Levulinic Acid Production from Deligniﬁed Rice Husk Waste over Manganese Catalysts: Heterogeneous Versus Homogeneous

: Deligniﬁed rice husk waste (25.66% (wt) cellulose) was converted to levulinic acid using three types of manganese catalysts, i.e., the Mn 3 O 4 / hierarchical ZSM-5 zeolite and Mn 3 O 4 heterogenous catalysts, as well as Mn(II) ion homogeneous counterpart. The hierarchical ZSM-5 zeolite was prepared using the double template method and modiﬁed with Mn 3 O 4 through wet-impregnation method. The structure and physicochemical properties of the catalyst materials were determined using several solid-state characterization techniques. The reaction was conducted in a 200 mL-three neck-round bottom ﬂask at 100 ◦ C and 130 ◦ C for a certain reaction time in the presence of 10% (v / v) phosphoric acid and 2% (v / v) H 2 O 2 aqueous solution, and the product was analyzed using HPLC. In general, furfural (5-HMF) as the intermediate product was produced after 2 h and decreased after 4 h reaction time. To conclude, the Mn 3 O 4 / hierarchical ZSM-5 heterogenous catalyst gave the highest yield (wt %) of levulinic acid (39.75% and 27.60%, respectively) as the main product, after 8 h reaction time.


Introduction
Levulinic acid (4-oxopentanoic acid) is an important compound used as a platform chemical in alternative energy production, such as the production of γ-valerolactone (GVL), 5-bromolevulinic acid, valeric acid, 2-methyl-THF, and methyl pyrrolidone [1]. It is also used as a renewable source for the production of polymers such as its derivative, GVL, which is used as a basis for precursor formation of polymers [2]. Kon et al. reported reacting levulinic acid with valeric acid and valeric biofuels via selective hydrogenation using a Pt/HMFI zeolite catalyst [3]. One of the important industrial heterogeneous catalysts used as support is micropores zeolite (>2 nm pore diameter). It is well-known that zeolite has thermal and chemical stability, as well as catalytic and adsorption activity [4]. However the micropores has limited the application of zeolite in conversion or adsorption of larger substrates. Therefore, some attempts have been reported to improve overcome this limitation including preparing hierarchical zeolite [5]. In contrast to conventional zeolites, hierarchical zeolites consist of both micropores (where the active sites located) and mesopores facilitating diffusion of reactants and products [6].
Preparation of hierarchical ZSM-5 and other zeolites structure using secondary template have been reported, such as hierarchical mesoporous Beta and ZSM-5 [7], also mesoporous FAU type zeolite [8]. ZSM-5, one type of zeolites has long been used as catalyst in the fine-chemical and petrochemicals industry for catalytic reactions involving cracking, isomerization, alkylation, acylation, purification for water adsorbing systems [9], and having active sites with diverse strength become the main attractions related to their application [10]. Hierarchical ZSM-5 zeolites used for partial oxidation of methane to methanol [11], catalytic hydrogenation of carbon dioxide [12], catalytic pyrolysis of biomass to aromatic hydrocarbons [13].
In many industries, the production of levulinic acid is relatively low because of the high cost of its synthetic raw materials. Hence, the importance of biomass raw materials, which are cheap and easy to make, cannot be overemphasized [14]. Indonesia is the third largest rice producer in the world, after China and India. However, the country has yet to attain optimal utilization of rice husk. To overcome the above limitations and make proper use of the abundant raw materials, rice husks can be used as organic (biomass) substitutes for synthetic raw materials in the production of levulinic acid, using the cellulose contained in biomass. It is necessary to optimize the delignification pre-treatment to obtain optimum amount of cellulose from lignocellulosic biomass. There are several methods of chemical delignification such as ozonolysis, oxidative delignification, acid and alkaline hydrolysis. The alkali pretreatment using NaOH is the most common method applied, where the NaOH concentration depends on the content of lignin in the plants. In example, diluted NaOH solution is effective to reduce lignin content (wt %) from 24-55% to 20% (wt) in hardwood [15]. In our previous work, pretreatment of mahogany wood with NaOH 10% (v/v) decreased the lignin content from 51.03 to 15.58% (wt) [16], whereas when applied to rice husk, the lignin content was decreased from 35.64 to 18.42% (wt) [17]. NaOH showed to be the best delignification agent because the obtained cellulose is more amorphous, compared to NaOCl giving more crystalline cellulose [16].
Chen et al. [17] studied the degradation of pure cotton cellulose into 5-(hydroxymethyl)furfural (5-HMF) and levulinic acid, using Mn/ZSM-5 agregate as catalyst through Fenton-like reaction. The degradation process investigated the use of Mn/ZSM-5 as catalyst for the degradation of pure cotton cellulose under phosphoric acid media and hydrogen peroxide, to yield 5-HMF as a main product and levulinic acid as a by-product. In other work, HCl has been used as a homogeneous catalyst while ZSM-5, β-zeolite, Y-zeolite were used as heterogeneous catalysts transform glucose into levulinic acid and 5-HMF [18]. In our previous study, Krisnandi et al. [17] used hierarchical MnOx/ZSM-5 as heterogeneous catalyst and Mn 2+ ions as homogeneous catalyst in conversion of delignified rice husk to levulinic acid. The conversion recorded a highest percentage yield (wt %) for levulinic acid (15.83%) at 100 • C, reaction time of 8 h, and in phosphoric acid media containing H 2 O 2 . It showed the stability of the heterogeneous catalysts after the reaction.
In order to obtain more information the role of manganese oxide and hierarchical ZSM-5 in Mn3O4/hierarchical ZSM-5 catalysts, conversion with manganese oxide Mn 3 O 4 (as heterogeneous catalyst) was also carried out as addition to experiments reported in [17]. Furthermore, identification of intermediate compound (HMF) and side product (formic acid) were also carried out. Therefore, mechanism in conversion cellulose to levulininc acid can be confirmed. To the best of the author knowledge, this is one among not many articles reporting the utilization of raw biomass as cellulose sources which contributed to the development of value-added chemical such as levulinic acid.

