Conversion of Furfural as a Bio-Oil Model Compound over Calcium-Based Materials as Sacrificial Low-Cost Catalysts for Bio-Oil Upgrading

Abstract

The stabilization and upgrading of biomass and waste-derived pyrolysis oils requires development of reliable, active and low-cost upgrading catalysts. Basic natural materials can act as such catalysts and convert reactive oxygenates present in biomass pyrolysis oils. The conversion of furfural as a model compound has been conducted in an autoclave reactor at 200 °C to 300 °C using different calcium-based materials. CaCO₃, Ca(OH)₂, CaO, cement raw meal (CRM) and calcined cement raw meal (cCRM) were screened for their catalytic activity and characterized using X-ray powder diffraction (XRD) and X-ray fluorescence (XRF), nitrogen physisorption, carbon dioxide temperature programmed desorption (CO₂-TPD) and thermogravimetric analysis (TGA). CaCO₃ and CRM had low basicity and showed no catalytic activity at 200 to 300 °C. Notably, 90% conversion of furfural was achieved at 200 °C using Ca(OH)₂ with products being mostly furfural di- and trimers. For the basic CaO and cCRM, a temperature of 250 °C or above caused rapid polymerization of furfural. The proposed mechanism follows the Cannizzaro reaction of furfural, catalyzed by basic sites, polymerization of furfuryl alcohol, decarboxylation of furoic acid and decarbonylation of furfural, releasing CO, CO₂ and H₂O. Calcined cement raw meal showed the most promise for application as low-cost, sacrificial, basic catalyst.

1. Introduction

Energy demands in many regions of the world are rising due to economic development, population growth and increases in standards of living [1]. At the same time, the impact of climate change forces the world to reduce the use of fossil fuels in the future [2]. Sectors like transportation, chemical industry, power generation and cement production are interested in utilizing renewable energy sources for power and fuels [3,4,5]. Biomass plays a key role being the sole renewable carbon feedstock. Fast pyrolysis has gained substantial interest both in industry and research and can convert different biomass feedstocks into liquid bio-oil [6,7]. Lignocellulosic biomass sources of many different kinds, like woody biomass, forest waste residues, energy crops and agricultural waste, can be utilized. The major biomass macromolecules hemicellulose, cellulose and lignin decompose into short-chained molecules, forming a highly diverse condensable mixture called bio-oil [8]. Bio-oil has a higher energy density than solid biomass and can be easily stored and transported [9]. Bio-oil consists of organic compounds including ketones, aldehydes, organic acids, alcohols and aromatics, and 15–30 wt% water [10]. This leads to a high level of oxygen and to many undesired properties like high acidity, viscosity and instability, as well as low energy density and poor miscibility with conventional fossil fuels [11].

Hence, upgrading through deoxygenation is favorable and has attracted the attention of researchers, particularly during the last two to three decades. The focus of many contributions is catalytic hydrotreating through HDO at 250–400 °C and high H₂ partial pressures up to 150 bar or cracking with solid acid catalysts like zeolites [11,12,13,14,15,16]. Encouraging works have been conducted using different kinds of catalysts, like promoted molybdenum sulfide-based hydrotreating catalysts or noble and transition metal-based catalysts [15,16,17,18]. Metal sites promote hydrogenation, whereas solid acid sites promote C-C cleavage and the removal of oxygen as CO₂ [10,12]. For these catalysts, high activity was often outweighed by deactivation due to coke formation [10,19]. At milder reaction conditions, deactivation showed to be less severe, especially for transition and noble metal catalysts [20], but deactivation is still significant and would require frequent catalyst regeneration or replacement. Economic considerations, together with the problem of deactivation, hinder the use of high-cost catalysts in the application of pyrolysis oil upgrading [19]. Two-step upgrading processes have been proposed to remove highly active species and stabilize the oil before further treatment [21,22], but a breakthrough is yet to be obtained. Research on low-cost alternative upgrading catalysts is still limited, but has recently received attention [23,24,25]. Important work has reported on catalytic pyrolysis using calcium containing catalysts in fixed and fluidized beds [26,27,28,29]. The amount of acetic acid in pyrolysis vapors of cedar wood was reduced by 65% with CaO as a catalyst in a fixed bed reactor at 600 °C [30]. Further, the effect of CaO in the pyrolysis of forest residues at 550 °C has been tested, showing an increase in the HHV of 20% and a decrease in the water content of the bio-oil to as low as 7.8 wt% [31]. The addition of CaO was reported in the pyrolysis of biomass and plastics, including PVC, to absorb HCl from the pyrolysis vapors by forming CaClOH and removing 92 wt% of the HCl this way [32]. Similarly, the effect of MgO as a basic catalyst has been investigated in a fixed bed reactor showing similar activity in the removal of acids and aldehydes compared to zeolite H-ZSM-5, but with the formation of more coke. Conversion of aldehydes over the basic materials was proposed to take place through aldol condensation reactions [33]. In particular, the aldol condensation of furfural with acetone is an often-investigated model experiment with catalysts such as hydrotalcite materials, acidic zeolites, activated dolomite, CaO/MgAl₂O₄ and MgO/MgAl₂O₄, all promoting the aldol condensation of the two reactants in a two-step process, as shown in Scheme 1 [34,35,36,37].

