Glycerol Dehydration to Acrolein over Supported Vanadyl Orthophosphates Catalysts

Giovanna Ruoppolo 1, Gianluca Landi 1,* and Almerinda Di Benedetto 2 1 Institute for Researches on Combustion—CNR P.le Tecchio 80, 80125 Naples, Italy; ruoppolo@irc.cnr.it 2 Dipartimento di Ingegneria Chimica, dei Materiali e della Produzione Industriale, University of Naples Federico II, P.le Tecchio 80, 80125 Naples, Italy; almerinda.dibenedetto@unina.it * Correspondence: gianluca.landi@cnr.it; Tel.: +39-08-1768-2235


Introduction
Biodiesel has been identified as a notable option for complementing conventional fuels. Its production is from renewable biological sources, mainly vegetable oils (rape, soybean, palm) and an alcohol (usually methanol) used to break down fatty acids from glycerol via a transesterification reaction, according to the reaction reported in Figure 1.

Introduction
Biodiesel has been identified as a notable option for complementing conventional fuels. Its production is from renewable biological sources, mainly vegetable oils (rape, soybean, palm) and an alcohol (usually methanol) used to break down fatty acids from glycerol via a transesterification reaction, according to the reaction reported in Figure 1. R1, R2, and R3 are hydrocarbons chains containing 15-21 C atoms [1][2][3][4].
Glycerol accounts for about 10 wt.% of the biodiesel produced [5]. Crude glycerol should be upgraded to obtain "technical grade" glycerol and, by a further purification, United States Glycerol accounts for about 10 wt.% of the biodiesel produced [5]. Crude glycerol should be upgraded to obtain "technical grade" glycerol and, by a further purification, United States Pharmacopeia (USP; appropriate for food and pharmaceutical products) and Food Chemicals Codex (FCC; appropriate for use in food) glycerin, both requiring ≈100% purity. However, the glycerol surplus together with its difficult upgrading in the presence of methanol [6] makes crude glycerol upgrading disadvantageous [7][8][9], especially if the price decrease of glycerol is considered [9][10][11].
Several catalysts have been proposed for glycerol dehydration, including heteropolyacids, zeolites, metal oxides, mixed metal oxides, phosphates and pyrophosphates, and mixed systems thereof [5,18,. Given the vast number of reactions potentially occurring during glycerol dehydration, the proper choice of catalyst is paramount. A non-exhaustive reaction scheme is reported in Figure 2. This scheme has been derived from schemes suggested by Katryniok et al. [47] and Jiang et al. [48], and the references within. The occurrence of each reaction depends on the reaction conditions (temperature, oxidizing or reducing environment) and the catalytic active sites.  [47,48]).
The key point for the development of catalysts is the balance of the acid sites. For instance, in H-ZSM5 based catalysts the co-presence of acid sites of different types (Lewis/Brønsted) and strength both improved the acrolein yield and decreased coke formation [5]. Coking is the major deactivating phenomenon, and it is still an open issue [48], despite several efforts to develop coking-resistant catalysts [49][50][51]. Coke is formed by cyclization, condensation, and polymerization reactions occurring on acid sites (see Figure 2). A strategy to reduce coking is to oxidize coke precursors by adding oxygen in the feed stream and by doping the catalyst with redox sites [48].
The key point for the development of catalysts is the balance of the acid sites. For instance, in H-ZSM5 based catalysts the co-presence of acid sites of different types (Lewis/Brønsted) and strength both improved the acrolein yield and decreased coke formation [5]. Coking is the major deactivating phenomenon, and it is still an open issue [48], despite several efforts to develop coking-resistant catalysts [49][50][51]. Coke is formed by cyclization, condensation, and polymerization reactions occurring on acid sites (see Figure 2). A strategy to reduce coking is to oxidize coke precursors by adding oxygen in the feed stream and by doping the catalyst with redox sites [48].
Vanadyl orthophosphates (VOPO 4 ) have been proposed for the oxidative dehydrogenation (ODH) of ethane and propane [52,53], and they both have acid and redox properties. In particular, VOPO 4 supported on γ-Al 2 O 3 , TiO 2 , and ZrO 2 showed the best catalytic properties towards propane ODH due to good active phase dispersion and improved redox properties, the best catalyst being Catalysts 2020, 10, 673 3 of 15 the TiO 2 -supported sample. In vanadyl phosphorus oxides [54][55][56][57], different vanadyl phosphate phases (mainly related to V 4+ and V 5+ , but also to V 3+ ) are typically detected in both as-prepared and used catalysts. In particular, V 4+ species have been detected in vanadyl orthophosphates under reducing reaction conditions [58].
In this work, VOPO 4 supported on γ-Al 2 O 3 , TiO 2 , and ZrO 2 were tested in the gas-phase glycerol dehydration. The catalyst performances were related to the site acid properties. Despite the possible co-presence of different vanadyl phosphates, in this work, the catalysts were labelled as VOP (vanadyl orthophosphate). Table 1 shows the specific surface areas (SSA) of the bare supports. Al 2 O 3 has the largest surface area (190 m 2 ·g −1 ), while the lowest exposed surface corresponds to ZrO 2 (49 m 2 ·g −1 ). Accordingly, the active phase load corresponding to the monolayer was directly related to the SSA and steadily increased from 3.6 to 14.0 wt. %. After the VOP deposition, the samples showed SSA values close to the corresponding supports, suggesting a good dispersion of the active phase. XRD spectra of both fresh and used samples (not reported) showed only the signals corresponding to the supports, and no peak due to vanadyl orthophosphate was detected, which was in agreement with previous results [53]. Thus, no VOP phase segregation occurs in these samples, suggesting a good active phase dispersion onto the support and no segregation phenomena occurring under the reaction conditions.