Catalyst Characterization
XRD analysis was conducted to confirm the structure of the as-synthesized ZSM-5 zeolite. Its diffraction pattern ( Figure 1) has typical peaks similar to those observed in the pattern of standard ZSM-5, in which there are two sharp peaks between 2θ = 7-10 • and three peaks between 2θ = 22-25 • [19].
The powder XRD pattern of the Mn 3 O 4 /hierarchical ZSM-5 zeolite (Figure 1b) show additional peaks which include, two sharp peaks between 2θ = 30-35 • , and three peaks between 2θ = 35-40 • [20]. According to JCPDS: 24-0734, peaks belong to Mn3O4 crystals. These results indicate the successful preparation of hierarchical Mn 3 O 4 /ZSM-5 samples. On the other hand, a sharp peak between 2θ = 15-20 • is also observed at XRD pattern of as-prepared Mn3O4 (Figure 1c), according to JCPDS: 44-0141, the peak belongs to MnO 2 [21]. FTIR analysis of ZSM-5 zeolites and manganese oxide-modified counterpart is summarized in Figure 2. The infrared spectra of hierarchical Mn3O4/ZSM-5 zeolite (Figure 2b) was compared to that of hierarchical ZSM-5 zeolite ( Figure 2c) and Mn3O4 (Figure 2a). The figures show that all the peaks observed in the hierarchical ZSM-5 zeolite were also observed in the Mn3O4/hierarchical ZSM-5 zeolite. However, there is a difference in the intensity of the broad band located at 3500-3000 cm −1 , which is a typical absorption band assigned to the stretching vibration of water-hydrogen bonded silanol group (ν-Si-OH) or internal silanol groups [22]. The intensity of the silanol band in Mn3O4/ZSM-5 decreased, suggesting that the Mn-oxide species were located on top of the silanol groups, thus they interacted with the -OH, thereby leading to a decrease in the water content of the water-hydrogen bonded -Si-OH group. In addition, Figure 2 shows the presence of Mn3O4 where the FTIR spectra indicate two characteristic emission bands located around 613 and 509 cm −1 , a typical absorption band coupling between Mn−O stretching at the finger print area [23]. In addition, there were more than two absorption band peaks around 615 and 513 cm −1 , which can be attributed to the coupling modes between the Mn-O stretching, while the band around 429 cm −1 can be assigned as a band-stretching mode [24]. AAS measurement for the Mn-oxide ZSM-5 indicates that the weight percent loading of Mn into the ZSM-5 structure was 1.91% (wt %), which is close to the expected value of 2.0%. It can also be seen that the insertion of Mn3O4 into the extra zeolite framework did not cause significant change to  However, there is a difference in the intensity of the broad band located at 3500-3000 cm −1 , which is a typical absorption band assigned to the stretching vibration of water-hydrogen bonded silanol group (ν-Si-OH) or internal silanol groups [22]. The intensity of the silanol band in Mn 3 O 4 /ZSM-5 decreased, suggesting that the Mn-oxide species were located on top of the silanol groups, thus they interacted with the -OH, thereby leading to a decrease in the water content of the water-hydrogen bonded -Si-OH group. In addition, Figure 2 shows the presence of Mn 3 O 4 where the FTIR spectra indicate two characteristic emission bands located around 613 and 509 cm −1 , a typical absorption band coupling between Mn−O stretching at the finger print area [23]. In addition, there were more than two absorption band peaks around 615 and 513 cm −1 , which can be attributed to the coupling modes between the Mn-O stretching, while the band around 429 cm −1 can be assigned as a band-stretching mode [24]. FTIR analysis of ZSM-5 zeolites and manganese oxide-modified counterpart is summarized in Figure 2. The infrared spectra of hierarchical Mn3O4/ZSM-5 zeolite (Figure 2b) was compared to that of hierarchical ZSM-5 zeolite ( Figure 2c) and Mn3O4 (Figure 2a). The figures show that all the peaks observed in the hierarchical ZSM-5 zeolite were also observed in the Mn3O4/hierarchical ZSM-5 zeolite. However, there is a difference in the intensity of the broad band located at 3500-3000 cm −1 , which is a typical absorption band assigned to the stretching vibration of water-hydrogen bonded silanol group (ν-Si-OH) or internal silanol groups [22]. The intensity of the silanol band in Mn3O4/ZSM-5 decreased, suggesting that the Mn-oxide species were located on top of the silanol groups, thus they interacted with the -OH, thereby leading to a decrease in the water content of the water-hydrogen bonded -Si-OH group. In addition, Figure 2 shows the presence of Mn3O4 where the FTIR spectra indicate two characteristic emission bands located around 613 and 509 cm −1 , a typical absorption band coupling between Mn−O stretching at the finger print area [23]. In addition, there were more than two absorption band peaks around 615 and 513 cm −1 , which can be attributed to the coupling modes between the Mn-O stretching, while the band around 429 cm −1 can be assigned as a band-stretching mode [24]. AAS measurement for the Mn-oxide ZSM-5 indicates that the weight percent loading of Mn into the ZSM-5 structure was 1.91% (wt %), which is close to the expected value of 2.0%. It can also be seen that the insertion of Mn3O4 into the extra zeolite framework did not cause significant change to AAS measurement for the Mn-oxide ZSM-5 indicates that the weight percent loading of Mn into the ZSM-5 structure was 1.91% (wt %), which is close to the expected value of 2.0%. It can also be seen that the insertion of Mn 3 O 4 into the extra zeolite framework did not cause significant change Catalysts 2020, 10, 327 4 of 13 to its XRD pattern. This indicates that the structure and crystallinity of the ZSM-5 zeolite were not destroyed during the Mn impregnation process.
