Quantitative evaluation of ball-milling effects on the hydrolysis of cellulose catalysed by activated carbon

Synthesis of glucose from cellulose is a critical roadblock for establishing a new sustainable cycle of biorefinery to produce bio-based and environmentally-benign chemicals. We have already demonstrated that a pre-treatment, ball-milling solid cellulose and solid catalyst together (mix-milling), drastically improves the yield of glucose and oligosaccharides; however, the effect of this type of ball-milling has not been quantitatively evaluated. In this study, we performed several model reactions and found that mix-milling method drastically enhanced solid-solid reactions such as hydrolysis of insoluble cellulose to soluble oligomers on the solid catalyst, but did not do liquid-solid reactions. The kinetic study indicated that the rate constant of hydrolysis of cellulose to oligomers using mix-milling increased 13-fold higher than that using individual milling. Owing to the fast depolymerisation of cellulose, we achieved 72% yield of glucose with 97% conversion of cellulose and 74% selectivity at 418 K.

(Scheme 1) should be the mainstream in the next-generation biorefinery, which substitutes for the current processes using food biomass; however, realising this vision has been hampered by the recalcitrance of cellulose.
Scheme 1 Hydrolysis of cellulose to glucose.
The hydrolysis of cellulose has been performed with various solid catalysts, e.g.
It is surprising that weakly-acidic carbons (pK a > 3) can hydrolyse cellulose, because usual acids with a pK a larger than 3 are ineffective for this reaction. 10 This unexpected ability of carbons may be ascribed to the good affinity between carbon and cellulose demonstrated in model adsorption experiments using cello-oligosaccharides. 11 It is reported that surface-immobilised β-glucans on silica and alumina are depolymerised regardless of their weak acidity (pK a ~ 7), whereas these weak acids do not work for the hydrolysis of free cellulose (that is, non-immobilised β-glucans). 12 These results indicate that even weak acids can function by making a good contact between the substrate and the catalyst. To increase the contact between solid catalyst and solid cellulose, cellulose and a weakly-acidic carbon catalyst (K26) were ball-milled together, denoted mix-milling. 6a The hydrolysis reaction of this mix-milled sample provided 90% yield of glucose and oligosaccharides in total at 453 K in water, while that of individually ball-milled cellulose and K26 gave only 13% yield under the same conditions. This enhancement was not due to the mechanocatalytic hydrolysis 13 during the ball-milling process, and thus the improvement of solid-solid contact was proposed.
Limited collision between solid catalyst and solid substrate is a common issue for this type of reactions; however, these results suggest that this drawback can be overcome in some cases.
Understanding how the mix-milling changes the reactions would be useful to design a more efficient system. Herein, the purpose of this study is the quantitative assessment of mix-milling effect on the hydrolysis by means of kinetics and model experiments.
Another remaining subject in the previous report is that the use of trace HCl is necessary for the high-yielding synthesis of glucose. 6a The yield of glucose was at most 57% (58% selectivity) in the reaction in pure aqueous medium under the severe conditions at 503 K with an autogenous pressure of 2.8 MPa. Accordingly, we also aimed for the selective synthesis of glucose, instead of a mixture of glucose and oligosaccharides, under mild conditions in pure water.

Reagents
Microcrystalline cellulose (Avicel, 102331) was purchased from Merck and distilled water

Milling of cellulose and cellobiose
Microcrystalline cellulose (10 g) was ball-milled with alumina balls (1.5 cm, 2 kg) in a ceramic pot (3.6 L) at 60 rpm for 48 h. Mix-milling of cellulose and solid catalysts was carried out in the same type of pots in the presence of alumina balls. Microcrystalline cellulose (10 g) and solid catalysts (1.54 g) [substrate/catalyst (S/C) ratio based on weight = 6.5] were added into the pots and were milled together at 60 rpm for 48 h. The amount of catalyst was reduced to 1.46 g for mix-milling of cellobiose (S/C = 6.8). The ball-milled samples were analysed by powder X-ray diffraction (XRD; Rigaku MiniFlex, Cu K ).

Catalytic reactions
Hydrolysis of cellulose was conducted in a hastelloy C22 high-pressure reactor (OM Conversion of cellulose was determined based on the weight difference of the solid part before and after reaction. The amount of organic carbons in reaction solution was quantified by total organic carbon (TOC; Shimadzu TOC-V CSN ) measurement for the determination of conversion when the catalyst was partially dissolved into water after the reaction (see section 3.1).
Hydrolysis of mix-milled samples at a lower temperature (≤ 423 K) was carried out in a pressure-resistant glass tube (15 mL, Ace Glass). Mix-milled sample 94 mg [containing cellulose (81 mg) and catalyst (13 mg)] and distilled water 10 mL were charged into the tube.
The tube was immersed in an oil bath at a certain temperature for a designated length of time.
The analysis of products was performed by the same procedure described above.
Hydrolysis of cellobiose was conducted in the hastelloy C22 high-pressure reactor.
Cellobiose 342 mg, catalyst 50 mg and distilled water 40 mL were charged into the reactor.
The temperature was raised to 463 K for 11 min and then the reactor was rapidly cooled down to 323 K by blowing air. Yield of product and conversion were determined by HPLC.

