1 Introduction

Conversion of biomass into renewable green energy is seen as an alternative substitute to the depleting fossil fuel resources. Cellulose and hemicellulose consist of carbon, hydrogen, and oxygen atoms in the form of linear polymeric saccharides chains but with varying homogeneity [1]. Cellulose is synthesized by living organisms, particularly plants, as the main substance in the cell wall [2]. As the most abundant organic compound, conversion of cellulose to value-added commodities has been investigated, for example as an additive in food and pharmaceutical industries [3]. Conversion of cellulose into glucose using methods such as hydrothermal treatment and hydrolysis also showed promising routes for conversion of cellulose to value-added commodities [4, 5]. Hydrothermal degradation of cellulose using subcritical water at 200–300 °C produced sugar monomers, in particular glucose and fructose [6]. Cellulose has also been investigated as alternative biomass for sustainable generation of H2 gas via high-temperature pyrolysis [7]. Pyrolysis required high-temperature decomposition of cellulose that produced H2 and various gasses such as CO2, CH4, and C2-C4 hydrocarbons [8]. High-temperature decomposition of cellulose also further converted cellulose to volatile organic compounds and char [9]. Conversion of cellulose to hydrogen at low temperatures can be carried out using methods such as dark fermentation [10, 11], photocatalysis [12, 13], and electrolysis [14, 15]. In all these methods, decomposition of cellulose occurred through random dissociation of β(1→4)-glycosidic bonds with 172 ± 2 kJ/mol of activation energy [16]. Dark fermentation has been extensively investigated as environmental friendly route for biohydrogen production from biomass waste [17]. The efficiency of H2 production depends on temperature, pH, and the ability of microorganism to break down carbohydrate into hydrogen [18]. However, dark fermentation often requires a long fermentation time to produce high yield of H2 [18, 19]. Biohydrogen production from electrolysis of lignocellulosic biomass exhibited high efficiency when using cellulose in comparison to holocellulose and lignin [14]; however, this method used-Pt based electrode and requires high energy consumption [20].

Photocatalysis provides a sustainable route for H2 gas generation at ambient conditions while taking the full advantage of energy from sunlight. Quantum efficiency of photocatalytic hydrogen production from methanol reforming while using Pd/TiO2 catalysts was higher in UV-B region at ~ 35% efficiency [21]. Photocatalytic decomposition of lignocellulosic biomass is another feasible and practical route with much higher efficiency in comparison to direct photocatalytic water splitting, with potential to store up to 12% of the light energy [22]. Cellulose becomes a sacrificial agent during photocatalytic water spitting, which degrades to produce CO2, thus allowing photoreduction of water for H2 production [12]. Studies on simultaneous cellulose conversion and hydrogen production using TiO2 photocatalysts suggest the feasibility of the process to utilize readily available biomass materials for isolating H2 from water [23]. Apart from acid hydrolysis, the ball milling method was also used to initiate cellulose decomposition into its glucose unit, which enhanced the photocatalytic decomposition of cellulose [24]. Formic acid and glucose, identified as intermediates [23], also decomposed in the photoreforming reaction to form CO2 and H2 gasses [25] on a variety of photocatalysts such as Pt/TiO2 [26], Pd/TiO2 [27], and other noble metals [28].

This research aimed to investigate sustainable H2 gas production from cellulose derived from biomass waste via photocatalysis. Cellulose isolated from coconut husk, fern fiber, and cotton linter with different crystallinity and degree of polymerization will be used to elucidate the effects of cellulose properties towards H2 production. H2 production will also be correlated with the composition of hemicellulose and α-cellulose. Photodecomposition of cellulose was carried out using Pd-, Cu-, Ni-, and Ce-loaded TiO2 photocatalysts. Finally, aqueous solutions containing glucose and fructose obtained from hydrothermal treatment of cellulose were also investigated for H2 production in order to provide insights into the mechanism of the reaction.

