Regenerable Acidity of Graphene Oxide in Promoting Multicomponent Organic Synthesis

The Brønsted acidity of graphene oxide (GO) materials has shown promising activity in organic synthesis. However, roles and functionality of Lewis acid sites remain elusive. Herein, we reported a carbocatalytic approach utilizing both Brønsted and Lewis acid sites in GOs as heterogeneous promoters in a series of multicomponent synthesis of triazoloquinazolinone compounds. The GOs possessing the highest degree of oxidation, also having the highest amounts of Lewis acid sites, enable optimal yields (up to 95%) under mild and non-toxic reaction conditions (85 °C in EtOH). The results of FT-IR spectroscopy, temperature-programed decomposition mass spectrometry, and X-ray photoelectron spectroscopy identified that the apparent Lewis acidity via basal plane epoxide ring opening, on top of the saturated Brønsted acidic carboxylic groups, is responsible for the enhanced carbocatalytic activities involving Knoevenagel condensation pathway. Recycled GO can be effectively regenerated to reach 97% activity of fresh GO, supporting the recognition of GO as pseudocatalyst in organic synthesis.

multicomponent synthesis of triazoloquinazolinone by systematically varying the amounts of oxygen groups. Extensive spectroscopic results identified that the enhanced activities are the results of additional carbocatalytic contribution due to the Lewis acid sites generated from basal plane epoxide groups, in addition to the contribution from the Brønsted acidic groups at the edges of GO sheets. The Brønsted acid groups showed an upper limit in quantity while the oxidation degrees of GO increase, revealing that the overall reactivity of GO is governed by Lewis acid sites when Brønsted sites reach saturation. We also reported a simple acid regeneration protocol to redeem the acidity of GO and showing the recovery of 97% of its activity compared to fresh ones.

Results and Discussion
Characterization of different GO. Due to the recognized importance of oxygen groups on the carbocatalytic activity of GO 2 , we hypothesized that higher degrees of oxygenation in GO could lead to a better activity in the multicomponent synthesis of triazoloquinazolinone. We recently reported a method using a preformed acidic oxidizing medium (PAOM) to achieve highly oxidized GO materials with higher quantity of oxygen groups than the conventional Hummers' method 11 . By pre-mixing concentrated sulfuric acid and permanganate prior to graphite addition, the as-yielded acidic oxidants establish the strong oxidizing environment to introduce higher oxygen groups than the Hummers' methods. In addition, the diffusion kinetics of PAOM was proven to be much faster in delivering oxidants into the interlayer spacing of graphite crystals. Herein, we followed PAOM method to vary oxidation degrees of GO products using different quantity of KMnO 4 oxidants, such as 3, 6, and 9 grams, hereby the products were denoted as 1xGO, 2xGO, and 3xGO respectively.
The X-ray diffraction (XRD) patterns of the GO materials using NaCl as internal standard are shown in Fig. 1a. The graphite powder precursor exhibits a characteristic peak at 26.7 (two theta), corresponding to (002) plane of graphite crystal 43 . All of the prepared GO samples show broad XRD peaks in a two theta range of 9-10 degrees, characteristic to GO materials and indicating a successful conversion of graphite to GO 10 . The interlayer distance of (001) peaks of 1xGO, 2xGO, and 3xGO corresponds to 0.881, 0.909, and 0.948 nm, respectively. The Hummers' method samples (H-GO) show a much smaller (001) d-spacing of 0.851 nm. By increasing the amounts of KMnO 4 oxidant and the reaction time, the (001) of the PAOM samples shifts to a larger d-spacing due to the higher degrees of oxidation leading to the expansion of interlayer spacing in GO 10,11 . The TGA results show the weight loss (150-280 °C) of 1xGO, 2xGO, and 3xGO to be 29.9%, 34.4%, and 38.8%, respectively (Fig. 1b). 3xGO with the largest weight loss among all GO materials demonstrates the highest quantity of oxygenated functional groups, well agreeing with the XRD data. The TEM images show the 2D characteristic of the as-produced GO samples (Fig. 1c). The AFM results exhibit a typical thickness of 0.9 nm of individual sheets, corresponding to single-layer GO materials (Fig. 1d). As shown in Fig. 1e, the photo comparison of the GO products were acquired right after the H 2 O 2 addition of the corresponding preparation. The much brighter yellow color in 3xGO suggests the higher oxidation degrees than H-GO, as indicated by the reported literature 11 .
