THE STATE OF NICKEL IN THE SILVER MODIFIED NiMg/SiO 2 VEGETABLE OIL HYDROGENATION CATALYSTS

Transcription

1 ЖУРНАЛ ФИЗИЧЕСКОЙ ХИМИИ, 2009, том 83, 9, с УДК CHEMICAL KINETICS AND CATALYSIS THE STATE OF NICKEL IN THE SILVER MODIFIED NiMg/SiO 2 VEGETABLE OIL HYDROGENATION CATALYSTS 2009 M. Gabrovska*, D. Nikolova*, J. Krsti c **, ' M. Stankovi c **, ' P. Stefanov***, R. Edreva-Kardjieva*, D. Jovanovi c ** ' *Institute of Catalysis, Bulgarian Academy of Sciences, Sofia, Bulgaria **Institute of Cemistry, Tecnology and Metallurgy, Department of Catalysis and Cemical Engineering, Belgrade, Serbia *** Institute of General and Inorganic Cemistry, Bulgarian Academy of Sciences, Sofia, Bulgaria margo@ic.bas.bg Abstract Two series of silver modified Ni Mg materials were syntesized by precipitation-deposition on SiO 2 support derived from two silica sources: diatomite activated at 800 C (Series a; Mg/Ni = 0.1 and SiO 2 /Ni = 1.07) and syntetic water glass (Series b; Mg/Ni = 0.1 and SiO 2 /Ni = 1.15). Te modification wit silver was made at tree molar Ag/Ni ratios, namely , and 0.1. Te effects of te source of te silica support and te silver presence and content on te nickel state in te silver modified reduced-passivated NiMg/SiO 2 precursors of te vegetable oil ydrogenation catalyst were establised by X-ray diffraction and X-ray potoelectron spectroscopy tecniques. Te passivation procedure was applied in order to protect te metallic nickel particles from furter oidation. Te crystallization of te formed nickel ydrosilicate pases depends on te source of te silica support, more epressed in te diatomite supported samples. It was sown tat te silver modification of te NiMg/SiO 2 precursors enances te reduction of te nickel ydrosilicates accompanied by formation of relatively smaller metallic nickel particles, more pronounced in te water glass supported precursors. Te increase of te silver content in te water glass deposited samples is responsible for te metallic nickel dispersion increase. Te iger content of te particles on te surface of te diatomite deposited samples is in accordance wit te iger stability of te larger metallic nickel crystallites to oidation during te passivation step. On contrary, iger dispersed particles on te surface of te water glass supported samples are more susceptible to te oidation. INTRODUCTION Te catalytic ydrogenation of vegetable oils continues being one of te most versatile ways to modify te pysical properties, oidative and termal stability, te melting caracteristics of te fat and its color. Te process is widely used in te elaboration of margarines, sortenings and products of bakery [1]. Te carbon double bonds are partially or fully saturated during te ydrogenation. Concomitantly, te isomerization of naturally occurring cis fatty acids (CFA) to trans fatty acids (TFA) takes place [2]. Te selectivity of te process represents a considerable callenge, aiming to enance te ydrogenation activity and simultaneously to suppress te isomerization [3]. Te composition and properties of final product depend on various operating factors, including catalyst type and concentration, agitation, ydrogen pressure and temperature. Altoug it as been eperienced wit several metals, including te noble metals, te most widely used commercial catalyst remains te metallic nickel supported on a SiO 2 carrier [2] because of its lower cost. It as been publised tat Ni Mg catalysts, supported on SiO 2 derived from two siliceous sources, eiter diatomite or water glass, demonstrate bot ig soybean oil ydrogenation activity and ig quantity of detrimental TFA production [4]. According to our data, te addition of Ag to te Ni-Mg/diatomite catalyst suppresses te TFA formation in te ydrogenated products [5]. Te aim of tis work was to estimate te effect of te source of te silica support (diatomite or water glass) as well as te silver presence and content on te nickel state in te silver modified reduced-passivated NiMg/SiO 2 catalyst precursors for vegetable oil ydrogenation. Te passivation procedure was applied in order to protect te metallic nickel particles from furter oidation. Te unreduced samples are studied for te sake of comparison. EXPERIMENTAL Sample preparation Te diatomite, designated as D, was obtained from Baroševac field B ( Kolubara coal basin, Lazarevac, Serbia). Te crude diatomite was mecanically, cemically (HCl) and termally (800 C) treated in order to obtain pure and activated support [6]. Te water glass material, designated as W, is a syntetic commercial product wit module (molar ratio) 1637

