On the direct use of CO2 in multicomponent reactions: introducing the Passerini four component reaction

We introduce a novel isocyanide-based multicomponent reaction, the Passerini four component reaction (P-4CR), by replacing the carboxylic acid component of a conventional Passerini three component reaction (P-3CR) with an alcohol and CO2. Key to this approach is the use of a switchable solvent system, allowing the synthesis of a variety of α-carbonate-amides. The reaction was first investigated and optimized using butanol, isobutyraldehyde, tert-butyl isocyanide and CO2. Parameters investigated included the effect of reactant equivalents, reactant concentration, solvent, catalyst, catalyst concentration and CO2 pressure. Of the other parameters, the purity of the aldehyde and its tendency to oxidize was one of the most critical parameters for a successful P-4CR. After optimization, a total of twelve (12) P-4CR compounds were synthesized with conversions ranging between 16 and 82% and isolated yields between 18 and 43%. Their structures were confirmed via1H and 13C NMR, FT-IR and high resolution mass spectrometry (ESI-MS). In addition, three (3) hydrolysis products of P-4CR (α-hydroxyl-amides) were successfully isolated with yields between 23 and 63% and fully characterized (1H, 13C NMR, FT-IR and ESI-MS) as well.


Introduction
Multicomponent reactions (MCRs) are dened as reactions involving more than two starting materials, while forming products in which most of the atoms of the starting materials are incorporated. 1 Already in 1850, Strecker reported the synthesis of a-amino-nitriles from an aldehyde, ammonia and hydrogen cyanide, one the rst reported MCRs. 2 Today, a large variety of different types of MCR exist. 3 In the context of this work, isocyanide-based multicomponent reactions (IMCRs) are of particular interest. Prominent examples of IMCRs are the Passerini three component reaction (P-3CR) and Ugi four component reaction . The P-3CR was discovered in 1921 by Mario Passerini and describes the reaction between a carboxylic acid, carbonyl component (aldehyde or ketone) and an isocyanide, forming a-acyloxyl-amides. 4 The Ugi four component reaction  was developed by Ivar Ugi in 1960 and involves a carboxylic acid, carbonyl component (aldehyde or ketone), an isocyanide and an amine as components, leading to the formation of a bis-amide. 5 Several variations of the IMCRs exist, one of them being the Ugi 5-CR that uses an alcohol (usually methanol) and CO 2 as acid component. 6 This strategy was also employed in combination with efficient thiol-ene polymerization to synthesize highly functionalized polycarbonates, polyamides, polyurethanes and polyhydantoins. 7 Besides polymer chemistry, MCRs have equally found application in high throughput synthesis, 8 as well as in combinatorial chemistry, 9 allowing the synthesis and screening of compound libraries. Furthermore, IMCRs are also very useful in the synthesis of heterocycles. 10 Most recently, the Ugi 4-CR of peruorinated acids was employed for the synthesis of a library of molecular keys that were applied for molecular cryptography. 11 Apart from the classic P-3CR, many variations have been reported. Taguchi and co-workers reported on the direct utilization of aliphatic alcohols alongside an isocyanide and a,bunsaturated aldehyde in the presence of an Indium(III) catalyst to form a-alkoxy-amide products. 12 El Kaim et al. employed the so-called Passerini-Smiles reaction of o-nitrophenol as a replacement of the acid component and synthesized a library of a-aryloxy-amide products. 13 Chatani et al. reported the reaction of cyclic and acyclic acetals with isocyanides in the presence of GaCl 3 as catalyst. 14 Here, the isocyanide inserts into the C-O bond of the acetals, nally resulting in a-alkoxy-imidates. Denmark and Yan reported the asymmetric a-addition of isocyanides to aldehydes. In this case, the reaction was catalyzed by a combination of a weak Lewis acid, SiCl 4 activated by a chiral Lewis base (bisphosphoramide). The desired ahydroxyl-amides were obtained aer basic workup. 