Hydrolysis of mix-milled cellulose
The activities and properties of various solid catalysts were studied in the hydrolysis of cellulose with using the mix-milling pre-treatment. As a control, the hydrolysis of ball-milled cellulose without catalyst gives a poor reaction result (7.9% yield of glucans, Table 1, entry 1), and all the catalysts tested in this study provide low yields of products without mix-milling (Table S1). Thus, hydrolysis of cellulose was performed after ball-milling together with various solid catalysts (Table 1, entries 2-10). K26 produced water-soluble glucans in 90% yield [glucose (20%) and oligosaccharides (70%)] with 97% selectivity (entry 2) as reported previously. 6a The other products were fructose (0.6%), mannose (0.7%), levoglucosan (0.7%), 5-hydroxymethylfurfural (1.0%) and unidentified compounds (0.1%). The solid containing the catalyst K26 was easily separated by filtration after the reaction (Fig. S1). Other carbons, BA and SX, produced glucans in 34% and 22% yields, respectively (entries 3 and 4), and they were less active than K26. Amberlyst gave the highest glucose yield (entry 5, 82%); however, the mechanocatalytic hydrolysis occurred in this case owing to strong acidity of this resin.
After ball-milling cellulose and Amberlyst together, >99% of cellulose was dissolved as various kinds of oligomers 13c in water at room temperature. Moreover, Amberlyst was completely degraded after the hydrolysis reaction; Amberlyst was not recovered by centrifugation at 4600g or filtration using a membrane (0.1 μm mesh) and the colour of the filtrate was brown (Fig. S2). Since the Tyndall effect was hardly observed, Amberlyst presumably dissolved in water during the reaction, which was contrastive to the behaviour of K26 described above (Fig. S1). The other catalysts tested (H-ZSM-5, H-MOR, SiO 2 -Al 2 O 3 , SiO 2 and TiO 2 ) were almost inactive (entries 6-10), and these catalysts except TiO 2 partially dissolved during the reaction (Fig. S3).  The correlation between crystallinity of cellulose and the result of respective catalytic reaction was estimated because the decomposition of crystalline structure of cellulose by milling treatment improves the reactivity of cellulose. 14 XRD measurements represented no peak of crystalline cellulose for all the milled samples ( Fig. 1), showing that cellulose was in the form of amorphous in these samples. Thus, the difference in the catalytic performance is not ascribed to the nature of cellulose but to the hydrolytic activities of catalysts and the contact between catalysts and cellulose. We conclude that K26 is the best solid catalyst for the hydrolysis of cellulose under the reaction conditions employed.  Table 1, entry 11) than that in the previous report 6a (503 K, 2.8 MPa, 57% yield, 58% selectivity), and the lower temperature suppressed the decomposition of glucose. The high glucose yield under the mild reaction conditions indicates the potential applicability of this catalytic process in the hydrolysis of cellulose.