2 Experimental details

2.1 Materials

Commercial TiO2 used in this study was supplied by Sigma-Aldrich. Metal precursors such as copper (II) nitrate hydrate (Cu(NO3)2·H2O, EMSURE), potassium tetrachloropalladate (K2PdCl4, Sigma-Aldrich), cerium nitrate hexahydrate (Ce(NO3)3·6H2O, EMSURE), and nickel nitrate hexahydrate (Ni(NO3)2·6H2O, EMSURE) were used to prepare metal/TiO2 catalysts. Cotton linter cellulose (medium fibers, 99.98%) was supplied by Sigma-Aldrich.

2.2 Preparation of catalysts

Incipient wetness impregnation method was used to prepare metal/TiO2 photocatalysts based on previously reported work that indicated high photocatalytic activity was achieved compared to the colloidal method [29]. The required amount of metal precursor at 0.3% of metal weight was dissolved in a small amount of water before being added to TiO2. The mixture was further ground until it formed a paste. Afterwards, the paste was dried at 120 °C for 2 h followed by calcination at 550 °C for 3 h.

2.3 Characterization studies

X-ray diffraction (XRD, Shimadzu) was used to determine the phase composition and analyze crystal structure of the samples. The patterns were recorded in the 2θ within range of 10 to 60° using a radiation source of λ = 1.54 Å at 40 kV and 30 mA. The crystallinity index (CI) of cellulose was calculated using equation below [26, 30]:

$$ \mathrm{CI}\ \left(\%\right)=\frac{{\mathrm{I}}_{200}\times {\mathrm{I}}_{\mathrm{AM}}}{{\mathrm{I}}_{200}}\times 100 $$
(1)

where I002 represents the maximum intensity of the (200) plane diffraction peak at 2θ = 23.1° and IAM represents the minimum intensity between (110) and (200) diffraction peak at 2θ = 19.0° [26, 30, 31]. The morphology of the samples was studied using JEOL scanning electron microscopy (SEM) equipped with energy dispersive X-ray (EDX). Surface area, pore volume, and pore size measurements were carried out by N2 adsorption-desorption at 77 K (Micromeritics ASAP 2020). Fourier-transform infrared spectroscopy (FTIR, CARY 630) was employed to determine the presence of functional group in the cellulose isolated from coconut husk, fern fiber, and cotton linter. All FTIR spectra were measured at 650–4000 cm−1 in transmittance mode. UV-Vis spectroscopy (CARY 3000) was used to study the optical absorption properties of the catalysts. The spectra were recorded at room temperature using wavelength from 200 to 800 nm.

2.4 Cellulose isolation from coconut husk and fern fiber

Holocellulose and alpha-cellulose were isolated from two raw biomass wastes, namely Malayan tall coconut (Cocos nucifera L.) husks and fern fiber (Dicranopteris linearis), using the “wood industry” method. The “wood industry” method is a standard method provided by the Technical Association of the Pulp and Paper Industry (TAPPI) that allowed efficient removal of lignin for extraction of cellulose [32,33,34]. The method is also known as acid-chlorite treatment. Raw biomass materials were collected and sun-dried for 2 weeks. Dried materials were cut to 2–3 cm, ground using a kitchen blender, and sieved using 212-μm sieve. The raw materials were then treated using Soxhlet extraction in acetone for 16 h. For holocellulose (a mixture of α-cellulose and hemicellulose) extraction using sodium chlorite, 4 g of extractive-free residue was treated with 160 cm3 0.2 M sodium acetate solution and heated to 75 °C for 5 h, followed by the addition of 4 cm3 of 20% (w/v) sodium chlorite. The mixture was then cooled, filtered, and washed with distilled water and acetone. The residue was then oven-dried at 105 °C. The dried residue (holocellulose) was further treated with sodium hydroxide to remove the remaining hemicellulose content. To obtain cellulose, 1 g of the extracted holocellulose was added to 20 cm3 17.5% (w/v) sodium hydroxide solution at room temperature for 30 min. The mixture was washed and filtered twice with 40 ml of distilled water followed by 3 ml of 10% (w/v) acetic acid solution, and finally with 100 ml of hot water. The residue (cellulose) was then oven-dried at 105 °C.

