- Email: [email protected]
TiO2 catalysts
Catalysis Communications 8 (2007) 1772–1780 www.elsevier.com/locate/catcom
Effect of surface sites of TiO2 support on the formation of cobalt-support compound in Co/TiO2 catalysts Kongkiat Suriye, Piyasan Praserthdam *, Bunjerd Jongsomjit Center of Excellence on Catalysis and Catalytic Reaction Engineering, Department of Chemical Engineering, Faculty of Engineering, Chulalongkorn University, Bangkok 10330, Thailand Received 25 October 2006; received in revised form 10 January 2007; accepted 12 January 2007 Available online 18 January 2007
Abstract This study has addressed the effect of surface sites (as Ti4+ and Ti3+) of TiO2 support on the formation of the cobalt-support compound on Co/TiO2 catalyst. A 20% cobalt was prepared on each TiO2 support having the different Ti4+/Ti3+ ratios covering on the surface (1.2, 1.3, 1.4, and 1.6). It can be concluded that the non-reducible compound as a Co-SCF preferred to form on the Co/TiO2 catalyst when the most proportional site of surface TiO2 became Ti4+. This is because the migration of cobalt cluster, which plays the significant role for creating Co-SCF, can be promoted by the increasing of Ti4+ sites resulting in the decrease of cobalt dispersion. The hyperfine patterns of ESR spectra demonstrated that the Co0(HxTiOy) was the simple chemical form of Co-SCF formed after standard reduction. Moreover, the structure of Co-SCF changed to be a higher non-uniform structure when the number of Ti4+ site on TiO2 surface increased. The possible mechanism to form this compound has also been discussed. Ó 2007 Elsevier B.V. All rights reserved. Keywords: Cobalt-support compound formation; Co-SCF; Ti4+; Ti3+; Supported cobalt catalyst; Titania
1. Introduction It has been known that the Co/TiO2 catalyst is considered to have a strong metal support interaction (SMSI) and shows high activity in CO hydrogenation [1,2]. This interaction is an important factor used for determining the properties of Co/TiO2 catalyst such as cobalt dispersion and reduction behavior [3]. In fact, the synthesis of highly dispersed cobalt on TiO2 support requires the strong interaction between cobalt and support. However, too strong interaction can produce the Co-support compound as a suboxide at an interface that is high resistant to reduction [4,5]. Jongsomjit et al. [6] reported that the Co-support compound formation (Co-SCF) during standard reduction resulted in a lower reducibility of Co/TiO2 catalyst. The *
Corresponding author. Tel.: +662 218 6883; fax: +662 218 6769. E-mail address: [email protected] (P. Praserthdam).
1566-7367/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.catcom.2007.01.011
compound of cobalt and TiO2 formed referred as ‘‘Co-titanate’’ was considered to be non-reducible form and it revealed that ‘‘Co-titanate’’ formed was likely to be different from CoTiO3. It was also reported that Co-SCF in SiO2 and Al2O3 can occur during standard reduction and resulted in a lower reducibility of catalyst [7]. In case of TiO2-supported cobalt, although, it has been known that the dominant surface sites of TiO2 support consist of two main sites as Ti4+ and Ti3+ [8–13], the effect of surface sites on the formation of Co-SCF has not yet been investigated as well. In our previous work, the impact of Ti3+ present in TiO2 support on catalytic properties of Co/TiO2 was discussed [14]. Therefore, in this work, the proposed research, we aim to extend the investigation to the cobalt-support compound formation on such catalyst. The effect of surface sites (as Ti4+ and Ti3+ on TiO2 support) on the cobalt-support compound formation has been investigated as the main goal. Furthermore, the simple chemical form and
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
1773
the mechanism to form the Co-SCF have also been proposed by means of electron spin resonance spectroscopy (ESR).
tivity detector was used to analyze CO and H2 in the feed and product streams. In all case, steady state was reached within 4 h.
