Utilization of hexagonal boron nitride as a solid acid–base bifunctional catalyst

Journal of Catalysis 355 (2017) 176–184

Contents lists available at ScienceDirect

Journal of Catalysis journal homepage: www.elsevier.com/locate/jcat

Utilization of hexagonal boron nitride as a solid acid–base bifunctional catalyst Shusaku Torii a, Keiko Jimura b, Shigenobu Hayashi b, Ryuji Kikuchi a, Atsushi Takagaki a,⇑ a

Department of Chemical Systems Engineering, School of Engineering, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan Research Institute for Material and Chemical Measurement, National Institute of Advanced Industrial Science and Technology (AIST), Central 5, 1-1-1 Higashi, Tsukuba, Ibaraki 305-8565, Japan b

a r t i c l e

i n f o

Article history: Received 6 July 2017 Revised 4 September 2017 Accepted 15 September 2017

Keywords: Solid acid–base catalyst Hexagonal boron nitride Ball milling Nitroaldol reaction Knoevenagel condensation

a b s t r a c t This work explores the use of hexagonal boron nitride (h-BN), a graphite-like compound, as a novel catalyst with base and acid functionalities. For use as a solid catalyst, the layered structure of h-BN was disrupted by ball-milling, exposing boron and nitrogen edge sites as well as increasing the surface area from 3 to ca. 400 m2 g 1. Fourier transform infrared spectroscopy, X-ray photoelectron spectroscopy, and proton magic-angle spinning nuclear magnetic resonance spectroscopy (1H MAS NMR) indicated simultaneous and adjacent formation of amino and hydroxyl groups by milling, which function as Brønsted base and acid sites, respectively. Analysis using color indicator reagents and pyrrole-adsorbed 1H MAS NMR results revealed that the ball-milled h-BN had basic sites of strength +9.3 > H  +7.2, comparable to those of KY zeolite. Measurements of 31P MAS NMR of adsorbed trimethylphosphine oxide indicated that the ball-milled h-BN had weak acid sites, comparable to those in HY zeolite. Despite its weak basicity, the ball-milled h-BN showed high activity and selectivity toward b-nitroalkenes for the nitroaldol reaction (Henry reaction) and the Knoevenagel condensation, whereas nontreated h-BN did not show activity. The nitroaldol reaction was considered to proceed in two steps: the abstraction of a proton from nitromethane by the amino group and the formation of an imine followed by a nucleophilic attack of the deprotonated nitromethane. Kinetic isotope effect experiments using D-substituted nitromethane revealed that the first step was the rate-determining step. Several nitroaldol reactions using a variety of monosubstituted benzaldehydes indicated that electron-donating groups enhanced the activity, suggesting that the formation of adjacent base and acid sites is responsible for it. This study shows the high catalytic activity of BN, a solid catalyst with moderate basicity and weak acidity. Ó 2017 Elsevier Inc. All rights reserved.

1. Introduction Design of well-defined catalytic active sites and understanding of their catalytic behavior are of significance for both homogenous and heterogeneous catalysts. The concept of frustrated Lewis pairs in which Lewis acid and base sites are rationally positioned in one molecule without neutralization is an excellent example of the application of homogeneous acid–base cooperative catalysts for the activation of small molecules, including hydrogen, alkynes, and carbon dioxide [1–3]. Another good example is the molecular catalyst proline, which is an amino acid consisting of a secondary amine-containing pyrrolidine with an attached carboxylic acid group [4]. The acid and base sites are appropriately positioned and efficiently catalyze asymmetric aldol reactions.

⇑ Corresponding author. E-mail address: [email protected] (A. Takagaki). https://doi.org/10.1016/j.jcat.2017.09.013 0021-9517/Ó 2017 Elsevier Inc. All rights reserved.

For heterogeneous acid–base catalysts, amine-tethered silica materials such as bulk silica [5], mesoporous silica [6–8], and silica-alumina [9] have been extensively studied as efficient acid–base cooperative catalysts that catalyze carbon–carbon bond formation reactions including aldol, nitroaldol, and Knoevenagel condensations and the Michael addition. For aminosilica materials, the coexistence of weak acid sites (e.g., silanols) with base sites (alkylamines) in spatially appropriate positions is a key to realizing high catalytic activity for these reactions. The neighboring weak acid sites are important because they help to stabilize the carbonyl groups of the ketone and aldehyde species, which efficiently promotes the condensation reactions. To maximize the cooperativity, there is an optimal distance between acid and base sites, which can be controlled by adequate concentration of acid and base sites [10], use of aminoalkyl silanes with appropriate alkyl linker lengths [11], and adjustment of pore size for mesoporous silica [8]. The spatial positioning of the acid–base pair with high precision is