Since earlier research by our group has shown that CaO and calcined cement raw meal (cCRM) exhibit cracking activity with vapors from slow plastic pyrolysis at elevated temperatures above 500 °C, they are also interesting candidates for bio-oil upgrading [38].

Integrating fast pyrolysis technology for the production of bio-oil in a cement plant could benefit from heat and material integration between the two processes. Heat for the pyrolysis could be supplied from the pre-heated CRM/cCRM mixture. The CRM/cCRM is returned to the cement calciner together with coke, char and pyrolysis gases, supplying fuel for the calcination. CRM/cCRM could further be used as upgrading catalyst for the pyrolysis vapors in a separate step. Once this CRM/cCRM is coked and deactivated, it is returned to the calciner, where combustion of the coke supplies further fuel to the calcination. This makes CRM and cCRM interesting candidates as sacrificial catalysts for bio-oil upgrading, as waste catalysts are converted to useful cement.

The transformation of calcium species follows Scheme 2, with reactions of water and carbon dioxide depending on the temperature and partial pressures of the gases present.

Although much is known about the catalytic conversion of furfural with acetone through aldol condensation as stated before, studies on pure furfural conversion as model compound are lacking. More research is needed to characterize calcium-based materials as catalysts for mild upgrading and stabilization of pyrolysis oils as a first step before deep HDO to obtain a final liquid hydrocarbon product.

Thus, in this work, the conversion of furfural as model compound for biomass pyrolysis oil with calcium-based materials was studied. It focuses on the behavior of calcium containing materials like CRM and cCRM, which are readily available at cement plants, and their main components CaO, Ca(OH)₂ and CaCO₃ in the conversion of pure furfural at moderate temperatures of 200 to 300 °C. The products were analyzed, and a chemical mechanism for the furfural conversion was suggested.

2. Results and Discussion

2.1. Characterization of the Catalyst

While calcium oxide, calcium hydroxide and calcium carbonate can be considered pure compounds as received, cement raw meal and its calcined form are a mixture of different minerals. X-ray fluorescence spectroscopy was used to determine the composition. The results can be seen in

Table 1. Composition of cement raw meal according to XRF analysis in wt%.

Material	SiO2	CaO	Fe2O3	K2O	Al2O3	MgO	SO3	Na2O	TiO2	Mn2O3	LOI, 975 °C
CRM	14.4	43.0	2.15	0.85	3.60	1.78	1.13	0.29	0.19	0.05	32.3

. As expected, the major components were SiO₂ with 14.4 wt% and CaO with 43.0 wt%. Smaller amounts of iron, aluminum, sulfur and magnesium, as well as traces of potassium and sodium, were also present. The loss on ignition was high, namely 32.3 wt%, due to calcination, as the calcium and magnesium detected by XRF was, in fact, mainly carbonates. The calcination temperature for the carbonates is in the range of 800–900 °C, where carbonates will become oxides under formation of gaseous carbon dioxide [39]. The silica and iron phases are not expected to be affected by the treatment.