2. Experimental section
3. Results and discussion
2.1. Materials
3.1. Effect of surface sites (Ti4+ and Ti3+) of TiO2 support on the formation of Co-SCF in Co/TiO2 catalyst
A 20 wt% of Co/TiO2 was synthesized by the incipient wetness impregnation [6]. A specific amount of cobalt nitrate [Co(NO3) Æ 6H2O] was dissolved in deionized water and then impregnated onto TiO2 which had a different Ti4+/Ti3+ ratios covering on the surface (the method for controlling the Ti4+/Ti3+ ratio was based on Ref. [14]). The resulting sample was dried at 383 K for 12 h and calcined in air at 723 K for 4 h. 2.2. Nomenclature The nomenclatures used for the Co/TiO2 catalysts in this study are following: A, B, C, D. The supported cobalt on TiO2 which had a Ti4+/Ti3+ molar ratio covering on the surface of 1.2, 1.3, 1.4, and 1.6, respectively. 2.3. Characterization CO2-temperature programmed desorption (TPD) was carried out using a 1 g of a TiO2 sample. The sample was dosed by using 1% CO2 in He for 1 h in liquid nitrogen and then desorbed at the range of temperature from 123 to 253 K by level controlling. A Gow-Mac (Series 150 thermal conductivity detector) gas chromatograph was used to analyze CO2. FT-IR spectra were collected under vacuum with a Bruker IFS 66v Spectrometer with a resolution of 2 cm 1. Measurements were performed by pressing powder between two KBr windows. Each reported spectrum is the average of 64 scans. Temperature programmed reduction (TPR), by 5% by volume H2 in Ar, was carried out using 50 mg of supported cobalt sample and a temperature ramp from 35 to 1073 K at 5 K/min. H2 chemisorption was carried out using a Micromeritics Pulse Chemisorb 2700 system. Prior to chemisorption, the supported cobalt sample was reduced at 623 K for 10 h after ramping at a rate of 1 K/min. Electron spin resonance spectroscopy (ESR) was conducted in vacuum at a 77 K and without illumination using a JEOL, JES-RE2X electron spin resonance spectrometer. CO hydrogenation was carried out at 493 K and 0.182 MPa with a flow rate of H2:CO:He = 20:2:8 ml/min in a fixed-bed flow reactor. Typically, 20 mg of a supported cobalt sample was reduced in situ in flowing H2 (30 ml/min) at 623 K for 10 h prior to the reaction. Reactor effluent samples were taken at 1 h intervals and analyzed by GC. The composition of hydrocarbons in the product stream was analyzed by a Shimadzu GC14B (VZ10) gas chromatograph equipped with a flame ionization detector. A Shimadzu GC8A (molecular sieve 5A) gas chromatography equipped with a thermal conduc-
TiO2 support having a difference of Ti4+ to Ti3+ covering ratios on the surface has been prepared following our previous work [14]. Based on CO2-TPD technique monitoring surface sites of TiO2 (original by Thomson et al. [15]), the Ti4+/Ti3+ ratios were estimated from the ratio between the CO2 desorption peak at temperature 175 K (Ti4+ site) and 200 K (Ti3+ site). The four TiO2 supports with a difference of Ti4+/Ti3+ ratios on the surface (1.2, 1.3, 1.4, and 1.6) were prepared as indicated in Table 1. The infrared spectroscopy (IR) (Fig. 1a) characterizing the OH group bonding to Ti4+ site and Ti3+ site of the three TiO2 supports of the sample A, B, and D was also used to confirm the Ti4+/Ti3+ ratios. As seen in Fig. 1b, IR spectra of these supports exhibited mainly two OH peaks at tOH = 3717 and 3673 cm 1. Based on the report of Szczepankiewicz et al. [16], the IR peak at tOH = 3717 cm 1 represented the OH bonded to Ti3+ site on the surface of TiO2 support. In case of the IR peak at tOH = 3673 cm 1, it was assigned to be the OH bonded to Ti4+ site [17,18]. Compared among the IR spectra of TiO2 supports used for preparing Co/ TiO2 in the sample A, B, and D (as seen in Fig. 1b), showing that the intensity of the band characterizing OH bonded to Ti3+ site (tOH = 3717 cm 1) decreased relatively