Catalysts Characterization
The results in terms of NH 3 -TPD profiles and the quantitative analysis are shown in Figures 3  and 4, respectively. The profile shapes are similar, suggesting that the same acid sites are detectable on each sample. This also suggested the negligible contribution of the supports (characterized by very different acidity) to overall acidity of the samples. This agreed with the good dispersion of the active phase. However, it is worth noting that the interaction between the VOP and support strongly affects the active phase features [59]. As shown in Figures 3a and 4, VOP/Al is characterized by the largest amount of acid sites. This sample was characterized by the highest VOP amount, too. Figure 3b shows the desorbed ammonia per VOP weight. Profiles of VOP/Al and VOP/Ti were similar but they did not overlap. The zirconia supported sample showed a larger ammonia desorption per VOP amount. This could be addressed to the higher coordination between ammonia and the acid sites. However, due to the low VOP content, the uncertainty of the ICP analysis could be higher. Nevertheless, VOP/Zr showed the lowest catalytic activity (see Section 2.2.2), and thus, further investigations were not carried out.

Effect of the Reaction Conditions.
VOP/Ti showed the best catalytic properties towards propane oxidative dehydrogenation (ODH) [53], and thus, it was used as the reference catalyst to evaluate the effect of the reaction conditions on the catalytic performance.

Effect of the Reaction Conditions.
VOP/Ti showed the best catalytic properties towards propane oxidative dehydrogenation (ODH) [53], and thus, it was used as the reference catalyst to evaluate the effect of the reaction conditions on the catalytic performance.
The effect of the furnace temperature was investigated in the range 210-255 °C at a fixed contact time τ = 0.057 g·s·cm −3 and feed composition Gly/H2O/O2/N2 = 2/41/6/51 vol. %.  From the quantitative point of view (Figure 4), VOP/Al showed the most significant amounts of both weak and medium acid sites, while the highest amount of strong acid sites was detected in VOP/Ti. Acrolein formation has been related to the acid sites and, in particular, to the weak and the medium acidity [27,31,35,36,60,61]. It is worth noting that the strong acid sites can be related to the catalyst deactivation by coking [27,35,36,60,61].
Redox properties of the samples were previously studied [53], and the corresponding results are given in Table 1 as H 2 /V. VOP/Al shows the lowest H 2 /V ratio, suggesting that vanadium is less reducible. Conversely, the VOP/Zr sample was the most reducible.

