Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study

international journal of hydrogen energy xxx (xxxx) xxx

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study Yoke Wang Cheng a, Kim Hoong Ng b, Su Shiung Lam c, Jun Wei Lim d, Suwimol Wongsakulphasatch e, Thongthai Witoon f, Chin Kui Cheng a,* a

Faculty of Chemical & Natural Resources Engineering, Lebuhraya Tun Razak, Universiti Malaysia Pahang, 26300, Gambang Kuantan, Pahang, Malaysia b Chemistry & Chemical Engineering, Xiamen University Malaysia, Jalan Sunsuria, Bandar Sunsuria, 43900, Sepang, Selangor, Malaysia c Eastern Corridor Renewable Energy Group (ECRE), School of Ocean Engineering, Universiti Malaysia Terengganu, 21030, Kuala Terengganu, Terengganu, Malaysia d Department of Fundamental and Applied Sciences, Universiti Teknologi PETRONAS, 32610, Seri Iskandar, Perak, Malaysia e Center of Ecomaterials and Cleaner Technology, Department of Chemical Engineering, Faculty of Engineering, King Mongkut's University of Technology North Bangkok, Bangkok 10800, Thailand f Center of Excellence on Petrochemical and Materials Technology, Department of Chemical Engineering, Faculty of Engineering, Kasetsart University, Bangkok 10900, Thailand

article info

abstract

Article history: Received 7 December 2018

In this work, the syngas production rate (FSyngas ) of LaNiO3-catalysed steam reforming of palm oil mill effluent (POME) was optimized with respect to POME flow rate (V_ POME ), catalyst

Received in revised form

weight (Wcat ), and particle size (dcat ). With a net acidity, the synthesized LaNiO3 catalysed

1 February 2019

POME steam reforming by cracking the bulky compounds and valorising simpler in-

Accepted 8 February 2019

termediates into syngas. The degradation efficiencies (XP ) were also evaluated by assessing

Available online xxx

wastewater parameters, viz. pH, chemical oxygen demand (COD), biochemical oxygen

Keywords:

reforming at 873 K, the liquid condensate has neutral pH and zero TSS. The parallel trend of

Palm oil mill effluent

FSyngas and XP verified syngas generation from degradation of POME's organics. At higher V_ POME (0.05e0.09 mL/min), greater POME partial pressure promoted its steam reforming and

demand (BOD5), total suspended solids (TSS), and colour intensity (A). After steam

Wastewater remediation Steam reforming

water gas shift, which enhanced catalytic performance. Beyond optimum V_ POME (0.09 mL/

Syngas production

min), coke-forming Boudouard reaction deteriorated catalytic activity. Catalytic performance was boosted for a longer residence time at higher Wcat (0.1e0.3 g); nonetheless, it was reduced by agglomerated catalyst when Wcat > 0.3 g. Finer LaNiO3 (dcat > 74 mm) with greater surface area to volume ratio exhibited better performance; however, ultrafine LaNiO3 (dcat < 74 mm) had poor performance because of occluded pores. Remarkably, optimized POME steam reforming over LaNiO3 (T ¼ 873 K, V_ POME ¼ 0.09 mL/min, Wcat ¼ 0.3 g, dcat ¼ 74e105 mm) has generated 132.47 mmoL/min of H2-rich syngas, whilst achieved 99.53% XCOD , 99.88% XA , 99.75% XBOD5 , and 100% XTSS . © 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

* Corresponding author. E-mail address: [email protected] (C.K. Cheng). https://doi.org/10.1016/j.ijhydene.2019.02.061 0360-3199/© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

2

international journal of hydrogen energy xxx (xxxx) xxx

Introduction Global palm oil production has skyrocketed from 11.86 million tons in 1991 to nearly 64 million tons in 2016 [1]. During this period, palm oil consumption was driven by its salubrious contents (antioxidant-rich and trans fat-free), broad oleochemical applications (soap, detergent, pharmaceuticals, and etc.), and also conversion into biodiesel [2e4]. However, this burgeoning agroindustry has resulted in deleterious threats to the hydrosphere (water), lithosphere (land), atmosphere (air), and ultimately biosphere (living things). The pollution issue stemmed from discharge of agro-residues, viz. palm oil fuel ash (POFA), palm press fibre (PPF), palm kernel shell (PKS), oil palm frond (OPF), oil palm trunk (OPT), empty fruit bunch (EFB), and palm oil mill effluent (POME). Excluding the POME liquor, the solid-form mill residues are either treated to landfills disposal or being incinerated. On the other hand, the acidic (pH of 3.4e5.2), hot (353e363 K) and brownish POME slurry is treated in a biological based ponding system before discharged into nearby hydrosphere in light of its high chemical oxygen demand (COD: 15,000e100,000 mg/L) and biochemical oxygen demand (BOD: 10,250e43,750 mg/L) [5]. Unfortunately, the ponding system is associated with numerous downsides, even though it requires low capital and operational cost to achieve 74e81% BOD reduction [6]. During wet milling process, POME is discharged at voluminous rates (2.5e3.75 tons POME/ton processed crude palm oil); hence, a huge area of the lithosphere is needed for creating the pond [7]. Furthermore, biogas (CH4 and CO2) emission from anaerobic biodegradation of POME is adding to the global warming [8]. According to the Malaysia Palm Oil Association (MPOA) [9], the performance of ponding system is inconsistent. This has resulted in a sluggish degradation performance, circa 66 days for entire process [8]. Recently, the application of hydrogen energy in fuel cells and internal combustion engines has garnered wide attention by virtue of its clean emission (only water and heat energy) [10]. By incorporating a carbon-neutral cycle, the valorisation of biomass into hydrogen either through thermochemical (gasification, pyrolysis, and liquefaction) or biological (dark fermentation and photofermentation) routes further intensified the endearment towards hydrogen energy [10e20]. Steam reforming of hydrocarbon (eg: natural gas, propane, liquefied petroleum gas, and naphtha feedstock) is the most wellestablished gasification process that was industrialized more than 83 years ago [21]. In the reviews authored by Huber et al. [22] and Guan et al. [23], they have envisaged the utilization of biomass waste as feedstock for renewable syngas production. To date, steam reforming of industrial effluents is very scarce in literature; only cheese whey and glycerol wastes have been reported [17,24]. To the best of our knowledge, the feasibility of using steam reforming to treat POME is relatively unknown. From our earlier thermodynamics study, the Gibbs free energy minimization approach has predicted the formation of hydrogen-rich syngas from POME steam reforming at 773e1173 K [25]. Through our recent investigations, POME (COD: ~27,000 mg/L and BOD5: ~17,000 mg/L) steam reforming catalysed by 20 wt%Ni/80 wt%Al2O3 catalyst has generated CO-

lean syngas. In addition, treated POME showed >90% COD degradation and BOD5 reduction [26,27]. In another independent study, raw POME (COD: ~70,000 mg/L, BOD5: ~11,000 mg/ L, and TSS: ~7700 mg/L) steam reforming over LaNiO3 catalyst at 773e1173 K had successfully harnessed renewable H2-rich syngas [28]. Based solely on syngas production rate, we have reported that the best POME steam reforming temperature over LaNiO3 catalyst was 873 K. As an extension of previous work, the current goal is optimizing syngas yield from POME steam reforming over LaNiO3 catalyst. Using one-factor-at-atime (OFAT) approach, the experiments were conducted by varying POME flow rate (V_ POME ), catalyst weight (Wcat ), and particle size (dcat ). Additionally, the degradation efficiencies (XP ) of POME steam reforming was determined to assess its effectiveness as a treatment approach.

Materials and methods POME sampling, preservation and characterization End-of-pipe raw POME was collected from a local palm oil mill located in Kuantan, Pahang. Upon collection, the bulk solids were screened off by one-time membrane filtration (average pore size of 11 mm) before kept at 277 K [28] until use. To assess wastewater quality, pretreated POME was measured for its pH, chemical oxygen demand (COD), 5-days biochemical oxygen demand (BOD5), total suspended solids (TSS), and absorbancebased colour intensity at 332 nm (A). The procedure of aforementioned analyses can be found elsewhere [28]. For characterization of POME's organics, the POME with high water content (96.33 wt%) was dehydrated by freeze drying at 193 K and 0.14 mbar [25]. Afterwards, the freeze-dried POME (organic contents of POME) was subjected to thermogravimetric analysis (TGA) and Fourier transform infrared spectroscopy (FTIR) to reveal its proximate contents (moisture, volatile matter, fixed carbon, and ash) and functional groups, respectively. By using a TGA analyser (Hitachi STA7200), non-isothermal weight loss profile of freeze-dried POME was recorded from 298 to 1273 K (ramping rate, x ¼ 10 K/min). Gas swapping from N2 to air was initiated at 873 K in line with proximate analysis standard (ASTM E1131) [29]. In an FTIR spectrometer (Perkin Elmer Spectrum 100), pelletized sample (mass ratio of sample:KBr ¼ 1:10) was irradiated by polychromatic infrared with wavenumber range of 4000e400 cm1. Since ash content represents inorganic residuals after complete combustion [30], two crucibles that prefilled with either pretreated POME or freeze-dried POME were heated in a muffle furnace at 1273 K for 5 h to determine its ash content on a wet and dry basis, respectively. Through this procedure, the value of dry basis ash content determined from proximate analysis via TGA could be validated.