2.5 Determination of degree of polymerization of cellulose

The viscosities of the cellulose samples were measured by using 0.5 M cupriethethylenediamine (CUEN) according to TAPPI method in a Cannon Fenske capillary viscometer. The viscosity average DP of the cellulose samples was calculated from the intrinsic viscosity [η] according to Eq. 2 [35]:

$$ {\mathrm{DP}}^{0.905}=0.75\left[\upeta \right] $$
(2)

2.6 Hydrothermal decomposition of cellulose

The conversion of cellulose to glucose and fructose was carried out using cellulose cotton linter and water mixture using hydrothermal reaction at 10 MPa and 250 and 270 °C for 15 min. At the end of 15 min, the reactor was cooled down immediately to terminate all reactions. The water-soluble products were filtered off from the cellulose residues using 0.45 μm membranes and analyzed using gas chromatography–mass spectrometry (GC-MS).

2.7 Photocatalytic H2 production

Photocatalytic H2 production was carried out in a 60-ml glass bottle equipped with a rubber septum for gas sampling. Fifty milligrams of photocatalyst and 0.1 g of cellulose cotton linter were suspended in 30 ml of distilled water and placed in an irradiation chamber BS-02 in which the interior was surrounded by mirrors, employing three MH-lamps of 150 W for UVA-visible light simulation. The suspension was stirred for 10 min to ensure homogeneous catalyst dispersion. Prior to light irradiation, the reaction mixture was purged with nitrogen gas for the removal of air and degassing of the solution. Afterwards, the light source was switched on and the mixture was stirred continuously for 3 h under light irradiation. The gas sample (3 ml) was collected at 30-min interval from the reactor and analyzed using gas chromatography (GC-2014, Shimadzu), where argon was used as carrier gas. The procedure was repeated three times to get the average yield of H2 and the standard deviation. The photocatalytic reaction outlined above was also carried out on an aqueous sugar solution derived from hydrothermal decomposition of cellulose cotton linter for comparison.

3 Results and discussion

3.1 Characterization of the catalysts

Figure 1 shows the XRD patterns of TiO2 and metal deposited onto TiO2. All catalysts exhibited high crystalline anatase structure peaks at 25.4°, 37.1°, 37.9°, 38.6°, 48.1°, 54.4°, and 55.2°. Furthermore, the patterns also showed comparatively small peaks ascribed to rutile structure [36, 37]. The peaks corresponding to the metals—Pd, Cu, Ce, and Ni—were not observed from XRD due to the formation of highly dispersed metal nanoparticles on TiO2 support [38]. The structure of TiO2 did not show any significant changes following impregnation with metal co-catalysts.

Fig. 1
figure 1

XRD patterns of commercial TiO2, 0.3% Pd/TiO2, 0.3% Cu/TiO2, 0.3% Ce/TiO2, and 0.3% Ni/TiO2

Full size image

Diffuse reflectance UV-Vis spectra of TiO2 in Fig. 2 showed the absorbance band edges of TiO2 were observed at 350 to 400 nm, corresponding to the band gap at approximately ~ 3.22 eV. Following impregnation with metal nanoparticles, the band gap of the catalyst was slightly reduced to 3.0 and 3.1 eV. Cu/TiO2 and Pd/TiO2 also showed additional absorption bands in visible region. Cu/TiO2 exhibited a broad absorption peak at 600–800 nm whereas Pd/TiO2 only showed a discrete absorption peak around 400–550 near UV region. The observation was correlated with the ability of the electron on the surface of metallic nanoparticles to oscillate during excitation with incident radiation, which is also known as localized surface plasmon resonance (LSPR) [39]. However, the plasmonic resonance was not observed on Ni and Ce. N2 analysis of the catalysts following deposition of metal onto TiO2 indicated the surface area of TiO2 was reduced from 47 m2/g to 41 m2/g after impregnation with Pd. Similar observation was also found when Ni was added to give 42 m2/g. Impregnation with Cu further reduced the surface area to 38 m2/g. On the other hand, Ce/TiO2 only showed slight reduction to give 45 m2/g of surface area. It is interesting to note that at the same loading content, all metal/TiO2 combinations showed different surface areas, presumably due to the differences in metal dispersion on TiO2.