S. Torii et al. / Journal of Catalysis 355 (2017) 176–184

necessary, but the active pairs formed on silica are average structures because of the amorphous nature of the silica, in which a variety of surface silanols including isolated, vicinal, and geminal silanols with slightly different bond lengths are present [12,13]. An alternative method of developing acid–base pairs on silica is the use of strained siloxane bridges as a source of active pairs. Opening of siloxane bridges by base molecules simultaneously affords acid–base sites [14,15]. One example is ammonia treatment at low temperature of a mesoporous silica that was pretreated at high temperature [14]. The high-temperature calcination led to formation of strained siloxane bridges, SiAO–Si. Ammonia dissociatively adsorbed onto the siloxane bridges by the reaction – SiAO–Si– + NH3 ? -Si–NH2 + –Si–OH. The thus formed acid–base pairs were well-defined and functioned as cooperative catalysts for the Knoevenagel condensation. Very recently, it was reported that mesoporous silica SBA15, which mostly has strained reactive siloxane rings, reacted with aniline to produce N-phenylsilana mine–silanol acid–base pairs, which were also effective for the Knoevenagel condensation. The SBA-15 was pretreated at very high temperature (1100 °C) under vacuum before reaction with aniline [15]. It is desirable to design such well-defined acid–base pairs for other materials composed of different elements. Utilization of the crystal structure of inorganic compounds is a promising approach. In this study, hexagonal boron nitride (h-BN) was investigated as a new bifunctional catalyst. h-BN is a layered compound isostructural with graphite and has been widely studied for diverse applications outside of catalysis, such as electronic and optical devices using heterostructures linked by van der Waals forces [16,17], coating materials with high-temperature oxidation resistance [18], adsorbents for organic compounds useful in water cleaning [19], and materials for hydrogen storage [20]. Unlike graphite, BN has unshared electron pairs localized on the nitrogen atoms, resulting in a polarized nature. Owing to strong chemical bonding between the atoms and physical bonding between the layers, BN is chemically and thermally stable, electrically insulating, and mechanically robust. However, very recently, h-BN was found to catalyze oxidative dehydrogenation of alkanes [21,22], and hydrogenation [23]. The present study focuses on the chemically polarized nature of h-BN. The nitrogen and boron composing h-BN are expected to function as base and acid sites, respectively, under appropriate treatment. A top-down ball-milling method was chosen to disrupt the structure of bulk h-BN [24–26], resulting in exposure of the edge sites of the planar h-BN. This study reveals that amino and hydroxyl groups were simultaneously formed at adjacent positions on the h-BN surface using the simple ballmilling method, and these groups functioned as efficient cooperative acid–base sites for the nitroaldol reaction. 2. Experimental 2.1. Preparation of ball-milled h-BN A quantity of 0.8 g of h-BN (Wako Pure Chemical Industries, Ltd.) was ball-milled at 400 rpm for 6–24 h using a planetary ball mill (Pulverisette 7, Fritsch, zirconia vessel with six zirconia balls (diameter 10 mm)). Ball milling was conducted for 12–48 cycles, in which each cycle was carried out for 30 min with 5 min intervals. The rotation direction was reversed each cycle. The samples prepared were denoted as h-BN bm6–24 h, respectively. 2.2. Characterization The specific surface area of the catalysts was evaluated by the Brunauer–Emmett–Teller (BET) method using nitrogen adsorption

177

(BELSORP-mini II, Microtrac-BEL). Samples were pretreated by evacuation at 200 °C for 3 h prior to the measurements. The crystal structure of the catalysts was determined by X-ray diffraction (XRD) (RINT-2700, Rigaku with CuKa radiation) at a voltage of 40 kV and a current of 100 mA with 2h values from 20° to 60°. Measurements were made on samples without pretreatment. Fourier transform infrared spectroscopy (FTIR) (FT/IR-6100, JASCO) measurements were performed to identify functional groups on the catalysts. Samples without pretreatment were pressed into pellets with KBr for the measurements. X-ray photoelectron spectroscopy (XPS) (JPS-9200, JEOL with MgKa radiation) measurements were carried out to investigate the oxidation state of elements and their compositions over the catalyst surface. The binding energies in each measurement were referenced to the core level of the C1s peak (284.8 eV), and all obtained peaks were fitted by Gaussian curves. The samples were not pretreated. Titrations using color indicator reagents were carried out to evaluate the base strength and base amounts of the catalysts. A quantity of 50 mg of catalyst was added to 1 mL of toluene solution containing 0.4 mg of bromothymol blue (pKa = 7.2) or phenolphthalein (pKa = 9.3) for 1 day. The base amounts of the catalyst were determined by titration using 0.01 M benzoic acid in toluene. Proton magic-angle spinning nuclear magnetic resonance (1H MAS NMR) measurements were conducted to quantify protoncontaining functional groups (hydroxyl and amino groups). Measurements were carried out as previously reported [27]. The 1H MAS NMR spectra were recorded with a Bruker Avance III HD 600WD spectrometer operating at a frequency of 600.39 MHz. A Bruker MAS probehead was used with a zirconia rotor of outer diameter 4 mm. All measurements were performed at room temperature with a spinning rate of 10 kHz. The 1H spectra were measured with an ordinary single pulse sequence. The flip angle of the pulse and the recycle delay were p/2 and 3 s, respectively. We did not measure the 31P spin–lattice relaxation times of the present samples. The 31P spin–lattice relaxation times are 6.3 and 1.7 s for crystalline and hydrated TMPO, respectively [28]. The 31P spins are relaxed mainly by the rotation of the methyl group. We did not quantitate the adsorbed TMPO, and thus full recovery was not necessary. To improve the efficiency of the signal accumulation, we set the recycle delay to 3 s. The 1H chemical shift was expressed with respect to neat tetramethylsilane (TMS). Experimentally, secondary standard compounds were used, such as adamantane (1.85 ppm at a spinning rate of 8 kHz). The 1H content in the samples was quantitated by comparing the integrated signal intensities of the samples with that of adamantane. The solid base properties of the catalyst were investigated by 1H MAS NMR using pyrrole (Wako Pure Chemical Industries, Ltd.) as a probe molecule. In a typical adsorption experiment, an appropriate volume of pyrrole (2–10 lL, 0.03–0.14 mmol) was injected into a glass tube containing h-BN (0.20 g) under nitrogen. Samples were heated to 100 °C for 3 h to ensure homogeneous distribution of the probe molecules. Samples were carefully transferred into a zirconia rotor in nitrogen atmosphere and capped tightly to avoid moisture. The solid acid properties of the catalysts were examined by 31P MAS NMR using trimethylphosphine oxide (Wako Pure Chemical Industries, Ltd) as a probe molecule. The NMR spectra were measured at room temperature using a Bruker Avance 400 spectrometer at a Larmor frequency of 161.98 MHz. A single-pulse sequence was employed with high-power proton decoupling at a sample spinning rate of 8 kHz. The flip angle of the pulse and the recycle delay were p/4 and 3 s, respectively. The 31P chemical shift was referenced to 85% H3PO4 at 0.0 ppm, with (NH4)2HPO4 used as a secondary reference material at 1.33 ppm. The samples with adsorbed