To verify the full extent of calcination and the composition of the minerals, X-ray diffraction was additionally conducted. Characteristic peaks can be associated with the presence of crystalline CaO, CaCO₃ and SiO₂. As can be seen in Figure 1, cement raw meal and the calcined form exhibit a characteristic reflection for silicon dioxide at 28 °2θ, and calcium carbonate and cement raw meal show distinctive reflections at around 30 and 44 °2θ, which are similar for calcium oxide and calcined cement raw meal at 37 and 53 °2θ. Peaks for CaCO₃ are present in the diffractogram of cement raw meal, but are not observed for calcined raw meal, verifying the calcination in the muffle furnace.

Thermogravimetric analysis was conducted both on the catalytic materials before the furfural upgrading experiment and after the experiment. Figure 2 shows the mass and temperature of the sample as function of time. CRM lost a significant amount of mass from around 650 °C, leading to a constant mass at 65 wt% after reaching 780 °C. This extent of weight reduction and this temperature range is consistent with calcination of the calcium carbonate fraction of CRM. The rate of thermal decomposition of calcium carbonate is dependent on heating rate and ambient conditions, but it is reported to start at around 600 °C and to be completed around 875 °C in comparable TGA [40]. Calcium oxide and cCRM show a small mass loss in this temperature range, possibly due to the recarbonation of a part of the samples when in contact with ambient CO₂ or water. Calcium hydroxide as a pure compound showed a loss of mass beginning at 450 °C due to the dehydration and evaporation of water [41]. Although the kinetics of dehydration is not yet fully understood, it is proposed to happen as a two-step mechanism, taking place between 400 and 500 °C, and the equilibrium is shifted completely towards CaO at 512 °C [42]. Calcium oxide, calcium hydroxide and calcium carbonate phases can, therefore, clearly be distinguished in TGA. The cCRM sample, although stored under airtight conditions, showed the presence of Ca(OH)₂ and CaCO₃ of up to 10 wt%, determined from Figure 2 by the mass loss at around 420 and 620 °C, respectively.

The particle properties of the materials were analyzed with laser diffraction and the results are shown in Figure 3. The particle size distribution (PSD) was quite narrow for all samples, but showed differences in the average particle size by approximately one order of magnitude. While calcium hydroxide and calcium oxide both exhibit a maximum in the distribution at around 3–4 µm, calcium carbonate shows a much larger particle size, with a maximum above 30 µm.

Smaller particle sizes can improve the mass and heat transfer within, as well as between, the particles and the reactant, and can have an influence on the conversion of furfural. However, as all particles were below 100 μm and with long residence times, transport limitations are expected to be minimal.

The specific surface area of the materials ranged from 2.5 m²/g for calcium carbonate to above 15 m²/g for calcium hydroxide, as stated in

Table 2. The physico-chemical properties of the BET surface area, total base sites density and average particle size of all used catalysts.

Catalyst Unit	Surface Area [m2/g]	Average Particle Size D [4;3] [µm]	Basic Site Density [μmol/m2]	CO2 Adsorption, <380 °C [μmol/g]
CRM	5.1	12.5	10.5	52.2
cCRM	7.0	16.0	42.8	299.9
CaO	14.7	6.2	42.1	619.3
CaCO3	2.5	33.1	0.4	1.1
Ca(OH)2	15.2	9.9	33.6	504.8

. CRM had a surface area of 5.0 m²/g, with the calcination resulting in an increase in surface area, as the cCRM showed as 7.0 m²/g.

