Identication of compounds adsorbed and deposited on the Ni-Mo catalysts surface of alumina and SBA-15 by DRIFTS, Raman and TPO Analyses before and after Guaiacol hydrodeoxygenation.

We studied and identied compounds adsorbed or deposited on the catalysts surface of the Ni-Mo supported on alumina and SBA-15, before and after hydrodeoxigenation of a bio-oil model (guaiacol). Marked differences were observed on both catalysts through DRIFTS and Raman spectroscopy showing that the alumina-supported catalyst contains deposits of aromatic and oxygenated organic substances, while the carbon deposits on the SBA-15 as aliphatic simple molecules. TPO analyses conrm that the carbon deposited on the NiMo/SBA-15 catalyst were light polymer types, mainly nanotubes and nano bers, while on the alumina catalyst the mainly carbon species formed were graphite type and heavier carbons. Post reaction analysis of the catalysts showed coke formation and carbon deposition rate of 1.14 mg coke .g cat−1 h − 1 for NiMo/SBA-15 and the deactivation was 44 % higher for the NiMo/Al 2 O 3 with 1.65 mg coke .g cat−1 h − 1 of carbon deposition rate.

moderate acidity and high surface area, such as HZSM-5 and Al-SBA-15, have better results for oxygen elimination and hydrogenation of aromatic rings, which makes them ideal catalysts for HDO process.
Recently, Hewer et al. [17] studied nickel and molybdenum nanoparticles on silicon aluminum phosphate  and mesoporous silica (SBA-15), respectively, and compared with γ-Al 2 O 3 support. These catalysts exhibited excellent catalytic activity for hydrodeoxygenation (HDO) of anisole and the hydrodeoxygenation and hydro-dearomatization reactions are strongly affected by the support nature due to the Brönsted and Lewis acid sites [17].
However, there are different causes that provoke route directions of the hydrogenation process for the formation of HDO and HDA or deoxygenation to aliphatic compounds affecting of the catalyst, which is related to the adsorbed species or intermediate compounds at the surface of the catalysts, which are really the precursors for the selectivity that depend signi cantly on the supports, in particular, the surface acid sites and dispersions of metallic phases. These species are precursors of carbon formation. Therefore, we study the nature of the compound deposition occurring on alumina and SBA-15 supports with Ni-Mo, and identi cation by the adsorption of intermediate compounds formation during the hydrogenation process of a model compound of bio-oils, analyzing the surface species deposited on the catalysts by comparing before and after reaction. For this purpose, we used spectroscopic analyses, in particular DRIFTS and Raman techniques, complemented by TGA and TPO analyzes.

2.1Catalysts preparation
The SBA-15 support was synthesized following the method described elsewhere [15] [18]. In short, 4 g of the polymer Pluronic P213 (PEO20PPO70PEO20) from BASF was dissolved in 122 g of HCl 2M solution and kept under stirring for 1 h. Then, 8.6 g of tetra orthosilicate (Fluka) was added and the solution was continuously stirred for 24 h. The mixture underwent a hydrothermal treatment during 48 h at 100 °C in an autoclave, washed thoroughly with deionized water and dried at room temperature. Finally, the polymeric support was calcined at 540 °C in N 2 ow for 2h and in air for 4 h. The alumina support (γ-Al 2 O 3 ) was acquired from Degussa and calcined at 500 °C for 20 h with synthetic air ow and at a rate of 5 °C/min. Then, nickel and molybdenum were added by wet impregnation in separate. For the required amount of Molybdenum oxide, the salt precursor ((NH 4 ) 6 Mo 7 O 24 .4H 2 O) was dissolved in deionized water and mixed with the support and the suspension was stirred for 2 h. Afterwards, the mixture was lyophilized for eliminating the residual solvent and calcined at 500 °C for 2 h in air ow of 50 ml min -1 . Then, the Ni(NO 3 ) 2 .6 H 2 O salt (ACROS Organics) was dissolved similarly and impregnated in sequence and calcined under similar conditions.