178

S. Torii et al. / Journal of Catalysis 355 (2017) 176–184

TMPO were prepared by dehydration at 250 °C for 1 h under vacuum followed by immersion in a dichloromethane solution containing TMPO at room temperature for 1 day in a glove bag filled with N2. An amount of 0.06 mmol of TMPO was exposed to 0.30 g of the BN sample. After evacuation to remove the dichloromethane solvent, the sample was packed in a rotor housed in a glove box under N2. 2.3. Nitroaldol reaction A typical reaction was conducted using 50 mg of catalyst in 2 mL of toluene solution with p-methoxybenzaldehyde (60.5 lL, 0.5 mmol), nitromethane (68.0 lL, 1.25 mmol), and n-decane (19.5 lL, 0.1 mmol) as an internal standard. The reaction was performed at 100 °C, and aliquots were taken by syringe and analyzed by gas chromatography (GC-FID; GC-2014, Column DB-1MS, Shimadzu). n-Butylamine (Kanto Chemical Co., Inc.) and boric acid (Wako Pure Chemical Industries, Ltd.) were used for comparison. The Hammett equation was used to analyze the results using different substrates for the nitroaldol reaction [29,30]. The reaction was conducted using monosubstituted benzaldehyde (0.50 mmol) and nitromethane (1.25 mmol) as substrates under the same conditions as above. The monosubstituted benzaldehydes used were p-methoxylbenzaldehyde (Wako Pure Chemical Industries, Ltd.), p-tolualdehyde (Tokyo Chemical Industry Co., Ltd.), benzaldehyde (Wako Pure Chemical Industries, Ltd.), pAchlorobenzaldehyde (Tokyo Chemical Industry Co., Ltd.), and m-methoxybenzaldehyde (Wako Pure Chemical Industries, Ltd.). The primary kinetic isotope effect (KIE) was carried out using p-methoxybenzaldehyde with nitromethane as well as d-substituted nitromethane (CD3CN) [31] (see the Supplementary Material). 2.4. Knoevenagel condensation A typical reaction was conducted using about 50 mg of catalyst in 2 mL of toluene with aldehyde (0.5 mmol), malononitrile (33 mg, 0.5 mmol, Sigma-Aldrich, Co.), and n-decane (19.5 lL, 0.1 mmol). The reaction was performed at room temperature, and aliquots were taken by syringe and analyzed by GC using n-decane as an internal standard. Aldehydes used were p-methoxybenzaldehyde, benzaldehyde, p-nitrobenzaldehyde, furfural (Tokyo Chemical Industry Co., Ltd.), propionaldehyde (Wako Pure Chemical Industries, Ltd.), and pivalaldehyde (Wako Pure Chemical Industries, Ltd.). 3. Results and discussion 3.1. Structure and physicochemical properties of ball-milled h-BN Table 1 shows BET surface areas and pore volumes of pristine and ball-milled h-BN samples. Ball milling increased the BET surface area of h-BN from 3 to 227 m2 g 1 for h-BN bm6h and 404 m2 g 1 for h-BN bm12h. Ball milling also increased pore volumes from 0.02 to 0.46 mL g 1 for h-BN bm6h and 0.50 mL g 1 for hBN bm12h. Fig. 1 shows nitrogen sorption isotherms and pore size distributions obtained from Barrett–Joyner–Halenda plots.

Table 1 BET surface areas and pore volumes of pristine and ball-milled h-BN. Sample

SBET/m2 g

h-BN h-BN bm6h h-BN bm12h h-BN bm24h

3 227 404 155

1

Pore volume/mL g 0.02 0.46 0.50 0.33

1

Increase of both micropores and mesopores was found for the ball-milled samples. The variations in surface area were mostly influenced by emergence of micropores, and those in pore volume by that of mesopores. It seems that both surface area and pore volume increased by exfoliation and successive disruption of bulk structure. Further ball-milling treatment decreased the surface area to 155 m2 g 1 and the pore volume to 0.33 mL g 1 for h-BN bm24h, likely due to agglomeration [32]. It should be noted that the reproducibility of the procedure was confirmed. For example, the samples of h-BN bm12 h were prepared three times. The BET surface areas of three samples were 404, 417, and 399 m2 g 1. Other physical properties determined from XRD and FTIR were almost the same. Fig. S1 in the Supplementary Material shows XRD patterns for pristine h-BN and h-BN bm12h (see Supplementary information). Both samples showed strong peaks centered at 26.6° (2h), corresponding to the (0 0 2) interlayer reflection of h-BN along with peaks attributed to in-plane structures. Ball milling resulted in weakening of all peaks, indicating that the bulk structure of h-BN was disrupted. Peaks corresponding to the layer structure ((0 0 2 and (0 0 4)) preferentially weakened compared with those of the in-plane structure ((1 0 0), (1 0 1) and (1 0 2)) (Table S1). In addition, the (0 0 2) peak position remained unchanged, indicating that the interlayer distance was maintained after ball milling. Fig. 2 shows FTIR spectra for pristine h-BN and ball-milled h-BN (h-BN bm12h). The IR spectrum of pristine h-BN exhibited only lattice vibration modes derived from covalent bonds between nitrogen and boron atoms (Fig. 2, bottom). A strong absorption band centered at 1378 cm 1 is assigned to an in-plane B–N bond stretching vibration. A sharp peak centered at 816 cm 1 is characteristic of an out-of-plane B–N–B bending vibration [33,34]. The IR spectrum of the ball-milled h-BN sample is shown at the top. The B– N in-plane vibrational mode broadened, suggesting a substantial change in lattice vibrations due to ball milling. In contrast, the out-of-plane B–N–B bending mode remained constant at 816 cm 1, indicating retention of the specific honeycomb structure of h-BN. However, a decrement of the absorbance of B–N–B bending mode was observed from that of the B–N in-plane vibrational mode, indicating that B–N–B bonds in ball-milled h-BN were cleaved by the ball-milling treatment. A significant difference between before and after ball-milling was the appearance of bands centered at 3400 and 3200 cm 1, which can be assigned to O–H and N–H stretching bands, respectively [35]. This clearly indicates formation of hydroxyl and amino groups [36]. From the results of the FTIR measurements, the surface structure of ball-milled h-BN is suggested to be as shown in Scheme 1, which involves cleavage of B–N–B bonds and simultaneous formation of B–OH and –NH2 groups. The cleavage of B–N–B bonds is expected to proceed by reaction with moisture in air. This oxidative cleavage of B–N–B bonds is in good agreement with the FTIR results previously reported [34]. Measurements of XPS spectra were performed to further investigate the chemical states of B, N, and O in the h-BN catalysts. XPS survey spectra of pristine h-BN and h-BN bm12h are shown in Fig. S2. The O1s peak of the samples was located at a binding energy of around 533 eV. In the spectrum of h-BN bm12h, a large increase in the signal intensity of the O1s peak was observed, whereas no significant changes in peak position or signal intensity were observed for N1s and B1s peaks. C1s detected was attributed to surface contamination. In addition, pristine h-BN has no oxygen species in its structure and no hydroxyl groups, as confirmed by 1H MAS NMR measurements (vide infra); hence the O1s peak detected in the pristine h-BN can also be attributed to surface contamination. Fig. 3 shows core-level spectra of B1s and N1s of h-BN samples before and after ball milling. The XPS spectrum of pristine h-BN

179

S. Torii et al. / Journal of Catalysis 355 (2017) 176–184

(A)

(B)

300 bm12h

0.12

200

0.10 dV/drp

-1

bm12h

0.14

250

V /mL g

0.16

bm6h

150 100

BN

0 0.0

0.2

0.4

0.6

0.8

0.06 0.04

bm24h

50

0.08 bm6h

1.0

0.02 bm24h 0.00 BN 1

10

P/P0

Pore size rp /nm

Fig. 1. N2 sorption isotherms and pore size distributions of pristine and ball-milled h-BN.