Scanning electron microscopy (SEM) characterization provided morphology information of the hierarchical ZSM-5 zeolite. Figure 3 clearly shows that the as-synthesized ZSM-5 zeolite had hexagonal structure and a coffin-like shape [25]. ZSM-5 zeolite synthesized using TPAOH template to form MFI framework has coffin-like morphology and has hexagonal crystal geometry [7]. However, the mesoporosity of the surface is not clearly visible, as reported by Wang et al. [7]. Therefore, it is necessary to perform surface area analysis to identify the ZSM-5 as a porous material.
Catalysts 2019, 9, x FOR PEER REVIEW 4 of 13 its XRD pattern. This indicates that the structure and crystallinity of the ZSM-5 zeolite were not destroyed during the Mn impregnation process. Scanning electron microscopy (SEM) characterization provided morphology information of the hierarchical ZSM-5 zeolite. Figure 3 clearly shows that the as-synthesized ZSM-5 zeolite had hexagonal structure and a coffin-like shape [25]. ZSM-5 zeolite synthesized using TPAOH template to form MFI framework has coffin-like morphology and has hexagonal crystal geometry [7]. However, the mesoporosity of the surface is not clearly visible, as reported by Wang et al. [7]. Therefore, it is necessary to SM-5 as a porous material. Transmission Electron Microscope (TEM) measurement ( Figure 4) of the hierarchical ZSM-5 zeolite and its modification with Mn3O4 indicates that the crystal morphology of the catalyst is not damaged after modification. There are some bright oval shapes with diameter sizes of 2.5-12 nm ( Figure 4a) that justify the presence of mesoporosity within the structure of hierarchical ZSM-5. This corresponds with the SEM images and SAA data. In the image of Mn3O4-modified ZSM-5 shown in Figure 4b, dark deposits were observed on the surface of hierarchical ZSM-5, which could be attributed to the presence of Mn3O4 cluster aggregation with diameter size of 6.8-20 nm. The analysis of the surface area and pore size were carried out using Brunauer-Emmett-Teller (BET) and Barrett-Joyner-Halenda (BJH) desorption methods, respectively. The results showed that the its XRD pattern. This indicates that the structure and crystallinity of the ZSM-5 zeolite were not destroyed during the Mn impregnation process. Scanning electron microscopy (SEM) characterization provided morphology information of the hierarchical ZSM-5 zeolite. Figure 3 clearly shows that the as-synthesized ZSM-5 zeolite had hexagonal structure and a coffin-like shape [25]. ZSM-5 zeolite synthesized using TPAOH template to form MFI framework has coffin-like morphology and has hexagonal crystal geometry [7]. However, the mesoporosity of the surface is not clearly visible, as reported by Wang et al. [7]. Therefore, it is necessary to perform surface area analysis to identify the ZSM-5 as a porous material.  The analysis of the surface area and pore size were carried out using Brunauer-Emmett-Teller (BET) and Barrett-Joyner-Halenda (BJH) desorption methods, respectively. The results showed that the  The analysis of the surface area and pore size were carried out using Brunauer-Emmett-Teller (BET) and Barrett-Joyner-Halenda (BJH) desorption methods, respectively. The results showed that the surface area of the as-synthesized hierarchical ZSM-5 zeolite was 348.6 m 2 /g. This property is important, because the greater the surface area, the easier it is to disperse metal oxides as active sites on the support surface [26]. Furthermore, Figure 5 shows the characteristic isotherm curves (type IV), which indicates a mesoporous structure. PDDA-Cl (polidialydimethyl ammonium chloride) aims to build this hierarchical ZSM-5 material so that the microporous and mesoporous material is formed in one structure. This results is different from the typical curve for microporous ZSM-5 isotherms which usually shows that specific surface area is developed from micropores [27], but it is consistent with the curve reported by Xu et al. [26] for hierarchical zeolites containing micro-and mesopores in one structure. In addition, the average diameter of the pores formed were 1.8842 nm and 2.7419 nm, which are within the range for microporous (d < 2 nm) and mesoporous (2 nm < d < 50 nm), respectively. The results for the surface area analysis are summarized in Table 1. surface area of the as-synthesized hierarchical ZSM-5 zeolite was 348.6 m 2 /g. This property is important, because the greater the surface area, the easier it is to disperse metal oxides as active sites on the support surface [26]. Furthermore, Figure 5 shows the characteristic isotherm curves (type IV), which indicates a mesoporous structure. PDDA-Cl (polidialydimethyl ammonium chloride) aims to build this hierarchical ZSM-5 material so that the microporous and mesoporous material is formed in one structure. This results is different from the typical curve for microporous ZSM-5 isotherms which usually shows that specific surface area is developed from micropores [27], but it is consistent with the curve reported by Xu et al. [26] for hierarchical zeolites containing micro-and mesopores in one structure. In addition, the average diameter of the pores formed were 1.8842 nm and 2.7419 nm, which are within the range for microporous (d < 2 nm) and mesoporous (2 nm < d < 50 nm), respectively. The results for the surface area analysis are summarized in Table 1.