Characterizations
The textural properties of the catalysts were analyzed by N 2 adsorption isotherm, using the Quantacrome volumetric adsorption analyzer (model 100E). The surface area was determined by BET technique and the pore distribution obtained by the BJH method [19,20]. Complete characterization of these materials were reported elsewhere [17].
The metal surface areas and dispersions of the catalysts were determined by H 2 chemisorption in an ASAP 2020C chemisorption system (Micrometrics). In a typical analysis, a 120 mg of sample was loaded in the reactor and dried with a He ow for 1h at 300 °C and at rate of 10°C min -1 . Then, switched to 5% H 2 /Ar ow and reduced for 4 h at 550 °C and 10 °C min -1 . After cooling down to 35 °C and evacuation the total and reversible isotherms and, consequently, the metallic surface areas and dispersions were obtained.
The total acidity of the supports and catalysts was measured by the temperature programmed desorption of NH3 (NH3-TPD). Approximately 200 mg of the sample was pretreated with a ow of He (30 ml min -1 ) at 250 °C for 1 h and at 5 °C min -1 . After cooling to room temperature, the gas was replaced by a ow of 4% NH 3 / He (60 ml min -1 ) at 150 °C for 1 h and cooled to room temperature under pure He ow. Then ammonia desorption took place, raising the temperature at 20 °C min -1 up to 700 °C. The amount of desorbed ammonia was measured on a quadrupole mass spectrometer (Balzers, PRISMA) The adsorption on the catalysts surface was studied by in-situ DRIFTS cell using a Nicolet Nexus 870 instrument with a DTGS-TEC detector and a Thermo Spectra-Tech reaction chamber with ZnSe windows. To reduce the sample, H 2 was owed (40mL/min) at 500 °C for a period of 1 h and the temperature was decreased to 200 °C while He was owed to record a spectrum for use as a background. Guaiacol adsorption involved bubbling He (40 mL/min) through a saturator with guaiacol/n-heptane solution (2% guaiacol, mass %) for 30 min, followed by a He purge. Spectra were recorded at 200, 250 and 300 °C. The scan resolution was 4 cm -1 and 512 scans were taken.
The Fourier-transform infrared spectroscopy (FTIR) measurements were performed in a Shimadzu IRPrestige-21 spectrometer using 20 mg of the sample pelletized with KBr. Spectra before and after reaction were recorded in a wavelength between 400-4000 cm -1 .
The Raman spectrometer, a Renishaw Confocal Raman Microscope model with excitation wavelength of 632 nm, was used for determining carbon deposition after reaction. .The crystallite sizes L a of the graphite deposited over the catalysts surface were obtained, according to the Equation 1 [21]: Where, λ laser is the laser wave length used in the Raman experiments, I D and I G are the intensity Raman D and G bands of the carbon depositions.
Thermogravimetric analysis (TGA) and differential thermal analysis (DTA) were performed under a synthetic air ow (100mL min -1 ), using a Shimadzu DTG-60 equipment. The samples were heated from room temperature up to 900 °C at 10 °C min -1 .
The carbon deposits formed on the catalysts surface were analyzed by temperature programmed oxidation technique (TPO) in an ASAP 2020C chemisorption system (Micromeritics). In a typical analysis, 120 mg of the sample was dried with He, at 90 mL min -1 ow and 300 °C for 30 minutes. Then, switched to the O 2 /Ar mixture (5 Vol. % O 2 ) ow at 60 mL min -1 and heated up to 600 °C at 10 °C min -1 .

