**1. Introduction**

The asymmetric intermolecular aldol addition catalysed by (*S*)-proline, proposed by List and co-workers in 2000 [1], is the prototype of enamine-based organocatalysis [2,3]. Proline is the smallest airand water-stable bifunctional catalyst; it is inexpensive, non-toxic, and available in both enantiomeric forms. It has been proven to catalyse enantioselective α-functionalizations of carbonyl compounds (aldol and Mannich reactions, Michael additions, α-halogenations, oxygenations, aminations, and so on), adopting reaction protocols that do not require inert atmosphere and anhydrous conditions and are carried out at room temperature [4,5].

The proline-catalysed aldol reaction [4,6–19] has been object of in-depth analyses after the first mechanistic rationale proposed by List and Houck [20], in particular, fundamental contributions have been given by Seebach [21], Armstrong and Blackmond [22,23], Sharma and Sunoj [24], Benaglia [25], and Gschwind [26,27]. However, proline scarce solubility in most organic solvents has limited its use in dimethylsulfoxide (DMSO), acetonitrile, or dimethylformamide (DMF). Moreover, proline often displays poor activity, requiring the use of high catalyst loadings and high reaction times, sometimes with unsatisfactory stereocontrol [1,28–36]. Because, in several cases, proline-catalysed aldol reactions are still underdeveloped, the last 15 years have witnessed an intense effort aimed at modifying the proline scaffold, following two directions: (1) the carboxylic group is replaced by a new hydrogen-bonding donor, such as a tetrazole, or by a sterically demanding group, such as the Hayashi–Jørgensen diarylmethanol and related compounds, as exhaustively reviewed by Trost [37]; and (2) the carboxylic group is retained and a supplementary substituent is bound to the proline scaffold. The new substituent, generally installed on the 4-OH group of 4-hydroxyproline, may play different roles: (*i*) it modifies the solubility profile of the parent amino acid, expanding the solvent choice to further classes [38]; and/or (*ii*) it enhances the catalyst activity and stereoselectivity, allowing a reduction of catalyst loading and reaction time; and/or (*iii*) it allows the catalyst immobilization on a solid support [39–56], adopting a biphasic condition reaction protocol. Despite that high levels of reactivity and selectivity have been achieved with these modified prolines, most of the aforementioned proline derivatives require multi-step syntheses, dramatically increasing the catalyst cost, a severe limitation particularly when the catalyst cannot be e fficiently recycled.

After having contributed to the synthesis of prolines, mostly modified with the incorporation of ionic tags on 4-position, and obtaining excellent results in terms of activity and stereoselectivity as well as catalyst recyclability [57–66], we decided to go back to the parent unmodified (*S*)-proline. We envisaged to improve the performance of this small, stable, inexpensive, non-toxic, and easily available organocatalyst exploring new experimental conditions.

The role of the solvent in determining the aldol reaction e fficiency was also addressed by other authors. Invariably, the use of unmodified proline (without additives) forced to choose polar aprotic solvents to obtain acceptable yields and selectivities [1,11,13–15,28–36,44,45]. A peculiar case was represented by ionic liquids (ILs) [67–72], which allowed in a few cases to decrease the catalyst loading (up to 1 mol%) and, during the work-up, to confine proline in a separate phase, enabling a simple product isolation and the reuse of the catalytic system. In recent literature, attempts are reported where proline is used in acetone/CHCl3 mixtures [73], in DMF at 4 ◦C (a condition that often requires several days) [74], in *tert*-butyl methyl ether (MTBE) [75], in deep eutectic solvents [76,77], or under solvent-free conditions, with [78–80] or without [81] the ball milling approach. However, many issues associated with the use of proline remain unsolved and polar aprotic solvents are characterized by several undesirable features (toxicity, high production cost, high environmental impact, di fficult product recovery) [82–84].

The use of unmodified proline can be also combined with additives [85,86], such as water [85–87], acids [85,86,88], diols [89,90], amines [85,86], or thioureas [91–93]. In these cases, the additive can tune the solubility, the reactivity, and/or the stereoselectivity of native proline, making the asymmetric aldol process more e fficient [18]. In one case, the addition of achiral guanidinium salts as additives allows to switch the diastereoselectivity as a function of the counterion, for example, tetrafluoroborate versus tetraphenylborate [94]. Nevertheless, the achieved performance is not optimal ye<sup>t</sup> (long reaction times, stereocontrol strongly depending on the substrate) and some drawbacks are still present, such as high proline loadings and the cost of the not recovered chiral additive. Significant advances were accomplished employing metal salts as additives [95–106]. In particular, Reiser and co-workers developed a strategy based on cobalt(II)-proline complexes, which ensured excellent results in direct aldol reactions involving aromatic aldehydes [106]. However, several disadvantages lead the avoidance of the use of metals, especially at an industrial level (costs, toxicity, environmental impact, limited sources) [107–109].