Pre-Treatment of Rice Husk Waste
The first stage in the pre-treatment process for rice husk is dewaxing to remove extractive substances contained in the biomass such as wax, fat, fatty acids, terpenes, and steroids [28] that may interfere with the conversion process. The second stage is the delignification process to improve the internal surface of lignocellulose, reduce the degree of polymerization and crystallinity, and damage the structure of lignin [29]. Table 2 shows that before pre-treatment, the lignin and α-cellulose content were 33.53% and 42.37% weight, respectively. After pre-treatment, the lignin content decreased almost 49% to 17.24% weight in Table 2. The α-cellulose and hemicellulose content was also

Pre-Treatment of Rice Husk Waste
The first stage in the pre-treatment process for rice husk is dewaxing to remove extractive substances contained in the biomass such as wax, fat, fatty acids, terpenes, and steroids [28] that may interfere with the conversion process. The second stage is the delignification process to improve the internal surface of lignocellulose, reduce the degree of polymerization and crystallinity, and damage the structure of lignin [29]. Table 2 shows that before pre-treatment, the lignin and α-cellulose content were 33.53% and 42.37% weight, respectively. After pre-treatment, the lignin content decreased almost 49% to 17.24% weight in Table 2. The α-cellulose and hemicellulose content was also decreased. This may be due to cellulose and hemicellulose removal during the delignification process, especially the part directly connected to lignin in the lignocellulosic bond [30]. The delignified rice husk is then used as a substrate in the catalytic test.  (1) and (2). Furthermore, the role of hierarchical ZSM-5 as support catalyst was also studied by observing the reaction profile when Mn 3 O 4 /ZSM-5 is used as a catalyst. The reaction using hierarchical ZSM-5 as catalyst has been discussed in significant detail in our other work [17]. Finally, reaction with a homogeneous solution of Mn 2+ catalyst was used as comparison, and the reaction without catalyst as control. All the reactions were carried out in phosphoric acid (H 3 PO 4 ) 40% (v/v) media with reflux, since cellulose is soluble in phosphoric acid (which is able to damage the intra-and inter-molecular hydrogen bonds in cellulose) but difficult to dissolve in water [33]. The Fenton-like species was used for the next step, with phosphoric acid as media and the Fenton-like reaction of manganese also giving H + to the increased acidic media. The cellulose can be degraded using a Fenton-like reagent such as H 2 O 2 together with a catalyst, and by a phosphoric acid solution [33].
The next step was breaking the β-(1,4)-glycosidic bonds, after the hydrogen bond termination by the solvent. The termination step occurred through Fenton-like reaction. Hydrogen peroxide was used in the presence of Mn(III) to produce a radical hydroxyl, HO• (Equation (1)). Therefore, hydrogen peroxide (H 2 O 2 ) acted as an initiating reagent. The Mn 3+ produced then reacts continuously with the H 2 O 2 , which in the presence of excess H 2 O 2 produces Mn 2+ and hydroperoxide radical, HO 2 • (Equation (2)). However, radical hydroperoxides are not as effective as hydroxyl radicals [31,34]. Thus, the quantity of H 2 O 2 used should not be too high to achieve optimum results [35]. The key mechanistic feature of the Fenton-like reaction is the generation of OH• radicals [34]. Figure 6 shows the percentage yield (wt %) of the reactions of delignified rice-husk over a 10 h-reaction time at 100 • C for three catalysts and without catalyst as control. The trends for yield (wt %) of produced levulinic acid were similar for all three catalysts, it was produced only after 2 h and kept increasing up to 8 h before decreasing after 10 h. In the reaction without catalysts, levulinic acid was only observed after the reaction has progressed for 4 h. This suggests that there is an initiation time for the reduction of β-1-4-glucosidic bonds in the cellulose to smaller fraction such as glucose, followed by oxidation of the hydroxyl radicals in both manganese and acidic media to produce levulinic acid.