2 1638 GABROVSKA et al. Table 1. Te cemical composition of te precursors, at. % Sample Ni Mg Ag Si Series a 1 Ni/D AgNi/D AgNi/D AgNi/D Series b 1' Ni/WG ' 0.1AgNi/WG ' 1.0AgNi/WG ' 4.0AgNi/WG SiO 2 /Na 2 O = 3.0 (GALENIKA - Magmasil A. D., Zemun (Belgrade, Serbia). Two series of silver modified Ni Mg materials were syntesized by precipitation-deposition of te corresponding metal nitrates wit 10% w/w Na 2 CO 3 aqueous solution under vigorous stirring at temperature of 90 C and constant value of ph 10.0 ± 0.1 on SiO 2 support derived from two silica sources: activated diatomite (Series a; Mg/Ni = 0.1 and SiO 2 /Ni = 1.07) and syntetic water glass (Series b; Mg/Ni = 0.1 and SiO 2 /Ni = 1.15). After ageing of te obtained slurry, te corresponding source of silica support as 2.0 wt % diatomite aqueous suspension or as 2.0 wt % water glass aqueous solution was added to te reaction miture to form te final catalyst precursor. Te resulting materials were aged again under constant stirring and tan filtered and torougly wased wit ot distilled SiO Fig. 1. XRD patterns of te unreduced precursors 1 4 (see Table 1) from Series a; () nickel ydrosilicate and () α- Ag 2 CO D water until absence of and Na + ions ave been obtained. Te precursors were dried for 24 in an oven at 105 C and ground to a powder. Te modification wit silver was made at tree molar Ag/Ni ratios, namely , and 0.1. Te unmodified reference Ni Mg samples were prepared by te same procedure on bot siliceous sources and denoted as Ni/D or Ni/WG. Te samples were referred as AgNi/Z, were is te atomic % of Ag (Table 1) and Z is te source of te support D or WG. Te reduction (activation) of te precursors was performed in a laboratory set-up by a dry reduction metod wit a gas miture of H 2 /N 2 (1/1 v/v) at te flow rate of 5 dm 3 1. Te reduction temperature was raised up to 430 C at te eating rate of 1.5 K min 1 and eld constant for 5. After cooling down to room temperature, te reduced precursors were passivated wit a miture of 350 ppm of O 2 in nitrogen for 2. For te sake of simplicity tese samples were denoted wit te prefi r, for eample, r-ni/d. NO 3 Sample caracterization Te cemical compositions of te precursors were determined by atomic absorption spectroscopy (Varian AA 775 spectropotometer). Te silica content was determined by a classical silicate analysis and a cemical procedure applicable in te analysis of sediment rocks. Te X-ray diffraction studies were accomplised on a Bruker D8 Advance powder diffractometer employing CuK α radiation, operated at U = 40 kv and I = =40 ma. Te X-ray potoelectron spectroscopy measurements were performed in a VG ESCALAB II electron spectrometer using AlK α radiation wit energy of ev. Te binding energies (E) were determined wit an accuracy of ±0.1 ev utilizing te C1s line at ev (from an adventitious) as a reference. Te composition and cemical state of te precursors were investigated on te basis of te areas and binding energies of Ni2p 3/2, Ag3d, Mg1s, O1s, and Si2p potoelectron peaks (after linear subtraction of te background) and Scofield s potoionization cross-sections. Te peak fitting software was Pimat program. RESULTS AND DISCUSSION Te results from te silicate and cemical analysis of te syntesized precursors are summarized in te Table 1. Te obtained data demonstrate at te samples posses a cemical composition very close to te teoretical values.

