol condensations are rare, but possible.[15]
When an acid catalyst is used, the initial step in the reaction mechanism involves acid-catalyzed tautomerization of the carbonyl compound to the enol. The acid also serves to activate the carbonyl group of another molecule by protonation, rendering it highly electrophilic. The enol is nucleophilic at the α-carbon, allowing it to attack the protonated carbonyl compound, leading to the aldol after deprotonation. This usually dehydrates to give the unsaturated carbonyl compound. The scheme shows a typical acid-catalyzed self-condensation of an aldehyde.
Acid-catalyzed aldol mechanism
Acid-catalyzed dehydration
If the catalyst is a moderate base such as hydroxide ion or an alkoxide, the aldol reaction occurs via nucleophilic attack by the resonance-stabilized enolate on the carbonyl group of another molecule. The product is the alkoxide salt of the aldol product. The aldol itself is then formed, and it may then undergo dehydration to give the unsaturated carbonyl compound. The scheme shows a simple mechanism for the base-catalyzed aldol reaction of an aldehyde with itself.
Base-catalyzed aldol reaction (shown using −OCH3 as base)
Base-catalyzed dehydration (frequently written incorrectly as a single step, see E1cB elimination reaction)
Although only a catalytic amount of base is required in some cases, the more usual procedure is to use a stoichiometric amount of a strong base such as LDA or NaHMDS. In this case, enolate formation is irreversible, and the aldol product is not formed until the metal alkoxide of the aldol product is protonated in a separate workup step.
More refined forms of the mechanism are known. In 1957, Zimmerman and Traxler proposed that some aldol reactions have "six-membered transition states having a chair conformation."[16] This is now known as the Zimmerman–Traxler model. E-enolates give rise to anti products, whereas Z-enolates give rise to syn products. The factors that control selectivity are the preference for placing substituents equatorially in six-membered transition states and the avoidance of syn-pentane interactions, respectively.[17] E and Z refer to the cis-trans stereochemical relationship between the enolate oxygen bearing the positive counterion and the highest priority group on the alpha carbon. In reality, only some metals such as lithium and boron reliably follow the Zimmerman–Traxler model. Thus, in some cases, the stereochemical outcome of the reaction may be unpredictable.
The problem of "control" in the aldol addition is best demonstrated by an example. Consider the outcome of this hypothetical reaction:
In this reaction, two unsymmetrical ketones are being condensed using sodium ethoxide. The basicity of sodium ethoxide is such that it cannot fully deprotonate either of the ketones, but can produce small amounts of the sodium enolate of both ketones. This means that, in addition to being potential aldol electrophiles, both ketones may also act as nucleophiles via their sodium enolate. Two electrophiles and two nucleophiles, then, have potential to result in four possible products:
Thus, if one wishes to obtain only one of the cross-products, one must control which carbonyl becomes the nucleophilic enol/enolate and which remains in its electrophilic carbonyl form.
The simplest control is if only one of the reactants has acidic protons, and only this molecule forms the enolate. For example, the addition of diethyl malonate into benzaldehyde would produce only one product. Only the malonate has α hydrogens, so it is the nucleophilic partner, whereas the non-enolizeable benzaldehyde can only be the electrophile:
The malonate is particularly easy to deprotonate because the α position is flanked by more than one carbonyl. Double-activation makes the enolate more stable, so not as strong a base is required to form it. An extension of this effect can allow control over which of the two carbonyl reactants becomes the enolate even if both do have α hydrogens. If one partner is considerably more acidic than the other, the most acidic proton is abstracted by the base and an enolate is formed at that carbonyl while the carbonyl that is less acidic is not affected by the base. This type of control works only if the difference in acidity is large enough and no excess of base is used for the reaction. A typical substrate for this situation is when the deprotonatable position is activated by more than one carbonyl-like group. Common examples include a CH2 group flanked by two carbonyls or nitriles (see for example the Knoevenagel condensation and the first steps of the Malonic ester synthesis).
One common solution is to form the enolate of one partner first, and then add the other partner under kinetic control.[18] Kinetic control means that the forward aldol addition reaction must be significantly faster than the reverse retro-aldol reaction. For this approach to succeed, two other conditions must also be satisfied; it must be possible to quantitatively form the enolate of one partner, and the forward aldol reaction must be significantly faster than the transfer of the enolate from one partner to another. Common kinetic control conditions involve the formation of the enolate of a ketone with LDA at −78 °C, followed by the slow addition of an aldehyde.
The enolate may be formed by using a strong base ("hard conditions") or using a Lewis acid and a weak base ("soft conditions"):
For deprotonation to occur, the stereoelectronic requirement is that the alpha-C-H sigma bond must be able to overlap with the pi* orbital of the carbonyl:
Extensive studies have been performed on the formation of enolates under many different conditions. It is now possible to generate, in most cases, the desired enolate geometry:[19]
For ketones, most enolization conditions give Z enolates. For esters, most enolization conditions give E enolates. The addition of HMPA is known to reverse the stereoselectivity of deprotonation.