15 Very important for the herein reported results, Jessop and coworkers introduced switchable solvent systems involving CO 2 alongside a super base in 2005. 16 The unique nature of this solvent system was its ability to switch from a non-polar to polar solvent in the absence or presence of CO 2 , respectively, by the reversible formation of a carbonate anion/protonated base complex. The system is very versatile and allows applications such as straight-forward product purication, 17 as CO 2 capturing agents, 18 for the selective extraction of hemicellulose from wood, 19 and also as sustainable solvents for cellulose solubilization. 20, 21 We recently reported an optimization of the DBU-CO 2 switchable solvent system and could unambiguously proof that carbonate anions are indeed formed in situ. 22 The formation of this in situ carbonate not only led to a mild solubilization of cellulose in DMSO, but also allowed an activation of the cellulose hydroxyl groups leading to a milder modication such as succinylation. 23 In this regard, a high DS value of 2.6 was reported for reaction carried out at room temperature in 30 min. 23 Tunge and co-workers reported a related approach of alcohol activation using CO 2 in the absence of any base catalyst. 24 In this case, allyl alcohols were reacted directly with CO 2 leading to the formation of an in situ allyl-carbonate, which forms a p-allyl complex with a Pd-precursor that can then reacts with nucleophiles (here derived from nitroalkanes, nitriles and aldehydes) in a Tsuji-Trost like fashion. 24 In the current contribution, the idea is to utilize this in situ generated carbonate anion as an acid component (i.e. nucleophile) in a typical P-3CR and is thus the starting point of the herein reported results. In this case, the CO 2 is able to activate the alcohols and is incorporated as C1 carbon source into the desired compound. In this work, we thus report, for the rst time, the Passerini four component reaction (P-4CR) as a variation of the P-3CR, achieved by replacing the acid component in the P-3CR by an alcohol and CO 2 . The utilization of CO 2 as a carbon source is both interesting from an environmental and sustainable perspective as well as to extend the scope and achievable structural variety of MCRs.

Instruments
IR spectroscopy. Infrared spectra of all samples were recorded on a Bruker alpha-p instrument using ATR technology within the range 4000 to 400 cm À1 with 24 scans.
Nuclear magnetic resonance spectroscopy (NMR). 1 H NMR spectra were recorded on a 500 MHz WB Bruker Avance I spectrometer operating at a frequency of 499.97 MHz for 1 Hand a frequency of 125.72 MHz for 13 C-measurement on a 8 mm TXI probe head with actively shielded z-gradients (at q ¼ 0 ) and on a 4 mm triple HCX MAS probe head (at ca. q ¼ 65 ) at 298 K, regulated with a Bruker VTU-3000. Measurements were done at ambient temperature. Measurements were done in CDCl 3 and data are reported in ppm relative to 7.26 ppm and 77.16 ppm for 1 H and 13 C, respectively.
Gas chromatography (GC-FID). For GC measurements, a GC-2010 Plus instrument from Shimadzu with a polar column (Rxi-642Sil MS, length: 30 m, diameter: 0.25 mm, lm thickness: 0.25 mm) and a ame-ionization detector (FID) was used. The sample (1 mL) was injected and vaporized at 250 C. The column was heated from 50 to 280 C at a rate of 10 K min À1 .
Gas chromatography-mass spectrometry (GC-MS). Electron impact (EI) analyses were conducted using a Varian 431-GC instrument with a capillary column Factor Four™ VF-5ms (30 m Â 0.25 mm Â 0.25 mm) and a Varian 210-MS ion trap mass detector. Scans were performed from 40 to 650 m/z at rate of 1 scan per second. The oven temperature program applied during the analysis was: initial temperature 95 C, hold for 1 min, ramp at 15 C min À1 to 200 C, hold for 2 min., ramp at 15 C min À1 to 300 C, hold for 5 min. The injector transfer line temperature was set to 250 C. Measurements were performed in the splitsplit mode (split ratio 50 : 1) using helium as carrier gas (ow rate 1.0 mL min À1 ).