Effect of the Reaction Conditions
VOP/Ti showed the best catalytic properties towards propane oxidative dehydrogenation (ODH) [53], and thus, it was used as the reference catalyst to evaluate the effect of the reaction conditions on the catalytic performance.
In Figure 5, the glycerol conversion ( Figure 5a) and the yields of several products (Figure 5b-d) are shown as a function of temperature.
Catalysts 2020, 10, x FOR PEER REVIEW 5 of 14 In Figure 5, the glycerol conversion ( Figure 5a) and the yields of several products (Figure 5b-d) are shown as a function of temperature. Glycerol conversion was high and quite complete in the whole temperature range investigated. Acrolein was the most abundant condensable product, while the acrylic acid yield was very low.
From these results, it appeared that in the presence of molecular oxygen, VOP was poorly active towards acrolein oxidation to acrylic acid or, more probably, too active towards acrylic acid oxidation. This behavior has been previously reported on vanadyl pyrophosphate [62]. The maximum yield (16%) has been obtained at 235 °C. Detected by-products included acetic acid, formic acid, and glycolic acid. Among these species, only the production of the acetic acid was significant, even if its highest yield was lower than 7% ( Figure 5d). Coke formation was negligible (Figure 5b), due to the good oxidation properties of the VOP catalyst. As reported in Section 1, resistance to coking is a major issue to be solved [48] and the intimate contact between the acid sites, related to phosphate, and the redox sites, related to vanadium, provides fast conversion of coke precursors into COx, which are the most abundant products (Figure 5b). CO increases at temperatures above 235 °C, while CO2 slightly decreases; therefore, oxidation of acrolein and acetic acid (where the yields decrease by  Glycerol conversion was high and quite complete in the whole temperature range inv Acrolein was the most abundant condensable product, while the acrylic acid yield was ver From these results, it appeared that in the presence of molecular oxygen, VOP was poo towards acrolein oxidation to acrylic acid or, more probably, too active towards ac oxidation. This behavior has been previously reported on vanadyl pyrophosphate maximum yield (16%) has been obtained at 235 °C. Detected by-products included acetic ac acid, and glycolic acid. Among these species, only the production of the acetic acid was s even if its highest yield was lower than 7% ( Figure 5d). Coke formation was negligible (F due to the good oxidation properties of the VOP catalyst. As reported in Section 1, resistance is a major issue to be solved [48] and the intimate contact between the acid sites, related to p and the redox sites, related to vanadium, provides fast conversion of coke precursors into C acid yield dropped down to zero at the highest contact time. These results suggested that oxidation of acids to COx was significantly faster than acrolein oxidation, while acrolein formation was faster than its oxidation at this temperature. Coke formation was still negligible. The combined effect of temperature and contact time on glycerol conversion and acrolein yield is shown in Figure 7. Independently from the contact time, temperatures higher than 200 °C were needed to get complete (or almost complete) conversion. Moreover, at low temperatures, incomplete carbon balance was obtained due to the formation of unknown products (not reported). The reaction temperatures providing the best acrolein production decreased. This behavior was less prominent for other products (not reported), but by decreasing the contact time from 0.088 g•s•cm −3 to 0.043 g•s•cm −3 , the production peak shifted at higher temperatures.

Effect of the Support.
The effect of the support has been studied at fixed contact time (τ = 0.057 g•s•cm −3 ) and feed composition (Gly/H2O/O2/N2 = 2/41/6/51 vol. %). As for VOP/Ti, acrylic acid, formic acid, and glycolic acid were detected. However, due to their very low amounts, the corresponding yields were not reported. In Figure 8, the glycerol conversion and the yields to acrolein, CO, CO2, acetic acid, and coke obtained on the VOP/x catalysts are compared. In Figure 5, the glycerol conversion ( Figure 5a) and the yields of several products (Figure 5b-d) are shown as a function of temperature. Glycerol conversion was high and quite complete in the whole temperature range investigated. Acrolein was the most abundant condensable product, while the acrylic acid yield was very low.
From these results, it appeared that in the presence of molecular oxygen, VOP was poorly active towards acrolein oxidation to acrylic acid or, more probably, too active towards acrylic acid oxidation. This behavior has been previously reported on vanadyl pyrophosphate [62]. The maximum yield (16%) has been obtained at 235 °C. Detected by-products included acetic acid, formic acid, and glycolic acid. Among these species, only the production of the acetic acid was significant, even if its highest yield was lower than 7% (Figure 5d). Coke formation was negligible (Figure 5b), due to the good oxidation properties of the VOP catalyst. As reported in Section 1, resistance to coking is a major issue to be solved [48] and the intimate contact between the acid sites, related to phosphate, and the redox sites, related to vanadium, provides fast conversion of coke precursors into COx, which are the most abundant products (Figure 5b). CO increases at temperatures above 235 °C, while CO2 slightly decreases; therefore, oxidation of acrolein and acetic acid (where the yields decrease by increasing the temperature above 235 °C) mainly produces carbon monoxide. in the whole temperature range investigated. while the acrylic acid yield was very low. of molecular oxygen, VOP was poorly active probably, too active towards acrylic acid rted on vanadyl pyrophosphate [62]. The cted by-products included acetic acid, formic production of the acetic acid was significant, ). Coke formation was negligible (Figure 5b), . As reported in Section 1, resistance to coking t between the acid sites, related to phosphate, onversion of coke precursors into COx, which ses at temperatures above 235 °C, while CO2 d acetic acid (where the yields decrease by Glycerol conversion was high and quite complete in the whole temperature range investigated. Acrolein was the most abundant condensable product, while the acrylic acid yield was very low. From these results, it appeared that in the presence of molecular oxygen, VOP was poorly active towards acrolein oxidation to acrylic acid or, more probably, too active towards acrylic acid oxidation. This behavior has been previously reported on vanadyl pyrophosphate [62]. The maximum yield (16%) has been obtained at 235 • C. Detected by-products included acetic acid, formic acid, and glycolic acid. Among these species, only the production of the acetic acid was significant, even if its highest yield was lower than 7% (Figure 5d). Coke formation was negligible (Figure 5b), due to the good oxidation properties of the VOP catalyst. As reported in Section 1, resistance to coking is a major issue Catalysts 2020, 10, 673 6 of 15 to be solved [48] and the intimate contact between the acid sites, related to phosphate, and the redox sites, related to vanadium, provides fast conversion of coke precursors into CO x , which are the most abundant products (Figure 5b). CO increases at temperatures above 235 • C, while CO 2 slightly decreases; therefore, oxidation of acrolein and acetic acid (where the yields decrease by increasing the temperature above 235 • C) mainly produces carbon monoxide.
The effect of contact time on the catalytic properties is shown in Figure 6, at 230 • C. Glycerol conversion slightly increases by increasing the contact time. As expected, acrolein yield increased, while acrylic acid, acetic acid, and glycolic acid yields showed a maximum. In particular, the acetic acid yield dropped down to zero at the highest contact time. These results suggested that oxidation of acids to CO x was significantly faster than acrolein oxidation, while acrolein formation was faster than its oxidation at this temperature. Coke formation was still negligible.
Catalysts 2020, 10, x FOR PEER REVIEW 6 of 14 acid yield dropped down to zero at the highest contact time. These results suggested that oxidation of acids to COx was significantly faster than acrolein oxidation, while acrolein formation was faster than its oxidation at this temperature. Coke formation was still negligible. The combined effect of temperature and contact time on glycerol conversion and acrolein yield is shown in Figure 7. Independently from the contact time, temperatures higher than 200 °C were needed to get complete (or almost complete) conversion. Moreover, at low temperatures, incomplete carbon balance was obtained due to the formation of unknown products (not reported). The reaction temperatures providing the best acrolein production decreased. This behavior was less prominent for other products (not reported), but by decreasing the contact time from 0.088 g·s·cm −3 to 0.043 g·s·cm −3 , the production peak shifted at higher temperatures.