Fig. 2
figure 2

(a) Diffuse reflectance UV-Vis absorption spectra of the photocatalysts. (b) Kubelka-Munk plot for the estimation of the band gap energies of the photocatalysts

Full size image

Figure 3(a) displays the surface morphology of Pd/TiO2 catalysts analyzed using SEM. Spherical shape agglomeration with relatively uniform sizes was analyzed on Pd/TiO2 catalyst. Similar aggregations of spherical nanoparticles were observed for TiO2 and Cu/TiO2 catalysts (Fig. S1 and Fig. S2). EDX analysis also showed the elemental mapping images of investigated components in the samples, as shown in Fig. 3(b). Analysis of each element showed even and uniform distributions of Pd, Ti, and O across the TiO2 surface. This is further evidence of the presence of metal co-catalysts on TiO2 and confirmed that the incipient-wetness impregnation method provides a good dispersion of metal throughout the TiO2 surface.

Fig. 3
figure 3

(a) SEM images Pd/TiO2. (b) Elemental mapping of /TiO2 photocatalysts

Full size image

3.2 Characterization of cellulose

3.2.1 XRD analysis of cellulose

Cellulose samples from coconut husk, fern, and cotton linter were characterized using XRD analysis (Fig. 4). XRD patterns of cellulose cotton linter showed diffraction peaks at 2θ = 15.5°, 17.1°, 23.1°, and 34.6° which represented the typical cellulose I type structure. For cellulose obtained from coconut husk, a broad peak centered at 2θ = 22.2° corresponded to type III of α-cellulose [40]. A peak at 27° also indicated the presence of impurities. Similar observations were also obtained from cellulose isolated from fern fiber. XRD analysis of holocellulose isolated from coconut fiber showed the broad diffraction peak centered at 22° with a small hump appearing at 16°. A similar pattern was observed on holocellulose from fern fiber; however, the intensity was slightly lower than coconut holocellulose.

Fig. 4
figure 4

XRD patterns of (a) cotton linter cellulose, (b) cotton linter cellulose (after photocatalytic reaction), (c) fern fiber holocellulose, (d) coconut husk holocellulose, (e) fern fiber cellulose, (f) coconut husk cellulose

Full size image

XRD analysis was also carried out on cellulose after photocatalytic reaction. It should be noted that only cotton linter cellulose can be filtered from the solution, while cellulose samples from coconut husk and fern fiber were dispersed in water, which prevented separation from the catalysts. The calculated degrees of crystallinity of cellulose before and after photocatalytic reaction were measured at 87% and 92% respectively. It is interesting to see that the crystallinity increased following H2 production. This implied the removal of readily dissolved amorphous components, such as hemicellulose during the reaction and consequently increased the crystallinity index [30, 41, 42]. The emergence of a new peak at 25.9° was due to the presence of remaining TiO2 catalysts that were trapped on the cellulose.