3400

ture, it is expected that it will show single peaks for each element. Analysis of the B/N atomic ratio indicated an elemental formula of BN0.97 for pristine h-BN. For all the ball-milled h-BN samples, the shape of the B1s spectra became asymmetric with a shoulder at higher binding energy (192.9 eV), indicating the appearance of oxidized boron on the surface. The main peaks of the ball-milled samples (h-BN bm6h, bm12h, and bm24h) were centered at 191.0, 191.1, and 191.1 eV, respectively, corresponding to that of pristine h-BN at 191 eV. A new peak, which appeared at 192.9 eV, for all ball-milled samples could be assigned to fully oxidized boron (B3+) on the edge sites, because this peak position was in good agreement with the previously reported binding energy for oxidized boron [38–40]. It is noteworthy that the ratio of oxidized boron increased with increasing ball-milling time. The ratios of oxidized boron to h-BN structural boron, B3+/B+, were 0.16 for h-BN bm6h, 0.19 for h-BN bm12h, and 0.33 for h-BN bm24h. For h-BN bm12h, analysis of the B/N/O atomic ratio indicated an elemental formula of BN0.94O0.27. The O/B3+ ratio was calculated to be around 1.7 for hBN bm12h. This indicates that the surface structure for h-BN bm12h consists of a mixture of mono- and di-hydrolyzed boron with hydroxyl and amino groups, as shown in Scheme 1. The N1s spectrum of h-BN bm12h was broad and can be deconvoluted to two peaks with the same width (Fig. 3B). The peak centered at 398.9 eV can be attributed to amino groups, which were formed by cleavage of B–N bonds. Measurements of 1H MAS NMR were conducted to estimate the concentrations of hydroxyl and amino groups. Fig. 4 shows 1H MAS NMR spectra for pristine h-BN and h-BN bm12h. The spectrum of pristine h-BN (Fig. 4b) was very similar to that of the empty rotor

3200

Absorbance

816 h-BN bm12h

1378

0.5

h-BN 3600

3200

2800

2400

2000

1600

1200

800

-1

Wavenumber /cm

Fig. 2. FTIR spectra for pristine h-BN and h-BN bm12h.

H2O

H2O

Scheme 1. Simultaneous formation of hydroxyl and amino groups via cleavage of B–N bonds.

showed single peaks for the core-level B1s at 191.1 eV and N1s at 398.9 eV. All peak positions were in close agreement with those reported for characterization of synthesized h-BN and modified h-BN substances [35,37,38]. Because BN has a delocalized struc-

(A) B 1s

(B) N 1s 398.9

191.1 192.9

h-BN bm12h B

3+

h-BN bm6h

399.8

Intensity /a.u.

Intensity /a.u.

h-BN bm24h

h-BN bm12h

+

B (Bh-BN)

h-BN

h-BN 196

194

192

190

Binding energy /eV

188

404 402 400 398 396 394 Binding energy /eV

Fig. 3. B1s and N1s core levels of pristine and ball-milled h-BN.

S. Torii et al. / Journal of Catalysis 355 (2017) 176–184

Intensity /a.u.

OH 6.7 3.7 NH2 SSB

SSB

SSB

SSB

(d) (c) - (a)

(c) h-BN bm12h (b) h-BN (a) vacant rotor 50 40 30 20 10

0 -10 -20 -30

δ /ppm Fig. 4. 1H MAS NMR spectra for (a) empty rotor, (b) pristine h-BN, (c) h-BN bm12h, and (d) h-BN bm 12h minus the empty rotor background. SSB indicates spinning sidebands.

(Fig. 4a), indicating that there were almost no protons in pristine hBN. On the other hand, the spectrum of h-BN bm12h consisted of two distinct peaks centered at 6.7 and 3.7 ppm. The peaks on each side of the main peaks (40, 24, 11, and 27 ppm) are spinning sidebands (SSB). Subtraction of the empty rotor background from the h-BN bm12h signal gives the spectrum shown in Fig. 4d. The two main peaks at 6.7 and 3.7 ppm can be tentatively attributed to hydroxyl and amino groups [41] formed by the ball-milling treatment, as discussed in the IR and XPS sections. From the quantitative analysis of 1H content in the sample, the amount of hydroxyl groups was estimated to be 3.91 mmol g 1 and of amino groups was 1.67 mmol g 1. The ratio of hydroxyl groups to amino groups was 2.34, which is close to 2, as shown in Scheme 1. To further investigate chemical bonding in these proton species, use was made of the spin–echo double resonance (SEDOR) method. By irradiating the 11B nucleus and observing the 1H nucleus, it is possible to determine whether 11B is present in the vicinity of 1 H. Fig. S3a shows the 1H MAS NMR spectrum obtained by the Hahn echo method (90°–s–180°–s–acquisition). Fig. S3b shows the spectrum obtained by irradiating a 11B pulse (192.62 MHz) during the interval between the 90° and 180° 1H pulses. Compared with Fig. S3a, the signal intensity decreased, indicating that only the 1H signal in the vicinity of 11B decreased. Fig. S3c is the SEDOR spectrum, which was obtained by subtraction of Fig. S3b from Fig. S3a. Both characteristic signals still remained unchanged. This clearly indicates that both protons are present equally near boron. Therefore, it is reasonable that these two proton species can be assigned to B-OH and B-NH2.