3 THE STATE OF NICKEL IN THE SILVER MODIFIED 1639 Ag 0 Ag 0 Ag X-ray diffraction (XRD) Unreduced precursors. Te diffractogram of te diatomite source (Fig. 1, curve D) sows reflections caracteristic of amorpous silica (silica alo peak centered at 2θ 21 ) and well-crystallized quartz pase (2θ 26 ; JCPDS ). Te dried water glass represents badly crystallized solid (Fig. 2, curve WG). Te precipitation deposition of Ni and Mg on te surface of bot siliceous sources results in obtaining of te unmodified reference samples Ni/D and Ni/WG. Teir XRD patterns are significantly different from te spectra corresponding to diatomite and water glass (Figs. 1 and 2). Te signals are weak and broad, denoting poorly crystallized, well-dispersed nickel compounds [7]. It as been establised tat two types of ydrosilicates seem to coeist in Ni/D and Ni/WG precursors: nickel silicate ydroide pase (2θ 10, 20, 24, 34, 37 and 60 ; JCPDS file ) and antigorite pase (2θ 12, 14, 15, 17, 19, 25, 35, 37, 41 and 54 ; JCPDS ). Similar XRD patterns for Ni/diatomite catalysts, prepared by deposition-precipitation ave been reported by Eceverria and Andres [7]. Te XRD pattern of te Ni/D sample (Fig. 1, curve 1) manifests more and better formed reflections of te nickel ydrosilicate pases in comparison wit its analogue Ni/WG from Series b. Te peak of te well crystallized quartz pase is preserve in te spectrum of Ni/D sample owever wit low intensity. Te registered ig background below 2θ = 10 on te diffractogram of Ni/WG precursor indicates advanced amorpisation of te observed pases tan in Ni/D one. It may be suggest tat te use of te water glass as a source of te silica support provokes iger dispersion of te nickel ydrosilicate pases. Te observed penomenon proves tat during te syntesis of te precursors, an interaction occurs between te nickel salt and te silica from te support resulting in formation of nickel ydrosilicate layers, wic covered te eternal surface of te silica particles. Te poorly crystallized nickel ydrosilicate compounds wit imperfect nickel antigorite and/or nickel montmorillonite-like structure are always formed during te co-precipitation of nickel nitrate and alkali silicate solutions at temperature under 100 C [8]. Te XRD patterns of nickel ydrosilicates are not clearly defined due to its turbostratic structure [8 10]. Te modification of te Ni/D and Ni/WG samples wit silver alters teir XRD patterns in different degree. Te main reflections of te nickel ydrosilicate pases are detected on te diffractograms of modified samples from Series a (Fig. 1, curves 2 4). Two additional lines at 2θ 33 and 39, caracteristics of α- Ag 2 CO 3 pase (JCPDS file ) is registered only for te sample wit te igest silver content (4.0AgNi/D). Contrariwise, te XRD patterns of te nickel ydrosilicate pases are not visible on te diffractograms of te modified samples from Series b (Fig. 2, curves 2' 4'). Moreover, te reflections of α-ag 2 CO 3 pase are more as number (2θ 19, 33, 51 and 61 ) and muc more intensive in all Ag-containing precursors from Series b. Supplementary, tese precursors display etra diffraction lines at 2θ 38, 44 and 64, typical of te cubic metallic silver pase (JCPDS file ). Its presence indicates tat te Ag + ions are weakly bounded to te silica from te water glass source in comparison to te diatomite. As a result, part of Ag + ions seems to be already reduced to te metallic state during te preparation of te precursors. It can be summarized tat te silver presence in te unreduced precursors from Series a causes partial amorpisation of te nickel ydrosilicates, wereas tese pases are probably fully amorpizated or idden by te silver entities in te samples from Series b. Reduced-passivated precursors. Te reduction-passivation procedure of te samples from bot series causes appearance of te diffraction lines at 2θ 44, 52, 76 and 93 (Figs. 3, 4). Tese reflections are caracteristics of te metallic nickel pase (JCPDS file ), better formed in te samples from Series a. Some reflections of nickel ydrosilicate pases are also observed in te samples from bot series, suggesting incomplete reduction of te Ni 2+ entities. It may be seen tat te presence of silver diminises te intensity of te nickel ydrosilicate pases indicating facilitated reduction of te Ni 2+ ions, more epressed wit te increase of te silver loading in te samples. Tis deduction was confirmed by our former study based on te TPR eperiments performed on te samples under study [11]. It was found tat te silver addition enances te reducibility of te bulk Ni 2+ entities 4' 3' 2' 1' WG Fig. 2. XRD patterns of te unreduced precursors 1' 4' (see Table 1) from Series b; () nickel ydrosilicate and () α-ag 2 CO 3.