3.2.2 FTIR analysis of cellulose

Figure 5 shows the FTIR spectra of cotton linter cellulose (before and after reaction), cellulose, and holocellulose extracted from coconut husk and fern fiber. The absorption peaks around 3299 cm−1, 2890 cm−1, 1427 cm−1, 1368 cm−1, and 892 cm−1 for all spectra indicated the typical characteristics of cellulose I. The peaks were visible on all cellulose and holocellulose samples but at different intensities. A broad band centered at 3299 cm−1 was observed on cotton linter cellulose, corresponding to the stretching vibration of the hydroxyl group. The hydroxyl band appeared broader on cellulose and holocellulose from coconut husk and fern fiber, suggesting the adsorption of water on the surface. The adsorption band at ~ 2770–3000 cm−1 corresponded to the stretching of asymmetric and symmetric -CH groups. For cellulose and holocellulose derived from coconut and fern biomass, the CH band appeared at 2926 cm−1 and 2849 cm−1 suggesting the presence of C-H band attached to a different environment. A well-defined absorption band around 1629 cm−1 in cellulose samples extracted from coconut husk and fern fiber was ascribed to OH bending of the adsorbed water, but the intensity of the band was reduced in cotton linter cellulose. The adsorption band situated at 1427 cm−1 was due to the -CH2 bending. The bending vibration peak detected at 1313 cm−1 was associated with the C-H and the C-O bonds in the polysaccharide aromatic rings. A small absorption peak at 1111 cm−1 was attributed to the bending of the OH group. C-O-C stretching vibrations of skeletal glucose rings and pyranose were located at 1152 cm−1 and 1056 cm−1 respectively. However, the peak intensities were visibly reduced for cellulose isolated from coconut and fern fibers. The stretching vibrations [n(CO)] of the COOOC glycosidic bridge appearing in 1175–1140 and 1000–970 cm−1 ranges were attributed to the difference in the glycosidic linkage configuration [43]. A small absorption band at 892 cm−1 corresponded to the cellulosic β-glycosidic linkages that consist of C1-H and O-H bending between glucose unit in cellulose [30, 41, 42, 44, 45]. There were no significant differences of the infrared spectra between cotton linter cellulose before and after photocatalysis reaction. However, the adsorption band at ~ 660 cm−1 that corresponded to the hydroxyl group out-of-plane bending gained more intensity after the reaction, which may indicate the presence of hydrolyzed fragments of cellulose [46]. Holocellulose extracted from coconut husk and fern fibers showed the characteristic bands of hemicellulose at 1732 cm−1, which corresponded to the C=O adsorption band of the acetyl group in hemicellulose [47].

Fig. 5
figure 5

FTIR spectra of (a) cotton linter cellulose (before reaction), (b) cotton linter cellulose (post-reaction), (c) coconut husk holocellulose, (d) fern fiber holocellulose, (e) coconut husk holocellulose, (f) fern fiber cellulose

Full size image

3.2.3 FESEM analysis of cellulose

Cotton linter cellulose was analyzed using SEM to investigate the morphology changes after photocatalytic reaction. Figure 6(a) showed a long rod-like shape morphology with smooth surface which was a typical structure of cotton linter cellulose. The cellulose fibril surface was reported to have non-fibrous components, e.g., lignin and hemicellulose, that were bound together to form a thick and smooth coating protecting the cellulose [30, 41]. After hydrolysis, the smooth surface of cellulose fiber was partially damaged, resulting in a corrugated texture. Although the rod-like shape structure remained unchanged, the corrugated surface indicated the removal of surface coating for further decomposition of cellulose to hydrogen gas [30].

Fig. 6
figure 6

SEM images of cotton linter cellulose (a) before photocatalysis and (b) after photocatalysis

Full size image

3.3 Photocatalytic H2 production

3.3.1 Effect of metal co-catalysts on H2 production

Photocatalytic hydrogen production from cellulose cotton linter using different types of metal co-catalysts (0.3% Pd, 0.3% Cu, 0.3%Ce, and 0.3% Ni) was investigated as shown in Fig. 7(a). No hydrogen was detected when the reaction was carried out without the presence of catalysts, indicating that cellulose did not self-decompose under light to generate hydrogen gas. Negligible hydrogen was also observed when the reaction was carried out in a dark environment, eliminating the possibility of H2 produced from mechanocatalysis reaction. Photodecomposition using TiO2 as catalysts only showed trace amounts of hydrogen, which indicated the importance of metals as co-catalyst to drive photocatalytic reaction. Pd/TiO2 appeared to be the most active catalyst, producing 131 μmol of H2 followed by Cu/TiO2 at 50 μmol in 3 h. However, not all metal co-catalysts enhanced the photocatalytic performance of TiO2; Ni/TiO2 and Ce/TiO2 only produced a very small amount of hydrogen over 3 h. Ce/TiO2 only started to produce hydrogen after 90 min and the rate increased to give ~ 10 μmol whereas for Ni/TiO2, only small traces of hydrogen were detected.