3.2. Solid acid–base properties of ball-milled h-BN Solid base strength was evaluated using color indicator reagents. Treatment of the three ball-milled h-BN samples (h-BN bm6h, 12 h, and 24 h) with a solution of toluene containing bromothymol blue (pKa = 7.2) resulted in a green color, whereas pristine h-BN gave a yellow color. However, these ball-milled h-BN samples did not become pink in phenolphthalein (pKa = 9.3). This indicates that pristine h-BN has no basic sites and ball-milled hBN samples have weak basic sites in the range of + 9.3 > H  +7.2, which are comparable to those in Mg(OH)2 and NaX zeolite, but weaker than those in hydroxyapatite and stronger than those in CaCO3 [42,43]. The number of basic sites in ball-milled h-BN

was estimated by titration using benzoic acid. The h-BN bm12h sample had the greatest number of basic sites, 0.37 mmol g 1. The basic site number in h-BN bm6h was 0.21 mmol g 1 and that in bm24h was 0.29 mmol g 1. The solid base properties of ball-milled h-BN were also investigated by 1H MAS NMR using pyrrole as a probe molecule. Fig. 5 shows the 1H MAS NMR spectra of h-BN bm12h before and after adsorption of pyrrole. There are two signals with peak positions at 6.7 and 3.7 ppm before pyrrole adsorption, as mentioned above (Fig. 4d). After adsorption of pyrrole, two peaks appeared at 6.1 and 1.2 ppm along with a small peak at ca. 9.5 ppm. The signal at 6.1 ppm can be ascribed to a mixture of a and b protons of the pyrrolate anion [44], and the signal at 1.2 ppm can be attributed to new proton species that are formed by dissociative adsorption of pyrrole [30]. These results imply that ball-milled h-BN has strong base sites that can dissociate pyrrole, although color indicator reagents composed of large molecules cannot interact with such sites because of steric hindrance. A small peak at around 9.5 ppm can be attributed to the N–H proton of pyrrole adsorbed on moderately basic sites of h-BN [45]. The signal for the N–H proton moves toward higher frequency after adsorption of pyrrole, with the chemical shift indicative of the base strength of the solids. For example, for LiY, KY, and KX zeolites, the signals for the N–H proton were observed at 8.4, ca. 9.7, and 11.5 ppm, respectively [45]. Corresponding signals appeared at 11.3 and 11.2 ppm for calcined hydrotalcite and c-Al2O3 [30]. Thus, the peak position of the ballmilled h-BN is comparable to that of KY zeolite [45]. The amount of adsorbed pyrrole (0.14 mmol g 1) was smaller than the amount of base sites (0.37 mmol g 1), suggesting that pyrrole molecules were preferentially adsorbed on relatively stronger base sites. The solid acid properties of the ball-milled h-BN were also investigated by 31P MAS NMR using trimethylphosphine oxide (TMPO) as a probe molecule. On the acidic sites of solids, the signal of TMPO moves toward higher frequency by adsorption to form a protonated species, TMPOH+. Thus, the chemical shift is indicative of the acid strength of the solids [46,47]. Fig. 6 shows the 31P MAS NMR spectra for pristine h-BN and h-BN bm12h after adsorption of

6.1

1.2 9.5 Intensity /a.u.

180

After adsorption

6.7 3.7 B-NH2

B-OH

Before adsorption 15

10

5

0

-5

δ /ppm Fig. 5. 1H MAS NMR spectra of h-BN bm12h before and after adsorption of pyrrole. Amount of pyrrole adsorbed: 0.14 mmol of pyrrole per gram of h-BN bm12h.

181

S. Torii et al. / Journal of Catalysis 355 (2017) 176–184

CH3

CH3 P

43 CH3

Intensity /a.u.

H3C H3C P = O H3C

63

O

ball-milled h-BN catalyzed the reaction; the pristine h-BN showed negligible activity (entry 1). The ball-milled h-BN catalysts afforded high selectivity (76–79%) toward p-methoxy-bnitrostyrene (3a) (entries 2, 3, and 6). For the three ball-milled catalysts, h-BN bm12h showed the highest activity (84% conv.) with high selectivity (77%) for 3a (entry 3). These results indicated that exposure of the side planes of h-BN resulted in the formation of reaction sites of the catalysts and led to significant improvement of the catalytic activity. In addition, the reaction using h-BN bm12h was not decelerated in the presence of water (0.11 mmol) (entry 4). The main reason for the selectivity losses is moderate carbon balance (90%). Some products were adsorbed onto the catalyst surface. Lack of mass transfer limitations was verified by the Weisz–Prater criterion (CWP) [51], which is based on the ratio of reaction rate to diffusion rate, and the calculations are summarized in Table S2. If the value of CWP is less than 0.3, the internal mass transfer effects can be neglected. The value of CWP for the case of the highest reaction rate (entry 4) was 0.02, much less than 0.3. The turnover frequencies (TOF) are also shown in Table 2. For example, the TOF of h-BN bm12h was 7.8 h 1 (entry 4). In the case of benzaldehyde as a reactant at 100 °C, the TOF was 2.4 h 1 (entry 5), which was roughly comparable to that of a primary amineimmobilized silica-alumina catalyst (2 h 1) and much lower than that of a primary and tertiary amines-immobilized silica-alumina catalyst (ca. 70 h 1), though the reactant concentrations were different [9]. The homogeneous counterpart boric acid, a Brønsted acid, showed no activity (entry 9), and n-butylamine, a Brønsted base, showed low selectivity toward b-nitrostyrene (entry 10). The main product with n-butylamine (73% selectivity) was N-benzylidene butylamine, an imine obtained by nucleophilic addition of the amine to the aldehyde group of benzaldehyde (Fig. S4). The coexistence of boric acid and n-buthylamine did not improve the activity (entry 11). This suggests that cooperative catalysis needs appropriate positioning of acid and base sites without neutralization. Conventional solid-base catalysts such as calcined MgO and calcined hydrotalcite were used for comparison and showed low conversion but high selectivity toward 4a, which is obtained via the formation of 3a followed by a Michael addition reaction with

physisorbed

h-BN bm12h 41 crystalline TMPO

90 SSB

h-BN 100

80

60

40

20

0

δ /ppm 31

Fig. 6. P MAS NMR spectra of trimethylphosphine oxide adsorbed onto pristine hBN and h-BN bm12h.