Electron spray ionization-mass spectrometer (ESI-MS). Spectra were recorded on a Q Exactive (Orbitrap) mass spectrometer (Thermo Fisher Scientic, San Jose, CA, USA) equipped with a HESI II probe to record high resolution electrospray ionization-MS (ESI-MS). Calibration was carried out in the m/z range 74-1.822 using premixed calibration solutions (Thermo Fisher Scientic). A constant spray voltage of 4.7 kV and a dimensionless sheath gas of 5 were employed. The S-lens RF level was set to 62.0, while the capillary temperature was set to 250 C. All samples were dissolved at a concentration range of 0.05-0.01 mg mL À1 in a mixture of THF and MeOH (3 : 2) doped with 100 mmol sodium triuoroacetate and injected with a ow of 5 mL min À1 .
General procedure for optimization study of P-4CR 0.41 g of Butanol (1 eq., 5.50 mmol) and 5 mol% tetradecane as internal GC standard (70 mL) were stirred in 1.5 mL of the solvent at room temperature for 1 to 2 minutes, aer which a sample was collected for GC analysis. The mixture was then saturated with CO 2 (5 bar) for 15 minutes. In the same manner, isobutyraldehyde was pre-saturated with CO 2 (5 bar) for 10 minutes, aer which both solutions were mixed and further saturated with 5 bar of CO 2 for 10 to 15 minutes. Subsequently, tert-butylisocyanide was added and the reaction was performed under 10 bar of CO 2 for 24 h at room temperature (22-24 C). Samples were then collected over the course of the reaction and analyzed by gas chromatography (GC) in order to calculate the conversion and the relative percentage between the observed products (P-4CR, P-3CR and hydrolysis product of P-4CR). Parameters investigated during the optimization study include: effect of reaction equivalents (1, 2 eq.), reaction concentration (1.84 and 3.68 M with respect to butanol), catalyst (triethylamine and DBU), catalyst concentration (5, 10 and 15 mol%), CO 2 pressure (5, 10 and 15 bar) and solvent (DMSO, chloroform, methyl-THF, DCM).

Results and discussions
As a starting point and to prove our hypothesis of a possible P-4CR, we investigated the reaction of butanol, isobutyraldehyde, tert-butylisocyanide and CO 2 in dichloromethane (DCM) as a solvent. In addition to the formation of the expected P-4CR product, the formation of a P-3CR by-product and the hydrolysis of the P-4CR product, resulting in the formation of an ahydroxyl-amide, was observed (compare Scheme 1). As we observed in the further course of our investigations, the formation of the P-3CR product is due to the presence of the respective carboxylic acids originating from the oxidation of the used aldehyde component. The isobutyraldehyde used for the optimizations had a lower purity (92% from 1 H NMR) than reported by the manufacturer (99.5%). We later observed that freshly distilled aldehydes gave the best results for a P-4CR and the P-3CR could be suppressed to less than 5%, although it could not be completely avoided. Nevertheless, this rst proofof-principle reaction clearly showed that the anticipated reactivity of the carbonate anion allows for a P-4CR. For optimizing the P-4CR, several parameters, such as the effect of solvent, reactant concentration, reactant equivalents, catalyst concentration and CO 2 pressure were investigated. Dichloromethane (DCM, 1.84 M) was employed as solvent for the rst trial experiments as it has been reported to be a suitable solvent for the P-3CR. 25 One equivalent of butanol, isobutyraldehyde and tert-butylisocyanide were employed, while the reaction was performed at 10 bar CO 2 for 48 h at room temperature. The conversion of butanol was followed via gas chromatography (GC) versus an internal standard (tetradecane). For the rst trial, a conversion of 38% was reached aer 48 h (see Fig. SI1 †). The presence of the desired P-4CR product (273.19 g mol À1 ) was indicated by gas chromatography mass spectrometry (GC-MS). In addition, the presence of the hydrolysis product of the P-4CR (174.10 g mol À1 ) as well as the P-3CR product (243.16 g mol À1 ) were conrmed by GC-MS. The formation of the three products was followed in time via GC and is illustrated in Fig. 1. Within the rst ve hours, the P-3CR accounted for most of the observed products. However, an increased formation of the P-4CR was observed as the reaction proceeded. In addition, some hydrolysis of the P-4CR product was observed with time. Considering the P-3CR side reaction, a two-fold excess of aldehyde and isocyanide components was employed in the next reaction, resulting in an improved butanol conversion of 56%.