Effect of the Support.
The effect of the support has been studied at fixed contact time (τ = 0.057 g·s·cm −3 ) and feed composition (Gly/H2O/O2/N2 = 2/41/6/51 vol. %). As for VOP/Ti, acrylic acid, formic acid, and glycolic acid were detected. However, due to their very low amounts, the corresponding yields were not  Glycerol conversion was high and quite complete in the whole temperature range invest Acrolein was the most abundant condensable product, while the acrylic acid yield was very l From these results, it appeared that in the presence of molecular oxygen, VOP was poorly towards acrolein oxidation to acrylic acid or, more probably, too active towards acryl oxidation. This behavior has been previously reported on vanadyl pyrophosphate [62 maximum yield (16%) has been obtained at 235 °C. Detected by-products included acetic acid, acid, and glycolic acid. Among these species, only the production of the acetic acid was sign even if its highest yield was lower than 7% (Figure 5d). Coke formation was negligible (Figu due to the good oxidation properties of the VOP catalyst. As reported in Section 1, resistance to is a major issue to be solved [48] and the intimate contact between the acid sites, related to pho acid yield dropped down to zero at the highest contact time. These results suggested that oxidation of acids to COx was significantly faster than acrolein oxidation, while acrolein formation was faster than its oxidation at this temperature. Coke formation was still negligible. The combined effect of temperature and contact time on glycerol conversion and acrolein yield is shown in Figure 7. Independently from the contact time, temperatures higher than 200 °C were needed to get complete (or almost complete) conversion. Moreover, at low temperatures, incomplete carbon balance was obtained due to the formation of unknown products (not reported). The reaction temperatures providing the best acrolein production decreased. This behavior was less prominent for other products (not reported), but by decreasing the contact time from 0.088 g•s•cm −3 to 0.043 g•s•cm −3 , the production peak shifted at higher temperatures.