The surface of these GO products is widely decorated by different oxygen functional groups such as hydroxyl groups (-OH), epoxide (C-O-C), and carboxylic acid (-C(O)OH), which can be observed with FT-IR spectroscopy 44 . Fig. 1f shows the FT-IR spectra of the prepared GO materials. The most characteristic FT-IR features are the absorption bands corresponding to C-O-C stretch of epoxide at 1046 cm −1 (designated as region a), the C-OH stretching at 1222 cm −1 (region b), and the O-H deformation vibration at 1411 cm −1 (region c). Besides the ubiquitous O-H stretch which appears as a broad and intense signal at 3300 cm −1 (See Fig. S1), another intense peak at 1621 cm −1 (designated as region d) can be associated to skeletal vibrations of un-oxidized graphitic domain. The absorption at 1730 cm −1 (designated as region e) corresponds to C=O stretching of undissociated carboxylic acid [44][45][46] . These IR peaks shown in different regions confirm the presence of different oxygen functional groups in the GO samples. Compared to 3xGO, the carbonyl (C=O) stretching peaks (region e) of 1xGO and 2xGO are relatively weaker by referencing the bands in the other regions, further suggesting the greatest oxidation of 3xGO.
To determine the effect of the increasing amount of KMnO 4 on the oxidation of GO, the elemental analysis data of the GO samples were acquired and summarized in Table 1. The oxygen-to-carbon (O/C) ratios of 1xGO, 2xGO, and 3xGO are 1.367, 1.458, and 1.494, respectively, which are much higher compared to O/C ratio of H-GO (1.257). Furthermore, this clearly shows the increase in oxygenation degrees with higher amount of KMnO 4 . This result can be corroborated with thermogravimetric analysis (TGA). The TGA curves (Fig. 1b) of 1xGO show a significant weight loss of 29.9% at 150-280 °C. This weight loss can be attributed to the thermal decomposition of labile oxygen functional groups such as hydroxyl, epoxide, and carbonyl groups [46][47][48] . In the same temperature range, both 2xGO and 3xGO exhibit weight loss of 34.4% and 38.8%, respectively. Peak deconvolution of XPS components of GO materials also shows increasing percentage of oxygenated components as the amount of oxidant is increased (see Table S1). All the results above confirm a successful preparation of GO with a decreasing quantity of oxygen functional groups in order of 3xGO > 2xGO > 1xGO > H-GO.
optimization of reaction condition. The activity of the GO materials in promoting multicomponent synthesis were studied for the three-component cyclocondensation reaction between 3-amino-1,2,4-triazole, benzaldehyde, and dimedone. This model multicomponent reaction was studied with respect to different solvents, temperatures, GO loadings, and oxidation degrees of GO materials to achieve the optimal yields (Table 2 and 3). The results show that the product yields are highly dependent on GO loadings as a higher yield was obtained with increasing amount of 1xGO.