Catalytic activity tests
The catalytic tests were carried out in a xed bed ow reactor system at 15 kgf.cm -2 pressure. The mass around 200 mg was pre-treated in situ with pure H 2 (AGA) at 50 mL min -1 ow, at 500 °C for 4 h, and then cooled to the reaction temperature (200, 250 or 300 °C) under hydrogen atmosphere. The feed mixture consisted of 2 % w/w Guaiacol (MERCK) in heptane P.A. (Aldrich) pumped at 14.8 mL min -1 , and carried with H 2 ow at 100 mL min -1 . The reactor input (reactants) and output (products) compounds were maintained in vapor phase, with a steal line heated at 125 °C, until the 6-way valve of the gas chromatograph. The organic compounds were analyzed by gas chromatography in a Shimadzu CG17A equipment, with FID and mass spectrometric detectors, capillary column CP sil 5CB and automatic injection. Conversion, selectivity and reaction rates were determined according to equations 2-4, respectively.
Where, F 0 is the molar ow rate of Guaiacol, M GUA is the Guaiacol molecular weight and m cat , the mass of the catalyst. Table 1 presents the surface areas of the supports Al 2 O 3 and SBA-15, 279 m² g − 1 and 630 m² g − 1 ,

Results And Discussion
respectively. The SBA-15 presents a total pore volume of 1.17 cm³ g − 1 and an average pore diameter of 7 nm. The surface area and the pore volume of the catalyst NiMo/SBA-15 decreased to 332 m²g − 1 and 0.68 cm³g − 1 , respectively. The surface area of NiMoAl 2 O 3 catalyst also decreased, as shown in Table 1. The total acidity of the supports and catalysts was measured by Temperature Programmed Desorption of NH 3 (NH 3 -TPD). Table 1 showed that SBA15 has a total acidity of 470 µmol g − 1 , while the alumina of 548 µmol g − 1 . The excess of acidity (> 890 µmol g − 1 ) causes an increase of polymerization products, resulting in the formation of coke at the catalyst surface [22].

IR spectra of Guaiacol adsorption
The IR spectra of Guaiacol adsorbed on NiMo-SBA15 and NiMo-Al 2 O 3 are compared in Fig with δ(OH) contribution. The poor resolved band at 1355 cm − 1 , corresponding to δ(OH) vibration, close to that observed for free gGuaiacol,Guaiacol, is present in its spectra [23]. For the NiMo-SBA15, the guaiacol adsorption spectrum at low frequencies was very different from that observed for NiMo-Al 2 O 3 . The band characteristic of the ν(CC ring ) vibration shifts to 1599 cm − 1 and the band at 1492 cm − 1 of the ν(CC ring ) is clearly observed. The bands of δ(CH 3 ) vibration, a triplet between 1463 − 1442 cm − 1 , are present in the NiMo-SBA15 spectra. The band at 1355 of δ(OH) vibration does not appear in the NiMo-SBA15 spectra, which indicates the absence of free phenol on its surface.
After guaiacol adsorption, the chamber was cleaned with He ow during 30 minutes and temperature increased up to 300°C, Fig. 2A. In the high frequency region, guaiacol was still adsorbed on the SiO 2 surface and it is possible to observe the low intensity bands in the range 3000 − 2800 cm − 1 . In fact, the guaiacol adsorption leads to a decrease of free Si-OH groups, as evidenced in the negative band at 3750 cm − 1 , and this behavior indicates the silanol groups were perturbed by the interaction with guaiacol through its aromatic ring [24,25]. and the band at 1454 cm − 1 , assigned to ν(C = C ring ) and CH 3 vibrational modes, respectively. In the same way, but less intense, new bands can be observed in the 1200-1300 cm − 1 region, which are attributed to the ν(CO) vibration.

hydrodeoxygenation of Guaiacol
The catalytic test results of conversion and reaction rates at 200, 250, 300°C are shown in Table 2 and Fig. 3.  [27,[28][29][30][31]. Figure 3B displays the reaction rate with increasing temperature and both catalysts. Note that the in uence of the support was signi cant, indicating that the NiMo/SBA-15 was higher than NiMo/Al 2 O 3 .