As solvents play a key role in multicomponent reactions, different solvents (dimethyl sulfoxide DMSO, methyl-THF and chloroform) were investigated next. While similar conversions were obtained for all the solvents investigated, different hydrolysis tendencies were noticed. By comparing the relative percentage between the P-4CR and its hydrolysis product, an increased formation of hydrolysis product in the order chloroform > DMSO > methyl-THF > DCM was observed. This trend might be explained considering the acidity of the solvents, e.g. chloroform is the most acidic solvent tested herein. For DMSO, the high hydrolysis rate observed is probably due to the presence of unavoidable water impurity in the solvent. Methyl-THF showed a relatively low tendency towards hydrolysis, but led to a higher degree of oxidation of the aldehyde and thus an increase of the P-3CR side product (see Fig. SI2 †), probably due to possible peroxide impurities. 26 DCM showed the lowest tendency towards hydrolysis and equally resulted in the highest selectivity towards the targeted P-4CR product. It should however be noted here that the observed hydrolysis products are valuable compounds as well. Furthermore, as multicomponent reactions usually provide higher efficiency at higher Scheme 1 Formation of P-4CR products in the absence (a) and presence (b) of a base catalyst. (c) Hydrolysis of the P-4CR product. (see Fig. 2 for synthesized structures). concentrations, this parameter was investigated by doubling the previously applied concentration. As expected and to our delight, the conversion of butanol increased from 56% (1.84 M with respect to butanol) to 73% (3.68 M with respect to butanol) in DCM.
Presumably, the mechanism of the P-4CR proceeds via the in situ formation of a carbonate anion initiated by a base catalyst 16,[20][21][22] (see Scheme 1b). The carbonate then reacts as the acid component with the isocyanide and aldehyde component in the typical fashion of the conventional Passerini 3-CR. 3 However, in the so far discussed set of experiments, no base catalyst was employed. It was therefore interesting to note that the P-4CR occurred nevertheless. This initially unexpected reactivity could be due to an activation of the alcohol by the isocyanide (Scheme 1a). Isocyanides were used as a weak Lewis base in previous reports in literature. 27 Equally, alcohol activation by other Lewis bases has been reported. 28 To verify the alcohol activation hypothesis, an in situ 1 H NMR study was performed (utilizing butanol in CDCl 3 ). In this context, the spectra of butanol, butanol in the presence of stoichiometric amounts of tert-butylisocyanide and butanol in the presence of stoichiometric amounts of Et 3 N were compared. The respective study is displayed in the ESI (Fig. SI3 †) and is described below.
From this NMR study, a slight downeld chemical shi of the OH proton (from 3.16 ppm to 3.20 ppm) and signicant signal broadening, characteristic of H-bonding, was observed for butanol in the presence of tert-butylisocyanide compared to the spectrum of butanol without isocyanide in CDCl 3 . Similar proton signal broadening of the OH proton, most probably caused by H-bonding, has also been reported for Schiff bases. 29 Comparing the spectrum of butanol with the spectrum of butanol in the presence of Et 3 N, a downeld chemical shi of the OH proton (from 3.16 ppm to 4.38 ppm) was observed, while displaying similar signal broadening (see Fig. SI3b †). These results indicate H-bonding of the OH proton and hence, activation of the alcohol by the isocyanide acting as a Lewis base.