Upon testing different solvents for the reaction ( Table 2, entry 1-6), only ethanol has shown considerable yields. The ethanol:water mixture results to low yield due to low solubility of the organic reactants. Ethanol could provide both sufficient polarity for homogeneous 1xGO dispersion and higher solubility for the reactants when compared to the other solvents. According to the previous reports, acid promoted synthesis of 4a was typically carried out in acetonitrile, a more toxic solvent compared to ethanol 36,37 . The amount of 1xGO used in the reaction affects the yields. A higher loading of 50 mg results in a better reaction yield of 37%. Formation of 4a was not observed in the absence of 1xGO in the synthesis, demonstrating the essential role of GO as the heterogeneous promoter (Table 2, entry 7-9). Carrying out the reaction at temperatures lower than 85 °C resulted in lower yields or no reaction at all. On the other hand, increasing the temperature to 100 °C did not improve the yield either ( Table 2, entry 10-13). With the optimal solvent and temperature conditions, the syntheses were further carried out using GO materials with higher oxygenation levels ( Table 3). By using 2xGO as the promoter, the yields (56%) are higher than the synthesis using 1xGO with only one-half the reaction time. 3xGO exhibits the highest yields of 95% in just 35 mins (Table 3, entry 3). These results clearly demonstrate the significant influence of GO's oxidation degrees on the yields of multicomponent synthesis. To further verify this concept, we conducted the reaction with relatively less oxidized GO samples, H-GO, reduced GO (r-GO), and pristine graphite that represents the oxidation-free promoter (see entry 4-6 in Table 3). After the 240 min reactions, the observed yields for H-GO and r-GO were only 17%, and less than 5% of 4a, respectively, and no formation of 4a can be observed for the graphite case. NaBH 4 , a weak reducing agent that cannot reduce carboxylic acids, was reported to result in decreased amount of epoxide groups 49 . Comparing 1xGO and its NaBH 4 reduced product r-GO, a significant lowering of 4a (Table 3) yields was observed for r-GO which suggests the high reactivity of the basal plane epoxide groups in the reaction. The critical role of oxygen functional groups in facilitating the organic synthesis has been clearly revealed.
Substrate scope. To evaluate the general applicability of the 3xGO in promoting multicomponent cyclocondensation, we carried out a series of substrate screening (summarized in Table 4) using various aromatic aldehydes (2b-f) with 3-amino-1,2,4-triazole (1) and dimedone (3) to produce the corresponding triazoloquinazolinone derivatives (4b-f). Single crystal X-ray of 4c (Fig. S4)  www.nature.com/scientificreports www.nature.com/scientificreports/ reaction time (e.g., from 35 to 60 mins) required in each synthesis varies with the electronic nature of the aromatic aldehydes (Table 4). All reactions show high yields (78-95%), demonstrating the general applicability of 3xGO in the synthesis of triazoloquinazolinones. oxygenated groups on Go sheets. To identify the oxygen species on GO involved in the multicomponent reaction, temperature-programed decomposition mass spectrometry (TPD-MS) measurement was conducted on all the prepared GO materials (Fig. 2). The amount of released species (Torr) is determined by integration of peak area (Table 5). Since GO materials are comprised of hydrogen, carbon, and oxygen atoms only, all the oxygenated groups in GO should be converted to CO and CO 2 after thermal decomposition. Thus, measurements were particularly focused on CO 2 and CO evolutions during the TPD-MS experiments as these represent two different oxygenation functional groups: carboxylic and non-carboxylic groups. The amount of these gases during the thermal decomposition of GO may be used to determine the relative population of the oxygen groups.     