In the present work, we aim to avoid the use of both polar aprotic solvents and additives (being sometimes expensive, mostly non-recoverable, and contaminants, used in non-generalizable procedures), in order to develop an efficient and sustainable organocatalyzed aldol condensation protocol, which can be interesting from a scale-up and an industrial point of view. In particular, our goals are as follows: (1) a reduction of the process costs, related to employed solvents and reagents, but also purification and waste disposal; and (2) an improved reactivity, especially for poorly reactive substrates. We planned to achieve these objectives by using the following: (*i*) the native proline, a small, stable, inexpensive, and non-toxic organocatalyst; and (*ii*) the minimum amount of a low-cost, non-toxic reaction solvent, enabling a good process e fficiency and a simple and inexpensive final reaction work-up.

A number of research groups noticed that protic media were not suitable for aldol condensations promoted exclusively by native proline [15,29–33,87]. However, despite a plethora of studies focused on the use and the role of water (as solvent, co-solvent, or additive) [31,32,34,87,100–105,110,111], very few authors extended their investigations to alcohols [15,29,31,42,102,106], discouraged by the generally observed poor diastereo- and enantioselectivity. Only when proline was used in combination with metal salts as additives, the use of methanol as solvent [106] or co-solvent [102] a fforded acceptable results.

Intrigued by the few data available on the proline-mediated aldol condensation employing methanol, a prototypical green solvent [84,112] also in terms of LCA (life-cycle assessment) [113], we decided to

explore in depth the impact of methanol on the asymmetric intermolecular aldol condensation promoted by unmodified (*S*)-proline. It should be stressed, however, that efficient organocatalyzed aldol condensations invariably require a large excess of a liquid donor ketone (5–10 equivalents) that must thus be considered as a part of the reaction solvent-system.

### **2. Results and Discussion**

#### *2.1. Optimization of the Reaction Protocol*

As model reaction, we selected the (*S*)-proline–catalysed aldol condensation between cyclohexanone **1a** and aromatic aldehydes **2** (Scheme 1). At the outset, we confirmed the low performance of proline in terms of stereocontrol in pure methanol, but soon we realized that the simple use of a hydroalcoholic solution as the reaction medium was highly profitable. Here, we report a comparison among (*S*)-proline–catalysed reactions between cyclohexanone **1a** and four different aromatic aldehydes **2a**–**d**, carried out in methanol/water (2/1 V/V), pure water, and pure methanol, respectively, all other parameters being kept identical (Table 1). The 2/1 V/V methanol/water mixture composition ensures that the aldol reaction takes place under homogeneous conditions.

**Scheme 1.** The benchmark aldol reaction.


**Table 1.** Comparison of different protic reaction media 1.

1 Reaction conditions: **1a** (5 equiv.), **2** (0.3 mmol), (*S*)-proline (10 mol%), rt, MeOH/H2O (20 μL/10 μL, 2/1 V/V) or H2O (10 μL), or MeOH (20 μL). 2 Determined by 1H NMR on the crude mixture. 3 Determined by chiral stationary phase (CSP)-HPLC on the crude mixture. 4 Here, 20 mol% of (*S*)-proline was used. rt = room temperature, h = hours, nd = not determined.

The data collected in Table 1 demonstrate the crucial role of water; if in pure water conversions are the lowest, enantioselectivity reaches the highest values. On the other hand, pure methanol displays the highest reactivity and the poorest stereocontrol. The 2/1 V/V methanol/water solution is able to combine the pros of the two pure solvents, providing the same conversions of pure methanol and almost the same *ee*s and good *dr*s observed in pure water.

In Table 2, the results are reported when the 2/1 V/V methanol/water solution was applied to aldol reactions between cyclohexanone **1a** and other aromatic aldehydes **2e**–**i** (Table 2).