Catalyst synthesis and characterization The LaNiO3 catalyst was synthesized via modified citrate solgel method [28]. In brief, the metal nitrate precursors (La(NO3)3$6H2O and Ni(NO3)2$6H2O) and chelating agent citric acid (C6H8O7) are dissolved in deionized water with a molar ratio of

Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

international journal of hydrogen energy xxx (xxxx) xxx

La3þ, Ni2þ, and C6H5O3 7 ions of 1:1:2. The mixture was agitated and heated at 353 K for 4 h to obtain a gel, which was later oven-dried at 403 K for 24 h before its calcination at 1123 K (x ¼ 5 K/min) for 4 h. Through crushing and sieving, the synthesized catalyst was partitioned into six particle size (dcat ) ranges, particularly <74 mm, 74e105 mm, 105e149 mm, 149e210 mm, 210e250 mm, and >250 mm (uncrushed). In a surface area and porosity analyser (Micromeritics ASAP 2020), fresh catalysts were degassed at 573 K for 4 h before analysed at 77 K. From N2 adsorption isotherms, their specific surface area was estimated through BrunauereEmmetteTeller (BET) method using data in relative pressure (P/P0) range of 0e0.30. For remaining characterization of fresh catalyst, fresh LaNiO3 (dcat ¼ 210e250 mm) was used by assuming no significant variation in other properties besides BET surface area. Special tubular quartz reactor (Thermo Scientific TPDRO 1100) designed for temperature programmed reduction (TPR) and desorption (TPD) was utilized to probe the reducibility, basicity, and acidity of fresh LaNiO3 through H2-TPR, CO2-TPD, and NH3-TPD, respectively. In H2-TPR, the catalyst was pretreated in N2 flow at 393 K for 2 h, cooled to ambient, and then heated to 1173 K (x ¼ 10 K/min) under 5% H2/N2 flow. For CO2or NH3-TPD, the catalyst was sparged with N2 at 313 K for 0.5 h, adsorbed with CO2 or NH3 at 393 K for 1 h, flushed with He for 0.5 h, cooled to ambient, and desorbed using He eluent by heating to 1173 K (x ¼ 10 K/min). To examine physicochemical changes after reaction, fresh and spent LaNiO3 catalysts were selectively analysed via X-ray photoelectron spectroscopy (XPS), temperature programmed oxidation (TPO), transmission electron microscopy (TEM), and scanning electron microscopy-energy dispersive X-ray analysis (SEM-EDX). With an XPS spectrometer (ULVAC-PHI Versaprobe II), the sample was irradiated by monochromatic Al Ka (hv ¼ 1486.6 eV) X-rays to facilitate XPS spectra acquisition, employing a background pressure of 9.8  104 Pa and passing energy of 117.40 eV (wide scan) or 29.35 eV (narrow scan). To quantify severity of carbon deposition, same TGA analyser was used to acquire TPO profiles of spent LaNiO3 from 298 to 1173 K (x ¼ 10 K/min) in a flow of compressed air. For visualization of morphological changes, 2D and 3D high resolution imaging technologies like TEM and SEM-EDX inspect the internal structure of catalysts. As TEM preparation, trace sample was dissolved in ethanol (99% purity) and sonicated for 1 h before the mixture was dropped on a Formvar/carbon coated copper grid. After air drying of grid, the sample was viewed in a transmission electron microscope

3

(Zeiss Libra 120) at an accelerating voltage of 120 kV. For SEMEDX, the sample that dispersed on a carbon conductive adhesive tape was sputter-coated with platinum before viewed in a scanning electron microscope (FEI Quanta 450) at an accelerating voltage of 10 kV. The auxiliary EDX analyser identified the elemental composition of samples from the emitting characteristic X-rays.

POME steam reforming The catalytic POME steam reforming at 873 K and 1 atm over LaNiO3 catalyst was further optimized for its syngas yield by manipulating POME flow rate (V_ POME ¼ 0.05e0.10 mL/min), catalyst weight (Wcat ¼ 0.1e0.5 g), and particle size (dcat ¼ <74 mm e >250 mm). For comparison purpose of V_ POME effect, blank runs were performed without catalyst and quartz wool. For a duration of 4 h, POME steam reforming was conducted in a stainless-steel fixed bed reactor (40 cm-long and 0.942-cm ID), with two replicates for every single run. Prior to reaction testing, accurately-weighed fresh LaNiO3 catalyst was transferred to quartz wool to serve as a catalyst bed. The catalyst was reduced in-situ by 50 mL/min of equimolar H2/N2 mixture at 1023 K for 1 h. Thereafter, the experimental setup was flushed with 50 mL/min of N2 for 0.5 h while cooling to 873 K. The pretreated POME was supplied by a syringe pump to the reactor, where it was vaporized in-situ and carried by a stream of N2 diluent to attain desired weighthourly-space-velocity (WHSV). In current study, the total feed rate of vaporized POME and N2 was fixed at 125 mL/min STP, therefore the WHSV (as defined in Equation (1)) was adjusted to the range of 15,000e75,000 mL/(g h) by altering catalyst weight.
WHSV ¼

QPOME þ QN2 Wcat

 
60 min 1h

(1)

where QPOME is the flow rate of vaporized POME (mL/min); QN2 is the flow rate of N2 (mL/min); and Wcat is the catalyst weight (g). A cold trap condenser connected to the reactor outlet would quench the product in an ice-water mixture. Noncondensable wet gas was desiccated through a drierite bed and hourly sampled as a gaseous product after the measurement of outlet gas flow rate. Gas chromatograph (Shimadzu GC 2014) was used for H2, CH4, CO2, and CO concentration determinations, whereby gas sample was eluted by He through three packed bed columns (Restek Rtx-

Fig. 1 e POME characterization e (a) TGA profile of freeze-dried POME and (b) FTIR spectrum of freeze-dried POME. Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

4

international journal of hydrogen energy xxx (xxxx) xxx

Table 1 e Functional groups of freeze-dried POME. No.

Wavenumber (cm1)

Functional group

a

Literature

Experimental

3436 (3708e3022)

v(O-H)

2922 (2990e2880) 2854 (2880e2800)

v(C-H)

2100 (2384e1866) 727 (883e671) 638 (671e404) 1740 (1770e1715)

v(C≡C) d(C-H) v(C¼O)

1633 (1696e1532)

v(C¼C)

1445 1258 1153 1080

v(C-O) & d(O-H) v(C-O) v(C-O)

1 2 3 4 5 6 7

Free hydroxyl (-OH) Intermolecular hydrogen-bonded OH (O-H) Hydroxyl group of carboxylic acid (-OH) Aliphatic methyl (-CH3) Acyclic methanediyl (-CH2-) Methoxy (-O-CH3) Monosubstituted alkyne (-C≡CH)

3670e3580 3550e3230 3300e2500 3000e2865 2870e2840 2880e2815 2150e2100 730e575

8 9 10 11 12

Carbonyl group of saturated aliphatic ketone (C¼O) Carbonyl group of carboxylic acid (C¼O) Isolated diene of alkene (C¼C) C¼C ring of 2,5-disubstituted furan (C¼C) Carboxyl group of carboxylic acid dimers (-COOH)

13 14

Saturated aliphatic secondary alcohols (-CHROH) Saturated aliphatic primary alcohols (-CH2OH)

1745e1715 1740e1700 1680e1620 1600e1570 1440e1395 1320e1210 1150e1075 1090e1000

a b c

FTIR vibrationc b

(1495e1313) (1303e1207) (1207e1129) (1129e932)

FTIR characteristic group frequencies compiled by Socrates [35]. Experimental wavenumber expressed as peak position (wavenumber range). Vibration modes: stretching (v) and bending (d).

Table 2 e BET surface area of fresh LaNiO3 from different particle size ranges. Particle size ranges (mm) <74 74e105 105e149 149e210 210e250 >250

BET surface area (m2/g) 5.71 24.78 21.32 18.51 15.44 13.11

Results and discussion POME characterization

1, Restek Rt-Q-BOND, and Restek Rt-Msieve 5A) that operated at 333 K. The gaseous product was assessed by the composition (yi ) and production rate (Fi ) of its individual gas species (Equations (2) e (3)). Xi ð%Þ yi ð%Þ ¼ P Xi ð%Þ

(2)

where yi and Xi are the composition (%) and concentration (%) of gas species i, respectively. 

6  P 1L 10 mmol RT 1000 mL mol


Fi ¼ Qo Xi

(3)

where Fi is the production rate of gas species i (mmol/min), Qo is the outlet gas flow rate (mL/min), P is  the ambient  pressure L$atm , and T is (1 atm), R is the universal gas constant 0:082057 ðK$molÞ the standard temperature (298 K). To evaluate the degradation efficiencies, condensed postreaction liquid product was collected and characterized for its pH, COD, BOD5, TSS, and colour intensity (A). Except for pH, the degradation efficiencies of wastewater parameters ðXP Þ is computed using Equation (4). XP ð%Þ ¼


 p 1  100% p0

where XP is the degradation efficiencies of wastewater parameters P (COD, BOD5, TSS, or A) while p0 and p are the values of wastewater parameters of POME before and after the reaction.