To study the effect of temperature on the yield (wt %), reactions were also conducted at 130 • C; the results are shown in Figure 7. The result follows a similar trend to those obtained at 100 • C, but the yield (wt %) of levulinic acid is higher for the same reaction time. A significant increase in yield (wt %) Catalysts 2020, 10, 327 7 of 13 was observed in the reaction using Mn 3 O 4 /hierarchical ZSM-5 as catalyst, while the increase of yield (wt %) from other catalysts was less obvious. Figures 6 and 7 show that the highest yield (wt %) of levulinic acid for each reaction was reached after 8 h reaction time. When using Mn 3 O 4 /hierarchical ZSM-5 as catalyst, the yield (wt %) obtained at 100 • C and 130 • C was 27.60% and 39.75%, respectively. The reaction using Mn 2+ gave yield of 17.69% (100 • C) and 19.34% (130 • C). When Mn 3 O 4 catalyst, the highest yield (wt %) of levulinic acid was similar, reaching 14.02% (100 • C) and 15.93% (130 • C), respectively. After 10 h reaction, the yield (wt %) of levulinic acid, for all catalysts, decreased. It indicates that the levulinic acid could decompose or undergo futher reaction. Based on this information, it is suggested that the reaction is selective to cellulose. This result recommends that cellulose degradation reactions requires higher temperatures (130 • C) to lower the activation energy [33]. The high reaction temperature give resulted in larger amount of products in the liquid hydrolysates, due to accelerated sugar structure decomposition and increased the severity of the treatment [36]. Longer reaction times and higher temperatures improve the yield of products by increasing the rate of degradation, which shows the relationship between the concentration and the reaction time. However, we do not go higher than 130 • C because we want it to be more selective to cellulose and keep the energy consumption as low as possible. In this reaction the remaining lignin mixed with other solid percipitate (char), as also reported in [17]. The fate of lignin was not discussed further, since usually lignin conversion, with or without the presence of catalyst, takes place at temperature above 150 • C with high preassure [37].  Figure 7 show that the highest yield (wt %) of levulinic acid for each reaction was reached after 8 h reaction time. When using Mn3O4/hierarchical ZSM-5 as catalyst, the yield (wt %) obtained at 100 o C and 130 o C was 27.60% and 39.75%, respectively. The reaction using Mn 2+ gave yield of 17.69% (100 o C) and 19.34% (130 o C). When Mn3O4 catalyst, the highest yield (wt %) of levulinic acid was similar, reaching 14.02% (100 o C) and 15.93% (130 o C), respectively. After 10 h reaction, the yield (wt %) of levulinic acid, for all catalysts, decreased. It indicates that the levulinic acid could decompose or undergo futher reaction. Based on this information, it is suggested that the reaction is selective to cellulose. This result recommends that cellulose degradation reactions requires higher temperatures (130 o C) to lower the activation energy [33]. The high reaction temperature give resulted in larger amount of products in the liquid hydrolysates, due to accelerated sugar structure decomposition and increased the severity of the treatment [36]. Longer reaction times and higher temperatures improve the yield of products by increasing the rate of degradation, which shows the relationship between the concentration and the reaction time. However, we do not go higher than 130 o C because we want it to be more selective to cellulose and keep the energy consumption as low as possible. In this reaction the remaining lignin mixed with other solid percipitate (char), as also reported in [17]. The fate of lignin was not discussed further, since usually lignin conversion, with or without the presence of catalyst, takes place at temperature above 150 o C with high preassure [37]. In the reaction using Mn 2+ , Mn3O4, and hierarchical Mn3O4/ZSM-5 as catalysts, Mn species affected the formation of hydroxyl radicals that being generated during Fenton-like reaction. For the reaction without a catalyst, there is no species to streamline the formation of hydroxyl radicals, hence, leading to a thermal decomposition of cellulose.