www.nature.com/scientificreports www.nature.com/scientificreports/ According to gas cracking pattern data 50 CO signals can be a cracking fragment of CO 2 with a relative percentage of 11.4%, this component has been subtracted from the measured intensity of m/z = 28. As shown in Fig. 2, for all the GO materials, CO 2 evolves at the temperature range of 90 to 170 °C. CO 2 evolving at this temperature range can be typically attributed to the decomposition of carboxylic acid on the edge of the GO material 51 . Both 2xGO and 3xGO have a similar amount of evolved CO 2 suggesting the comparable quantity of carboxylic acid groups ( Table 5). The saturated quantity of Brønsted carboxylic acid is plausibly due to the limited amounts of accessible carbon at the edges. The CO evolved can be observed at the temperature range of 90 to 160 °C. CO gas detected at this temperature range is due to thermal decomposition of basal plane oxygen groups such as epoxides and alcohols 52,53 . The integrated area for CO evolving from 3xGO (7.76 × 10 −8 Torr) is greater than that of 2xGO (5.06 × 10 −8 Torr), indicating the relatively higher quantity of epoxide and alcohol groups on the surface of 3xGO. The same result was obtained from peak deconvolution of XPS spectra for 2xGO and 3xGO which shows an increase in epoxide (C-O-C) component with increasing oxidation (Table S1). The results suggest that the edge groups of carboxylic acid already reached saturation limit of oxidation in 2xGO, which forced the oxidation to occur on the basal plane when the amount of KMnO 4 was further increased to 9 g (3xGO). The selective increase of epoxide over alcohol groups on basal plane may suggest the occurrence of epoxidation prior to ring-opening that results in alcohol groups. In terms of carbocatalytic activity on the synthesis of triazoloquinazolinones, these observations support that the higher activities of 3xGO than 2xGO and 1xGO are due to the greater number of basal plane oxygen groups of epoxide and alcohols on the surface of GO sheets. In addition, the basal plane epoxide and alcohol groups are experimentally supported to be involved in the reaction pathway yielding triazoloquinazolinone compounds based on the observed trend for yield of 4a using GO materials with different oxidations (Table 3). Peak deconvolution of the XPS spectra (Table S1) shows 3xGO, which gives the highest yield of 4a, has the highest percentage of epoxide (C-O-C) peak component. The non-saturated characteristics of basal plane groups increasing with the oxidation degrees of GO indicates its dominant role in promoting reaction yields when Brønsted acid sites reaches the upper limits.
To further confirm the role of oxygen groups in the multicomponent synthesis, we conducted characterization for the comparison between the fresh 3xGO and that after reusing for 5 times (hereby denoted as 3xGO-5c) in the synthesis of 4a. It is generally accepted that the oxygen functional groups on the surface of GO sheets are the key to hold carbocatalytic activities 54 . As shown in Fig. 3, the X-ray photoelectron spectroscopy (XPS) data of the fresh and used 3xGO show the C 1 s signals with peak fitting and deconvolution results. The relative amounts of aromatic carbon (C−C sp 2 , 284.3 eV), aliphatic carbon (C−C sp 3 , 284.9 eV), hydroxyl (C−OH, 285.6 eV), epoxy (C−O−C, 286.9 eV), carbonyl (CO, 287.8 eV), and carboxyl (O−CO, 289.15 eV) were quantified and summarized in Table 6 26,55 .
The relative abundance of epoxy groups in 3xGO-5c decreased substantially from 36.8% to 30.3%, while the relative abundance of hydroxyl groups increased from 6.2% to 10.5% suggesting a partial ring opening of epoxide units during the course of the reaction. In addition, the FT-IR spectra of 3xGO and 3xGO-5c (Fig. 4a) shows that, in 3xGO-5c, the weakening (looking at the relative intensity compared to the other peaks) of the stretching band for C=O stretching of undissociated carboxylic acid at 1730 cm −1 (region c) accompanied by appearance of new peak at 1690 cm −1 (region b) which corresponds to C=O stretching of dissociated carboxylic acid (carboxylate ion) 44 . This suggests the deprotonation of carboxylic acid leading to the formation of carboxylate during the course of the

Mechanism.