(4)

The raw POME was acidic (pH ¼ 5), opaque (colour intensity of 1.945), and pollutant-laden (COD of ~70,000 mg/L, BOD5 of ~11,000 mg/L, and TSS of ~7700 mg/L). The TGA profile in Fig. 1(a) depicts the thermal (298e873 K) and oxidativethermal (873e1273 K) degradation behaviours of freeze-dried POME (POME's organics). The TGA profile unveils the proximate contents of POME's organics, viz. 6.13 wt% moisture, 79.57 wt% volatile matter, 7.66 wt% fixed carbon, and 6.64 wt% ash. Meanwhile, the weight changes of pretreated and freezedried POME after 5 h combustion at 1273 K (cf. Fig. S1 in supplementary data) revealed the ash content of POME as 1178.75 mg/L, 0.11 wt% wet basis, or 5.98 wt% dry basis. For dry basis ash content, the value difference between both approaches was insignificant (0.66 wt%), thus validating the accuracy of proximate analysis for ash content determination. Interestingly, the POME's organics possessed comparable ash content with other oil palm wastes, such as OPF (3.40 wt%), OPT (5.27 wt%), EFB (6.87 wt%), and PKS (4.94 wt%) [13,31e33]. Purwandari et al. [34] had reported 75.93% ash removal from EFB after its pretreatment with concentrated N-methylmorpholine-N-oxide, attributed to the leaching of soluble metal ions. From economic and environmental viewpoints, ash leaching of POME's organics is impractical as drying incurs additional cost and leaching inflicts another wastewater source. Nevertheless, the reforming feedstock here was POME instead of POME's organics, wherefore the quantification of its

Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

international journal of hydrogen energy xxx (xxxx) xxx

ash content on wet basis (0.11 wt%) was more constructive. These results suggest high feasibility of POME steam reforming by virtue of appreciable volatile matters (self-vaporized at elevated temperature), inconsiderable fixed carbon (easily oxidized by superheated steam), and negligible ash content (unlikely to cause blockage in the reactor) of POME.

5

The FTIR spectrum of freeze-dried POME shown in Fig. 1(b) is almost similar with that of oven-dried POME in the report by Ng et al. [26] albeit different raw POME sources. Based on previous GC-MS results of freeze-dried POME [25], functional groups of POME's organics were assigned using FTIR characteristic group frequencies compiled by Socrates [35]. Due to

Fig. 2 e Fresh LaNiO3 characterization e (a) H2-TPR profile, (b) CO2-TPD profile, (c) NH3-TPD profile, and (d) SEM-EDX image at 30 k£ magnification.

Fig. 3 e TEM images of LaNiO3 catalysts at 50 k£ magnification - (a) fresh LaNiO3, (b) spent LaNiO3 (773 K), (c) spent LaNiO3 (973 K), and (d) spent LaNiO3 (1173 K). Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

6

international journal of hydrogen energy xxx (xxxx) xxx

the complexity of POME, a total of 15 functional groups had been identified as tabulated in Table 1. No functional groups of organonitrogen and organosulfur were discovered, owing to low N and S contents of POME [25]. The hydroxyl group that found as a broad peak at 3436 cm1 was likely superimposed by free OH, intermolecular hydrogen-bonded OH, and carboxylic acid OH, thereby signifying the possibility of alcohols, carboxylic acids, and other OH-containing compounds (i.e. phenol) in POME. The assignments of saturated aliphatic primary and secondary alcohols at 1080 cm1 and 1153 cm1, respectively, had affirmed the existence of alcohols, such as 3furanmethanol, 2-nonadecanol, 2-propyl-tetrahydropyran-3ol, and n-heptadecanol [25]. Besides the hydroxyl group, the presence of carboxylic acids (n-hexadecenoic acid, cyclopentaneundecanoic acid, and undecylenic acid) was corroborated through the discovery of corresponding carboxyl (-COOH) and carbonyl (C¼O) groups [25]. Since sterilizer condensate constitutes part of POME, high proportion of n-hexadecenoic (palmitic) acid was anticipated due to unavoidable palm oil leaching during fruit sterilization. Despite -CH3 and -CH2- groups are common for organic compounds, both stringent criteria of aliphatic and acyclic were met by ketones (3-methyl-2-butanone and 4-hydroxy-5methyl-2-hexanone) that found in POME, which also concurs with assigned carbonyl group (C¼O) of saturated aliphatic ketone [25]. Meanwhile, the revelation of functional groups like methoxy (-O-CH3), monosubstituted alkyne (-C≡CH), isolated diene of alkene (C¼C), and C¼C ring of 2,5-disubstituted furan (C¼C) eventually proved the existence of methyl 9oxononanoate, 1-ethoxy-but-1-ene-3-yne, squalene, and 5methylfurfural, respectively [25]. Based on the peak integrals, FTIR discovers that POME's organics mainly composed of carboxylic acids, phenol, and alcohols, with minor quantities of ketones, ester, alkyne, alkene, and furan.

Fresh catalyst characterization Table 2 summarizes the BET surface area of fresh LaNiO3 from different particle size (dcat ) ranges. Basically, fresh catalyst with smaller dcat has greater BET surface area; nevertheless, previous statement only applicable for dcat range of 74e105 mm to >250 mm. In current study, fresh LaNiO3 (74e105 mm) recorded greatest BET surface area (24.78 m2/g). With further

dcat reduction to below 74 mm, a drastic diminution of BET surface area to 5.71 m2/g was observed as over-crushing possibly instigated pore occlusion of LaNiO3. As the pores are occluded, fresh LaNiO3 (<74 mm) exhibits a lower effective BET surface area. Deconvoluted H2-TPR profile of fresh LaNiO3 in Fig. 2(a) allowed a comprehensive inspection of its stepwise reduction mechanism. From the XRD pattern in Fig. S2(a) [28], the fresh LaNiO3 also contained trace phases of lanthanum oxide (La2O3) and nickel (II) oxide (NiO). Mile et al. [36] reported the reduction of bulk unsupported NiO by 5% H2/N2 produced two TPR peaks; a small peak at 523 K (Ni2O3 þ H2 / 2NiO þ H2O) and a relatively intense peak at 673 K (NiO þ H2 / Ni þ H2O). The TPR peaks at 531 K and 655 K can be ascribed to the reduction of Ni3þ of trace Ni2O3 (undetected by XRD presumably due to small crystallite size) into Ni2þ and the reduction of Ni2þ of NiO into Ni0, respectively. From past LaNiO3 reducibility studies, two-step reduction of LaNiO3 (2LaNiO3 þ H2 / La2Ni2O5 þ H2O and La2Ni2O5 þ 2H2 / La2O3 þ 2Ni þ 2H2O) characterized by two successive H2-TPR peaks (area ratio z 1:2 and temperature interval ¼ 128e210 K) regardless of catalyst synthesis route [37e41]. Since the peak integral ratio of 757 Ke928 K was approximate ½, the TPR peak at 757 K ascribed to the reduction of Ni3þ of LaNiO3 to Ni2þ of La2Ni2O5 and the peak at 928 K indicated its subsequent reduction to Ni0. Due to stronger active metal-support (Ni-La2O3) interaction of synthesized LaNiO3, these two characteristic TPR peaks were shifted to higher temperature region as compared to literature range of 585e873 K (refer Table S1) [42,43]. In conformity with XRD result of reduced LaNiO3 (cf. Fig. S2(a)) [28], the H2-reduction at 1023 K was capable to convert fresh LaNiO3 into well dispersed Ni active metal on a La2O3 support (2LaNiO3 þ 3H2 / La2O3 þ 2Ni þ 3H2O). The surface basicity and acidity of fresh LaNiO3 were analysed from CO2- and NH3-TPD profiles. Theoretically, the desorption ease of probe molecules measured by desorption temperature could serves as a strength indicator of its complementary active sites [44]. Daza et al. [45] classified active sites into weak sites (<573 K), moderate sites (573e873 K), and strong sites (>873 K). Sugunan and Meera [46] studied the acidbase properties of selected perovskites and their constituent metal oxides via maximum Hammett acidity constant (H0, max), where strong basic sites and weak acid sites

Fig. 4 e XPS spectra of LaNiO3 catalysts - (a) survey spectra of LaNiO3 catalysts and (b) La 3d þ Ni 2p3/2, (c) Ni 2p1/2, and (d) O 1s narrow scan spectra of fresh LaNiO3. Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