The effect of a catalyst was thus evaluated next. As mentioned above, a basic catalyst is able to activate the hydroxyl group, thereby increasing the carbonate anion formation. Thus, triethylamine (10 mol%) was tested. The obtained results were compared with the reaction without addition of the catalyst (see Fig. SI4a †). As expected, the presence of the catalyst further accelerated the reaction. A conversion of almost 70% was reached within 24 h (compared to 48 h required to achieve similar conversion in the absence of the catalyst). In addition, the overall selectivity towards P-4CR increased (see Fig. SI4b and c †). However, a slight increase in hydrolysis product was also observed in the presence of the base catalyst, as one could expect (see Fig. SI4b and c †). In another set of experiments, the two catalysts Et 3 N and diazabicyclo [5.4.0]undec-7-ene (DBU) were compared. The obtained results showed similar conversion of butanol (about 70%) aer 24 h. However, an increased hydrolysis was observed for the reaction with DBU compared to Et 3 N (Fig. SI5 †), probably due to the higher basicity of DBU compared to Et 3 N. In addition, the formation of the P-3CR side product was less pronounced for Et 3 N. Hence, Et 3 N was used for the following experiments, screening for catalyst concentration employing 5, 10 and 15 mol%. This study revealed that the conversion of butanol remained at about 70% for both 5 and 10 mol%. However, the quasi conversion decreased to 52% when 15 mol% catalyst were used. This decrease can be explained by the observed increased hydrolysis of the P-4CR product, which reforms butanol and thus counterfeits a lower butanol conversion (see Scheme 1c and Fig. SI6 †). As 10 mol% catalyst loading gave slightly better results, this catalyst loading was employed for investigating the effect of CO 2 pressure. Three CO 2 pressures were investigated (5, 10 and 15 bar) and the results are presented in Fig. SI7, † revealing a slight increase in butanol conversion from 58% to 69% as the CO 2 pressure was increased from 5 to 10 bar. However, a slight decrease was observed at 15 bar (62% conversion). Furthermore, the highest relative percentage of P-4CR was obtained at 10 bar of CO 2 . Therefore, 10 bar of CO 2 was selected for further experiments.
Using the described optimized conditions (3.84 M with respect to butanol, DCM as solvent, 1 eq. of alcohol, 2 eq. of aldehyde and isocyanide with respect to butanol, 10 bar CO 2 at room temperature (22-25 C)), the scope of the P-4CR was investigated by varying the used components (see Fig. 2). For the synthesis of 1a, butanol, isobutyraldehyde and tert-butylisocyanide were used. A conversion of about 70% was reached within 24 h, longer reaction times (up to 45 h) led to an increase of selectivity of the targeted P-4CR compound 1a. Compound 1a was isolated in 43% yield aer column chromatography. This value is within the range of previously reported Ugi 5-CR products. 6,7 Keating et al. reported similar results when other alcohols apart from methanol were used in this Ugi-5CR. 6 It is also important to point out that in the case of methanol, an excess (over 10 times) was utilized in order to achieve high conversion and yields. The structure of 1a was conrmed via 1 H and 13 C NMR performed in CDCl 3 as shown in Fig. 3.
The attribution of the proton and carbon peaks are similar to previous reports on the P-3CR, 25 as well as Ugi-5CR. 7 In the 1 H NMR spectrum, the amide proton is visible at 5.87 ppm. The proton signal at the tertiary carbon at 4.83 ppm and the 9H from the tert-butyl side chain at 1.35 ppm conrm the proposed structure. From the 13 C NMR, two characteristic carbonyl carbon chemical shis are observed at 168.78 ppm (amide) and 154.93 ppm (carbonate). Furthermore, the tertiary carbon in the isocyanide side chain is observed at 51.70 ppm. In addition, from the FT-IR spectrum (see Fig. SI8 †), the characteristic C]O stretching absorbance band of the carbonate (1743 cm À1 ) and amide (1654 cm À1 ) were visible. Finally, the mass of the molecule was conrmed by ESI-MS ([C 14 H 27 NO 4 Na] + ¼ 296.19 g mol À1 , obtained ¼ 296.18 g mol À1 ).