A temperature-programmed desorption analysis of CO₂ (CO₂-TPD) was carried out to investigate the surface basicity of the samples; see Figure 4. In this analysis, the desorption temperature and amount of desorbed CO₂ correlates with the basic character of the sample’s active sites. It shows different temperature ranges for weak, medium and strongly adsorbed CO₂ on the surface [43]. Peaks desorbed at high temperature regions of 500 °C and above indicate a high bonding strength of CO₂ as the results of bulk carbonation of some of the CaO and Ca(OH)₂ as other researchers have connected this peak with the CO₃²⁻ formation of the material [44]. Peaks desorbed at temperatures lower than 400 °C suggest sites with medium to low bonding strength, indicating surface-bound CO₂ [45]. The small peaks at 400 to 450 °C observed for Ca(OH)₂ and cCRM were mainly water, formed by dehydration, which interfered with the CO₂ signal in the FTIR gas analysis.

The CO₂-TPD profiles of the samples CaO, cCRM and Ca(OH)₂ showed surface basicity according to the broad peak at 100 to 400 °C. CaCO₃ and CRM (also mostly consisting of CaCO₃) exhibited no peaks before a strong signal from around 650 °C associated with calcination of bulk carbonates; hence, they showed little to no surface basicity. The onset of the high temperature peak is consistent with calcination temperature across the materials. Calibration of the FTIR gas analyzer with calcium oxalate was used to determine the total basicity shown in Table 2.

The basic site density of CaO was measured to be 42 µmol/m² or CO₂ uptake of 619 µmol/g, being close to the values published by Thitsartarn et al. [46] and Kumar et al. [47]. The data shows the strong basicity of activated calcium species with the order of increasing basic site density of Ca(OH)₂ < cCRM = CaO, while CaCO₃ and CRM show no or very low basicity. The similar basic site density of cCRM and CaO is to be expected since cCRM consists to large extent of CaO itself. The hydroxide form shows slightly lower basicity.

2.2. Conversion of Furfural

An overview of the different conditions, the mass balances and the physical properties of the products from the batch reactor tests with furfural as a model compound are given in

Table 3. Experimental conditions; indicators of experimental results [furfural/catalyst: 5:1; initial pressure: 1 bar nitrogen].

Exp. Number	Temperature	Catalyst	Mass Balance Closure wt%	Conversion %	Run-Away
1	200 °C	No catalyst	99.0	0.7	No
2	250 °C	No catalyst	98.7	1.3	No
3	300 °C	No catalyst	98.2	1.6	No
4	200 °C	CRM	99.7	2.1	No
5	250 °C	CRM	99.3	2.2	No
6	300 °C	CRM	99.8	2.0	No
7	200 °C	cCRM	99.5	13.3	No
8	250 °C	cCRM	98.5	100 *	Yes
9	300 °C	cCRM	96.9	100 *	Yes
10	200 °C	CaO	96.7	64.3	No
11	225 °C	CaO	98.4	100 *	Yes
12	250 °C	CaO	98.1	100 *	Yes
13	300 °C	CaO	97.0	100 *	Yes
14	200 °C	Ca(OH)2	98.3	89.2	No
15	225 °C	Ca(OH)2	99.0	100 *	Yes
16	250 °C	Ca(OH)2	99.1	100 *	Yes
17	300 °C	Ca(OH)2	98.2	100 *	Yes

* Assumption of total conversion based on runaway.

. The test conditions such as temperature and feed to catalyst ratio were chosen based on previous experiments using calcium oxide as catalyst in biomass pyrolysis summarized by Li et al. [48]. Mass balances for the conducted experiments closed to more than 95 wt%. This deviation might be due to product loss during filtration, remaining products on the stirrer and reactor walls, and measurement errors.

All experiments which showed a temperature increase of 25 °C above the set temperature during the reaction are labeled as run-away experiments in Table 3. When this happened, the heating was switched off and the experiment was stopped.