with the number of OH bonded to Ti4+ site (tOH = 3673 cm 1). The Ti4+/Ti3+ ratios covering on the surface of each TiO2 support, estimated from the ratio between the IR peak at tOH = 3671 and tOH = 3717 cm 1, are also summarized in the Table 1. It shows that the trend of the Ti4+/Ti3+ ratios covering on TiO2 support has a good agreement with those from CO2-TPD results. The electron spin resonance spectroscopy (ESR) as seen in Fig. 2a was also used for confirming the quantitative data from CO2-TPD and IR. ESR spectra, performing at 77 K without illumination on samples, exhibited mainly one signal at the g value of 1.996 for all samples. According to the results as reported by Nakaoka et al., this peak is attributed to the Ti3+ site on TiO2 [19]. This ESR result is consistent with works of Nakamura et al. and Serwicka et al., they reported two signals occurred on the surface of TiO2 during ESR analysis without illumination including: (i) the surface Ti3+ and (ii) the bulk defect [20,13]. However, one signal of ESR result in this work agrees with the works of Howe and Gratzel [21] and Lee et al. [22]. Based on intensity of ESR spectrum as summarized in Table 1 (which corresponds to the amount of Ti3+ present in TiO2 surface), it shows that a decrease in number of Ti3+ corresponded well with an increase of the Ti4+/Ti3+ ratio obtained from CO2-TPD and IR.
1.7 1.6 1.1 0.9 (Ti4+–OH) and tOH = 3716 cm
1
(Ti3+–OH).
16.8 16.2 11.4 9.6 738 729 800 859 66 51 48 35
1
f
g
e
d
c
3631 3024 2898 2885 1.4
0.8 1.2
Calculated from CO2-TPD spectra. The ratio between the intensity of IR spectra at tOH = 3670 cm The intensity of ESR spectra at g value 1.996. Calculated from temperature-programmed-reduction (TPR). The temperature at the maximum point of TPR. Calculated from H2-chemisorption. Rate of CO hydrogenation after 4 h. a
Ti /Ti ratioa
1.2 1.3 1.4 1.6 A B C D
b
% Co dispersionf H2-chemisorbtion (107 molecule/g) TPR temperature (K)e % Reducibilityd Ti –OH/Ti –OH ratio (IR)b
3+ 4+ 3+ 4+
TiO2 support Sample
Table 1 Characteristics of TiO2 supports and Co/TiO2 catalysts
Ti
3+
(ESR)
c
Co/TiO2 catalyst
18 15 13 11
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
Reaction rateg (mmol g 1 s 1)
1774
A 20% cobalt was prepared on each TiO2 support which had a difference of Ti4+/Ti3+ ratios covering on the surface (1.2, 1.3, 1.4, and 1.6 based on CO2-TPD). The characteristics of each Co/TiO2 catalyst are summarized in Table 1. Considered on the% reducibility, when the most proportion of surface site on TiO2 support became Ti4+ site (from A to B, C, and D, consequently), the Co/TiO2 catalyst changed to be a harder reductive catalyst (Co3+ reduced to Co2+, and then reduced to Co0 in final step) as seen by the decrease in number of % reducibility. This is because more amount of non-reducible compound as a Co-SCF was formed on the Co/TiO2 catalyst [6,7]. The reducibility loss (based on sample A) during the standard reduction process was found to be in the range of 8% (for sample B) to 47% (for sample D). Based on the TPR profiles as shown in Fig. 3, it indicates that the reduction peaks (overlap of Co3+ > Co2+ > Co0) shifted to higher temperature when the most proportion on the surface of TiO2 support became Ti4+ site (the changing of temperature is summarized in the Table 1) confirming the existence of Co-SCF on Co/TiO2. Combined between the % reducibility loss and the moving to higher temperature of TPR profile, it can be concluded that the non-reducible compound as a Co-SCF preferred to form on the surface of Co/TiO2 catalyst, when the TiO2 support had more Ti4+ site covering on the surface. Based on the two dominant surface sites of TiO2 support as Ti4+ and Ti3+ [8–13], on the other hand, it can be referred that the Co-SCF preferred to form at the Ti4+ site, but more characterization is still needed to