Employing similar conditions, a total of twelve molecules were synthesized and equally characterized (see Fig. 2). On the example of butanol, ve variations of the aldehyde and four isocyanide variations were demonstrated. In addition, a secondary alcohol (2-butanol), an unsaturated alcohol (allylalcohol), benzyl-alcohol, octanol and cyclohexanol were utilized. The respective P-4CR products were obtained with conversions ranging from 16 to 82% and isolated yields between 18 and 43%. Lower conversions were obtained for reactions involving sterically demanding side chains such as 1-adamantyl-7a (19%) and 2-morpholinoethyl-8a (16%). In the course of these investigations, we observed that the P3-CR sidereaction can be suppressed to less than 5% by immediate utilization of a freshly distilled aldehyde component.
The obtained P-4CR compounds were fully characterized via 1 H and 13 C NMR (see ESI †), FT-IR and high-resolution mass spectrometry (ESI-MS). In the case of 1b, employing phenyl acetaldehyde, the P-3CR side reaction was not observed. However, in this case (phenyl acetaldehyde), the presence of a proton in a-position to the aromatic ring led to an aldol condensation side-product. The aldol side reaction was favoured when the base catalyst was used, as expected. Hence, for this reaction, no catalyst was utilized and a butanol conversion of 36% was reached aer 24 h (without much improvement with longer reaction time). Noteworthy, the P-4CR products from benzyl-alcohol 4a and octanol 5a were less prone to hydrolysis. Their higher hydrophobicity compared to butanol might explain this tendency. The reaction performed with allyl alcohol 3a required longer reaction times to achieve suitable conversions (30 h, conversion 50%) and showed the highest tendency towards hydrolysis, which also increased over time. Running the reaction even longer (48 h) resulted in 82% conversion. The P-4CR product was isolated in a yield of 33%, while 23% of the hydrolysis product were isolated (and characterized via the same techniques). This result conrms that hydrolysis can also be used on purpose to obtain equally useful a-hydroxyl-amide products. Our attempt to introduce an aromatic side chain via the aldehyde component (benzaldehyde) was unsuccessful, as only the P-3CR and hydrolysis products were formed. In addition, benzaldehyde is very easily oxidized to the corresponding acid (despite distillation before usage). Thus, 2,4-dinitrobenzaldehyde was used instead to obtain 9a. A conversion of 40% was achieved aer 48 h with 22% isolated yield. In a similar manner, we could not observe the formation of the P-4CR product when tert-butanol was used. Again, only P-3CR and the hydrolysis product of P-4CR were observed via GC-MS. Nevertheless, this demonstrate that also tertiary alcohols can be used in this reaction, whereby only one of the two possible products (i.e. the hydrolysis product) is accessible so far. Finally, the hydrolysis products of 1c, 3a and 9a could be isolated in a yield of 30%, 23% and 63%, respectively. Their structures were conrmed via 1 H and 13 C NMR (see ESI †).

Conclusion
We report a novel variant of the Passerini reaction, the Passerini four component reaction (P-4CR), by replacing the carboxylic acid component in a conventional P-3CR with an alcohol and CO 2 . Upon optimization of the reaction parameters (reaction time, reactant equivalents, reactant concentration, solvent, catalyst concentration and CO 2 pressure), twelve P-4CR products were successfully synthesized with conversions ranging from 16 to 82% and isolated yields between 18 and 43%. In addition, hydrolysis of the P-4CR products, leading to the formation of a-hydroxyl-amides, was observed. Three of these hydrolysis products were isolated with yields between 23 and 63%. Furthermore, the formation of P-3CR products was observed, which occurred due to the oxidation of the employed aldehyde components. The success of our reported P-4CR does not only expand the structural diversity of multicomponent reactions, but the direct utilization and activation of CO 2 as a C1 building block is equally noteworthy.