The product distribution is shown for all experiments in Figure 5 with differentiation into liquid-, solid-, and gas-phase products, as well as unconverted furfural. The variation in temperature for the experiments without a catalyst only showed furfural conversions <2%, and no gas-phase products were formed. With CRM and calcium carbonate, no significant catalytic effect was visible, even at 300 °C, resulting in <5% furfural conversion and only liquid products. With the introduction of basic catalysts like cCRM, the formation of solid- and gas-phase products became evident. As expected, an increase in temperature translated to a higher conversion of furfural and formation of more solid and gaseous products. Looking at cCRM, the amount of solid products increased from 2 wt% at 200 °C to 6 wt% at 225 °C and above 90 wt% at 250 °C, where furfural was completely converted and no liquid products were collected. At 300 °C, even more gaseous and less solid products were formed with 79 wt% solids and 21 wt% gaseous products. Using CaO and Ca(OH)₂ at 200 °C, mainly liquid products were present and furfural conversion ranged from 60 to 80%, but an increase of just 25 °C showed complete furfural conversion and the formation of only solid and gaseous products. For CaO, the product phase composition did not change with further temperature increase, as the solid phase amounted to 65 wt% to 68 wt% at 250 and 300 °C, respectively. For Ca(OH)₂, an increase in temperature from 225 °C to 250 °C and 300 °C led to an increase in gas-phase from 6.3 wt% to 15.1 wt%/26.0 wt% and a decrease in solid-phase products, respectively.

Investigating the gas phase after the reaction using gas GC-FID/TCD (Figure 6) revealed the presence of mostly CO, CO₂, methane and minor amounts of C2+ hydrocarbons. This might indicate that cracking or reforming reactions take place, breaking up the oligomers into smaller compounds. Traces of oxygen amounting to <0.1 wt% could be found, most likely from contamination during sampling or insufficient flushing of the autoclave with nitrogen.

At 200 °C for CaO, cCRM and Ca(OH)₂, the liquid products were 2-furanmethanol and multiple di- and trimers, with yields of 64% and 53%, respectively. For Ca(OH)₂ at 200 °C, the selectivity towards 2-furanmethanol was 86% with the remaining being water. For cCRM and CaO, the product composition became more diverse, as the data in

Table 4. Liquid product selectivity of selected experiments.

Products	Ca(OH)2 at 200 °C [%]	cCRM at 200 °C [%]	cCRM at 225 °C [%]	CaO at 200 °C [%]
2-furanmethanol	85.79	69.64	74.29	74.19
2.2′-methylenebis-furan	0.13	0.84	1.68	1.94
2,2′-[oxybis(methylene)]bis-furan	0.58	0.00	0.62	2.27
1.2-di-2-furanyl-2-hydroxy-ethanone	0.20	0.86	1.47	13.13
2.5-bis(2-furanylmethyl)-furan	0.16	0.52	0.99	0.50
2.6-di(2-furylmethylidene)cyclohexan-1-one	0.00	0.17	0.00	0.11
Water	13.14	27.98	20.95	7.65

show. At 200 °C, cCRM shows a high water selectivity of 28%, which might be due to a low conversion at this temperature level, and most of the liquid is still unconverted furfural. At 225 °C, cCRM as the catalyst leads to the formation of multiple dimers with a selectivity of around 5% and trimers with about 1%. CaO at 200 °C shows especially high amounts of dimers, with more than 17% selectivity. Dehydration of alcohols at high temperatures leads to ethers like the dimer 2.2′-oxybis(methylene)]bis-furan which can split off formaldehyde and react to 2.2′-methylenebis-furan.

A degree of deoxygenation, as shown in

Table 5. Degree of deoxygenation of selected experiments based on liquid GC-MS analysis.

	Ca(OH)2 at 200 °C [%]	cCRM at 200 °C [%]	cCRM at 225 °C [%]	CaO at 200 °C [%]
Degree of deoxygenation	0.42	0.89	1.62	1.98

, can be calculated based on the liquid-phase analysis and amounts to 0.42% to 1.98% in the experiments using Ca(OH)₂ at 200 °C and CaO at 200 °C, respectively. The main product in these experiments is 2-furanmethanol, which does not lead directly to a reduction of oxygen, but the formation of dimers and trimers.

Since condensation reactions of furfural form water as a side product, a reaction of H₂O with CaO forming Ca(OH)₂ is likely, causing a change in basicity and catalytic activity. The results indicate that basic surface sites of calcium catalysts catalyzed condensation of furfural to form di- and trimer species, with further polymerization ultimately leading to coke formation.