Side Product Analysis
Cellulose conversion, in acidic condition, to levulinic acid follows a complex reaction pathways [35], which involves the production of glucose, fructose, and 5-HMF as intermediates and formic acid as the by-product. In this work, only formation the of 5-HMF and formic acid were studied. Figures 6 and 7 shows that the 5-HMF was not formed until 2 h of reaction, reached optimum 4 h, then decreased when the reaction carried out longer than 4 h. The decrease in 5-HMF yield indicates it has been converted futher into levulinic acid and formic acid. This trend is observed for all the manganese catalysts used in this study. In line with levulinic production, the reaction gives higher 5-HMF yield (wt %) at 130 • C than the reaction at 100 • C.

5-HMF Intermediate
Catalysts 2019, 9, x FOR PEER REVIEW 8 of 13 reaction without a catalyst, there is no species to streamline the formation of hydroxyl radicals, hence, leading to a thermal decomposition of cellulose.

Side Product Analysis
Cellulose conversion, in acidic condition, to levulinic acid follows a complex reaction pathways [35], which involves the production of glucose, fructose, and 5-HMF as intermediates and formic acid as the by-product. In this work, only formation the of 5-HMF and formic acid were studied.

Formic Acid
The results in Figure 6; Figure 7 show that formic acid is formed as a by-product because of two pathways from HMF intermediate compounds which produce levulinic acid as a main product and formic acid as a by-product [38]. The yield of formic acid follows the same trend as that of

Formic Acid
The results in Figures 6 and 7 show that formic acid is formed as a by-product because of two pathways from HMF intermediate compounds which produce levulinic acid as a main product and formic acid as a by-product [38]. The yield of formic acid follows the same trend as that of levulinic acid, which is the largest product obtained after 8 h of reaction. However, the yield of formic acid is lower than that of levulinic acid, due to evaporation at temperature above 100 • C.

Discussion
The presence of Mn catalysts has accelerated the process of delignified biomass conversion to levulinic acid. Comparing the activity of Mn 2+ ion to solid Mn 3 O 4 , it was found that it is a better catalyst than the solid Mn 3 O 4 , giving a slightly higher yield (wt %) of levulinic acid. This is expected since in fenton like system, Mn 2+ produced directly OH•, from H 2 O 2 , that will catalyze cellulose conversion to yield (wt %) of levulinic acid. Meanwhile, solid Mn 3 O 4 is insoluble in water, hence, the amount of Mn 2+ and Mn 3+ produced will be less than that in the reaction using free Mn 2+ catalyst. On the other hand, Mn 3 O 4 impregnated to hierarchical ZSM-5 gives the highest % yield of levulinic acid among the three catalysts. It shows that, the ZSM-5, with high surface area, caused the Mn 3 O 4 species distributed as small agregates inside the pores (as shown by TEM), thus it has more active sites than in bulk Mn 3 O 4 , producing more OH• species. This indicates that the hierarchical ZSM-5 plays an important role as a support catalyst. This result is consistent with our previous results [17] and results obtained by Chen et.al [33]. Unfortunately, the zeolite catalyst used in our reaction did not survive the acidic aqueous, as confirmed in our previous work [17]. This is a disadvantage using zeolite in severe aqueous phase reaction [39,40]. On the other hand, there are some advantages that can be gained in this reaction, such as small amount of catalyst used (10% of substrate weight), aqueous phase reaction, feasible separation of liquid product from unwanted solid char and lignin, less concentrated H 3 PO 4 (40% v/v) and low reaction temperature (100 to 130 • C). They correspond to green chemistry and UN Sustainable Development Goals (SDGs) number 9 and 12 [41].
Analysis of the products is in agreement with the proposed mechanism for the production of levulinic acid from cellulose is illustrated in Figure 8. The cellulose from delignified rice husk first will be hydrolyzed to yield glucose, which is isomerized to form fructose. The fructose is then converted first to 5-HMF (intermediate product) and finally to levulinic acid, with formic acid as by-product [42]. Finally, a temperature of 130 • C give higher yield (wt %) results, since it speeds up the dehydration and rehydration processes from fructose to levulinic acid. The proposed mechanism is in agreement with mechanism reported by Chen et al. using cotton cellulose [33] and Singh et al. using aromatic crop [43].
Catalysts 2019, 9, x FOR PEER REVIEW 9 of 13 levulinic acid, which is the largest product obtained after 8 h of reaction. However, the yield of formic acid is lower than that of levulinic acid, due to evaporation at temperature above 100 °C.