4 1640 GABROVSKA et al. Ag 0 Ag 0 SiO 2 Ag 0 Ag Ag 0 Ag 0 4 ' 3 ' 2 2 ' Fig. 3. XRD patterns of te reduced-passivated precursors 1 4 (Table 1) from Series a; () nickel ydrosilicate. 1 ' Fig. 4. XRD patterns of te reduced-passivated precursors 1' 4' (see Table 1) from Series b; () nickel ydrosilicate. depending on te source of bot carriers and te silver concentration. Moreover, te reduction temperatures decrease wit te increase of Ag content. Te results indicate tat te use of te water glass as a silica source ensures reduction of te Ni 2+ species in all modified samples at temperatures lower tan in teir analogues prepared on diatomite. Te results obtained are in agreement wit te scarce studies publised in te literature on te reducibility of supported NiAg oides. Ertl and Knözinger [12] reported tat te presence of silver decreased te reduction temperature of nickel supported on SiO 2. Te role of Ag in promoting te Ni 2+ reduction is likely associated wit te fact tat Ag 2 O can easily be reduced to metal state at temperature lower tan NiO. Taking into consideration tat silver and nickel do not form a solid solution at any composition under equilibrium conditions [13] it is assumed tat metallic Ag migt plays an essential role in provoking te activation of H 2 and te removal of oygen atom from te Ni-containing pases via weakening vicinal Ni O bonds. Te appearance of te nickel ydrosilicate pase after reduction of te samples from Series b (Fig. 4) supports our assumption tat te silver pases idden te nickel ydrosilicates in te unreduced samples (see Fig. 2). A plausible eplanation is tat te interaction of Ni 2+ ions is presumably different wit diatomite and water glass, during te precipitationdeposition of te Ni 2+ ions onto te supports, wic can cause differences in nucleation and growt of te nickel species. Te well-formed signals of te metallic silver pase at 2θ 38, 64 and 77 appear only in te samples wit te igest silver content (r-4.0agni/d and r-4.0agni/wg), better organized in te precursor from Series a. Te XRD study sows no influence of silver on te bulk structure of nickel, tus te metallic nickel and metallic silver present as separate pases. Te reflections of NiO, AgO and Ag 2 O pases are not detected as results of te passivation of te reduced samples. It is necessary to mention tat te most intensive peak of te pase (2θ 44.5 ) overlaps te second in intensity diffraction line of te Ag 0 pase (2θ 44.3 ). On tis reason, te mean crystallite size is estimated from te reflection at 2θ 52. Te results sow tat te values of te metallic nickel particles of te precursors from Series a are located in te nanosize region, namely: 7.0 nm (r-ni/d), 6 nm (r-0.1agni/d), 5.4 nm (r-1.0agni/d) and 4.5 nm (r-4.0agni/d), respectively. Tis trend demonstrates te increase of te pase dispersion wit te increase of te silver content in te diatomite supported samples [5]. Te reflection at 2θ 52 on te XRD patterns of te samples from Series b is very broad and terefore, te determination of te mean particle size is practically nonsense. Te broadening indicates te presence of smaller particles of te tan in te precursors from Series a. Te metallic nickel appears as more dispersed in te modified precursors from Series b in comparison wit tis one from Series a. It can be summarized tat te usage of diatomite as silica source provokes better crystallization of te and Ag 0 pases after te reduction-passivation procedure tan te water glass one. Te presence of silver and its amount increase te reducibility of te Ni 2+ entities in te samples from bot series. Moreover, te silver modification affects te degree of te pase crystallization and leads to te formation of relatively smaller particles. Wit addition of te silver modifier, te dispersion etent of te metallic Ni increase. Tis effect is more pronounced in te reduced-passivated precursors from Series b.