7

international journal of hydrogen energy xxx (xxxx) xxx

anticipated in solids with high H0, max and vice versa. The reported H0, max values of LaNiO3, NiO, and La2O3 were 4.80, 5.22, and 9.80 [46], so their basicity can be ranked as La2O3>NiO > LaNiO3 and their acidity followed the reverse orders. From the XRD diffractogram of fresh LaNiO3 (cf. Fig. S2(a)), overlapping of LaNiO3, La2O3, and NiO phases was witnessed, which probably forms mixture phases of different basicity and acidity. Coincidentally, the devolution of both TPD profiles give rise to seven individual peaks. On grounds of acidity/basicity and XRD result, potential phases were speculated by presuming dissimilar environments for different peaks. Detailed information such as peak temperature, strength and quantity of basic/acid sites, and potential phase were well presented in CO2-TPD (Fig. 2(b)) and NH3-TPD (Fig. 2(c)) profiles, respectively. Overall, the synthesized LaNiO3 was a net-acidic catalyst, which exhibited a greater surface acidity (750.61 mmoL/g) as compared to its surface basicity (632.35 mmoL/g). As observed in SEM-EDX and TEM images (cf. Fig. 2(d) and Fig. 3(a)), the modified citrate sol-gel route produced a LaNiO3 catalyst with irregular sized and strip-like shaped particles. The EDX analysis confirmed that fresh LaNiO3 was impuritiesfree, with its elemental composition equals to 19.76 mol% La, 19.30 mol% Ni, and 60.94 mol% O. Almost concurs with desired molecular formula of LaNiO3, the fresh LaNiO3 had a bulk empirical formula of La1.024NiO3.157. From XPS survey spectra of fresh LaNiO3 in Fig. 4(a), non-interfering peaks with greatest peak area were selected for XPS quantification of elements [47]. During XPS quantification, both La 3d and Ni 2p transitions were omitted because complicated overlapping of La 3d3/ 2 and Ni 2p3/2 transitions in LaNiO3 was reported [48e50]. The resulting elemental composition of fresh LaNiO3 was 24.91 mol% La (from La 4d), 19.23 mol% Ni (from Ni 3p), and 55.86 mol% O (from O 1s), which corresponds to a surface empirical formula of La1.30NiO2.90. The consolidation of EDX and XPS results informed a uniform elemental distribution of La, Ni, and O in fresh LaNiO3 owing to similar bulk and surface empirical formula. To elucidate chemical states of elements, narrow scan spectra (La 3d þ Ni 2p3/2, Ni 2p1/2, and O 1s) of fresh LaNiO3 were deconvoluted as shown in Fig. 4(b)-(d). Through deconvolution, the underlying Ni 2p3/2 transition in Fig. 4(b) was located by resolving the superimposed La 3d3/2 and Ni 2p3/2 region into six chemical states. The XPS peaks at 833.57 eV and 850.32 eV can be assigned to La2O3 species, accompanied by their satellite peaks at 837.15 eV and 853.99 eV [48]. According

to Wu et al. [51] and Wei et al. [50], the XPS peaks at 854.38 eV and 855.47 eV can be linked to Ni2þ and Ni3þ oxidation states. As illustrated in Fig. 4(c), the XPS peak at 872.10 eV was NiO species [52] and the remaining peaks (861.40 eV, 863.61 eV, and 866.08 eV) were Ni 2p3/2 satellite peaks [48]. Meanwhile, deconvoluted O 1s transition in Fig. 4(d) has two peaks; the 527.99 eV and 530.06 eV peaks correspond to lattice oxygen (O2) of La2O3 and lattice oxygen (O) of NiO [48,51,53].

Validation of optimum temperature From our preliminary study [28], the temperature effect of POME steam reforming over LaNiO3 was investigated from 573 to 1173 K (V_ POME ¼ 0.08 mL/min, Wcat ¼ 0.3 g, and dcat ¼ 210e250 mm) at a WHSV of 25,000 mL/(g$h). Table S2 shows the variation of syngas production rate (FSyngas ) with temperature, which increased from 573 to 873 K, declined significantly at 973 K, but rose again from 973 to 1173 K. This unexpected FSyngas trend was correlated with the catalytic behaviour, wherein LaNiO3 severely coked at 973 K and its coking deactivation was gradually mitigated at higher temperature (>973 K). Despite this, high reforming temperatures (>973 K) tend to promote progressive sintering of LaNiO3 into dilanthanum nickel tetroxide (La2NiO4). To further validate 873 K as optimum temperature, physicochemical changes of spent LaNiO3 were scrutinized through TEM and XPS analysis for better understanding regards its deactivation mechanism. TEM images of spent LaNiO3 catalysts (773 K, 973 K, and 1173 K) are depicted in Fig. 3(b)-(d). On account of precedent TEM images, the spent LaNiO3 (773 K) comprised of lanthanum oxycarbonate (La2O2CO3), lanthanum oxide (La2O3), nickel (Ni), and carbon (C). The clusters of strip-like particles (smaller than LaNiO3 in size) in a pseudo-rectangular envelope are La2O2CO3 phase [54e57], in line with our XRD analysis (cf. Fig. S2(b)) [28]. Moreover, the rod-like shaped particles, spherical particles, and dark spots are attributed to La2O3 phase [58,59], Ni [41,60,61], and deposited carbon [26], respectively. For spent LaNiO3 (973 K), widespread dark spots and enlarged strip-like shaped particles are noticed, engendered by severe coking deactivation and sintering of LaNiO3 into La2NiO4 (refer XRD pattern in Fig. S2(b)) [28]. As compared to spent LaNiO3 (973 K), the spent LaNiO3 (1173 K) has reduced dark spots and more conspicuous sintering. From the XPS survey spectra in Fig. 4(a), a C 1s peak is observed at approximately 284.8 eV for all the LaNiO3 catalysts. However, the small C 1s peak of fresh LaNiO3 was most likely

Table 3 e Partial pressures of vaporized POME in the study of POME flow rate effect. Flow rate (mL/min) POME (V_ POME ) 0.05 0.06 0.07 0.08 0.09 0.10 a

Partial pressure (atm)

Vaporized POME (QPOME ) 59.22 71.07 82.91 94.76 106.60 118.45

a

Feed ratio of vaporized POME: N2

N2 (QN2 )

Vaporized POME (PPOME )

N2 (PN2 )

65.78 53.93 42.09 30.24 18.40 6.55

0.47 0.57 0.66 0.76 0.85 0.95

0.53 0.43 0.34 0.24 0.15 0.05

0.90 1.32 1.97 3.13 5.79 18.08

QPOME ¼ rPOME V_ POME RT=ðMWPOME PÞ, where rPOME ¼ 985 kg/m.3 and MWPOME ¼ 18.639 g/mol [28].

Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

8

international journal of hydrogen energy xxx (xxxx) xxx

Fig. 5 e Syngas properties and degradation efficiencies of non-catalytic and catalytic POME steam reforming with respect to POME flow rate e (a) syngas production rate, (b) syngas composition, (c) syngas ratio, (d) COD removal, (e) decolourization, and (f) BOD5 removal.

due to deposited adventitious carbon from the atmosphere [62e64]. After POME steam reforming, the intensified C 1s peak confirmed carbon deposition on spent LaNiO3 catalysts; particularly, the intensity of C 1s peak in spent LaNiO3 can be ranked as 973 K > 773 K > 1173 K. From this observation, progressive coke formation on LaNiO3 from 773 to 973 K and mitigation of its coking deactivation beyond 973 K were inferred. Plausible reactions that responsible for coke suppression are sequential carbonatation-decarbonatation of La2O3 supand port ðLa2 O3 þ CO2 /La2 O2 CO3 La2 O2 CO3 þ C/La2 O3 þ 2 COÞ and char gasification (C þ H2 O/CO þ H2 ) [25,28,65]. Conclusively, integrated TEM and XPS analysis of spent LaNiO3 strengthened our previous finding [28] by assuring that a reforming temperature >973 K could lessen coking of LaNiO3 with a drawback, viz. accelerated sintering into a less favourable La2NiO4. The doubt of temperature influence on catalytic performance was successfully resolved; henceforth, subsequent POME steam reforming was carried out at 873 K and 1 atm.

Catalytic performance evaluation Syngas production

To study the effect of POME flow rate (V_ POME ), the experiments (T ¼ 873 K, Wcat ¼ 0.3 g, and dcat ¼ 210e250 mm) were conducted at a WHSV of 25,000 mL/(g h) by varying V_ POME . The partial

Table 4 e Weight-hourly-space-velocity (WHSV) in the study of catalyst weight effect. Catalyst weight, Wcat (g) 0.1 0.2 0.3 0.4 0.5

Weight-hourly-space-velocity, WHSV [mL/(g$h)] 75,000 37,500 25,000 18,750 15,000

Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

international journal of hydrogen energy xxx (xxxx) xxx

pressure of vaporized POME (PPOME ) used was summarized in Table 3, specifically, the V_ POME is directly proportional to PPOME . For non-catalytic runs, its yi and Fi profiles obtained via identical condition are illustrated in Figs. S3(a)e(b) to investigate the V_ POME effect. As visualized in Fig. S3(a), the yH2 and yCO2 nearly invariant whereas yCH4 and yCO steadily increased from 0.05 to 0.09 mL/min; nonetheless, yH2 and yCO greatly augmented with declining yCO2 and yCH4 from 0.09 to 0.10 mL/ min. Most likely, POME steam reforming (Equation (5)) outweighed its thermal decomposition (Equation (6)) when PPOME > 0.85 atm. Theoretically, higher V_ POME produced more gaseous products; notwithstanding, the left-skewed Fi profiles with respect to V_ POME (Fig. S3(b)) showed 0.08 mL/min as the optimum POME feed rate of non-catalytic run.   y Cx Hy Oz þ ðx  zÞH2 O/ x þ  z H2 þ x CO 2

(5)

Cx Hy Oz /H2 þ CO2 þ CH4 þ CO þ Ca Hb Oc þ C

(6)

D

9

In relation to V_ POME variation, the yi and Fi profiles of catalytic POME steam reforming are presented in Figs. S3(c)e(d). Compared with non-catalytic run, the gaseous product of catalytic run generally exhibited higher yH2 and lower yCO2 , yCH4 and yCO within stipulated V_ POME range. After the incorporation of LaNiO3 catalyst, thermal decomposition of POME was superseded by simultaneous POME steam reforming and water gas shift reaction (Equation (7)) owing to its high steam-to carbon (S/C) ratio, viz. 39.91 [25]. It can be inferred that acid sites of LaNiO3 catalysed the cracking of bulky organic compounds (i.e. carboxylic acids, phenol, alcohols, ketones, and esters) before steam reforming [44,66], leading to easier valorisation of these cracked intermediate compounds (i.e. shorter carboxylic acids, ketones, alkenes, alkanes, and benzene) into syngas [67e69]. Withal, the net acidity of LaNiO3 conceivably assists the adsorption of Lewis base molecules (H2O and CO), subsequently stimulating water gas shift reaction.