TMPO. For pristine h-BN, a sharp signal was observed at 41 ppm, along with a spinning side band at 90 ppm. The signal at 41 ppm is ascribed to crystallized TMPO [48], indicating that there was no interaction between pristine h-BN and TMPO because there were no acid sites on pristine h-BN. In contrast, for h-BN bm12h, two different signals were obtained at 63 and 43 ppm. The former can be attributed to TMPO adsorbed onto low-acid-strength sites of h-BN bm12h, which are comparable to those in HY zeolite [48]. The latter can be ascribed to TMPO physisorbed on the sample [49,50]. 3.3. Nitroaldol reaction The catalytic activity of the pristine h-BN and the ball-milled hBN catalysts was evaluated for the nitroaldol reaction using pmethoxybenzaldehyde (1a) and nitromethane (2a) as substrates (Table 2). Boric acid and an amine, which are homogenous counterparts, were also used for comparison. It is noteworthy that only

Table 2 Results of nitroaldol reaction using h-BN catalysts.a

a b c d e f g h

Entry

Catalyst

Base amount /mmol g

1 2 3 4d 5e 6 7 8 9 10 11

h-BN h-BN bm6h h-BN bm12h h-BN bm12h h-BN bm12h h-BN bm24h MgOf Hydrotalciteg H3BO3h n-Butylamineh H3BO3 + n-Butylamineh

0 0.21 0.37 0.37 0.37 0.29 0.31 2.03 – – –

1b

1a Conv. /%

9 25 84 90 51 45 13 16 6 9 9

Selec. /%

TOF /h

3a

4a

<1 79 77 79 72 76 38 31 0 27 8

0 <1 <1 4 <1 <1 62 69 0 0 0

Reaction conditions: 1a (0.5 mmol), 2a (1.25 mmol), catalyst (50 mg), toluene (2 mL), 100 °C, 8 h. Determined by titration of benzoic acid for bromothymol blue-adsorbed h-BN. Determined by CO2-TPD for MgO and hydrotalcite. Turnover frequency. Added water (0.11 mmol). Reaction conditions: benzaldehyde (0.5 mmol), 2a (0.5 mmol), catalyst (50 mg), toluene (2 mL), 100 °C, 12 h. Pretreated at 700 °C for 3 h. Pretreated at 500 °C for 3 h. 0.05 mmol.

– 1.7 6.2 7.8 2.4 2.6 0.6 0.1 – 0.1 0.1

1c

182

S. Torii et al. / Journal of Catalysis 355 (2017) 176–184

nitromethane, reactions favored by the strong basicity of the solids (entries 7 and 8) [9]. It is reported that the activity over MgO was not influenced by CO2 and H2O in the nitroaldol reaction because nitromethane is so acidic that it preferentially adsorbs on the catalyst when CO2 and H2O are preadsorbed on the catalyst [52]. The ball-milled h-BN could catalyze the nitroaldol reaction. Despite its weak basicity, the h-BN exhibited much higher activity than calcined MgO and hydrotalcite. The suggested reaction steps for the nitroaldol reaction using the ball-milled h-BN catalysts are shown in Scheme 2 [5,9]. The reaction steps involve formation of imine intermediates with surface NH2 groups. There are three steps in the suggested sequence. The first step is the abstraction of a proton from nitromethane by an –NH2 group of the ballmilled h-BN catalyst. The second step is formation of an imine by nucleophilic attack by an –NH2 group of the catalyst, but this time on the aldehyde group of the benzaldehyde. The coexisting –OH groups on the catalyst polarize the carbonyl group of the aldehyde to facilitate the formation of the imine intermediate [53,54]. Formation of an iminum ion creates a carbon center that is significantly more electrophilic than the carbonyl compound [55]. This facilitates attack by the carbon nucleophile (CH2NO2) in the reac-

tion. The first and second steps are expected to be interchanged on the catalyst. The final step is a nucleophilic attack by the deprotonated nitromethane. In this sequence, both imine formation [56] and imine protonation to form iminum ions [55] are promoted by the acid–base bifunctionality. Two different kinetic isotope effect (KIE) experiments were carried out using p-methoxybenzaldehyde with nitromethane (CH3NO2) as well as D-substituted nitromethane (CD3NO2) [31]. The first KIE experiment was conducted by measuring the two individual rate constants of the reaction using CH3NO2 and CD3NO2, kH and kD. Fig. S5 shows the results of the nitroaldol reaction using the two nitromethanes at 100 and 80 °C. The KIE values calculated from the relative ratio of rate constants (kH/kD) were 3.1 and 3.5 for measurements conducted at 100 and 80 °C, respectively. These were greater than 1, indicating that abstraction of a proton from nitromethane is the rate-determining step. The second KIE experiment was conducted by measuring intermolecular competition between CH3NO2 and CD3NO2 in the same reaction vessel. The KIE value was calculated from the relative amounts of products formed by CH3NO2 and CD3NO2 ([PH]/[PD]), as shown in Table S3. The KIE value was found to be almost constant at 3.4 for reaction times 0.5 to 1.5 h, which is in good agreement with the first KIE experiment. This demonstrates that abstraction of a proton from nitromethane is the rate-determining step for the reaction (Scheme 2, Step1). The reaction mechanism of the nitroaldol reaction over an active h-BN catalyst was further investigated by applying the Hammett equation to data obtained using various monosubstituted benzaldehyde substrates (R-C6H4CHO, R = p-OMe, p-CH3, H, p-Cl, and m-OMe). Fig. 7 shows the results of a Hammett plot for the nitroaldol reaction using h-BN bm12h at 100 °C. A negative value of slope against the Hammett constant [29], which is the substituent constant for aromatic substitution, was obtained. Electron-donating groups (EDG) in the para-position of benzaldehyde (p-OMe, p-CH3) enhanced reactivity with nitromethane, and the corresponding b-nitrostyrene was formed in excellent yields. On the other hand, electron-withdrawing groups (EWG) in the para-position of benzaldehyde (p-Cl) and electron-donating groups in the meta-position (m-OMe) decreased reactivity, and the corresponding b-nitrostyrene was obtained in low yields. These results indicate that electron-donating groups stimulated polarization of the carbonyl group of benzaldehyde, leading to enhanced interaction with the –OH group of the catalyst (Scheme 2 Step 2). These were different from those of aminosilica catalysts [57] and alkaline cation-exchanged zeolites [58]. While a negative value of slope, 1.8, was obtained for h-BN catalyst in this study, a slightly positive value of slope, 0.18, was observed for the aminosilica catalyst

0.6 p-OMe

Log (k/kH)

0.4

p-CH3

Slope = -1.8

0.2 H

0.0 -0.2 -0.4 -0.6 -0.3

m-OMe p-Cl

-0.2

-0.1

0.0

0.1

0.2

0.3

Hammett constant σ Scheme 2. Proposed overall mechanism of nitroaldol reaction on h-BN catalyst surface.