Selectivity and products distribution
The selectivity and products are presented in Table 2 and Fig. 4  OH groups, which act as Lewis acid sites. The SBA-15 has higher ability for direct demethoxylation of Guaiacol resulting in phenol, while the alumina support has had more ability for demethylation, forming catechol, which is transformed to phenol through hydrogenolysis [5,[29][30][31][32].
It depends on presence of the metallic oxide sites Ni and Mo on the surface. The rst for activation of hydrogen and the second for adsorption of oxygenated compounds, due to the oxo facility [33]. Therefore, the molybdenum on the surface of SBA-15 favored the demethyloxylation of the guaiacol molecule, forming phenol.
In fact, the alumina support presents greater amount and higher acid forces than the SBA-15, thus directing the reaction to a different route. The strong Lewis sites adsorb guaiacol molecules through the oxygen of the methoxyl groups, and consequently, favors the rupture of methyl and adsorption on the surface [34,35]. Therefore, the main product is catechol and its methyl derivatives.
In fact, phenol can also be formed through hydrogenolysis of catechol due to the Mo sites on the surface of the supports, which explains the higher amount of the phenol formed on the alumina catalyst compared to the SBA-15 catalyst, in accordance to [30]. Indeed, the energy needed for total elimination of oxygen for both catalysts is different. On silica, it needs the rupture of the ether bindings C(sp 2 )-OCH 3 (356 kJ/mol) and the phenolic bonding C(sp 2 )¬OH (414 kJ/mol), while for the alumina catalyst, occurs the rupture of the methyl group C(sp 3 )-OC(sp 2 ) (247 kJ/mol), followed by two phenolic bonds. Therefore, the SBA-15 catalyst has greater facility to generate non-oxygenated compounds or with less oxygen atoms, following a reaction energetically more favorable than on the alumina catalyst. Results evidence that the selectivity depends on the support but signi cantly on the product distribution.
Analyses after reaction IR analyses cm − 1 to the silanol groups of Si-OH and hydroxyls of the water adsorbed on the catalyst surface [36,37].The band 1640 cm − 1 is attributed to the adsorbed water [45]. In this spectrum it was not possible to detect the Ni-O bond, which occurs between 400 and 500cm 1 , as well as the bands relative to the bonding of Mo = O of the MoO 3 , which would appear between 800 a 1000 cm − 1 , probably because the silica bands overlap the nickel and molybdenum oxide band [38,39].The Mo-O-Mo bonding, that occurs in this range [40],shows similar behavior. Finally, the small band at 704 cm − 1 can be attributed to the stretching symmetric band of Ni-Mo-O present on the mixed oxide NiMoO 4 [41].
After reaction it is noteworthy to observe signi cant modi cation in the IR spectrum of the silica catalyst.
New bands appeared at 1390 and 1470 cm -1 , which are assigned to the rotation vibration and folding of C-H bonds of methyl groups [42]. The formation of -CH 3 groups at the catalysts surface during the reaction is closely related to the deactivation due to the carbon deposition on its surface [43]. The band intensity is low, in agreement with the literature, since on silica supports the carbonaceous species formation is low [44].
The infrared spectra of the NiMo/Al 2 O 3 catalyst before and after reaction are shown in Fig. 5B. indicates that this oxide is a well-de ned crystalline structure, in agreement with the literature [51][52][53].
The Raman spectrum of the NiMo/SBA-15after reaction between 1000 and 1800 cm − 1 indicates the broad carbonaceous species present on the catalyst surface, from graphite to amorphous carbons. In fact, the band at 1324 cm − 1 is assigned as D1 carbon structures that correspond to unordered and poorly organized carbon [54]. On the other hand, at 1589 cm − 1 is the G band, indicating the well-organized structure like graphite, carbon nanotubes or carbon bers, according to SHANMUGAN et al. [55]. Besides, the band at 1265 cm − 1 shows the D4 band and the bands 1485 cm − 1 and 1552 cm − 1 , identi ed as D3 are assigned to disordered graphite matrices and amorphous carbons, respectively [54,56]. It seems that on the silica support the carbon structure is less organized (D1 band) and exhibits defects (D4 band), and therefore, less organized structures like graphite (G band). These results corroborate with the TPO analyses, which demonstrated the deposition of light carbons deposited on NiMo/SBA-15. It can also be attributed to defects of polymeric structures, like carbon nanotubes of bers. The organization degree of the carbon structure formed at catalysts surface was measured from the ratio between D and G band intensities. For NiMo/SBA-15, this ratio is 2.05 suggesting higher disordered carbon structure formed on the catalysts surface during the HDO reaction. The crystallite size L a of the carbon deposits was calculated according to Eq. 1. For NiMo/SBA-15spent catalysts it was 18.8 nm, Table 3. This value is greater than the mean pore sizes of the SBA-15 support, around 8.77nm, as observed by TEM analyses [17], which suggests that most carbon was deposited on the surface and blocked most of the pores of the support. In fact, it decreased the surface area of the NiMo/SBA-15 after reaction, and provoked high deactivation due to the carbon deposition. and D3. The calculated ratio between I D /I G was 0.99, indicating that the carbon formed deposits with similar proportion of graphitic allotrope and disordered carbon, Table 3. However, the presence of D3 and D4 bands denotes the amorphous and disordered carbon on the carbon deposits. The crystallite size of the carbon deposits formed on NiMo/Al 2 O 3 was 38.5nm, which is 2.5 higher compared to NiMo/SBA-15.
Probably, the deactivation was more pronounced since the sizes are much bigger than the pore sizes of alumina support, around 7.9nm. Thermogravimetric and Thermodifferential analyses Thermogravimetric analyses of the spent catalysts were performed with air ow and increasing temperature at 10 ºCmin -1 up to 900 ºC. The TGA and DTA results are presented in Table 4, that shows water elimination of 11.1 % and physisorbed water [57,58]. At higher temperature the weight loss was 1.8 % due to the decomposition of silanol and siloxane groups, which suggests high stability of the support at high temperatures [58,59].The spent NiMo/SBA-15catalysts showed less water elimination (5.6 %).
However, the exothermal peak at 305°C and the endothermic peaks at 431 and 536°C indicate coke burn. Then, comparing the mass loss of the fresh and spent catalysts, there is a loss of 3.1 %, and the carbon deposition rate was 1,14 mg coke g cat -1 h -1 . The carbon deposition rate was 1.65 mg coke g cat −1 h − 1 for NiMo/Al 2 O 3 . In addition, the sublimation of MoO 3 occurred at 836°C, corresponding to a mass loss of 6.5 % as observed for the non-used sample.