**Table 2.** MeOH/H2O-based protocol applied to di fferent aromatic aldehydes **2** 1.

1 Reaction conditions: **1a** (5 equiv.), **2** (0.3 mmol), (*S*)-proline (10 mol%), MeOH/H2O (20 μL/10 μL), rt. 2 Determined by 1H NMR on the crude mixture. 3 Determined by CSP-HPLC on the crude mixture. 4 Here, 20 mol% of (*S*)-proline was used.

With the most reactive electron-poor aldehydes (**2a**, **2b**, **2e**, and **2f**), high conversion and high stereocontrol were achieved in only 19 h. Moreover, these results are excellent if compared with those reported in the literature for analogous transformations promoted by unmodified (*S*)-proline and exploiting more complex protocols [114–117]. Unfortunately, the limitations of the proline-catalysed aldol reactions were not completely overcome. In fact, electron-rich aromatic aldehydes were confirmed to be less reactive, requiring longer reaction times. More in detail, for substrates **2g** and **2h**, the conversions reached after 19 and 24 h, respectively, were modest; nevertheless, it is worth mentioning that the enantio- and the diastereoselectivities were both higher than those reported so far by proline-based protocols [118,119]. As far as the electron-rich *p*-methoxy benzaldehyde **2i** is concerned, the only example with proline (20 mol%) in DMSO reported a low conversion (15%) and absence of diastereoselection [105]. The e ffect on product conversion was even poorer when a Lewis acid and water were added. Exploiting our MeOH/H2O-based protocol, the product conversion remained poor (18%), but the reaction proceeded with good enantio- (90% *ee*) and diastereoselectivity (*anti*/*syn* = 86:14).

Once the performance of native proline in 2/1 V/V methanol/water solution had been examined, we explored the e ffect of a more methanol-rich aqueous mixture. In Table 3, aldol reactions of cyclohexanone **1a** and di fferent aldehydes **2** in 2/1 V/V and 4/1 V/V solutions are compared.

Doubling the methanol volume (40 μL), the conversions significantly improved with all the tested aldehydes, while maintaining an excellent to remarkable level of enantiocontrol (Table 3). The most reactive substrates (**2a**, **2b**, **2e**, and **2f**) provided excellent conversions in only 4 h, demonstrating an unprecedented reactivity of proline. Moreover, interesting amounts of product were obtained exploiting these reaction conditions for less electrophilic aldehydes as well (**2c,2d**, **2g,2i**; Table 3). Concerning the diastereocontrol, a slight drop of the *anti*/*syn* ratio was observed with some aldehydes when the volume of methanol was increased. Conversely, for benzaldehyde **2d** and 2-naphthyl aldehyde **2h**, the diastereoselection lightly improved. In the case of benzaldehyde **2d**, the better performance could be the result of the reduced amount of catalyst (10 mol%), exploitable thanks to the higher proline reactivity reached with larger amounts of methanol. In the case of the most reactive 4-nitrobenzaldehyde **2a**, we solved the problem of diastereoselectivity drop by adding the methanol amount in two portions (one half after 2 h), completely restoring the diastereocontrol, while maintaining a high reaction rate. In general, a good diastereoselectivity level is retained with this protocol (4/1 V/V methanol/water) compared with the literature data [114–119]. At the same time, reaction rates are significantly enhanced. Therefore, these reaction conditions represent the best trade-o ff between reactivity and stereoselectivity. In Table 3, this optimized protocol was extended to some other aldehydes (**2j**–**<sup>m</sup>**), with good results compared with the literature data [120–123]. In particular, aliphatic aldehyde **2j**, known as poorly responsive in this kind of organocatalysed reaction, reached an interesting conversion (63%) and remarkably high stereochemical results (>99% *ee* and *anti*/*syn* > 99:1), superior to those reported by other authors for unmodified proline [120].


**Table 3.** Optimization of the MeOH/H2O-based protocol 1.

1 Reaction conditions: **1a** (5 equiv.), **2** (0.3 mmol), (*S*)-proline (10 mol%), rt. 2 Determined by 1H NMR on the crude mixture. 3 Determined by CSP-HPLC on the crude mixture. 4 Here, 20 μL of MeOH was added after 2 h. 5 Here, 20 mol% of (*S*)-proline was used.

The next step of our investigation was directed to the ketone partner **1** of the asymmetric aldol condensation. In proline-catalysed aldol reactions, a typical drawback is represented by the excess of ketone over the limiting aldehyde generally required to achieve good yields. To increase the sustainability of the process, we planned to lower the ketone excess (Table 4).