Discussion
The presence of Mn catalysts has accelerated the process of delignified biomass conversion to levulinic acid. Comparing the activity of Mn 2+ ion to solid Mn3O4, it was found that it is a better catalyst than the solid Mn3O4, giving a slightly higher yield (wt %) of levulinic acid. This is expected since in fenton like system, Mn 2+ produced directly OH•, from H2O2, that will catalyze cellulose conversion to yield (wt %) of levulinic acid. Meanwhile, solid Mn3O4 is insoluble in water, hence, the amount of Mn 2+ and Mn 3+ produced will be less than that in the reaction using free Mn 2+ catalyst. On the other hand, Mn3O4 impregnated to hierarchical ZSM-5 gives the highest % yield of levulinic acid among the three catalysts. It shows that, the ZSM-5, with high surface area, caused the Mn3O4 species distributed as small agregates inside the pores (as shown by TEM), thus it has more active sites than in bulk Mn3O4, producing more OH• species. This indicates that the hierarchical ZSM-5 plays an important role as a support catalyst. This result is consistent with our previous results [17] and results obtained by Chen et.al [33]. Unfortunately, the zeolite catalyst used in our reaction did not survive the acidic aqueous, as confirmed in our previous work [17]. This is a disadvantage using zeolite in severe aqueous phase reaction [39,40]. On the other hand, there are some advantages that can be gained in this reaction, such as small amount of catalyst used (10% of substrate weight), aqueous phase reaction, feasible separation of liquid product from unwanted solid char and lignin, less concentrated H3PO4 (40% v/v) and low reaction temperature (100 to 130 °C). They correspond to green chemistry and UN Sustainable Development Goals (SDGs) number 9 and 12 [41].
Analysis of the products is in agreement with the proposed mechanism for the production of levulinic acid from cellulose is illustrated in Figure 8. The cellulose from delignified rice husk first will be hydrolyzed to yield glucose, which is isomerized to form fructose. The fructose is then converted first to 5-HMF (intermediate product) and finally to levulinic acid, with formic acid as byproduct [42]. Finally, a temperature of 130 o C give higher yield (wt %) results, since it speeds up the dehydration and rehydration processes from fructose to levulinic acid. The proposed mechanism is in agreement with mechanism reported by Chen et al. using cotton cellulose [33] and Singh et al. using aromatic crop [43].

Materials and Chemicals
Rice husk was supplied from Dramaga area, Bogor, Indonesia with the geographical coordinates -6.577672, 106.741376, harvested in dry season. The chemicals used in this experiment were all of

Materials and Chemicals
Rice husk was supplied from Dramaga area, Bogor, Indonesia with the geographical coordinates -6.577672, 106.741376, harvested in dry season. The chemicals used in this experiment were all of analytical grade: hydrogen peroxide (