TPO analyses
Temperature programmed oxidation was employed for catalysts before and after reaction, using a mixture of 5% O 2 /He and exit gases were analyzed by mass spectroscopy, as shown in Fig. 7.
The results showed the mass signals of CO, CO 2 and H 2 before and after reaction. For the fresh NiMo/SBA-15 the pro les didn't show any important signal of CO or CO 2 , indicating that all organic products present were eliminated during the calcination. But, the H 2 pro le presented two broad peaks at 250 and 560 ºC, which are attributed to elimination of hydroxyls groups present at the silica surface [61].
After reaction we observe three broad peaks from room temperature up to 600 ºC, with maximum at 371 ºC for H 2 , at 400 ºC for CO and at 276 ºC for CO 2 , which indicates decomposition of the coke formed at the surface during the HDO reaction.
Considering CO 2 as reference, due to the total combustion of carbon, we suggest that most coke deposited on the NiMo/SBA-15 are of polymeric types, such as carbon nanotubes, carbon nano bers with low weight, because the production of CO 2 occurred at low temperature (276°C). According to Lee et al. [62], this behavior could be assigned to the burning of light and well-organized carbonaceous species. However, the broad peak indicates also the presence of different carbon materials of polymerization products, as also observed by Kim et al. [63] for ethanol reforming on Ni/SBA-15 e NiMo/SBA-15 catalysts. In fact, nickel promoted the polymerization of ethylene, while molybdenum the partial oxidation, resulting in CO adsorption, which was deposited as coke around metallic particles, like small donuts circling the metallic active sites. In fact, Omoregbe et al. and Wang et al. [64,65] proved that graphitic carbon on metallic sites burns at temperatures above 550 ºC, while polymeric carbon burnt at lower temperatures. These authors observed the in uence of the support with higher Lewis sites concentrations tend to improve the adsorption of acids, and carbon dioxide, which can reduce the carbon deposition.
On the other hand, the fresh NiMo/Al 2 O 3 showed higher H 2 formation than the NiMo/SBA-15 sample, which are assigned to the higher hydroxyl content at the alumina surface [66]. The low CO and CO 2 formation at lower temperature are organic impurities adsorbed on the alumina support. However, the spent catalyst displayed high amounts of de H 2 , CO e CO 2 formation, certainly due to the carbon deposition at the surface. For all signals it is possible to see two peaks, the rst one at 50 ºC extended to 250 ºC and the second one from 250 to 600 ºC, with maximum around 415 and 450 ºC. These results suggest two different carbon species deposited, one less organized and with low weight for lower temperatures and the other one well-structured and higher weight burning at higher temperatures. The rst coke burnt up to 250 ºC are assigned to aromatic organic compounds with low polymerization degree and strongly adsorbed at the surface. These results reinforce the FTIR results indicating aromatic bonds of type C = C, C-H and C-O adsorbed on the surface of the materials, which agrees with Olcese et al. [60], for the HDO on a Fe/SiO 2 catalyst. Indeed, these carbons are quite different from those deposited on SBA-15 materials. The carbon deposits formed on NiMo/Al 2 O 3 are predominantly graphite carbons.
These carbons allotropes have well organized and dense structures when compared to the carbon nanotubes and nano bers observed for the NiMo/SBA-15 catalyst [16,55,62].
Paivi et al 67] studied the Hydro deoxygenation (HDO) of bio-oils, lignin and the reaction catalyst stability with time-on-stream for understanding the industrial utilization of biomass in HDO to produce fuels and chemicals. Our results showed that more oxygenated feedstock, as well as presence of certain catalyst poisons in uence the HDO performance and the results evidence carbon allotropes.