The mechanism in Scheme 3 and Scheme 4 is suggested with furfural undergoing the Cannizzaro reaction, catalyzed by a base, towards furfuryl alcohol and furanoic acid. Furfuryl alcohol then follows a di- and oligomerization, in sequential dehydration and hydrogenation reactions, eventually leading to formation of furfuryl alcohol polymer, which is recovered as coke. The furanoic acid can undergo dimerization and decarboxylation, resulting in CO₂ release and further water formation. The resulting furan containing products can further be incorporated into the furfuryl alcohol polymer, for which the reaction pathways are complex and not yet fully understood.

Overall, the reactions lead to a decrease in oxygen content of the product by separation of CO₂ and water, as well as an increase in molecular size. Both are beneficial in the application of biofuels upgrading with oxygen removal leading to better storage stability and longer chain length increasing the diesel range share of the oils.

TGA of the spent catalysts showed that they, dependent on the reaction temperature, underwent recarbonization and rehydration (Figure 7). Very low amounts of organic volatile deposits were present and no significant change with higher temperatures was visible. CRM showed a TGA signal similar to pure CaCO₃. cCRM at 200 °C showed a signal similar to the fresh catalyst. At temperatures above 250 °C, on the other hand, a weight loss in the range of 200–350 °C can be correlated with remaining oligo- or polymers on the coked catalyst being evaporated or pyrolyzed. A sharp weight loss at 400 °C showed the presence of Ca(OH)₂ for cCRM, which indicates that substantial parts of the former CaO phase have absorbed water, which was formed in the furfural condensation reactions. Furthermore, the decrease at 750 °C showed the recarbonation of substantial amount of the CaO phase. Analysis of the Ca(OH)₂ sample in Figure 7c indicates that much of the hydrate was converted into carbonate by adsorbing CO₂ over the course of the reaction, even at low temperatures of 200 °C. Further, for Ca(OH)₂, at a process temperature of 250 °C, the weight loss in the low temperature range up to 350 °C is potentially caused by solid oligomer deposits which evaporate or react further. A spent CaO catalyst exhibited similar behavior to cCRM with small amounts of hydroxide and carbonates at 200 °C, but increased with higher temperatures. Higher partial pressures of H₂O and CO₂ at high conversion led to the formation of more CaCO₃ and Ca(OH)₂ according to the principle of Le Chatelier. At 300 °C, both CaO and cCRM show a smaller weight loss in the 400 °C and 700 °C range, which indicates less formation of CaCO₃ and Ca(OH)₂. This can be based on the shorter process time at these conditions with the thermal runaway happening rapidly and causing stopping, or the higher temperatures favoring the formation of CaO over hydration and carbonation.

An extraction with n-octane was conducted on the solid samples, and just in the case of Ca(OH)₂ at 250 °C, showed small amounts of trimers (1.7 wt% of the solid mass). This finding is consistent with the TGA of the solid of the same experiment showing volatile organic compounds evaporating before 350 °C. It indicates that there is a layer of oligomers formed which, at higher temperatures, further react to coke.

Overall, carbonation of CaO, Ca(OH)₂ and cCRM is expected to result in catalyst deactivation, as CaCO₃ and CRM were catalytically inactive. However, when coupled to a cement production facility, spent cCRM catalyst should be returned to the preheating and calcination process, to be converted into cement and potentially fresh catalyst.

3. Materials and Methods

3.1. Materials and Preparation

Furfural (99%), calcium oxide, calcium hydroxide and calcium carbonate were purchased from VWR (Radnor, PA, USA). Cement raw meal (CRM) was delivered by FLSmidth & Co. A/S (Copenhagen, Denmark). No further purification was performed on these chemicals.

CRM was calcined in a muffle furnace for 2 h at 800 °C with a heating rate of 10 °C/min under ambient atmosphere. For this, the powder was spread out on an alumina crucible to a thickness of around 1 cm. The produced calcined cement raw meal (cCRM) was stored under airtight conditions.