Fig. 7
figure 7

(a) H2 production from direct photocatalytic degradation of cellulose using 0.3% metal/TiO2 catalysts in comparison to TiO2. (b) H2 production from photodecomposition of cellulose as a function of time over Pd/TiO2 at different metal loading

Full size image

Since Pd generated high H2 yield from cellulose, the effect of metal loading was carried out using Pd/TiO2 at 0.3%, 0.05%, and 0.03% in order to minimize the amount of expensive Pd metal used as co-catalysts (Fig. 7(b)). The reaction was carried out for 3 h and the rate of hydrogen production showed no significant reduction despite Pd content being reduced from 0.3% to 0.03%. The presence of highly dispersed Pd nanoparticles on TiO2 surface increased the number of active sites for H2; however, high Pd loading caused particle size agglomeration that reduced the rate of H2 production [48]. We suggest that at low Pd loading ~ 0.03%, particle agglomeration was less likely to occur. A similar number of active sites were produced despite different amounts of metal loading, hence resulting in no detrimental effect on the H2 production. The amount of hydrogen produced at 0.03% of Pd loading was also significantly higher in comparison to TiO2 alone. The results suggested that although Pd is considered a precious metal that contributed to the high operational cost of catalyst production, the use of only 0.03% Pd is practical and cost-effective in producing relatively large amounts of hydrogen.

3.3.2 H2 production from cellulose cotton linter, coconut husk, and fern fiber on Pd/TiO2

Decomposition of cellulose for H2 production was carried out using cotton linter cellulose, and cellulose isolated from raw biomass wastes (coconut husk and fern fiber) on 0.3% Pd/TiO2. Figure 8(a) showed a high rate of hydrogen production at ~ 130 μmol when using cellulose cotton linter. Cellulose from coconut fiber produced 38 μmol of H2, which indicated significant reduction of H2 yield when compared to cotton linter cellulose. Hydrogen production evolved from fern-derived cellulose occurred at a much slower rate to give 6 μmol of H2 in 3 h. The results demonstrated the potential use of cellulose from natural biomass resources as a sustainable feedstock for hydrogen production via photocatalysis. The level of H2 production also varied depending on the type of biomass used for cellulose isolation.

Fig. 8
figure 8

(a) H2 production from photodecomposition of cellulose from cotton linter, coconut husk, and fern fiber using 0.3% Pd/TiO2. (b) H2 production from photodecomposition of cellulose at different concentrations as a function of time over 0.3% Pd/TiO2

Full size image

Holocellulose isolated from coconut husk and fern fiber was then used for H2 production on 0.3% Pd/TiO2. As can be seen in Table 1, high production of H2 gas was observed on both holocellulose isolated from coconut husk and fern fiber in comparison to their respective cellulose counterparts. After 3 h of photocatalytic reaction, ~ 65 μmol of H2 was measured from coconut holocellulose, which was almost double the H2 produced from coconut cellulose (~ 38 μmol). For holocellulose isolated from fern fiber, significant H2 production was also observed at 24 μmol, which is almost four times higher than the corresponding cellulose. Holocellulose was extracted from biomass waste containing high hemicellulose fractions (Table 1), and therefore, high H2 production observed from holocellulose suggested that hemicellulose could enhance H2 production.