Fig. 6 e Syngas properties and degradation efficiencies of catalytic POME steam reforming with respect to catalyst weight e (a) syngas production rate, (b) syngas composition & syngas ratio, (c) COD removal, (d) decolourization, and (e) BOD5 removal.

Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

10

CO þ H2 O/H2 þ CO2

international journal of hydrogen energy xxx (xxxx) xxx

(7)

Transcending from 0.05 to 0.09 mL/min, higher PPOME escalated the chemisorption rate of POME vapor on active sites (Ni) of LaNiO3 and thereby enhanced the rate of POME steam reforming, supported by a slight increment of yH2 and yCO and slight decrement of yCH4 . Fresh LaNiO3 also possesses basic sites that capture CO2 for methane dry reforming (Equation (8)) [70,71]. However, yCO2 was relatively stable for a V_ POME range of 0.05e0.09 mL/min, possibly alluded to the compensation of CO2 loss via water gas shift. Seemingly, the water gas shift gradually diminished at higher V_ POME , conjectured from increasing yCO and decreasing yCO2 trends from 0.05 to 0.09 mL/min. When a V_ POME of 0.10 mL/min was used, the yH2 further raised with intensified POME steam reforming, surprisingly with a drop in yCO . As signified by the drastic drop of yCO , Boudouard reaction (Equation (9)), an inevitable cokeforming side reaction in most reforming processes became conspicuous when PPOME > 0.85 atm. The increment of V_ POME from 0.09 to 0.10 mL/min induced a slight increase of yCH4 , indicating the prevailing effect of carbon methanation

(Equation (10)) over methane dry and steam reforming (Equation (8) and (11)). Additionally, this marginal yCH4 increment rules out the possibility of coke formation via methane thermal decomposition (Equation (12)), which is thermodynamically feasible at 873 K [25]. Dealing with higher optimum V_ POME (0.09 mL/min), the catalytic run offers higher steamreform capability than non-catalytic run, as evidenced in Fig. S3(d). CH4 þ CO2 /2H2 þ 2CO

(8)

2CO/CO2 þ C

(9)

C þ 2H2 /CH4

(10)

CH4 þ H2 O/CO þ 3H2

(11)

CH4 /C þ 2H2

(12)

The FSyngas profile of non-catalytic and catalytic runs (Fig. 5(a)) was alike, both displayed a rising trend with V_ POME increment followed by an appreciable decline beyond

Fig. 7 e Syngas properties and degradation efficiencies of catalytic POME steam reforming with respect to particle size e (a) syngas production rate, (b) syngas composition & syngas ratio, (c) COD removal, (d) decolourization, and (e) BOD5 removal. Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

international journal of hydrogen energy xxx (xxxx) xxx

optimum V_ POME . Upon introduction of LaNiO3 catalyst, the FSyngas greatly improved from 4.41 to 13.12 mmoL/min to 42.45e106.14 mmoL/min due to its catalytic effect. Fig. 5(b)(c) illustrates the ySyngas and syngas ratio (H2:CO) profiles of both systems in variation with V_ POME . For non-catalytic run, its fluctuating ySyngas from 0.05 to 0.09 mL/min indicates competition between POME steam reforming and thermal decomposition while the uptrend from 0.09 to 0.10 mL/min marked the surpassing of thermal decomposition by POME steam reforming. In contrast, the ySyngas of catalytic run steadily increased from 0.05 to 0.10 mL/min since LaNiO3

11

selectively catalysed POME steam reforming. By increasing V_ POME , H2:CO ratio of non-catalytic run progressively reduced with higher CO formation from more favoured POME steam reforming at higher PPOME . The syngas of catalytic run had an exceptional high H2:CO ratio (46.86e999.95) because of water gas shift, where its depreciated effect caused the downward trend in Fig. 5(c). At 0.10 mL/min, notable CO-consuming Boudouard reaction prompted a slight increase of its H2:CO ratio. Thereafter, the catalytic performance of LaNiO3 was evaluated at the optimum V_ POME of 0.09 mL/min.

Fig. 8 e SEM image (30 k£ magnification) and TPO profile of spent LaNiO3 catalysts with respect to POME flow rate e (a) & (b) 0.06 mL/min, (c) & (d) 0.08 mL/min, and (e) & (f) 0.10 mL/min. Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

12

international journal of hydrogen energy xxx (xxxx) xxx

Table 5 e Weight percent of deposited minerals on spent LaNiO3 as detected by EDX analysis. Catalyst 0.06 mL/min 0.08 mL/min 0.10 mL/min 0.1 g 0.3 g 0.5 g 74e105 mm 149e210 mm >250 mm

Weight percent (wt %) Potassium (K) 1.17 0.83 0.23 0.25 0.52 0.11 0.85 3.52 3.54

Chlorine (Cl) 0.42 0.05 0 0.06 0.09 0 0.05 0.40 2.99

Sulphur (S) 0.14 0 0 0 0 0 0 0 0.12

Magnesium (Mg) 0.61 0.16 0 0 0 0 0 0.05 0.08

Total 2.34 1.04 0.23 0.31 0.61 0.11 0.90 3.97 6.73

Fig. 9 e SEM image (30 k£ magnification) and TPO profile of spent LaNiO3 catalysts with respect to catalyst weight e (a) & (b) 0.1 g, (c) & (d) 0.3 g, and (e) & (f) 0.5 g. Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

international journal of hydrogen energy xxx (xxxx) xxx

To explore the influence of catalyst weight (Wcat ), the experiments (T ¼ 873 K, V_ POME of 0.09 mL/min, and dcat ¼ 210e250 mm) were carried out by manipulating WHSV through adjustment of Wcat . As tabulated in Table 4, a higher Wcat corresponds to a lower WHSV that provides a greater residence time for reactants. Noticeably from Fig. S3(e), the addition of 0.1 g LaNiO3 caused an appreciable yH2 increment with a great reduction of yCO2 , yCH4 , and yCO . The finding ultimately evinced the catalytic role of LaNiO3 in simultaneous POME steam reforming and water gas shift reaction. Accompanied by a slight drop of yCO2 , yCH4 , and yCO , the yH2 continue to rise slightly with increasing Wcat from 0.1 to 0.3 g. Hence, for

13

a WHSV range of 25,000e75,000 mL/(g$h), lower WHSV facilitated both POME steam reforming and water gas shift owing to longer residence time. As the Wcat further raised to 0.5 g, both yH2 and yCO2 marginally inclined with concomitant yCO reduction while the yCH4 almost remained unchanged. Evidently, if the WHSV<25,000 mL/(g$h), only water gas shift was further promoted by a prolonged residence time. With respect to Wcat , the Fi , FSyngas , and ySyngas profiles depicted in Fig. S3(f), Fig. 6(a), and Fig. 6(b) could be dissected into two parts. Explicitly, their increments from 0.1 to 0.3 g associated with improved catalytic reaction at lower WHSV while their decrements beyond 0.3 g possibly hinted at the occurrence of

Fig. 10 e SEM image (30 k£ magnification) and TPO profile of spent LaNiO3 catalysts with respect to particle size e (a) & (b) 74e105 mm, (c) & (d) 149e210 mm, and (e) & (f) > 250 mm. Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