3.2. Catalyst Characterization

The crystallinity of the samples and the chemical composition were determined using X-ray diffraction (XRD) on a Malvern Panalytical Empyrean diffractometer (Malvern, United Kingdom; diffraction patterns were recorded for the 2θ angle range from 3.5° to 90° with 0.008° step size at a scan speed of 10.8 2θ/min). In addition, the elemental composition of cement raw meal was analyzed using X-ray fluorescence (XRF) following ASTM C114 (m) for analysis of cements, conducted at FLSmidth & Co. A/S (Copenhagen, Denmark). Nitrogen adsorption–desorption analysis was performed on a Quantachrome (Boynton Beach, FL, USA) Nova Touch LX2. Thehe specific surface area was calculated based on the equation of Brunauer–Emmett–Teller (BET).

The basicity of the catalysts was determined by CO₂ temperature-programmed desorption (CO₂-TPD) in a TGA-FTIR setup from Netzsch GmbH and Co. KG (Selb, Germany). In a typical experiment, a sample weighing about 20 mg was placed in a crucible and pretreated by drying at 10 °C/min from room temperature, increasing to 80 °C, purged by nitrogen gas flow of 100 mL/min, kept at 80 °C for 1 h with a purified mixture of 40 mL/min CO₂ and 60 m/min nitrogen gas to adsorption saturation. It was switched to pure nitrogen gas flow of 100 mL/min for 1.5 h to remove physisorbed CO₂ on the surface. Samples were heated to 800 °C in a nitrogen gas flow of 100 mL/min at a heating rate of 10 °C/min and held at 800 °C for 1 h. The effluent gas was analyzed by a Bruker FTIR alpha detector.

The particle size distribution was determined using a Malvern (Malvern, United Kingdom) Mastersizer 3000 HydroR on powders dispersed in water (CaCO₃, CRM, Ca(OH)₂) or ethanol (CaO, cCRM). Ethanol was chosen due to its low reactivity with the basic sites of the activated materials and previous tests have shown a low tendency to form agglomerates [51]. The solid powders were used as received since further grinding and sieving to achieve a smaller particle size was not possible for these already very fine powders. Using a STA 449 F3 apparatus from Netzsch GmbH and Co. KG (Selb, Germany) thermogravimetric analysis (TGA) was performed on the catalysts before and after the experiments. The samples were first heated up under nitrogen at 10 °C/min to 800 °C, cooled down to 200 °C and subsequently heated up with a 10:90 air to nitrogen mixture to 800 °C again. In this way, the non-oxidative pyrolysis and the combustion curve of the samples could be obtained.

3.3. Catalytic Activity Test and Product Analysis

The upgrading reaction was carried out with furfural as biomass model compound in all cases. An amount of 50 g of furfural and 10 g of catalyst in powder form were charged into a Parr (Moline, IL, USA) 300 mL autoclave (Series 4560 Mini Bench Top Reactors) equipped with electrical heating, an internal mechanical stirrer, a pressure gauge, a thermocouple and a controller system. The reactor was flushed ten times with nitrogen and then put under 1 bar of absolute nitrogen pressure. It was heated up to the desired temperature with a heating rate of around 10 °C/min, and the stirring was set to 200 rpm. Subsequently, the reactor was stirred for 60 min at the set temperature. After this, the heating jacket was removed and the reactor cooled down by air to 40 °C. Heating-up and cooling-down time was not considered as reaction time. After reaching 40 °C, a gas sample was taken with a gas bag, the pressure at this level was noted and the reactor was opened to collect the products. The total mass of the autoclave with solid and liquid contents was determined by weighing, spent catalyst and solid products were separated from liquids by filtration, solids were dried at 40 °C for 4 h and the mass of the solid phase was measured; the mass of the liquid phase was calculated by difference. The mass of the gas phase was both calculated based on the pressure at 40 °C with the measured composition of the gas phase and by difference. The deviation between both methods was below 5% relative. In this way, a gas, solid and liquid sample could be taken for further analysis.