Gathering leads from all the previous results, a mechanism for the formation of 4a in the presence of carboxylic acid and epoxides on 3xGO is proposed (Fig. 5). It was previously reported in a similar reaction by Dam et. al., that the carboxylic acids of GO are the only functional group involved in the mechanism through: (1) increasing the electrophilicity of the aldehyde and the β-dicarbonyl compound and; (2) promoting the dehydration steps 56 . The XPS and FTIR data, however, suggest that the role of GO is not only limited to the participation of its carboxylic acid group, but also includes the epoxide groups. As shown on Scheme 1a, 3xGO is initially involved as a Brønsted acid initiating the reaction by promoting keto-enol tautomerization of dimedone and activation of benzaldehyde. Reaction of the enol tautomer and the protonated benzaldehyde leads to the formation of intermediate i. The role of the epoxides in the mechanism is proposed and summarized in Fig. 5b. Due to the inherent Brønsted acidity of 3xGO, it forms an acidic dispersion which allows the protonation of the epoxide groups at the reaction temperature. Ring opening of the epoxide groups relieves ring strain and forms an alpha-carbocation. The carbocation could act as a Lewis acid aiding dehydration of the beta-hydroxy intermediate i to form the Knoevenagel intermediate ii 25,26,57,58 . Nucleophilic attack of 3-amino-1,2,4-triazole to ii followed by 3xGO promoted dehydration of iii and cyclization reaction of iv leads to the formation of 4a. This mechanism not only justifies the overall increase of alcohol functionalities and eventual decrease in epoxide group observed in XPS but also explains the formation of carbonyl peak corresponding to carboxylate ion in FT-IR. In addition, the small increase in the O/C ratio of 3xGO vs 2xGO (Table 1) is mainly associated with the basal plane oxidation increasing the number of Lewis acid sites and enhancing the carbocatalytic activity of 3xGO. This observation also supports the proposed role of GO and the importance of epoxide groups in the mechanism.
To further verify the proposed mechanism where the Brønsted acidic functional groups and the Lewis acidic sites generated from epoxides on 3xGO surface responsible for the activities during the multicomponent reaction, samples of 3xGO treated with a base (denoted as 3xGO-Base) and another subjected to a sequential base-then-acid treatment (3xGO-BaseAcid) were prepared 59 . By subjecting 3xGO to the base treatment, the epoxide groups of the material may undergo ring opening reaction upon nucleophilic attack of OH − , leading to the formation of alcohol units as product on the surface of 3xGO-Base, while the subsequent acid treatment may lead to epoxide ring formation via acid catalyzed dehydration (Fig. 6) 60 . The 3xGO was first sonicated in the presence of 1 M NaOH to get 3xGO-Base. Compared to the 3xGO, the FT-IR spectra of 3xGO-Base (Fig. 4b) shows a relatively lowered intensity of C=O of undissociated carboxylic acid at ~1730 cm −1 (region c) and the appearance of carboxylate C=O peak at around 1670 cm −1 (region b), supporting the deprotonation of carboxylic acid groups. The intense peak corresponding to C-O-C stretch of epoxide which appears at ~1040 cm −1 (region a) is absent in the FT-IR spectra of 3xGO-Base. The broad -OH peak at ~3300 cm −1 also disappeared after base treatment due to the abstraction of the protons by the strong hydroxide base (Fig. S3). To recover the Brønsted   Table 6. Abundance percentage of different functional groups on the surface of 3xGO before and after reactions based on Fig. 3.
www.nature.com/scientificreports www.nature.com/scientificreports/ acidic groups and epoxide groups, 3xGO-Base was sonicated in the presence of 1 M HCl to yield 3xGO-BaseAcid. Interestingly, the C=O signal of carboxylic acid at ~1730 cm −1 can be regenerated with the weakened signal of carboxylate C=O peaks, revealing a successful reprotonation of the carboxylic acid groups. Broad hydroxyl peak in the FT-IR was also regenerated after acid treatment (Fig. S3). Furthermore, the epoxide C-O-C stretch at ~1040 cm −1 was restored signifying the recovery of epoxide units after acid treatment. We carried out the synthesis of 4a using 3xGO-Base, but it only gave 26% yield of 4a. By using 3xGO-BaseAcid in the multicomponent reaction, 81% yield of 4a can be achieved. Current methods for the selective protection of the carboxylic acid groups employs reaction conditions that will lead to ring opening of the epoxide groups which makes isolation of the effect of epoxide groups in the reaction difficult. However, all the obtained results support the importance of both Brønsted acidic groups and generated Lewis acid sites of GO in promoting the multicomponent reaction.