Effect of the Support.
The effect of the support has been studied at fixed contact time (τ = 0.057 g•s•cm −3 ) and feed composition (Gly/H2O/O2/N2 = 2/41/6/51 vol. %). As for VOP/Ti, acrylic acid, formic acid, and glycolic acid were detected. However, due to their very low amounts, the corresponding yields were not reported. In Figure 8, the glycerol conversion and the yields to acrolein, CO, CO2, acetic acid, and In Figure 5, the glycerol conversion ( Figure 5a) and the yields of several products (Figure 5b-d) are shown as a function of temperature. Glycerol conversion was high and quite complete in the whole temperature range investigated. Acrolein was the most abundant condensable product, while the acrylic acid yield was very low.
From these results, it appeared that in the presence of molecular oxygen, VOP was poorly active towards acrolein oxidation to acrylic acid or, more probably, too active towards acrylic acid oxidation. This behavior has been previously reported on vanadyl pyrophosphate [62]. The maximum yield (16%) has been obtained at 235 °C. Detected by-products included acetic acid, formic acid, and glycolic acid. Among these species, only the production of the acetic acid was significant, even if its highest yield was lower than 7% (Figure 5d). Coke formation was negligible (Figure 5b), due to the good oxidation properties of the VOP catalyst. As reported in Section 1, resistance to coking is a major issue to be solved [48] and the intimate contact between the acid sites, related to phosphate, and the redox sites, related to vanadium, provides fast conversion of coke precursors into COx, which are the most abundant products (Figure 5b). CO increases at temperatures above 235 °C, while CO2 slightly decreases; therefore, oxidation of acrolein and acetic acid (where the yields decrease by In Figure 5, the glycerol conversion ( Figure 5a) and the yields of several products (Figure 5b-d) are shown as a function of temperature. Glycerol conversion was high and quite complete in the whole temperature range investigated. Acrolein was the most abundant condensable product, while the acrylic acid yield was very low.
From these results, it appeared that in the presence of molecular oxygen, VOP was poorly active towards acrolein oxidation to acrylic acid or, more probably, too active towards acrylic acid oxidation. This behavior has been previously reported on vanadyl pyrophosphate [62]. The maximum yield (16%) has been obtained at 235 °C. Detected by-products included acetic acid, formic acid, and glycolic acid. Among these species, only the production of the acetic acid was significant, even if its highest yield was lower than 7% (Figure 5d). Coke formation was negligible (Figure 5b), due to the good oxidation properties of the VOP catalyst. As reported in Section 1, resistance to coking is a major issue to be solved [48] and the intimate contact between the acid sites, related to phosphate, and the redox sites, related to vanadium, provides fast conversion of coke precursors into COx, which are the most abundant products (Figure 5b). CO increases at temperatures above 235 °C, while CO2 slightly decreases; therefore, oxidation of acrolein and acetic acid (where the yields decrease by The combined effect of temperature and contact time on glycerol conversion and acrolein yield is shown in Figure 7. Independently from the contact time, temperatures higher than 200 • C were needed to get complete (or almost complete) conversion. Moreover, at low temperatures, incomplete carbon balance was obtained due to the formation of unknown products (not reported). The reaction temperatures providing the best acrolein production decreased. This behavior was less prominent for other products (not reported), but by decreasing the contact time from 0.088 g·s·cm −3 to 0.043 g·s·cm −3 , the production peak shifted at higher temperatures.