The stereoselective formation of enolates has been rationalized with the Ireland model,[20][21][22][23] although its validity is somewhat questionable. In most cases, it is not known which, if any, intermediates are monomeric or oligomeric in nature; nonetheless, the Ireland model remains a useful tool for understanding enolates.
In the Ireland model, the deprotonation is assumed to proceed by a six-membered or cyclic[24] monomeric transition state. The larger of the two substituents on the electrophile (in the case above, methyl is larger than proton) adopts an equatorial disposition in the favored transition state, leading to a preference for E enolates. The model clearly fails in many cases; for example, if the solvent mixture is changed from THF to 23% HMPA-THF (as seen above), the enolate geometry is reversed, which is inconsistent with this model and its cyclic transition state.
If an unsymmetrical ketone is subjected to base, it has the potential to form two regioisomeric enolates (ignoring enolate geometry). For example:
The trisubstituted enolate is considered the kinetic enolate, while the tetrasubstituted enolate is considered the thermodynamic enolate. The alpha hydrogen deprotonated to form the kinetic enolate is less hindered, and therefore deprotonated more quickly. In general, tetrasubstituted olefins are more stable than trisubstituted olefins due to hyperconjugative stabilization. The ratio of enolate regioisomers is heavily influenced by the choice of base. For the above example, kinetic control may be established with LDA at −78 °C, giving 99:1 selectivity of kinetic: thermodynamic enolate, while thermodynamic control may be established with triphenylmethyllithium at room temperature, giving 10:90 selectivity.
In general, kinetic enolates are favored by cold temperatures, conditions that give relatively ionic metal–oxygen bonding, and rapid deprotonation using a slight excess of a strong, sterically hindered base. The large base only deprotonates the more accessible hydrogen, and the low temperatures and excess base help avoid equilibration to the more stable alternate enolate after initial enolate formation. Thermodynamic enolates are favored by longer equilibration times at higher temperatures, conditions that give relatively covalent metal–oxygen bonding, and use of a slight sub-stoichiometric amount of strong base. By using insufficient base to deprotonate all of the carbonyl molecules, the enolates and carbonyls can exchange protons with each other and equilibrate to their more stable isomer. Using various metals and solvents can provide control over the amount of ionic character in the metal–oxygen bond.
The aldol reaction is particularly useful because two new stereogenic centers are generated in one reaction. Extensive research has been performed to understand the reaction mechanism and improve the selectivity observed under many different conditions. The syn/anti convention is commonly used to denote the relative stereochemistry at the α- and β-carbon.
The convention applies when propionate (or higher order) nucleophiles are added to aldehydes. The R group of the ketone and the R' group of the aldehyde are aligned in a "zig zag" pattern in the plane of the paper (or screen), and the disposition of the formed stereocenters is deemed syn or anti, depending if they are on the same or opposite sides of the main chain.
Older papers use the erythro/threo nomenclature familiar from carbohydrate chemistry.
There is no significant difference between the level of stereoinduction observed with E and Z enolates. Each alkene geometry leads primarily to one specific relative stereochemistry in the product, E giving anti and Z giving syn:[19]
The enolate metal cation may play a large role in determining the level of stereoselectivity in the aldol reaction. Boron is often used[25][26] because its bond lengths are significantly shorter than that of other metals such as lithium, aluminium, or magnesium.
For example, boron–carbon and boron–oxygen bonds are 1.4–1.5 Å and 1.5–1.6 Å in length, respectively, whereas typical metal-carbon and metal-oxygen bonds are 1.9–2.2 Å and 2.0–2.2 Å in length, respectively. The use of boron rather than a metal "tightens" the transition state and gives greater stereoselectivity in the reaction.[27] Thus the above reaction gives a syn:anti ratio of 80:20 using a lithium enolate compared to 97:3 using a bibutylboron enolate.
The aldol reaction may exhibit "substrate-based stereocontrol", in which existing chirality on either reactant influences the stereochemical outcome of the reaction. This has been extensively studied, and in many cases, one can predict the sense of asymmetric induction, if not the absolute level of diastereoselectivity. If the enolate contains a stereocenter in the alpha position, excellent stereocontrol may be realized.