The gas-phase composition was analyzed using a Thermo Scientific (Waltham, MA, USA) Trace 1300 gas chromatograph with flame ionization and thermal conductivity detectors (GC-FID/TCD). An Agilent (Santa Clara, CA, USA) J&W HP-PLOT Al₂O₃ S column, measured 25 m long with an inner diameter of 0.53 mm and a layer thickness of 10 µm, together with a backflush column Restek Rtx-1, 5 m long, 0.53 mm ID and 3 µm layer were used for hydrocarbon analysis on the FID. Gases were detected on two TCD detectors dedicated to CO/CO₂/O₂/N₂ and H₂, respectively, using micro-packed columns. Liquid samples were analyzed using an Agilent GC-MS/FID. Helium was used as carrier gas, the split ratio was set to 80:1 and the injection volume 1 µL. The GC-MS was used to identify the compounds. GC-FID was calibrated to quantify the expected products with chemical standards. For the liquids, the standards were furfural, furfuryl alcohol, furan and tetrahydro furan (THF). Using the calculated FID response factors, di- and trimer quantification was conducted according to ref. [52].

The following expressions were used to determine the conversion of furfural ($X_{Furfural}$), yields ($Y_{producti}$) and selectivities ($S_{producti}$) of products, mass and carbon balance.

$$X_{Furfural}\left(\mathrm{\%}\right)=(1-{\displaystyle \frac{n_{Furfural,product}}{n_{Furfural,feed}}})\times 100$$

$$Y_{producti}\left(\mathrm{\%}\right)={\displaystyle \frac{n_{producti}}{n_{Furfural,feed}\times z_{monomers,producti}}}\times 100$$

$$S_{producti}\left(\mathrm{\%}\right)={\displaystyle \frac{n_{producti}}{n_{Furfural,converted}\times z_{monomers,producti}}}\times 100$$

where $z_{monomers,producti}$ refers to the number of monomers needed to form the respective product from a single furfural unit.

4. Conclusions

Catalytic conversion of furfural as a biomass model compound was conducted in a batch autoclave reactor at temperatures from 200 °C to 300 °C under nitrogen atmosphere at initial pressure of 1 bar using different calcium containing catalysts: CaO, CaCO₃, Ca(OH)₂, cement raw meal and calcined cement raw meal. XRF, XRD and TGA were used to determine the composition of the cement raw meal before and after calcination and verified the full calcination of the CaCO₃ phase to CaO. CO₂-TPD showed significant basicity for CaO, cCRM and Ca(OH)₂. The basic site density of CaO was measured to be 42 µmol/m² or CO₂ uptake of 619 µmol/g The conversion of furfural was dependent on temperature and type of catalyst, especially its basicity. Catalysts with no measurable basicity had almost no catalytic activity, as determined by the conversion of furfural being <5% with CRM at 300 °C, and the formation of only minor amounts of liquid products. Experiments above 250 °C with the highly basic catalysts showed full conversion of furfural and mostly solid and gaseous products. The presence of basic sites favors the Cannizaro reaction of the aldehyde group, followed by condensation, dehydration, decarbonylation and decarboxylation under production of H₂O, CO and CO₂, leading to oligomers and polymers. Analysis of the solid-phase products and used catalysts with TGA showed different deactivation patterns. The solid products from experiments above 250 °C were largely carbon deposits, presumably as a result of polymerization. Both pure CaO and the CaO fraction of cCRM were partially recarbonated with the CO₂ formed by the conversion of furfural, which reduced their basic functionality causing catalyst deactivation. Below 250 °C, furfural reacts in the presence of the basic materials towards dimers and trimers following the proposed Cannizzaro reaction.

Calcined cement raw meal showed high conversion of furfural with up to 90% at a temperature level of 250 °C, with comparable level of solid products and CO₂ formation as pure CaO; therefore, it could be a promising candidate for the conversion of oxygenated compounds in pyrolysis oil upgrading. It can lead towards a reduced amount of oxygen through formation of water and increased chain length. Deactivation through coking and recarbonization can be handled by using cCRM as a sacrificial catalyst, when the pyrolysis and upgrading are integrated at a cement plant, by recycling the cCRM to the cement process. Ongoing research in our group investigates the application of cCRM as a sacrificial catalyst in the mild upgrading of biomass pyrolysis vapors.