Effect of the Support
The effect of the support has been studied at fixed contact time (τ = 0.057 g·s·cm −3 ) and feed composition (Gly/H 2 O/O 2 /N 2 = 2/41/6/51 vol. %). As for VOP/Ti, acrylic acid, formic acid, and glycolic acid were detected. However, due to their very low amounts, the corresponding yields were not reported. In Figure 8, the glycerol conversion and the yields to acrolein, CO, CO 2 , acetic acid, and coke obtained on the VOP/x catalysts are compared.  Glycerol conversion was about 60% on both VOP/Zr and VOP/Al at 180 • C. By increasing the temperature, glycerol conversion became quite complete at 200 • C for VOP/Zr, while higher temperatures were needed for VOP/Al (250 • C).
On VOP/Zr yields to acrolein and acetic acid show a maximum (7.4% and 3.3%, respectively) at about 200 • C. Their decrease is mainly compensated by a corresponding increase of the CO yield, in agreement with the results on VOP/Ti. Coke yield is very low in the whole temperature range, becoming entirely negligible above 200 • C. Carbon balance (not reported) is about 100% except at 180 • C, where it is about 40%. It is worth noting that no COx production was detected at this temperature. Thus, VOP/Zr could activate and convert glycerol into condensable species, but oxidation activity was negligible. As stated above, VOP/Al needs higher temperatures to convert glycerol fully. Interestingly, significant peak production of acrolein was obtained (35% yield) at about 300 • C, with about 30% yield in the temperature range 265-325 • C. Yield to acetic acid showed a maximum at 285 • C. The CO yield continuously increased while increasing the temperature, while the CO 2 yield showed a sharp increase between 200 and 265 • C, and then it increased moderately. Coke formation was higher than that obtained on the other samples, but it was still very low, being below 0.4% in the whole temperature range. Carbon balance is reported in Figure 9. It increased by increasing the temperature; however, it never reached 100%, with the highest value being 95%. This meant that (i) unknown products are produced in the whole temperature range (in low amounts starting from 300 • C) and (ii) the CO yield mainly increases due to the oxidation of the unknown species.
Interestingly, significant peak production of acrolein was obtained (35% yield) at about 300 °C, with about 30% yield in the temperature range 265-325 °C. Yield to acetic acid showed a maximum at 285 °C. The CO yield continuously increased while increasing the temperature, while the CO2 yield showed a sharp increase between 200 and 265 °C, and then it increased moderately. Coke formation was higher than that obtained on the other samples, but it was still very low, being below 0.4% in the whole temperature range. Carbon balance is reported in Figure 9. It increased by increasing the temperature; however, it never reached 100%, with the highest value being 95%. This meant that i) unknown products are produced in the whole temperature range (in low amounts starting from 300 °C) and ii) the CO yield mainly increases due to the oxidation of the unknown species.
The behavior of the catalytic performance in terms of glycerol conversion and products distribution as a function of the reaction temperature is not directly related to the VOP amount. This result suggests the role of the support, as evidenced by NH3-TPD characterization in terms of the acid sites distribution.
In this sample, the effect of oxygen concentration in the feed was studied ( Figure 10) at 325 °C. With respect to the reaction conditions reported above (O2 = 6 vol. %), a slight oxygen decrease provided improved performance, with the yield to acrolein increasing to about 40%, while a decrease in the acrolein yield was found when increasing O2 concentration, mainly due to acrolein oxidation.
In the presence of oxygen in the feed, glycerol conversion was 100%, while carbon balance was >90%, reaching 100% at 10 vol. % O2 in the feed. On the contrary, in the absence of oxygen in the feed, glycerol conversion was incomplete, where the yield to acrolein decreased down to 15.7%, no COx production was detected (not reported) and the carbon balance declined to about 30%. Coke formation significantly increased, with coke yield being 1.5%. These results suggest that in the absence of molecular oxygen, oxygenated compounds other than acrolein, acetic acid acrylic acid, glycolic acid, and formic acid, and high boiling point compounds, such as glycerol oligomers, acrolein oligomers, and aldol condensation products, are preferentially produced. The behavior of the catalytic performance in terms of glycerol conversion and products distribution as a function of the reaction temperature is not directly related to the VOP amount. This result suggests the role of the support, as evidenced by NH 3 -TPD characterization in terms of the acid sites distribution.
In this sample, the effect of oxygen concentration in the feed was studied ( Figure 10) at 325 • C. With respect to the reaction conditions reported above (O 2 = 6 vol. %), a slight oxygen decrease provided improved performance, with the yield to acrolein increasing to about 40%, while a decrease in the acrolein yield was found when increasing O 2 concentration, mainly due to acrolein oxidation.
In the presence of oxygen in the feed, glycerol conversion was 100%, while carbon balance was >90%, reaching 100% at 10 vol. % O 2 in the feed. On the contrary, in the absence of oxygen in the feed, glycerol conversion was incomplete, where the yield to acrolein decreased down to 15.7%, no CO x production was detected (not reported) and the carbon balance declined to about 30%. Coke formation significantly increased, with coke yield being 1.5%. These results suggest that in the absence of

Discussion
In this work, the dehydration of glycerol to acrolein over supported VOPO4 (VOP) catalysts was studied. A theoretical VOP monolayer was supported on TiO2, Al2O3, and ZrO2; where deposition onto these supports guaranteed good active phase dispersion. All the samples show low coke formation in the presence of molecular oxygen in the feed, due to their redox properties. However, the higher the redox ability, the lower the acrolein yield due to the oxidation of the desired product to COx. Acrolein is the main condensable product, and low yields to acrylic acid, formic acid, acetic acid, and glycolic acid were measured. Unknown products are rarely detected; the corresponding reaction conditions are not interesting due to incomplete glycerol conversion and/or negligible acrolein production.
VOP supported onto alumina provided the best catalytic performances. Dehydration activity is related to the abundance of weak and medium acid sites, while its lowest redox activity preserves acrolein from oxidation. The catalysts investigated exhibit excellent resistance to coke formation, thereby suggesting it as a good candidate for the glycerol dehydration to acrolein.
Differently from the literature tested samples, the VOP/Al catalyst has shown very high resistance to coking. As a consequence, it is a good candidate for the glycerol dehydration to acrolein, provided that the active phase loading and the operating conditions are not optimized.
Future works will be devoted to the optimization of the active phase loading and the operating conditions to maximize the acrolein yield. Moreover, the effect of crude glycerol, containing several organic and inorganic impurities, will be investigated. Meanwhile, the presence of other organic compounds can affect the reaction pathway, and the inorganic compounds (mainly alkali metal ions, such as Na + and K + ) can poison the catalyst by affecting the number and the strength of the acid sites.