In the case of an E enolate, the dominant control element is allylic 1,3-strain whereas in the case of a Z enolate, the dominant control element is the avoidance of 1,3-diaxial interactions. The general model is presented below:
For clarity, the stereocenter on the enolate has been epimerized; in reality, the opposite diastereoface of the aldehyde would have been attacked. In both cases, the 1,3-syn diastereomer is favored. There are many examples of this type of stereocontrol:[28]
When enolates attacks aldehydes with an alpha stereocenter, excellent stereocontrol is also possible. The general observation is that E enolates exhibit Felkin diastereoface selection, while Z enolates exhibit anti-Felkin selectivity. The general model[29][30] is presented below:
Since Z enolates must react through a transition state that contains either a destabilizing syn-pentane interaction or an anti-Felkin rotamer, Z-enolates exhibit lower levels of diastereoselectivity in this case. Some examples are presented below:[31][32]
If both the enolate and the aldehyde both contain pre-existing chirality, then the outcome of the "double stereodifferentiating" aldol reaction may be predicted using a merged stereochemical model that takes into account the enolate facial bias, enolate geometry, and aldehyde facial bias.[33] Several examples of the application of this model are given below:[32]
Modern organic syntheses now require the synthesis of compounds in enantiopure form. Since the aldol addition reaction creates two new stereocenters, up to four stereoisomers may result.
Many methods which control both relative stereochemistry (i.e., syn or anti, as discussed above) and absolute stereochemistry (i.e., R or S) have been developed.
A widely used method is the Evans' acyl oxazolidinone method.[34][35] Developed in the late 1970s and 1980s by David A. Evans and coworkers, the method works by temporarily creating a chiral enolate by appending a chiral auxiliary. The pre-existing chirality from the auxiliary is then transferred to the aldol adduct by performing a diastereoselective aldol reaction. Upon subsequent removal of the auxiliary, the desired aldol stereoisomer is revealed.
In the case of the Evans' method, the chiral auxiliary appended is an oxazolidinone, and the resulting carbonyl compound is an imide. A number of oxazolidinones are now readily available in both enantiomeric forms. These may cost roughly $10–$20 US dollars per gram, rendering them relatively expensive.
The acylation of an oxazolidinone is a convenient procedure, and is informally referred to as "loading done". Z-enolates, leading to syn-aldol adducts, can be reliably formed using boron-mediated soft enolization:[36]
Often, a single diastereomer may be obtained by one crystallization of the aldol adduct. However, anti-aldol adducts cannot be obtained reliably with the Evans method. Despite the cost and the limitation to give only syn adducts, the method's superior reliability, ease of use, and versatility render it the method of choice in many situations. Many methods are available for the cleavage of the auxiliary:[37]
Upon construction of the imide, both syn- and anti-selective aldol addition reactions may be performed, allowing the assemblage of three of the four possible stereoarrays: syn selective:[38] and anti selective:[39]
In the syn-selective reactions, both enolization methods give the Z enolate, as expected; however, the stereochemical outcome of the reaction is controlled by the methyl stereocenter, rather than the chirality of the oxazolidinone. The methods described allow the stereoselective assembly of polyketides, a class of natural products that often feature the aldol retron.
Recent methodology now allows a much wider variety of aldol reactions to be conducted, often with a catalytic amount of chiral ligand. When reactions employ small amounts of enantiomerically pure ligands to induce the formation of enantiomerically pure products, the reactions are typically termed "catalytic, asymmetric"; for example, many different catalytic, asymmetric aldol reactions are now available.
A key limitation to the chiral auxiliary approach described previously is the failure of N-acetyl imides to react selectively. An early approach was to use a temporary thioether group:[37][40]
The Mukaiyama aldol reaction is the nucleophilic addition of silyl enol ethers to aldehydes catalyzed by a Lewis acid such as boron trifluoride or titanium tetrachloride.[41][42] The Mukaiyama aldol reaction does not follow the Zimmerman-Traxler model. Carreira has described particularly useful asymmetric methodology with silyl ketene acetals, noteworthy for its high levels of enantioselectivity and wide substrate scope.[43]
The method works on unbranched aliphatic aldehydes, which are often poor electrophiles for catalytic, asymmetric processes. This may be due to poor electronic and steric differentiation between their enantiofaces.
The analogous vinylogous Mukaiyama aldol process can also be rendered catalytic and asymmetric. The example shown below works efficiently for aromatic (but not aliphatic) aldehydes and the mechanism is believed to involve a chiral, metal-bound dienolate.[44][45]
A more recent version of the Evans' auxiliary is the Crimmins thiazolidinethione.[46][47] The yields, diastereoselectivities, and enantioselectivities of the reaction are, in general, high, although not as high as in comparable Evans cases. Unlike the Evans auxiliary, however, the thiazoldinethione can perform acetate aldol reactions (ref: Crimmins, Org. Lett. 2007, 9(1), 149–152.) and can produce the "Evans syn" or "non-Evans syn" adducts by simply varying the amount of (−)-sparteine. The reaction is believed to proceed via six-membered, titanium-bound transition states, analogous to the proposed transition states for the Evans auxiliary. NOTE: the structure of sparteine shown below is missing a N atom.