5 THE STATE OF NICKEL IN THE SILVER MODIFIED 1641 Table 2. XPS data of te unreduced (I) and reduced-passivated (II) precursors from bot series Sample Surface concentrations, at. % E, ev (Ni2p 3/2 ) Ni Ag I II Nickel oidation state contribution, % (II) I II I II Ni 2+ Ni 2+ Ni 2+ Series a Ni/D AgNi/D AgNi/D AgNi/D Series b Ni/WG AgNi/WG AgNi/WG AgNi/WG X-ray potoelectron spectroscopy (XPS) Unreduced precursors. Te binding energy (E) values of te main Ni2p 3/2 peak in unreduced precursors are presented in Table 2. As epected, nickel is presented in Ni 2+ oidation state on te surface of te samples from bot series. Te differences in te E values of te unmodified Ni/D (855.3 ev) and Ni/WG (856.1 ev) samples may be attributed to te different Ni 2+ state in te nickel ydrosilicates on te surface of te samples [7, 14]. Te silver modification of te samples from Series a leads to te sifting of Ni2p 3/2 peak towards iger E values (Table 2), more clearly epressed in te sample wit te lowest silver content (0.1AgNi/D). Te sifting may be assigned to te stronger interaction between te nickel and te diatomite surface. However, te silver presence in te samples from Series b does not arouse te sifting of te Ni2p 3/2 peak, on te analogy of Series a. It may be observed tat te increase of te silver amount provokes te sifting of Ni2p 3/2 peak towards lower binding energies, wic suggested weakening of te interaction between te nickel and te water glass surface. Te data obtained corroborate te facilitated reduction of te Ni 2+ ions, more epressed wit te increase of te silver loading in te samples. Te E values of Ag3d 5/2 peak of te modified samples from Series a are situated in te range of ev; te corresponding ones from Series b in te range of ev. Tese energies can not be attributed to te definite Ag O species or Ag 0, because of te narrow E interval of te different Ag entities, namely ev for Ag 0, ev for Ag 2 CO 3, ev for AgO and ev for Ag 2 O respectively [15]. However, te iger E values of Ag3d 5/2 peak in Series b confirm te presence of te metallic silver as it as been registered by XRD (Fig. 2) It may be seen from Table 2 tat a small amount of silver raises te nickel surface concentration in 0.1AgNi/D sample from Series a. Te increase of te silver concentration leads to te lessening of Ni content in te order: 0.1AgNi/D > 1.0AgNi/D > 4.0AgNi/D. It is interesting tat te nickel surface concentration on te water glass supported samples is iger ( at. %), comparing wit teir analogues deposited on diatomite ( at. %). Te detected surface concentrations of nickel on te samples from Series b indicate a iger dispersion of Ni entities [16]. Reduced-passivated precursors. Te reduction-passivation procedure provokes a strong impoverisment of te nickel on te surface of te samples from Series a (Table 2). Te nickel content decrease may be associated wit te diminised dispersion [17]. Te surface nickel concentration remains practically uncanged on te surface of te samples from Series b. Obviously te use of te water glass as a silica source preserves te iger dispersion of te Ni entities on te sample surface. Te silver is not registered on te samples wit lowest Ag content from bot series (0.1AgNi/D and 0.1AgNi/WG). Te general view of te Ni2p level reveals te asymmetry of te main Ni2p 3/2 peak towards lowers E (Fig. 5 and Table 2). Te peak-fitting of te composite Ni2p level evidences te presence of te two types of te nickel oidation states, and Ni 2+. Te values of te E in te range of ev for Series a and ev for Series b caracterize te. Te presence of metallic nickel after