Discussion
In this work, the dehydration of glycerol to acrolein over supported VOPO4 (VOP) catalysts was studied. A theoretical VOP monolayer was supported on TiO2, Al2O3, and ZrO2; where deposition onto these supports guaranteed good active phase dispersion. All the samples show low coke formation in the presence of molecular oxygen in the feed, due to their redox properties. However, the higher the redox ability, the lower the acrolein yield due to the oxidation of the desired product to COx. Acrolein is the main condensable product, and low yields to acrylic acid, formic acid, acetic acid, and glycolic acid were measured. Unknown products are rarely detected; the corresponding reaction conditions are not interesting due to incomplete glycerol conversion and/or negligible acrolein production.
VOP supported onto alumina provided the best catalytic performances. Dehydration activity is related to the abundance of weak and medium acid sites, while its lowest redox activity preserves acrolein from oxidation. The catalysts investigated exhibit excellent resistance to coke formation, thereby suggesting it as a good candidate for the glycerol dehydration to acrolein.
Differently from the literature tested samples, the VOP/Al catalyst has shown very high resistance to coking. As a consequence, it is a good candidate for the glycerol dehydration to acrolein, provided that the active phase loading and the operating conditions are not optimized.
Future works will be devoted to the optimization of the active phase loading and the operating conditions to maximize the acrolein yield. Moreover, the effect of crude glycerol, containing several organic and inorganic impurities, will be investigated. Meanwhile, the presence of other organic compounds can affect the reaction pathway, and the inorganic compounds (mainly alkali metal ions, such as Na + and K + ) can poison the catalyst by affecting the number and the strength of the acid sites.

Discussion
In this work, the dehydration of glycerol to acrolein over supported VOPO 4 (VOP) catalysts was studied. A theoretical VOP monolayer was supported on TiO 2 , Al 2 O 3 , and ZrO 2 ; where deposition onto these supports guaranteed good active phase dispersion. All the samples show low coke formation in the presence of molecular oxygen in the feed, due to their redox properties. However, the higher the redox ability, the lower the acrolein yield due to the oxidation of the desired product to CO x . Acrolein is the main condensable product, and low yields to acrylic acid, formic acid, acetic acid, and glycolic acid were measured. Unknown products are rarely detected; the corresponding reaction conditions are not interesting due to incomplete glycerol conversion and/or negligible acrolein production.
VOP supported onto alumina provided the best catalytic performances. Dehydration activity is related to the abundance of weak and medium acid sites, while its lowest redox activity preserves acrolein from oxidation. The catalysts investigated exhibit excellent resistance to coke formation, thereby suggesting it as a good candidate for the glycerol dehydration to acrolein.
Differently from the literature tested samples, the VOP/Al catalyst has shown very high resistance to coking. As a consequence, it is a good candidate for the glycerol dehydration to acrolein, provided that the active phase loading and the operating conditions are not optimized.
Future works will be devoted to the optimization of the active phase loading and the operating conditions to maximize the acrolein yield. Moreover, the effect of crude glycerol, containing several organic and inorganic impurities, will be investigated. Meanwhile, the presence of other organic compounds can affect the reaction pathway, and the inorganic compounds (mainly alkali metal ions, such as Na + and K + ) can poison the catalyst by affecting the number and the strength of the acid sites.
Each support was loaded with a VOP amount corresponding to a theoretical monolayer and, thus, depended on the support specific surface area, as reported in Table 1. The VOP was loaded onto the supports by the impregnation method, as described in Reference [53].
All these catalysts were indicated as VOP/M, where M was the metal ion of the oxide support.