14

international journal of hydrogen energy xxx (xxxx) xxx

catalyst agglomeration at high catalyst loading. Since lower WHSV facilitates CO-consuming water gas shift, the H2:CO ratio (Fig. 6(b)) of syngas surged exponentially with increasing Wcat from 0.1 to 0.5 g. The ever-changing ySyngas and H2:CO ratio profiles with WHSV variation also reflects a significant external diffusion resistance in POME steam reforming over LaNiO3. At the optimum Wcat of 0.3 g, the syngas with ySyngas of 73.43% and H2:CO ratio of 46.86 was generated at a FSyngas of 106.14 mmoL/min. Thereupon, the particle size (dcat ) effect of LaNiO3 in POME steam reforming was studied with several aims: (1) to assess the performance of uncrushed LaNiO3, (2) to determine the required extent of crushing, (3) to evaluate the significance of internal diffusion resistance, and (4) to ascertain its optimum dcat for highest catalytic performance. For future process application, mechanical crushing operation in scale-up production of LaNiO3 might be energy-intensive. Poor catalytic performance could be predicted from either over-crushing or significant internal diffusion resistance; explicitly, the former induces formation of occluded pores with lower effective surface area whereas the latter creates difficulty for reactants on approaching active sites within the pores [72]. Ideally, finer LaNiO3 catalyst should display a higher catalytic activity ascribed to its larger surface area to volume (SA/V) ratio. To elucidate the impact of particle size, the experiments (T ¼ 873 K, V_ POME of 0.09 mL/min, and Wcat ¼ 0.3 g) were performed using fresh LaNiO3 from different particle sizes (dcat ) ranges. For dcat range of 74e105 mm to >250 mm, all the yi shown in Fig. S3(g) are invariant towards different dcat ; thus, negligible diffusion resistance. Nonetheless, the yH2 substantially dropped with both yCO2 and yCH4 when dcat was further reduced to <74 mm. If dcat < 74 mm, thermal decomposition may have become more prominent due to the loss of catalytic surface area. As depicted in Fig. S3(h), Fi increased with decreasing dcat from >250 mm to 74e105 mm because a greater SA/V ratio improves the catalytic activity of LaNiO3. A higher SA/V ratio of LaNiO3 increases the exposure of active sites towards impinging POME vapor, which ultimately hasten sequential adsorption-surface reaction-desorption to a faster rate. Howbeit, when dcat < 74 mm, deteriorated POME steam reforming and water gas shift caused noteworthy thermal decomposition, which subsequently provoked FCH4 increment with reduction of FH2 , FCO2 and FCO . Fig. 7(a)-(b) illustrate the FSyngas , ySyngas , and H2:CO ratio profiles as a function of dcat : In response to decreasing dcat , the increment and decrement of FSyngas plausibly related to higher SA/V ratio and clogged pores with reduced catalytic area; respectively. Highest FSyngas (132.47 mmoL/min) was obtained at optimum dcat (74e105 mm), which critically increased by 30.62% as compared to that of uncrushed LaNiO3 (dcat > 250 mm). Because of negligible internal diffusion resistance, the ySyngas and H2:CO ratio almost invariant in the dcat range of 74e105 mm to >250 mm. When dcat < 74 mm, occluded pores of LaNiO3 induced conspicuous thermal decomposition that reduced its ySyngas and H2:CO ratio.

Degradation efficiencies After POME steam reforming treatment, all the liquid condensates have neutral pH (pH ¼ 7) and zero TSS concentration (XTSS ¼ 100%) regardless of V_ POME , Wcat , and dcat employed.

Remarkably, non-catalytic POME steam reforming also achieved the same performance for neutralization and complete TSS removal. The neutralization of acidic POME was related to the thermal destruction of carboxylic acids (i.e. n-hexadecenoic acid, cyclopentaneundecanoic acid, and undecylenic acid). Complete TSS removal indicates that the suspended solids were successfully valorised into hydrogen-rich syngas. Despite this, all liquid condensates have non-zero concentrations of COD and BOD5, informing that the organic pollutants exist as dissolved solids. After vertically settling for one day, no immiscible layers were observed in all the liquid condensates, thus reinforcing our previous presumption. Unlike pH and TSS, other wastewater quality parameters (COD, A, and BOD5) of liquid condensates were influenced by the changes of V_ POME , Wcat , and dcat . To illustrate these influences, line graphs and bar charts were used to present the degradation efficiencies (XCOD , XA , and XBOD5 ) and final values of wastewater parameters (COD, A, and BOD5), respectively. Fig. 5 (d)-(f) depict the degradation profiles of non-catalytic and catalytic POME steam reforming with changing V_ POME . For non-catalytic run, its XCOD , XA , and XBOD5 substantially declined from 91.69%, 98.05% and 98.26%e85.56%, 89.79%, and 97.50% with increasing V_ POME from 0.05 to 0.10 mL/min. This deterioration of degradation efficiencies was plausibly related to increasing number of unreformed/partially reformed organic pollutants at higher V_ POME . In the absence of LaNiO3 catalyst, the high degradation efficiencies (XCOD , XA , and XBOD5 > 80%) achieved hints at high susceptibility of chromophoric organic pollutants to thermal degradation. Ostensibly, non-catalytic run demonstrated acceptable degradation efficiencies; however, even with lowest V_ POME (0.05 mL/min), treated POME was still highly contaminated (COD ¼ 5817 mg/L and BOD5 ¼ 191 mg/L). Regarding V_ POME effect on catalytic run, its XCOD , XA , and XBOD5 profiles followed its FSyngas trend in Fig. 5(a). With increasing V_ POME , the degradation efficiencies were enhanced by more prevailing POME steam reforming but offset by Boudouard reaction at 0.10 mL/min. Utilizing optimum V_ POME (0.09 mL/min), the final COD and BOD5 concentrations of liquid condensate from catalytic run were 544 mg/L and 86 mg/L. For Wcat influence, its XCOD , XA , and XBOD5 in Fig. 6(c) e (e) exhibited a similar trend with its FSyngas in Fig. 6(a). If Wcat  0.3 g, greater degradation efficiencies attained at higher Wcat because a lower WHSV grants a longer residence time for POME vapor. Beyond a Wcat of 0.3 g, undesired catalyst agglomeration occurs and reduces its catalytic degradation. In analogous with V_ POME and Wcat , the XCOD , XA , and XBOD5 trends of changing dcat in Fig. 7(c) e (e) were in conformity with its FSyngas profile in Fig. 7(a). Interestingly, the mutual relation between syngas generation and degradation efficiencies confirm that POME steam reforming abated organic pollutants by ameliorating them into syngas. With a greater SA/V ratio, finer LaNiO3 particles were more effective for promoting degradation but over-crushing might provoke pore occlusion that associated with poor degradation. Using optimum (74e105 mm), POME steam reforming accomplished 99.53% XCOD , 99.88% XA , and 99.75% XBOD5 . Under optimized conditions, detrimental POME was successfully purified to a more environmental-benign liquid (COD ¼ 326 mg/L and BOD5 ¼ 27 mg/L).

Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

international journal of hydrogen energy xxx (xxxx) xxx

Spent catalyst characterization In the previous sections, critical findings such as the occurrence of Boudouard reaction and catalyst agglomeration were postulated based on the interpretation of gaseous profiles. After manual removal of quartz wool and agglomerated biochar, the powdered spent LaNiO3 catalysts were analysed via SEM-EDX and TPO to corroborate our preceding hypothesis regards deactivation mechanism of LaNiO3. Evidently, all the spent LaNiO3 catalysts (cf. Fig. 8 e 10(a), (c), (e)) displayed a rougher surface topography than fresh LaNiO3 (cf. Fig. 2 (d)). Through EDX results, the foreign miniature particles deposited on spent LaNiO3 were positively identified as coke. Besides carbon, EDX analysis of spent LaNiO3 catalysts also discovered several minerals such as potassium (K), chlorine (Cl), sulphur (S), and magnesium (Mg), with their weight percent tabulated in Table 5. In all cases, potassium (K) represented the major impurity from POME, in parallel with the finding of Ng et al. [26]. Although coking was discernible in Fig. 8(a) and (c), spent LaNiO3 (0.06 mL/min and 0.08 mL/min) still somehow retained the pristine shape of LaNiO3. Nevertheless, spent LaNiO3 (0.10 mL/min) shown in Fig. 8(e) was heavily-coked, and fullyengulfed by carbonaceous species that formed via Boudouard reaction, resulting in lower FSyngas and poorer degradations. Through the TPO analysis (Fig. 8(b), (d), (f)), the carbon contents of spent LaNiO3 were quantified as 4.86% (0.06 mL/min), 12.52% (0.08 mL/min), and 15.97% (0.10 mL/min), thus justifying that higher V_ POME eventually rendered higher coke deposition. The weight percent of total minerals decreased with increasing V_ POME , which likely diluted by higher carbon contents at higher V_ POME . For spent LaNiO3 (0.1 g and 0.3 g), the rod-like structure of LaNiO3 still visible in Fig. 9(a) and (c). Nonetheless, excessive Wcat inflicted undesirable agglomeration of rodlike particles into a plate-like structure, judging from SEM image of spent LaNiO3 (0.5 g) in Fig. 9(e). By stacking in a plate-like structure, active sites that accessible by POME vapor greatly reduced; therefore, a lower catalytic activity of agglomerated LaNiO3 particles was expected. From the TPO profiles (Fig. 9(b), (d), and (f)), the carbon contents of spent LaNiO3 were measured as 37.49% (0.1 g), 22.39% (0.3 g), and 8.85% (0.5 g). Apparently, coking issue of LaNiO3 are alleviated at higher Wcat , possibly linked to its lower WHSV and catalyst agglomeration. At low Wcat range (0.1e0.3 g), lower WHSV renders a prolonged residence time of POME vapor in reactor, which promotes CO-utilizing water gas shift (CO þ H2 O/H2 þ CO2 ) to suppress coke-forming Boudouard reaction (2CO/CO2 þ C). At high catalyst loading (Wcat > 0.3 g), agglomerated LaNiO3 has fewer active sites that could adsorb POME vapor, wherein the declined catalytic activity contributes to a less severe coke deposition. Due to dwindling coking, the weight percent of total minerals raised with increasing Wcat from 0.1 to 0.3 g. It is believed that agglomerated LaNiO3 has lower susceptibility towards mineral accumulation since the total mineral contents decreased beyond a Wcat of 0.3 g. Regarding dcat effect, the rod-like structure of LaNiO3 was still perceptible in spent LaNiO3 catalysts (74e105 mm,

15

149e210 mm, and >250 mm) as presented in Fig. 10(a), (c), and (e). As the dcat reduced from >250 mm to 74e105 mm, coking become more observable whereby the void spaces between LaNiO3 particles progressively filled up by carbonaceous species. This result suggests that the greater SA/V ratio of finer LaNiO3 inevitably induces coking via CO-utilizing Boudouard reaction because it boosts CO-producing POME steam reforming. Evidently, TPO profiles of spent LaNiO3 catalysts (cf. Fig. 10(b), (d), (f)) identified their carbon contents as 25.91% (>250 mm), 30.77% (149e210 mm), and 46.42% (74e105 mm). Likewise, the total mineral contents decreased with reducing dcat as diluted by higher carbon contents.