and Mn(II) ion
The synthesis of hierarchical ZSM-5 zeolite was done following the procedure published in Krisnandi et al. [44], using a double template method with NaAlO 2 and TEOS as sources of Al and Si, and TPAOH and PDDA-Cl (35% wt) as first and second template sources. The crystallization process was carried out at 150 • C for 6 days. The product was filtered and calcined at 550 • C for 5 h to remove the organic template, followed by characterization using X-ray diffraction (XRD), Fourier transformed infra-red spectroscopy (FT-IR), Brunauer-Emmett-Teller (BET), and scanning electron microscopy-energy dispersive X-ray spectroscopy (SEM-EDS).
Hierarchical Mn 3 O 4 /ZSM-5 zeolite was prepared using the incipient wetness impregnation method with manganese nitrate solution for 24 h at 25 • C into a calcined hierarchical ZSM-5 zeolite. The mixture was dried at 110 • C for 5 h followed by calcination at 550 • C. For comparison, Mn 3 O 4 and Mn(II) solutions were also prepared as catalysts. The Mn catalysts were then characterized accordingly using XRD, FTIR, BET, and atomic absorption spectroscopy (AAS).
XRD measurements of the samples were performed at the Physics Laboratory of BATAN PUSPIPTEK, Serpong, Indonesia. The crystal structures of the samples were measured at room temperature using XRD PANalytical Instrument with copper at λ Kα1 1.54096 Å. The X-ray source was a conventional sealed 2500-Watt X-ray tube operated at 45 kV and 40 mA in a 2θ range between 5 • and 50 • . The FT-IR sample measurements were performed using a Shimadzu IR Prestige 21. A mixture of dried solid sample and KBr was pelletized to form a transparent film for the measurement of the FT-IR at wavenumber range from 450 to 4000 cm −1 . SEM-EDS for analysis of the chemical composition at the Chemistry Laboratory, BATAN PUSPIPTEK, Serpong, Indonesia. The surface area and pore size were analysed using a Quantachrome Quadrawin v3.21 and ASAP 2400. Samples were measured by N 2 adsorption at 196 • C. The adsorption-desorption measurements for the samples were degassed at 300 • C for 13 h while the surface area was measured using the BET equation. The micropore and mesopore size distributions were determined using Barrett-Joyner-Halenda (BJH) methods. AAS was performed using a AA 6300. Destruction by digestion with nitric and perchloric acids and by low-temperature ashing with dissolution of the ash in a hydrofluoric acid mixture with solid sample. TEM measurement on ZSM-5 before and after Mn 3 O 4 impregnation was carried out using Tecnai F20, FEI Company, KAIST.

Pre-Treatment of Biomass
Rice husk waste was subjected to dewaxing and delignification pre-treatment. Dewaxing was carried out by extracting the mash of the rice husk waste using n-hexane: ethanol (2:1, v/v) with soxhlet extraction for 6 h at 60 • C. Delignification was then carried out by immersing the residue from dewaxing process into 10% NaOH solution (1:25, m/v) at 55 • C for 90 min under vigorous stirring (200 rpm) The lignin and cellulose content were determined before and after the pre-treatment using the procedure outlined in our previous study [17].

Catalytic Test: Conversion of Delignified Rice Husk
The conversion of delignified rice husk (substrate) was carried out following Chen et al. [33] with modification. In general, it was conducted in Pyrex three neck-round bottom flask at 100 • C and 130 • C. Approximately 1 g of substrate was dispersed into 20 mL of phosphoric acid (40%, v/v) and 0.5 mL of hydrogen peroxide (30%, v/v). Hierarchical Mn 3 O 4 /ZSM-5 zeolite, Mn 3 O 4 , and Mn(II) ions were added in different amount as catalysts, adjusted to 10% of substrate weight. The mixture was heated at 100 or 130 • C, and sampled after 0, 2, 4, 6, 8, and 10 h reaction, quenched in ice-water bucket.
The liquid product was then separated from the precipitate and quantitatively measured using high performance liquid chromatography (HPLC), performed at the Department of Biology, UI using HPLC Ultimate 3000 on a silica gel column (RezexTM ROA-Organic Acid H + (8%)) equipped with a pump, autosampler, Rs column compartment, diode array detector UV at 210 nm for formic acid [45]; 220 nm for levulinic acid [17]; 284 nm for 5-HMF [33]. The eluent (H 2 SO 4 0.0025 M) and distilled water were pumped at a flow rate of 1 mL/min.

Product Analysis
The yield (wt%) of products was calculated using the following Equation: % yield of product = total concentration mg L × solution volume (L) cellulose mass (mg) × 100% (3)

Conclusions
The main role of hierarchical Mn 3 O 4 /ZSM-5 zeolite catalyst in this delignified rice husk conversion was its hierarchical porous ZSM-5 zeolite as porous material which accelerated the degradation of cellulose into levulinic acid. Mn oxide species also assisted in the Fenton-like reaction to form hydroxyl groups. The Mn 3 O 4 / hierarchical ZSM-5 zeolite proved the best catalyst for the reaction, offering the highest percentage yield (wt %) of levulinic acid at 130 • C (39.75%) and 100 • C (27.60%) after 8 h as well as good selectivity toward levulinic acid than 5-HMF. It was also shown that the reaction temperature affects the yield (wt %) of levulinic acid.