Table 1 The properties of cellulose of cotton linter, coconut husk, and fern fiber and the amount of H2 produced within 3 h using 0.3% Pd/TiO2 catalysts
Full size table

Hydrogen production at different cellulose concentrations was carried out to investigate the dependency of hydrogen production on the amount of cellulose (Fig. 8(b)). Cellulose cotton linter was varied between 5 and 100 mg, which showed the rate of H2 production increased with the amount of cellulose. The presence of cellulose was crucial to scavenge the photogenerated hole. The irreversible reaction between cellulose and the hole generated in the valence band of TiO2 improved the separation of charge carriers during photocatalytic excitation, resulting in efficient electron and hole separation. Apart from the role as a sacrificial agent that inhibited the recombination of photogenerated electron-hole pairs [49], cellulose also simultaneously acted as a reactant for generation of hydrogen [26] [50]. Higher cellulose concentrations increased the amount of available substrates for decomposition, which resulted in a higher photocatalytic performance.

3.3.3 H2 production from sugar derived from hydrothermal treatment of cellulose

Cellulose is a polymeric biomaterial consisting of sugar monomers such as sucrose and glucose. High-pressure and high-temperature conditions via hydrothermal treatment were employed to disintegrate cellulose into aqueous sugar mixtures. GC-MS analysis of the aqueous solution recovered after 35 min of cellulose hydrothermal treatment showed the presence of a mixture of glucose and fructose. Cellulose in its natural form was stable towards decomposition. Only 6.8% of cellulose converted to monosaccharides at 250 °C, with 91.1% selectivity towards glucose, and 8.8% selectivity towards fructose. Aqueous solution containing glucose and fructose was subsequently used for photocatalytic H2 production (Fig. 9). A control experiment carried out in the absence of a photocatalyst showed only trace amounts of hydrogen produced. The photocatalytic activity for Pd/TiO2, Cu/TiO2, and TiO2 for 3 h under light illumination exhibited significantly higher H2 production from cellulose cotton linter. It is interesting to note that Cu/TiO2 was active for hydrogen production from glucose and sucrose to give ~ 100 μmol of H2, in comparison to reaction with cellulose at only ~ 10 μmol of H2.

Fig. 9
figure 9

H2 production from the aqueous solution containing glucose and fructose derived from hydrothermal treatment of cellulose using TiO2, 0.3% Pd/TiO2, and 0.3% Cu/TiO2

Full size image

4 Discussion

Cellulose are lignocellulosic materials with polysaccharide chains that can be utilized as renewable feedstocks for H2 production via photocatalytic decomposition. Pd/TiO2 was found to be the most active catalyst for H2 followed by Cu/TiO2, Ce/TiO2, and Ni/TiO2. Reducing the Pd loading to only 0.03 wt.% showed no significant reduction in the H2 production and therefore can mitigate the concern associated with the cost of Pd as an expensive metal. The photocatalytic activity of the catalysts was found to depend on the reducibility of metals in the photocatalytic environment. Metal/TiO2 photocatalysts were calcined at 550 °C and therefore oxidized into metal oxides. PdO was photoreduced under UV irradiation to form Pdo, which can then accept the photogenerated electron from TiO2 [39]. In contrast, NiO, CuO, and CeO were stable towards photoreduction and existed in various oxidation states. Another plausible explanation of the high activity of Pd/TiO2 was the formation of Schottky barrier. At the metal-TiO2 interface, a Schottky barrier formed when the work functions of the metal were greater than that of TiO2. The presence of this barrier minimized the electron-hole recombination upon photoexcitation, resulting in prolonged charge carriers’ lifetime and greatly enhancing the photocatalytic performance [28]. The work functions of Pd and Cu were 5.12 eV and 4.65 eV respectively, both of which were relatively higher than TiO2 (4.2 eV) [51]. This implied that Pd has a larger work function that increased the Schottky barrier effect.