6 1642 GABROVSKA et al. Ni2p ' 3' Ni 2+ surface of te water glass supported samples are more susceptible to oidation. CONCLUSIONS It may be concluded tat: Te crystallization of te nickel ydrosilicate pases depends on te source of te silica support, more epressed in te diatomite supported samples. Te modification wit silver enances te reduction of te nickel ydrosilicates in bot series. Te iger content of te particles on te surface of te diatomite deposited samples is in accordance wit te iger stability of te larger metallic nickel crystallites to oidation during te passivation step. On te contrary, te iger dispersed particles on te surface of te water glass supported samples are more susceptible to oidation. Te increase of te silver content increase te dispersion of te metallic nickel more pronounced in te water glass deposited samples. It is assumed tat metallic Ag provokes te activation of H 2 and te removal of oygen atom from te Ni-containing pases via weakening vicinal Ni O bonds E, ev Fig. 5. Peak-fitting of te Ni2p potoelectron spectra of te reduced-passivated precursors 1 4 and 1' 4' (see Table 1); E is te binding energy. passivation is supported by X-ray diffraction data (Figs. 3, 4). Te E values in te range of ev for Series a and ev for Series b, respectively, caracterize Ni 2+ state in Ni O species originated bot from te unreduced nickel pylosilicate pase and dispersed two-dimensional NiO -like pase on te metallic nickel particles. Te oidation state contribution on te samples from Series a increases wit te silver presence and content. However, tis effect is not clearly pronounced in te samples from Series b (Table 2). Te iger content of te particles on te surface of te diatomite deposited samples is in accordance wit te iger stability of te larger metallic nickel crystallites to oidation during te passivation step (XRD data). On contrary, te iger dispersed particles on te 2' 1' ACKNOWLEDGMENTS Te autors gratefully acknowledge te partial financial support of te Ministry of Education and Science of Bulgaria (Project X-1411) and te Serbian Ministry of Science (Project B and tank Assist. Prof. MSc. Genoveva Atanasova for XPS spectra recording (Institute of General and Inorganic Cemistry of Bulgarian Academy of Sciences). REFERENCES 1. M. Fernández, G. Tonetto, G. Crapiste, D. Damiani, J. Food Engineering 82, 199 (2007). 2. J. Weldsink, M. Bouma, N. Scöön, A. Beenackers, Catal. Rev. Sci. Eng. 39, 253 (1997). 3. K. Belkacemi, A. Boulmerka, J. Arul, S. Hamoudi, Topics in Catalysis 37, 113 (2006). 4. M. Gabrovska, J. Krsti c,' R. Edreva-Kardjieva, M. Stankovi c,' D. Jovanovi c,' Appl. Catal. A: General 299, 73 (2006). 5. M. Stankovi c,' M. Gabrovska, J. Krsti c,' P. Tzvetkov, M. Sopska, Ts. Tsaceva, P. Bankovi c,' R. Edreva- Kardjieva, D. Jovanovi c,' J. Mol. Catal. A: Cemical 297, 54 (2009). 6. D. Jovanovic, R. Radovic, L. Mares, M. Stankovic, Br. Markovic, Catal. Today 43, 21 (1998). 7. S. Eceverria and V. Andres, Appl. Catal. 66, 73 (1990). 8. J. van Eijk van Voortuijsen, P. Franzen, Rec. Trav. Cim. 70, 793 (1951). 9. K. Guge, A. Bat, G. Babu, Appl. Catal. A: Gen. 103, 183 (1993). 10. P. Burattin, M. Ce, C. Louis, J. Pys. Cem. B 101, 7060 (1997).

7 THE STATE OF NICKEL IN THE SILVER MODIFIED M. Gabrovska, I. Stereva, J. Krsti c,' M. Stankovi c,' R. Edreva-Kardjieva, D. M. Jovanovi c,' 9t International Conference on Fundamental and Applied Aspects of Pysical Cemistry, PHYSICAL CHEMIS- TRY 2008, September 24 26, 2008, Belgrade, Serbia, Vol. 1, p G. Ertl, H. Knözinger, J. Weitkemp (Eds.), Handbook of Heterogeneous Catalysis, VCH, Weineim, vol. 2, p. 439 (1997). 13. Binary Alloy Pase Diagrams, 2nd ed., edited by T. B. Massalski, H. Okamoto, P. Subramanian, and Kacprzak (ASM International, Metals Park, OH, 1990). 14. M. Rodrigo, L. Daza and S. Mendioroz, Appl. Catal. A: Cemical 88, 101 (1992). 15. ttp://srdata.nist.gov/xps/ 16. K. Hadjiivanov, M. Miaylov, D. Klissurski, P. Stefanov, N. Abadjieva, E. Vasileva, L. Mincev, J. Catal. 185, 314 (1999). 17. M. Miaylov, K. Hadjiivanov, D. Klissurski, L. Petrov, C. Bonev, G. Kadinov (Eds.), Heterogeneous Catalysis, Proceedings of te 9t International Symposium, Varna, Bulgaria, 2000, p. 429.