Physico-Chemical Characterizations
The actual composition of the investigated samples was determined using inductively coupled plasma mass spectrometry (ICP-MS) analysis using an Agilent 7500CE instrument (Agilent, Santa Clara, CA, USA). Results showed differences with the nominal content within the experimental error (± 5%).
The specific surface areas (SSA) of supports and catalysts were measured by N 2 adsorption at 77 K with a Quantachrome Autosorb-1C instrument (Quantachrome, Boynton Beach, FL, USA) after degassing the samples at 150 • C for 1.5 h and calculated according to the Brunauer-Emmett-Teller (BET) method. No significant differences were detected between fresh and used samples.
Temperature programmed desorption of ammonia (NH 3 -TPD) were carried out using a Micromeritics Autochem II 2020 (Micromeritics, Norcross, GA, USA) equipped with a TC detector (Norcross, GA, USA). The samples (≈ 150 mg) were pre-treated in helium for 15 min at 450 • C, then saturated with a 5 vol. % NH 3 /N 2 mixture for 1 h at 30 • C and washed in nitrogen for 0.5 h. The reactor was then heated at 10 • C/min up to 550 • C. NH 3 -TPD profiles were deconvoluted using Origin software (OriginPro 2015 version) (OriginLab Corporation, Northampton, MA, USA, 2015). Three contributions were detected and assigned to weak, medium, and strong acid sites. NH 3 -TPD tests were carried out on both the fresh and the used samples. No significant differences were detected.
Temperature programmed reduction in hydrogen (H 2 -TPR) are reported in Reference [53], and here, they were used to discuss the relationship between catalytic activity and redox properties.

Catalytic Tests
Catalytic tests were conducted in a lab-scale test rig. The experimental apparatus was developed to carry out activity tests under stationary conditions, varying several operating parameters, such as temperature, reactant flow rates, and concentrations.
The flow rates of permanent gases (O 2 and N 2 ) were independently controlled using mass flow controllers (Brooks model SLA5850S), while a liquid 20 wt. % glycerol-water solution [5,18,31,64,65] was fed through a volumetric pump (KNF SIMDOS ® 02), vaporized using a heating tape, set at about 160 • C, and mixed with the permanent gases before entering the reactor. Pure glycerol (Sigma-Aldrich, purity ≥ 99.5%) was used as purchased. Crude glycerol from an alkali transesterification plant contains some impurities, such as methanol, soap, free fatty acids (FFAs), fatty acid methyl ester (FAMEs), glycerides. In addition, alkali metal ions (as Na + and K + ) are generally found in crude glycerol [66]; these alkali metals can poison the catalyst [66].
The catalyst (180-300 µm) was placed in a vertical quartz tube. The reaction temperature was set using a heating jacket (Tyco Thermal Controls) equipped with a PID controller. A K-type thermocouple was placed inside the reactor, thereby providing the reaction temperature.
Downstream the reactor, condensable species were separated from the gas mixture in three flasks in series placed in an ice bath. The collected liquid was analyzed off-line in a high-pressure liquid chromatography (HPLC) system (Agilent1100), provided with a refractive index detector (RID) and a diode Array type of UV/VIS Detector (DAD). Measurements were repeated three times, and the average values were used for the mass balance. However, values differed within the experimental error (±3%).
Incondensable species (CO, CO 2 , O 2 ), dried by a CaCl 2 trap, were analyzed in a continuous analysis system (ABB AO2000), provided with a paramagnetic detector for O 2 measurement and a nondispersive infrared (NDIR) detector for CO and CO 2 measurements. The absolute uncertainty of the analyzer depends on the measurement range and corresponds to 50 ppm, 250 ppm, and 1000 ppm for CO, CO 2 , and O 2, respectively. These values are much lower than the corresponding measurements during the reaction tests, thereby suggesting good reliability and accuracy of the measurements.
Gas-phase glycerol concentration was set at 2 vol. %, while oxygen concentration was in the range 0-10 vol. %. Nitrogen was used as a balance. The temperature was in the range 170-360 • C and the contact time in the range 0.04-0.09 g·s·cm −3 . It is worth noting that glycerol concentration was below its lower flammability limit (3 vol. %).
After each catalytic test, temperature-programmed oxidation (TPO) was carried out to (i) verify the formation of carbonaceous deposits onto the catalyst surface and (ii) clean the catalyst surface for the following catalytic test.

Conclusions
Supported vanadyl orthophosphate (VOP) were tested in the glycerol dehydration to acrolein. VOP supported onto γ-alumina showed the best performance, related to the highest amount of weak and medium acid sites. The oxygen co-feeding provides a good resistance to coking, which is, indeed, the main open issue of the catalytic systems for this reaction. The main by-products are CO and CO 2 ; while high selectivity towards acrolein is detected among the condensable products.
Catalytic performance is comparable to the performance of other catalytic systems but lower than that of the best catalysts. In this work, no reaction conditions nor active phase loading were optimized, suggesting that VOP supported onto γ-alumina are good candidates for glycerol dehydration to acrolein.