Fig. 7. Results of the Hammett plot for the nitroaldol reaction.

S. Torii et al. / Journal of Catalysis 355 (2017) 176–184 Table 3 Results of the Knoevenagel reaction with h-BN bm12h.a

Entry

Substrate 1b

Product 3b

3b Yield /%

1

92

1ba

3ba 100

2

1bb

38

1bc

3bc 100

4

1bd

3bd 70

1be 3be

6b

34

1bf

could respectively function as acid and base sites. The results of solid state NMR of adsorbed pyrrole and trimethylphosphine oxide indicated that the ball-milled h-BN had basic sites comparable to those of KY zeolite, as well as weak acid sites comparable to those in HY zeolite. The ball-milled h-BN showed high activity and selectivity for the nitroaldol reaction, whereas nontreated h-BN did not show activity. The abstraction of a proton from nitromethane was the rate-determining step, and the reaction mechanism involves formation of imine intermediates with surface amino groups. The coexistence of acid and base sites on the catalyst enhanced its activity. Further investigation of the catalyst may open up new possibilities for design of new acid–base pairs based on metal nitride compounds.

3bb

3

5

183

Acknowledgments The authors thank Professor S. Ted Oyama for valuable discussions. This work was supported by a Grant-in-Aid for Challenging Exploratory Research (No. 16 K14474) and Young Scientists (A) (No. 25709077) of JSPS, Japan. A part of this work was supported by the AIST Nanocharacterization Facility (ANCF) platform as a program of the Nanotechnology Platform of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.

3bf

a

Reaction conditions: 1b (0.5 mmol), 2b (0.5 mmol), h-BN bm12h (50 mg), toluene (2 mL), r.t., 90 min. Yields were determined by GC using n-decane as an internal standard. b 120 min.

[57]. Acceleration of the reaction rate by electron-donating groups indicates that acid sites interact with the carbonyl group of benzaldehyde and stabilize the intermediates. This demonstrates the importance of weak acid sites B–OH for promoting the nitroaldol reaction. 3.4. Knoevenagel reaction The Knoevenagel reaction was also conducted using different aldehydes (1b) and malononitrile (2b) as substrates at room temperature for 90 min (Table 3). Yields of the corresponding alkene 3b were high for benzaldehyde (1ba) (92%, entry 1) and p-methoxybenzaldehyde (1bb), 1ba with electron-donating group (EDG) at the para-position (100%, entry 2). In contrast, an electron-withdrawing group (EWG) at the para-position (pnitrobenzaldehyde (1bc)) lowered the electron density of the reacting carbonyl group and gave a low yield of p-nitrobenzylidenemalononitrile (3bc) (38%, entry 3). Reactions using nonaromatic aldehydes were also conducted. Highly reactive furfural (1bd) gave a 100% yield for the corresponding alkene (entry 4). In addition, the h-BN catalyst also functioned with the less reactive aliphatic aldehydes (butanal (1be) and pivalaldehye (1bf)), and corresponding alkenes (butylidenemalononitrile (3be) (70%, entry 5) and neopentylidenemalononitrile (3bf) (34%, entry 6)) were successfully produced at room temperature. 4. Conclusions The layered structure of a commercial hexagonal boron nitride (h-BN) of low surface area (3 m2 g 1) was successfully disrupted by planetary ball milling to produce a solid of high surface area (ca. 400 m2 g 1). The results of Fourier transform infrared spectroscopy, X-ray photoelectron spectroscopy, and solid state nuclear magnetic resonance confirmed the cleavage of B–N bonds by reaction with moisture in air, forming hydroxyl and amino groups that

Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.jcat.2017.09.013. References [1] G.C. Welch, D.W. Stephan, J. Am. Chem. Soc. 129 (2007) 1880. [2] M.A. Dureen, D.W. Stephan, J. Am. Chem. Soc. 131 (2009) 8396. [3] C.M. Mömming, E. Otten, G. Kehr, R. Fröhlich, S. Grimme, D.W. Stephan, G. Erker, Angew. Chem. Int. Ed. 48 (2009) 6643. [4] B. List, R.A. Lerner, C.F. Barbas III, J. Am. Chem. Soc. 122 (2000) 2395. [5] J.D. Bass, A. Solovyov, A.J. Pascall, A. Katz, J. Am. Chem. Soc. 128 (2006) 3737. [6] R.K. Zeidan, M.E. Davis, J. Catal. 247 (2007) 379. [7] K.K. Sharma, T. Asefa, Angew. Chem. Int. Ed. 46 (2007) 2879. [8] N.A. Brunelli, C.W. Jones, J. Catal. 308 (2013) 60. [9] K. Motokura, M. Tada, Y. Iwasawa, Angew. Chem. Int. Ed. 47 (2008) 9230. [10] J. Lauwaert, E. de Canck, D. Esquivel, J.W. Thybaut, P. van der Voort, G.B. Marin, ChemCatChem 6 (2014) 255. [11] N.A. Brunelli, S.A. Dias, K. Venkatasubbaiah, C.W. Jones, J. Am. Chem. Soc. 134 (2012) 13950. [12] A.J. van Roosmalen, J.C. Mol, J. Phys. Chem. 82 (1978) 2748. [13] X.S. Zhao, G.Q. Li, A.K. Whittaker, G.J. Millar, H.Y. Zhu, J. Phys. Chem. B 101 (1997) 6525. [14] Y. Inaki, Y. Kajita, H. Yoshida, K. Ito, T. Hattori, Chem. Commun. (2001) 2358. [15] B. Hamzaoui, A. Bendjeriou-Sedjerari, E. Pump, E. Abou-Hamad, R. Shougrat, A. Gurinov, K.-W. Huang, D. Gajan, A. Lesage, L. Emsley, J.-M. Basset, Chem. Sci. 7 (2016) 6099. [16] A.K. Geim, I.V. Grigorieva, Nature 499 (2014) 419. [17] F. Xia, H. Wang, D. Xiao, M. Dubey, A. Ramasubramaniam, Nat. Photon. 8 (2014) 899. [18] Z. Liu, Y. Gong, W. Zhou, L. Ma, J. Yu, J.C. Idrobo, J. Jung, A.J. MacDonald, R. Vajtai, J. Lou, P.M. Ajayan, Nat. Comm. 4 (2013) 2541. [19] W. Lei, D. Portehault, D. Liu, S. Qin, Y. Chen, Nat. Comm. 4 (2013) 1777. [20] Q. Weng, X. Wang, C. Zhi, Y. Bando, D. Golberg, ACS Nano 7 (2013) 1558. [21] J.T. Grant, C.A. Carrero, F. Goeltl, J. Venegas, P. Mueller, S.P. Burt, S.E. Specht, W. P. McDermott, A. Chieregato, I. Hermans, Science 354 (2017) 1570. [22] J.M. Venegas, J.T. Grant, W.P. McDermott, S.P. Burt, J. Micka, C.A. Carrero, I. Hermans, ChemCatChem 9 (2017) 2118. [23] D.J. Nash, D.T. Restrepo, N.S. Parra, K.E. Giesler, R.A. Penabade, M. Aminpour, D. Le, Z. Li, O.K. Farha, J.K. Harper, T.S. Rahman, R.G. Blair, ACS Omega 1 (2016) 1343. [24] Y. Yao, Lin, Z. Li, X. Song, K.-S. Moon, C.-P. Wong, J. Mater. Chem. 22 (2012) 13494. [25] Y. Li, T.V. Williams, W. Cao, H.E. Elsayed-Ali, J.W. Connell, J. Phys. Chem. C 114 (2010) 17434. [26] L.H. Li, Y. Chen, G. Behan, H. Zhang, M. Petravicc, A.M. Glushenkov, J. Mater. Chem. 21 (2011) 11862. [27] Y. Wang, X.W. Han, A. Ji, L.Y. Shi, S. Hayashi, Micropor. Mesopor. Mater. 77 (2005) 139. [28] S. Hayashi, Anal. Sci. 25 (2009) 133. [29] L.P. Hammett, J. Am. Chem. Soc. 59 (1937) 96.