Significant amounts of hydrogen evolved from cellulose were observed despite having no pre-treatment procedures carried out in order to improve the solubility prior to the photocatalytic reaction. Cellulose is insoluble in water, which is attributed to its rigid long chain structure with strong hydrogen bonds that gives cellulose its unique strength and stability [50]. The presence of numerous hydroxyl groups in cellulose and the β-1,4-glycosidic bonds form intra- and inter-molecular hydrogen bonds which resulted in stability and insolubility [52]. Pd/TiO2 photocatalysts facilitated the process of breaking down cellulose via generation of active hydroxyl radicals. Active hydroxyl radicals dissociated the β-1,4-glycosidic bonds in cellulose into monosaccharides, followed by subsequent decomposition into H2 and CO2 [16]. H2 produced from cellulose cotton linter and coconut husk in this study were comparable with the H2 yield obtained when using Pt/TiO2 as summarized in Table 2 [53]. To determine the viability of photocatalysis as a sustainable method for biohydrogen production, the hydrogen yields obtained from photocatalysis were compared with hydrogen yields from dark fermentation. Dark fermentation was carried out using bacterial microorganisms for fermentation of biomass substrates to biohydrogen in the absence of light [57]. The rate of hydrogen production for every gram of biomass via photocatalytic reaction was relatively higher compared to H2 produced from dark fermentation method (Table 2). Dark fermentation required a long fermentation time for decomposition of cellulose to H2, whereas the presence of a catalyst in a photocatalytic system accelerated the decomposition of cellulose. Our results demonstrated that biohydrogen production from lignocellulosic biomass via photocatalysis is an efficient and green process for energy generation.

Table 2 Comparison of reported H2 production from cellulose derived from different biomass substrates via dark fermentation method with present work
Full size table

Photodecomposition of cellulose at ambient conditions is an uphill reaction due to the stability of the glycosidic bonds. Thermodynamic analysis of cellulosic depolymerization via hydrolysis indicated that large concentrations of glucose were obtained from amorphous cellulose at equilibrium compared to crystalline cellulose [52]. Studies on enzymatic conversion of cellulose to glucose indicated the importance of reducing cellulose crystallinity in order to increase the glycosidic bond accessibility towards dissociation, which can occur via rearrangement of hydrogen bonding [58]. It is interesting to note that in our studies, cellulose from cotton linter with high crystallinity showed a higher rate of H2 production than cellulose isolated from coconut husk and fern fiber. However direct comparison between the cellulose samples was not ideal since the celluloses were obtained from different sources of biomass waste. Cellulose obtained from coconut husk and fern fiber exhibited a lower degree of polymerization in comparison to cotton linter cellulose. The differences of polysaccharide units have affected the volume of H2 production as given in Table 1. [5]. Holocellulose from coconut husk and fern fibers produced high yield of H2 due to the presence of hemicellulose. Hemicellulose is an amorphous polysaccharide comprised of different monomeric units with acetyl groups originally linked to the xylose unit [59]. The presence of hemicellulose was suggested to enhance the hydrolysis of cellulosic materials due to the elimination of acetic acid that further accelerated β-1,4-glycosidic bond dissociation [60]. High H2 production from glucose and fructose mixtures derived from hydrothermal treatment of cotton linter cellulose implied the rate determining step of the reaction may be the dissociation of β-1,4-glycosidic bonds. H2 production from glucose was initiated by OH radicals to form gluconic acid, which subsequently underwent further decarboxylation to form H2 and CO2 gasses [61].

5 Conclusion

Photocatalytic decomposition of cellulose extracted from biomass waste for H2 production demonstrated the feasibility of harvesting raw biomass for sustainable production of hydrogen. High activity of Pd/TiO2 compared to Cu/TiO2, Ni/TiO2, and Ce/TiO2 indicated the important role of metal co-catalysts to decompose cellulose into H2. The efficiency of H2 generation also depended on the crystallinity and the degree of polymerization of cellulose. Holocellulose extracted from coconut husk and fern fiber was more susceptible towards H2 generation due to the release of acetic acid from hemicellulose, which accelerated hydrolysis. Aqueous solution containing glucose and fructose from hydrothermal treatment of cellulose generated a large volume of H2 at 350 μmol in comparison to 130 μmol from direct photodecomposition of cellulose, suggesting dissociation of glycosidic bond in cellulose as the rate determining step of the reaction.