Conclusions XPS and TEM analysis of spent LaNiO3 verified 873 K as the optimum temperature of LaNiO3-catalysed POME steam reforming. The strongest H2-TPR peak of fresh LaNiO3 was found at 928 K, therefore a reduction temperature of 1023 K could turn LaNiO3 into Ni and La2O3 phases. The CO2-TPD and NH3-TPD imparted the coexistence of acid and basic sites with various strengths on net-acidic LaNiO3. Through one-factor-ata-time approach, degradation efficiencies and syngas yield were optimized with respect to POME flow rate (V_ POME ), catalyst weight (Wcat ), and particle size (dcat ). With increasing V_ POME from 0.05 to 0.09 mL/min, higher PPOME facilitated simultaneous POME steam reforming and water gas shift, with a diminishing effect of latter. Beyond optimum V_ POME (0.09 mL/min), LaNiO3 severely coked by Boudouard reaction. When Wcat raised from 0.1 to 0.3 g, both POME steam reforming and water gas shift was enhanced by a lower WHSV. For the case of Wcat > 0.3 g, only water gas shift was further promoted, accompanied by undesired catalyst agglomeration. Applicable to dcat range of 74e105 mm to >250 mm, finer LaNiO3 had higher surface area to volume ratio which granted it higher degradation efficiencies and syngas yield. However, catalytic performance of ultrafine LaNiO3 (dcat < 74 mm) deteriorated owing to occluded pores. At optimized process conditions (T ¼ 873 K, V_ POME ¼ 0.09 mL/min, Wcat ¼ 0.3 g, and dcat ¼ 74e105 mm), POME steam reforming over LaNiO3 generated 132.47 mmoL/min of syngas while accomplished 99.53% XCOD , 99.88% XA , 99.75% XBOD5 , and 100% XTSS .

Acknowledgment This work is funded by the Malaysia’s Ministry of Education via Fundamental Research Grant Scheme (FRGS) with a grant number of RDU170116. YWC would like to acknowledge Malaysia Toray Science Foundation for financial support via Science & Technology Research Grant (STRG), RDU181501.

Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.ijhydene.2019.02.061.

Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

16

international journal of hydrogen energy xxx (xxxx) xxx

references

[1] Mielke T. Global oil supply, demand, and price outlook with special emphasis on palm oil. Oil World; 2017. [2] Mekhilef S, Siga S, Saidur R. A review on palm oil biodiesel as a source of renewable fuel. Renew Sustain Energy Rev 2011;15:1937e49. [3] Islam MR, Beg MDH, Jamari SS. Dispersion of montmorillonite nanoclays and their effects on the thermomechanical, structural and drying properties of palm oil based coating. Prog Org Coating 2016;91:17e24. [4] Kodali DR. Trans fats replacement solutions. USA: Academic Press and AOCS Press; 2014. [5] MPOB. Oil palm & the environment. 2014. [6] Wong KK. Application of ponding systems in the treatment of palm oil mill and rubber mill eflluents. Pertanika 1980;3:133e41. [7] Ding GT, Yaakob Z, Takriff MS, Salihon J, Abd Rahaman MS. Biomass production and nutrients removal by a newlyisolated microalgal strain Chlamydomonas sp in palm oil mill effluent (POME). Int J Hydrog Energy 2016;41:4888e95. [8] Cheng YW, Chang YS, Ng KH, Wu TY, Cheng CK. Photocatalytic restoration of liquid effluent from oil palm agroindustry in Malaysia using tungsten oxides catalyst. J Clean Prod 2017;162:205e19. [9] MPOA. BOD Level and POME Tertiary Treatment Technologies. 2014. http://mpoa.org.my/bod-level-andpome-tertiary-treatment-technologies [Accessed 12 December 2018]. [10] Balat H, Kırtay E. Hydrogen from biomass e present scenario and future prospects. Int J Hydrog Energy 2010;35:7416e26. [11] Cheng CK, Foo SY, Adesina AA. Glycerol steam reforming over bimetallic CoNi/Al2O3. Ind Eng Chem Res 2010;49:10804e17. [12] Xu X, Jiang E, Wang M, Xu Y. Dry and steam reforming of biomass pyrolysis gas for rich hydrogen gas. Biomass Bioenergy 2015;78:6e16. [13] Atnaw SM, Sulaiman SA, Yusup S. Syngas production from downdraft gasification of oil palm fronds. Energy 2013;61:491e501. [14] Shoja M, Babatabar MA, Tavasoli A, Ataei A. Production of hydrogen and syngas via pyrolysis of bagasse in a dual bed reactor. J. Energy Chem. 2013;22:639e44. [15] Rennard D, French R, Czernik S, Josephson T, Schmidt L. Production of synthesis gas by partial oxidation and steam reforming of biomass pyrolysis oils. Int J Hydrog Energy 2010;35:4048e59. [16] Lai M-P, Lai W-H, Horng R-F. Dry autothermal reforming from biomass derived gas under excess enthalpy with porous medium. J Power Sources 2012;217:407e16.  n J, Garcı´a L, Arauzo J. Cheese whey management by [17] Remo catalytic steam reforming and aqueous phase reforming. Fuel Process Technol 2016;154:66e81. [18] Nanda S, Rana R, Hunter HN, Fang Z, Dalai AK, Kozinski JA. Hydrothermal catalytic processing of waste cooking oil for hydrogen-rich syngas production. Chem Eng Sci 2019;195:935e45. [19] Sitthikitpanya S, Reungsang A, Prasertsan P, Khanal SK. Two-stage thermophilic bio-hydrogen and methane production from oil palm trunk hydrolysate using Thermoanaerobacterium thermosaccharolyticum KKU19. Int J Hydrog Energy 2017;42:28222e32. [20] Keskin T, Hallenbeck PC. Hydrogen production from sugar industry wastes using single-stage photofermentation. Bioresour Technol 2012;112:131e6.

[21] Rostrup-Nielsen J, Christiansen LJ. Concepts in syngas manufacture. London: Imperial College Press; 2011. [22] Huber GW, Iborra S, Corma A. Synthesis of transportation fuels from Biomass: chemistry, catalysts, and engineering. Chem Rev 2006;106:4044e98. [23] Guan G, Kaewpanha M, Hao X, Abudula A. Catalytic steam reforming of biomass tar: prospects and challenges. Renew Sustain Energy Rev 2016;58:450e61.  CL, Apesteguı´a CR. [24] Sad ME, Duarte HA, Vignatti C, Padro Steam reforming of glycerol: hydrogen production optimization. Int J Hydrog Energy 2015;40:6097e106. [25] Cheng YW, Lee ZS, Chong CC, Khan MR, Cheng CK, Ng KH, Hossain SS. Hydrogen-rich syngas production via steam reforming of palm oil mill effluent (POME) e a thermodynamics analysis. Int J Hydrog Energy 2018. [in Press]. [26] Ng KH, Cheng YW, Lee ZS, Cheng CK. A study into syngas production from catalytic steam reforming of palm oil mill effluent (POME): a new treatment approach. Int J Hydrog Energy 2018. [in Press]. [27] Ng KH, Cheng YW, Lee ZS, Khan MR, Lam SS, Cheng CK. Experimental evaluation and empirical modelling of palm oil mill effluent steam reforming. Int J Hydrog Energy 2018;43:15784e93. [28] Cheng YW, Khan MR, Ng KH, Wongsakulphasatch S, Cheng CK. Harnessing renewable hydrogen-rich syngas from valorization of palm oil mill effluent (POME) using steam reforming technique. Renew Energy 2019;138:1114e26. [29] ASTM. Standard test method for compositional analysis by thermogravimetry (ASTM E1131). West Conshohocken, PA: ASTM International; 2014. [30] Peavy HS, Rowe DR, Tchobanoglous G. Environmental engineering. Singapore: McGraw-Hill; 1985. [31] Nipattummakul N, Ahmed II, Kerdsuwan S, Gupta AK. Steam gasification of oil palm trunk waste for clean syngas production. Appl Energy 2012;92:778e82. [32] Abnisa F, Arami-Niya A, Wan Daud WMA, Sahu JN, Noor IM. Utilization of oil palm tree residues to produce bio-oil and bio-char via pyrolysis. Energy Convers Manag 2013;76:1073e82. [33] Shahbaz M, Yusup S, Inayat A, Ammar M, Patrick DO, Pratama A, Naqvi SR. Syngas production from steam gasification of palm kernel shell with subsequent CO2 capture using CaO sorbent: an aspen plus modeling. Energy Fuel. 2017;31:12350e7. [34] Purwandari FA, Sanjaya AP, Millati R, Cahyanto MN,  th IS, Niklasson C, Taherzadeh MJ. Pretreatment of oil Horva palm empty fruit bunch (OPEFB) by N-methylmorpholine-Noxide (NMMO) for biogas production: structural changes and digestion improvement. Bioresour Technol 2013;128:461e6. [35] Socrates G. Infrared and Raman characteristic group frequencies: tables and charts. 3rd ed. England: John Wiley & Sons Ltd; 2001. [36] Mile B, Stirling D, Zammitt MA, Lovell A, Webb M. The location of nickel oxide and nickel in silica-supported catalysts: two forms of “NiO” and the assignment of temperature-programmed reduction profiles. J Catal 1988;114:217e29. [37] Lima SMd, Silva AMd, Costa LOOd, Assaf JM, Jacobs G, Davis BH, Mattos LV, Noronha FB. Evaluation of the performance of Ni/La2O3 catalyst prepared from LaNiO3 perovskite-type oxides for the production of hydrogen through steam reforming and oxidative steam reforming of ethanol. Appl. Catal. A-Gen. 2010;377:181e90. [38] Marinho ALA, Rabelo-Neto RC, Noronha FB, Mattos LV. Steam reforming of ethanol over Ni-based catalysts obtained from LaNiO3 and LaNiO3/CeSiO2 perovskite-type oxides for the production of hydrogen. Appl. Catal. A-Gen. 2016;520:53e64.

Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061

international journal of hydrogen energy xxx (xxxx) xxx

[39] Nair MM, Kaliaguine S, Kleitz F. Nanocast LaNiO3 perovskites as precursors for the preparation of coke-resistant dry reforming catalysts. ACS Catal 2014;4:3837e46. [40] Singh S, Prestat E, Huang L-F, Rondinelli JM, Haigh SJ, Rosen BA. Role of 2D and 3D defects on the reduction of LaNiO3 nanoparticles for catalysis. Sci Rep 2017;7:10080. [41] Chen H, Yu H, Peng F, Yang G, Wang H, Yang J, Tang Y. Autothermal reforming of ethanol for hydrogen production over perovskite LaNiO3. Chem Eng J 2010;160:333e9. [42] Zheng T, He J, Zhao Y, Xia W, He J. Precious metal-support interaction in automotive exhaust catalysts. J Rare Earths 2014;32:97e107. [43] Ivanova AS, Slavinskaya EM, Gulyaev RV, Zaikovskii VI, Stonkus ОА, Danilova IG, Plyasova LM, Polukhina IA, Boronin AI. Metalesupport interactions in Pt/Al2O3 and Pd/ Al2O3 catalysts for CO oxidation. Appl Catal B: Environ 2010;97:57e71. [44] Tanabe K, Misono M, Hattori H, Ono Y. New solid acids and bases: their catalytic properties. Elsevier Science; 1989.  n F, Moreno S, Molina R. High [45] Daza CE, Gallego J, Mondrago stability of Ce-promoted Ni/MgeAl catalysts derived from hydrotalcites in dry reforming of methane. Fuel 2010;89:592e603. [46] Sugunan S, Meera V. Acid-base properties and catalytic activity of ABO3 (perovskite-type) oxides consisting of rare earth and 3d transition metals. React Kinet Catal Lett 1997;62:327e32. [47] Fairley N. CasaXPS manual 2.3.15: introduction to XPS and AES. Casa Software Ltd; 2009. ius S, Grebinskij S, Bondarenka V, Vengalis B, [48] Mickevic  z iene_ K, Orlowski BA, Osinniy V, Drube W. Investigation Sliu of epitaxial LaNiO3x thin films by high-energy XPS. J Alloy Comp 2006;423:107e11. ^ncio CVM, Varga E, Oszko A, [49] Rabelo-Neto RC, Sales HBE, Inoce Erdohelyi A, Noronha FB, Mattos LV. CO2 reforming of methane over supported LaNiO3 perovskite-type oxides. Appl Catal B: Environ 2018;221:349e61. [50] Wei Z, Zhao T, Zhu X, An L, Tan P. Integrated porous cathode made of pure perovskite lanthanum nickel oxide for nonaqueous lithiumeoxygen batteries. Energy Technol 2015;3:1093e100. [51] Wu Y, Wang T, Zhang Y, Xin S, He X, Zhang D, Shui J. Electrocatalytic performances of g-C3N4-LaNiO3 composite as bi-functional catalysts for lithium-oxygen batteries. Sci Rep 2016;6:24314. [52] Naumkin AV, Kraut-Vass A, Gaarenstroom SW, Powell CJ. NIST X-ray photoelectron spectroscopy database: NIST standard reference database 20, version 4.1. Gaithersburg MD, United States of America: National Institute of Standards and Technology (NIST); 2012. [53] Srinivas B, Rao VRS. Physico-chemical and catalytic properties of gamma-irradiated lanthanum nickelate (LaNiO3). J. Radioan. Nucl. Chem. Ar. 1996;210:3e14. [54] Park C-y, Nguyen-Phu H, Shin EW. Glycerol carbonation with CO2 and La2O2CO3/ZnO catalysts prepared by two different

[55]

[56]

[57]

[58]

[59]

[60]

[61]

[62]

[63] [64] [65]

[66]

[67]

[68] [69] [70]

[71]

[72]

17

methods: preferred reaction route depending on crystalline structure. Mol. Catal. 2017;435:99e109. Huang X, Dang C, Yu H, Wang H, Peng F. Morphology effect of Ir/ La2O2CO3 nanorods with selectively exposed {110} facets in catalytic steam reforming of glycerol. ACS Catal 2015;5:1155e63. Wang F, Ta N, Li Y, Shen W. La(OH)3 and La2O2CO3 nanorod catalysts for Claisen-Schmidt condensation. Chin J Catal 2014;35:437e43. Bakiz B, Guinneton F, Arab M, Benlhachemi A, Gavarria J-R. Elaboration, characterization of LaOHCO3, La2O2CO3 and La2O3 phases and their gas solid interactions with CH4 and CO gases. Moroc J Of Condens Matter 2010;12:60e7. Huang P, Zhao Y, Zhang J, Zhu Y, Sun Y. Exploiting shape effects of La2O3 nanocatalysts for oxidative coupling of methane reaction. Nanoscale 2013;5:10844e8. Li C, Liu H, Yang J. A facile hydrothermal approach to the synthesis of nanoscale rare earth hydroxides. Nanoscale Res. Lett. 2015;10:144. Chong CC, Abdullah N, Bukhari SN, Ainirazali N, Teh LP, Setiabudi HD. Hydrogen production via CO2 reforming of CH4 over low-cost Ni/SBA-15 from silica-rich palm oil fuel ash (POFA) waste. Int J Hydrog Energy 2018. [in Press]. Bukhari SN, Chong CC, Teh LP, Vo D-VN, Ainirazali N, Triwahyono S, Jalil AA, Setiabudi HD. Promising hydrothermal technique for efficient CO2 methanation over Ni/SBA-15. Int J Hydrog Energy 2018. [in Press]. Moulder JF, Chastain J. Handbook of X-ray photoelectron spectroscopy: a reference book of standard spectra for identification and interpretation of XPS data. Physical Electronics Division, Perkin-Elmer Corporation; 1992. Watts JF, Wolstenholme J. An introduction to surface analysis by XPS and AES. Wiley; 2003. Heide Pvd. X-ray photoelectron spectroscopy: an introduction to principles and practices. Wiley; 2011. Bakiz B, Guinneton F, Arab M, Benlhachemi A, Villain S, Satre P, Gavarri J-R. Carbonatation and decarbonatation kinetics in the La2O3-La2O2CO3 system under CO2 gas flows. Ann Mater Sci Eng 2010;2010.  nchez-Marı´n J, Toma  s F. The role of Corma A, Planelles J, Sa different types of acid site in the cracking of alkanes on zeolite catalysts. J Catal 1985;93:30e7. Leung A, Boocock DGB, Konar SK. Pathway for the catalytic conversion of carboxylic acids to hydrocarbons over activated alumina. Energy Fuel. 1995;9:913e20. Brezinsky K, Pecullan M, Glassman I. Pyrolysis and oxidation of phenol. J Phys Chem A 1998;102:8614e9. Mo N, Savage PE. Hydrothermal catalytic cracking of fatty acids with HZSM-5. ACS Sustain Chem Eng 2014;2:88e94. Gao W, Zhou T, Wang Q. Controlled synthesis of MgO with diverse basic sites and its CO2 capture mechanism under different adsorption conditions. Chem Eng J 2018;336:710e20. Arora S, Prasad R. An overview on dry reforming of methane: strategies to reduce carbonaceous deactivation of catalysts. RSC Adv 2016;6:108668e88. Fogler HS. Elements of chemical reaction engineering. 4th ed. Pearson Education International; 2010.

Please cite this article as: Cheng YW et al., Syngas from catalytic steam reforming of palm oil mill effluent: An optimization study, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.02.061