Some aldehydes characterized by high or medium reactivity were selected for this study, in which the ketone amount was reduced to 2 equivalents. With almost all the tested substrates, high conversions and excellent *ee* values were again obtained. Although longer reaction times were required, the reaction rates remained worthy of note, especially when compared with the performance of other protocols in similar conditions. The main drawback of this procedure was a slight decrease of diastereoselectivity, an effect that is not immediately obvious. Benaglia, using the reaction progress kinetic analysis (RPKA) approach [25], a technique that allowed Blackmond et al. to define the kinetic rate law of proline-catalysed aldol reactions [23], proved the reversibility of the aldol reaction. Lowering the ketone excess leads to the following: (*i*) longer reaction times to preserve the same level of product conversion; and (*ii*) a less efficient opposition to the retroaldol reaction, which is a slow process within the time scale of our reactions. Both factors promote equilibrium on a little extent, likely accounting for the slightly decreased diastereocontrol observed when reduced amounts of cyclohexanone **1a** were used (Table 4). In conclusion, the high efficiency achieved by MeOH/H2O/(*S*)-proline-based protocol

allows to reduce the ketone excess, involving (*i*) slight adverse effects on aldol reaction performance; and (*ii*) benefits, such as lower costs and easier work up and product purification.


**Table 4.** Effects of cyclohexanone **1a** amount in the MeOH/H2O/(*S*)-proline-based protocol 1.

1 Reaction conditions: **2** (0.3 mmol), (*S*)-proline (10 mol%), MeOH/H2O (40 μL/10 μL), rt. 2 Determined by 1H NMR on the crude mixture. 3 Determined by CSP-HPLC on the crude mixture. 4 Here, 20 μL of MeOH was added after 2 h.

#### *2.2. Application of the Protocol to Other Ketones*

Afterwards, we focused on the application of the developed catalytic protocol to two different donor ketones **1b,1c** (Table 5). Considering the excellent performance (stereoselectivity and reaction rate) achieved employing the MeOH/H2O/(*S*)-proline protocol in the presence of 5 equivalents of **1a**, we decided for convenience to apply these conditions to the ketones investigation.


**Table 5.** MeOH/H2O/(*S*)-proline-based protocol applied to ketones **1b,c** 1.

1 Reaction conditions: **2** (0.3 mmol), **1** (5 eq.), (*S*)-proline (10 mol%), MeOH/H2O (40 μL/10 μL), rt. 2 Determined by 1H NMR on the crude mixture. 3 Determined by CSP-HPLC on the crude mixture. 4 Reaction carried out at 0 ◦C.

At first, we tested cyclopentanone **1b** with highly reactive 4-nitrobenzaldehyde **2a**, observing a particularly high reaction rate, with the transformation being complete in only 4 hours. This result is unprecedented in the presence of unmodified proline or most of its derivatives [124–127], confirming once again the high reactivity achievable employing the MeOH/H2O protocol. The corresponding product **3ba** was obtained with excellent *ee*, but low diastereoselectivity. This behaviour was expected because poorly diastereoselective aldol reactions catalyzed by proline or its derivatives were regularly reported for cyclopentanone **1b** [87,124–127]. To improve the diastereoselctivity, we lowered the reaction temperature to 0 ◦C and we obtained a good 78:22 *anti*/*syn* ratio, maintaining a high conversion in a reasonable time.

Considering the excellent performance achievable with MeOH/H2O/(*S*)-proline-based protocol, we were particularly interested in the results obtainable with less reactive aldehydes. In fact, 4-Br benzaldehyde **2g** and benzaldehyde **2d** provided the corresponding products (**3bg** and **3bd**, respectively) with good conversions and diastereoselectivities, and, noteworthy, with the best enantioselectivities ever achieved employing unmodified proline as catalyst [128].

As further confirmation, 2,2-dimethyl-1,3-dioxan-5-one **1c** (Table 5) also displayed good reactivity and stereoselectivity when reacted with less reactive aldehydes **2g** and **2d**. In particular, our results represent the first examples of organocatalyzed aldol condensation between 2,2-dimethyl-1, 3-dioxan-5-one **1c** and 4-Br benzaldehyde **2g** or benzaldehyde **2d**, promoted by only 10 mol% of proline [129–133].

At last, we applied our protocol to acetone **1d** as simple aliphatic ketone (Table S3, Section 2, Supplementary Materials). Although with 4-NO2 benzaldehyde **2a**, we obtained an unprecedented high reaction rate if compared with the published corresponding transformations, the enantioselectivity was poor, as commonly reported for the proline-catalysed aldol additions involving these substrates.