To date, the mechanism of multicomponent reactions remains questionable 27,61,62 . We were able to isolate the enol 5′ as the major side product formed during reaction optimization (Table S2) giving an important insight on the mechanism of the reaction. The formation of 5′ could arise from the Michael-addition of the enol tautomer of dimedone to the Knoevenagel intermediate (Fig. 7) similar to previously reported mechanisms 63,64 . The reaction mechanism for the formation of 5′ is consistent with our proposed mechanism involving the formation of a Knoevenagel intermediate.  www.nature.com/scientificreports www.nature.com/scientificreports/ Reusability study. An important advantage of using heterogeneous acid promoters is that they could be conveniently reused without a drastic drop in activity 65 . Reusability studies of 3xGO toward the synthesis of 4a were conducted and the highest yields for each runs are reported on Fig. 8. 3xGO can be reused for at least 5 times with a slight decrease in the yield (e.g., from 95% for fresh 3xGO to 85% for recycled 3xGO after 5 runs) of 4a. The consumption of acidic protons and structural changes of 3xGO during the reaction may gradually deactivate 3xGO after each uses suggesting that, contrary to the claims on other carbocatalytic systems, 3xGO is not strictly acting as a catalyst in the reaction 66 . The GO samples are more suitable to be considered as a promoter in the organic synthesis. The observed regenerable acidity of 3xGO in the mechanism studies implies that the catalytic activity of used 3xGO may be recovered by acid treatment. A similar acid treatment protocol (sonication with 1 M HCl) was carried out in an attempt to regenerate the used 3xGO. The regenerated 3xGO after 2 runs shows a yield of 92% in the third run, while only 84% yield can be obtained using 3xGO for successive 3 runs of the reaction (Fig. 8). The yield recovery of the acid-regenerated 3xGO is 97% that of the fresh 3xGO. In addition, a slightly higher yield was observed for the acid-regenerated 3xGO (85%) than non-regenerated ones (83%) in the fourth runs. As the structural changes (consumption of acidic protons and ring opening of epoxide groups) on 3xGO excludes it from being a "catalyst" by following conventional definition, the ex-situ acid treatment is capable of regenerating the recycled 3xGO to achieve the reaction yields comparable with a fresh one. Together with the acid treatment, 3xGO can nearly recover to its original state ready for the next reaction. In this case, 3xGO cannot be precisely described either as a catalyst or a consumable reagent in organic synthesis. Thus we adopt "pseudocatalyst" to describe GO as a catalyst-like material that require ex-situ regeneration to complete its catalytic cycle.  www.nature.com/scientificreports www.nature.com/scientificreports/ conclusion In this work, both the Lewis-acid sites and Bronsted acidic edges of GO are identified to facilitate triazoloquinazolinone synthesis. Since the quantity of edge carboxylic groups has an upper limit, the increased amount of epoxide Lewis acid sites become the determining factor in achieving highly active GO promoters in organic synthesis, rather than the Bronsted acid sites that tend to be saturated. The 3xGO promoter can be easily regenerated by simple acid treatment, suggesting the high stability of both Lewis and Bronsted acid sites for diverse acid-promoted organic synthesis. The reviewer suggested extending the reaction scope to heterocyclic aldehydes. Preliminary tests using pyridine-3-carbaldehyde yields the corresponding triazoloquinazolinone with the promising yields (see SI). This indicates the wide potential application of 3xGO in heterocyclic aldehyde systems for the multicomponent synthesis of triazoloquinazolinones. Mild reaction conditions, easy work up, and good reusability of 3xGO may fortify the future applications as heterogeneous pseudocatalyst highly active for metal-free, sustainable, less toxic, and green pathway.