Conclusions
Comparing the selectivity of both catalysts, three main aspects may be analyzed: the selectivity of nonoxygenated compounds is higher on NiMo/SBA-15 than on NiMo/Al 2 O 3 ; the selectivity to phenol is higher on NiMo/Al 2 O 3 and the deoxygenated compounds (catechol, methyl guaiacol and derivatives). The SBA-15 support presents basically Si-OH groups, which act as Lewis acid sites and has higher ability for direct demethoxylation of guaiacol resulting in phenol.
After reaction it is noteworthy to observe signi cant modi cation in the IR spectrum of the silica catalyst.
New bands appeared at 1390 and 1470 cm -1 , which are assigned to the rotation vibration and folding of C-H bonds of methyl groups. The formation of CH 3 groups at the catalysts surface during the reaction is closely related to the deactivation due to the carbon deposition on its surface. After reaction the spectrum of the alumina supported catalyst presented signi cant differences in the range 1000 and 1600 cm -1 , with appearance of new bands which are attributed to the unfolding C-H band bonded to the aromatic rings.
After reaction, the infrared spectroscopy showed that the alumina-supported catalyst contained deposits of aromatic and oxygenated organic substances, while the carbon deposited on SBA-15 materials are mainly the aliphatic type of simpler molecules. TPO analyses showed that the carbons deposited on the NiMo/SBA-15 catalyst were light polymer types, mainly nanotubes and nano bers, while on the alumina catalyst it was of graphite type and heavier carbons were formed. This data was con rmed by Raman spectroscopy, displaying disorganized and defective carbonaceous structures on NiMo/SBA-15. On the other hand, in the alumina, the graphitic carbon was found in a greater proportion than in the previous one, besides structures of amorphous carbon and graphite defective. TGA analyses presented the amount of coke deposition after HDO, 1.14 and 1. 65  Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author as requested.      Raman spectrum of the catalysts NiMo/Al2O3 and NiMo/SBA-15 before and after hydro deoxygenation of guaiacol.