#### *2.3. Large-Scale Application of the Protocol*

Our aim is the development of an efficient and sustainable organocatalyzed aldol condensation protocol, which can be interesting from a scale-up perspective. Therefore, as a first step, we confirmed the excellent performance of the MeOH/H2O/(*S*)-proline-based protocol by carrying out the aldol condensation between moderately reactive benzaldehyde **2d** and cyclohexanone **1a** on a 10 mmol scale (gram scale). The desired product **3ad** was isolated in 78% yield and with high stereocontrol (90:10 *dr*, 95% *ee*), fully confirming the data obtained on small scale (Table 3).

The next step was the accomplishment of the same reaction on a 100 mmol scale of the limiting reagen<sup>t</sup> benzaldehyde **2d** (Scheme 2) in order to study some aspects in more detail.

**Scheme 2.** Process scale-up on 100 mmol of limiting aldehyde **2d**.

At first, we investigated the impact of the aldehyde addition rate on the reaction outcome (Table 6). With benzaldehyde **2d** not being very reactive, good conversions were recorded only after 23 h and we did not observe a significant difference depending on the addition rate of benzaldehyde (Table 6).

Then, we monitored product conversion and stereoselectivity for a longer reaction time (Table 6), to establish if a high conversion could be achieved without a significant loss in stereocontrol exploiting our reaction conditions. Indeed, as previously mentioned, aldol reaction is reversible and longer reaction times could make the retroaldol process competitive, providing a decreased diastereomeric ratio. Actually, we observed a slow increase of the product conversion, achieving 85% after 2 days, without a significant erosion of *anti*/*syn* ratio (in comparison with small scale-reactions, a slightly lower *dr* was recorded, which remained constant for the first 48 h). We confirmed that even the enantiomeric excess of the product remained at high levels (94% *ee* after 47 h).


**Table 6.** Process scale-up study 1.

1 Reaction conditions: **2d** (100 mmol), **1a** (500 mmol), (*S*)-proline (10 mol%), MeOH/H2O (13.33 mL/3.33 mL), rt. Total volume = 79 mL. 2 Determined by 1H NMR on the crude mixture. 3 Determined by CSP-HPLC on the crude mixture.

The results reported in Table 6 clearly show that the best reaction outcome is obtained at a reaction time representing the best balance between product conversion and stereocontrol. To further explore this effect, we compared the data obtained with different moderately or poorly reactive aldehydes (Table 7).


**Table 7.** Study of the reaction outcome as a function of the reaction time 1.

1 Reaction conditions: **2** (50 mmol), **1a** (5 eq.), (*S*)-proline (10 mol%), MeOH/H2O (6.67 mL/1.67 mL), rt. 2 Determined by 1H NMR on the crude mixture. 3 Here, 20 mol% of (*S*)-proline was used.

Concerning the stereoselectivity, in this study, we focused our attention on diastereoselectivity variation, which is much more impaired by retroaldol reaction (see Supplementary Materials for a study on enantioselectivity variation). The data collected in Table 7 demonstrate that the aldol transformation reaches a position, after which the product conversion no longer grows, while the *dr* continues to drop. The time required to achieve this situation depends on the aldehyde reactivity. On the other hand, the rate of retroaldol process is less affected by the aldehyde nature; therefore, the retroaldol effects are less troublesome for reactive aldehydes (high conversion in short time with high *dr*) and more marked for poorly reactive aldehydes (long time required to reach acceptable conversion with low *dr*). This study proves that, in the asymmetric aldol process promoted by proline, the reaction time providing the best performance (balance between conversion and stereoselectivity) strongly depends on the substrate; therefore, a careful investigation should be done before tackling a large-scale application.

A further point that we evaluated to increase the sustainability of our large-scale protocol was the reduction of the ketone excess. For this purpose, we applied the MeOH/H2O/(*S*)-proline-based protocol to moderately reactive benzaldehyde **2d** (50 mmol) in the presence of only 2 equivalents of cyclohexanone **1a**, monitoring the results over the time. After 71 h, we achieved the highest product conversion (83%) with an excellent 89:11 *dr*. Prolonging the reaction time (98 h) only led to a drop in *dr* (84:16). These findings sugges<sup>t</sup> that, exploiting our protocol, a large excess of ketone (5 equivalents) only enhances the initial reaction rate, but it is not necessary for the achievement of an excellent final performance.