Effect of mix-milling pre-treatment
The roles of mix-milling were evaluated in model reactions using soluble catalysts or substrates. If the major role of mix-milling pre-treatment is the improvement of solid-solid contact, the promotional effect of mix-milling will disappear since the use of soluble substrate or catalyst gives a liquid-solid reaction. Therefore, we carried out the mix-milling and hydrolysis for two types of substrate-catalyst combinations: (i) cellobiose (water-soluble substrate) and K26 (insoluble catalyst) and (ii) cellulose (insoluble substrate) and benzoic acid (soluble catalyst). We chose benzoic acid as a soluble catalyst because this compound is a typical model for the active sites 15 of weakly-acidic carbons. Note that a typical soluble catalyst, H 2 SO 4 , is not suitable for this solid-solid mixing because H 2 SO 4 itself is liquid and furthermore this strong acid depolymerises cellulose during the mix-milling treatment 13b,c . For the combination (i), the hydrolysis of cellobiose by K26 was performed with or without the mix-milling pre-treatment. As expected, the reactions provided almost the same yield of glucose (Table 2, entries 12 and 13). For the both cases, cellobiose of ca. 90% dissolved into water at room temperature, and the remaining part (ca. 10%) adsorbed onto K26. 11c Therefore, the contacts between K26 and cellobiose in the both cases were the same during the hydrolysis reaction. Likewise, the combination (ii) indicated no positive effect of mix-milling, in which almost no difference in product yields was observed with or without mix-milling (entries 14 and 15). Benzoic acid completely dissolved into water at the reaction temperature.
In contrast, the mix-milling pre-treatment drastically enhanced the hydrolysis with the combination of cellulose and K26; the yield of glucans was increased seven times from 13% to 90% (entries 2 and 16). Mix-milling pre-treatment accelerates solid-solid reactions but does not liquid-solid ones. These results indicate that the predominant role of the mix-milling is improvement of the solid-solid contact. Kinetic study of the hydrolysis of cellulose was conducted to quantitatively estimate the effect of mix-milling, in which 418 K was chosen as the optimised temperature to synthesise glucose (see section 3.1) and accurately estimate the kinetic parameters under the steady state. The amount of cellulose (black circles) decreased, whereas that of oligosaccharides (blue squares) contrastively increased in the initial period. The yield of oligomers was maximised at 6 h (44%) and then gradually decreased as oligomers were intermediates in this reaction. With regard to glucose, a small amount of glucose (red diamonds) was simultaneously produced with oligosaccharides from cellulose, but glucose formed after accumulating oligosaccharides.
The yield of glucose reached 72% at 24 h with 97% conversion of cellulose and 74% selectivity, as noted in section 3.1. Yield of glucose started to decrease after 24 h due to the decomposition as by-products (green triangles) increased. Accordingly, the hydrolysis of cellulose consists of three steps as shown in eqn 1. 16 products -By Glucose Oligomers Cellulose where k 1-3 are rate constants. We hypothesised that all steps were first-order reactions as reported elsewhere. 14,17 The reaction rate for each step was represented as eqns (2) (9) where [Cellulose] 0 is the initial concentration of cellulose and k 1 k 2 k 3 .
Four lines in Fig. 2 represent the curve fitting with using eqns (6) (9), which reproduced the experimental data. The determined rate constants were k 1 = 0.17 h -1 , k 2 = 0.16 h -1 and k 3 = 0.017 h -1 . The rate constant of hydrolysis of cellulose to soluble oligosaccharides was as high as that of oligosaccharides to glucose (k 1 /k 2 = 1.1). It is surprising that k 1 /k 2 exceeds 1 as the rate-determining step of the hydrolysis of cellulose to glucose is generally the first step, 16 indicating that the conversion of solid cellulose to soluble oligomers was selectively accelerated by the mix-milling. In addition, high k 1 /k 3 (10) and k 2 /k 3 (9.4) ratios provided the good yield of glucose because the decomposition of glucose was limited.
We also performed the hydrolysis of individually-milled cellulose by K26 at 418 K instead of the mix-milled sample. Note that the reactivity itself of individually-milled cellulose should be similar to that of mix-milled cellulose, as their polymerisation degrees determined by viscometry 18 (640 690), median particle diameters (12 13 m) and crystallinity indexes (< 5%) are almost the same. 6a Since the hydrolysis reaction was slow in this case, we analysed products by a sampling method to improve the accuracy. Formation of oligomers, subsequent production of glucose and successive decomposition of glucose were observed in this reaction. Therefore, the same curve fitting was applied to this reaction, which provided rate constants of k 1 = 0.013, k 2 = 0.16 and k 3 = 0.017 h -1 . k 1 was greatly decreased by changing the pre-treatment from mix-milling to individual milling, while the values of k 2 and k 3 were similar regardless of the pre-treatment method. As a result, the ratio of k 1 /k 2 was as small as 0.081, showing that the hydrolysis of cellulose to soluble oligomers was the rate-determining step. Consequently, we demonstrated that the mix-milling selectively and drastically increase k 1 (13-fold) in the solid-solid reaction.
A typical soluble acid H 2 SO 4 was also tested as a catalyst for the hydrolysis of 13 individually-milled cellulose in order to compare solid-solid and solid-liquid reactions (Fig. 3).
The concentration of H 2 SO 4 was 50 mM (0.49%) as a usual value for the diluted H 2 SO 4 processes. 2a,19 The rate constants obtained by a curve fitting were k 1 = 0.5 h -1 , k 2 = 17 h -1 and k 3 = 0.12 h -1 . The first step is slow (k 1 /k 2 = 0.029), indicating that hydrolysis of cellulose to soluble oligosaccharides is the rate-determining step for the production of glucose. This result is reasonable as the hydrolysis of cellulose is significantly more difficult than that of soluble oligomers. 16 Therefore, the high ratio of k 1 /k 2 for the hydrolysis using K26 and mix-milling is specific to this reaction, and hence we propose that the solid-solid contact created by mix-milling can selectively accelerate the solid reaction.

Conclusions
We evaluated the effect of mix-milling by three types of combinations: (i) insoluble cellulose and insoluble K26, (ii) soluble cellobiose and insoluble K26 and (iii) insoluble cellulose and soluble benzoic acid. Mix-milling enhanced the hydrolysis of glycosidic bonds only when the combination (i) was employed. These results indicated that solid-solid reactions were accelerated by creating better contact between solid cellulose and solid catalyst. The rate constant of hydrolysis of mix-milled cellulose to oligomers increased 13 times higher than that of individually-milled one. This high rate constant contributed to the high-yielding synthesis of glucose from cellulose (72% yield, 97% conversion and 74% selectivity) under mild conditions. We believe that the mix-milling is a promising technique for accelerating the reactions that occur at the solid-solid interface.

Electronic Supplementary Information
Quantitative evaluation of ball-milling effect on hydrolysis of cellulose catalysed by activated carbons Mizuho Yabushita, a,b