A more recent development is the use of chiral secondary amine catalysts. These secondary amines form transient enamines when exposed to ketones, which may react enantioselectively[48] with suitable aldehyde electrophiles. The amine reacts with the carbonyl to form an enamine, the enamine acts as an enol-like nucleophile, and then the amine is released from the product all—the amine itself is a catalyst. This enamine catalysis method is a type of organocatalysis, since the catalyst is entirely based on a small organic molecule. In a seminal example, proline efficiently catalyzed the cyclization of a triketone:
This reaction is known as the Hajos-Parrish reaction[49][50] (also known as the Hajos-Parrish-Eder-Sauer-Wiechert reaction, referring to a contemporaneous report from Schering of the reaction under harsher conditions).[51] Under the Hajos-Parrish conditions only a catalytic amount of proline is necessary (3 mol%). There is no danger of an achiral background reaction because the transient enamine intermediates are much more nucleophilic than their parent ketone enols. This strategy is particularly powerful because it offers a simple way of generating enantioselectivity in reactions without using transition metals, which have the possible disadvantages of being toxic or expensive.
It is interesting to note that proline-catalyzed aldol reactions do not show any non-linear effects (the enantioselectivity of the products is directly proportional to the enantiopurity of the catalyst). Combined with isotopic labelling evidence and computational studies, the proposed reaction mechanism for proline-catalyzed aldol reactions is as follows:[52]
This strategy allows the otherwise challenging cross-aldol reaction between two aldehydes. In general, cross-aldol reactions between aldehydes are typically challenging because they can polymerize easily or react unselectively to give a statistical mixture of products. The first example is shown below:[53]
In contrast to the preference for syn adducts typically observed in enolate-based aldol additions, these organocatalyzed aldol additions are anti-selective. In many cases, the organocatalytic conditions are mild enough to avoid polymerization. However, selectivity requires the slow syringe-pump controlled addition of the desired electrophilic partner because both reacting partners typically have enolizable protons. If one aldehyde has no enolizable protons or alpha- or beta-branching, additional control can be achieved.
An elegant demonstration of the power of asymmetric organocatalytic aldol reactions was disclosed by MacMillan and coworkers in 2004 in their synthesis of differentially protected carbohydrates. While traditional synthetic methods accomplish the synthesis of hexoses using variations of iterative protection-deprotection strategies, requiring 8–14 steps, organocatalysis can access many of the same substrates using an efficient two-step protocol involving the proline-catalyzed dimerization of alpha-oxyaldehydes followed by tandem Mukaiyama aldol cyclization.
The aldol dimerization of alpha-oxyaldehydes requires that the aldol adduct, itself an aldehyde, be inert to further aldol reactions.[54] Earlier studies revealed that aldehydes bearing alpha-alkyloxy or alpha-silyloxy substituents were suitable for this reaction, while aldehydes bearing Electron-withdrawing groups such as acetoxy were unreactive. The protected erythrose product could then be converted to four possible sugars via Mukaiyama aldol addition followed by lactol formation. This requires appropriate diastereocontrol in the Mukaiyama aldol addition and the product silyloxycarbenium ion to preferentially cyclize, rather than undergo further aldol reaction. In the end, glucose, mannose, and allose were synthesized:
In the usual aldol addition, a carbonyl compound is deprotonated to form the enolate. The enolate is added to an aldehyde or ketone, which forms an alkoxide, which is then protonated on workup. A superior method, in principle, would avoid the requirement for a multistep sequence in favor of a "direct" reaction that could be done in a single process step. One idea is to generate the enolate using a metal catalyst that is released after the aldol addition mechanism. The general problem is that the addition generates an alkoxide, which is much more basic than the starting materials. This product binds tightly to the metal, preventing it from reacting with additional carbonyl reactants.
One approach, demonstrated by Evans, is to silylate the aldol adduct.[55][56] A silicon reagent such as TMSCl is added in the reaction, which replaces the metal on the alkoxide, allowing turnover of the metal catalyst. Minimizing the number of reaction steps and amount of reactive chemicals used leads to a cost-effective and industrially useful reaction.
A more recent biomimetic approach by Shair uses beta-thioketoacids as the nucleophile.[57] The ketoacid moiety is decarboxylated in situ. The process is similar to the way malonyl-CoA is used by Polyketide synthases. The chiral ligand is case is a bisoxazoline). Interestingly, aromatic and branched aliphatic aldehydes are typically poor substrates.
Examples of aldol reactions in biochemistry include the splitting of fructose-1,6-bisphosphate into dihydroxyacetone and glyceraldehyde-3-phosphate in the second stage of glycolysis, which is an example of a reverse ("retro") aldol reaction catalyzed by the enzyme aldolase A.