184

S. Torii et al. / Journal of Catalysis 355 (2017) 176–184

[30] C. Hansch, A. Leo, R.W. Taft, Chem. Rev. 91 (1991) 165. [31] E.M. Simmons, J.F. Hartwig, Angew. Chem. Int. Ed. 51 (2012) 3066. [32] P.J. Sánchez-Soto, M. Del Carmen, Jiménez de Haro, L.A. Pérez-Maqueda, I. Varona, J.L. Pérez-Rodríguez, J. Am. Ceram. Soc. 83 (2000) 1649. [33] R. Geick, C.H. Perry, G. Rupprecht, Phys. Rev. 146 (1966) 543. [34] M.I. Baraton, T. Merle, P. Quintard, V. Lorenzelli, Langmuir 9 (1993) 1486. [35] T. Sainsbury, A. Satti, P. May, Z. Wang, I. McGovern, Y.K. Gun’ko, J. Coleman J. Am. Chem. Soc. 134 (2012) 18758. [36] G. Huang, C. Chen, X. Ye, W. Ye, J. Hu, C. Xu, X. Qiu, J. Mater. Chem. A 1 (2013) 12192. [37] W. Lei, D. Portehault, R. Dimova, M. Antonietti, J. Am. Chem. Soc. 133 (2011) 7121. [38] L. Ci, L. Song, C. Jin, D. Jariwala, D. Wu, Y. Li, A. Srivastava, Z.F. Wang, K. Storr, L. Balicas, F. Liu, P.M. Ajayan, Nat. Mater. 9 (2010) 430. [39] C.Y. Zhi, Y. Bando, T. Terao, C.C. Tang, H. Kuwahara, D. Golberg, Chem. Asia J. 4 (2009) 1536. [40] J.P.L. Perez, B.W. McMahon, J. Yu, S. Schneider, J.A. Boatz, T.W. Hawkins, P.D. McCrary, L.A. Flores, R.D. Rogers, S.L. Anderson, A.C.S. Appl, Mater. Interfaces 6 (2014) 8513. [41] J. Wu, Q. Li, C. Wang, B. Lv, L. Wang, J. Chen, Y. Xu, RSC Adv. 6 (2016) 38356. [42] K. Higuchi, M. Onaka, Y. Izumi, Bull. Chem. Soc. Jpn. 66 (1993) 2016. [43] U. Meier, W.F. Hoelderich, J. Mol. Catal. A 142 (1998) 213.

[44] J.A. Dean (Ed.), Lange’s Handbook of Chemistry, 14th ed., McGraw-Hill Inc, New York, 1992, p. p799. [45] M. Sánchez-Sánchez, T. Blasco, Chem. Commun. (2000) 491. [46] A. Zheng, S. Li, S.B. Liu, F. Deng, Acc. Chem. Res. 49 (2016) 655. [47] A. Zheng, S.J. Huang, S.B. Liu, F. Deng, Phys. Chem. Chem. Phys. 13 (2011) 14889. [48] E.F. Rakiewicz, A.W. Peters, R.F. Wormsbecher, J. Phys. Chem. B 102 (1998) 2890. [49] L. Baltusis, J.S. Frye, G.E. Maciel, J. Am. Chem. Soc. 108 (1986) 7119. [50] K.J. Sutovich, A.W. Peters, E.F. Rakiewicz, R.F. Wormsbecher, S.M. Mattingly, K. T. Mueller, J. Catal. 183 (1999) 155. [51] P.B. Weisz, C.D. Prater, Adv. Catal. 6 (1954) 143. [52] K. Akutu, H. Kabashima, T. Seki, H. Hattori, Appl. Catal. A 247 (2003) 65. [53] G. Sartori, F. Bigi, R. Maggi, R. Sartorio, D.J. Macquarrie, M. Lenarda, L. Storaro, S. Coluccia, G. Martra, J. Catal. 222 (2004) 410. [54] S.L. Hruby, B.H. Shanks, J. Catal. 263 (2009) 181. [55] T.I. Crowell, D.W. Peck, J. Am. Chem. Soc. 75 (1953) 1075. [56] J. Hine, M.S. Cholod, W.K. Chess, J. Am. Chem. Soc. 95 (1973) 4270. [57] V.E. Collier, N.C. Ellebracht, G.I. Lindy, E.G. Moschetta, C.W. Jones, ACS Catal. 6 (2016) 460. [58] R. Devi, R. Borah, R.C. Deka, Appl. Catal. A 433–434 (2012) 122.