confirm this proposal. There are many reports showing the effect of surface site as Ti4+ and Ti3+ on the adsorption of metal and oxide, as summarized by Diebold [8]. It has been known that the bond energy for anchoring the metal is stronger at the Ti3+ site than that of Ti4+ site [23]. This is because the Ti3+ site has one lone pair electron in d orbital resulting in the higher reactive with the metal and oxide loading [8]. Based on this bond energy, Wallace et al. summarized the role of Ti3+ site on the nucleation, growth, and stability of Au cluster on such site of TiO2 support [24]. It showed that an increase of Ti3+ site strongly decreased the rate of Au cluster migration during high temperature condition due to the strongly bonding between Ti3+ site and Au cluster. The literatures showing the strong bond between Gr. VIII metals and Ti3+ have also been reported by many researchers [8,25,26]. In this work, based on this bond energy, Ti3+ site on the surface of TiO2 support should have the responsibility for fixing the cobalt cluster on the support resulting in a decrease of cobalt migration for forming a bigger size of cobalt. This suggestion can be confirmed by the changing of the cobalt dispersion on the surface of TiO2 support as indicated in Table 1. The increase of cobalt dispersion was observed when the surface of TiO2 support has more Ti3+ sites. This increase of cobalt dispersion due to the Ti3+ site can inhibit the agglomeration between cobalt clusters by the decrease of the migration rate of clusters. Zhang et al. [5] mentioned the effect of cobalt migration on the
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
a
1775
b 3717 A
B
3673
B
C
D
A
4000
3100
2200
1300
-1
Wavenumber, cm
3760
3740
3720
3700
3680
3660
3640
3620
-1
Wavenumber, cm
Fig. 1. The infrared spectroscopy: (a) characterizing the TiO2 supports (prior to cobalt loading) used for sample A, B, D; (b) considering on the OH region.
formation of Co-SCF, they investigated the reducibility of Co–Ru/c-Al2O3 during standard reduction in the presence of added water vapor. They reported that the migration of cobalt played the significant role for the forming of Co-SCF. In this work, the results showing the existence of Co-SCF with the observing the decrease of cobalt dispersion corresponded well with the conclusion of Zhang et al. [5]. Therefore, it can be suggested that the nonreducible compound, as the Co-SCF, preferred to form on Co/TiO2 catalyst having more Ti4+ site on the support due to the increasing of cobalt migration rate. 3.2. Determination of Co-SCF characteristics Based on the previous work in our laboratory, Jongsomjit et al. also investigated to identify the nature of Co-SCF on Co/TiO2 catalyst using several characterization techniques including SEM-EDX, XRD, and Raman spectroscopy [6]. In case of SEM-EDX and XRD, it cannot detect a Co-SCF on the Co/TiO2 because this compound has highly dispersed form. However, Raman spectroscopy revealed that this highly dispersed Co-SCF formed was likely to be different from CoTiO3 and present as the more complex derivatives on Co/TiO2 catalyst. It should be noted that, however, the Co-SCF has still never been directly detected by any techniques. In order to determine the nature of cobalt compound as more complex derivatives from the interaction between cobalt and the other metal compound, the electron spin resonance has been used by many researchers [27–31]. Schmidt et al. [31] investigated the nature of cobalt complexes in Ziegler type catalytic system using ESR and reported that the reduction of cationic cobalt to the zero-valent state can facilitate the formation of paramagnetic Co(II) and Co(0) complexes. However, in case of Co/TiO2 catalyst used for hydrogenation reaction, the evidence of Co-SCF by moni-
toring of ESR has never been done so far. Therefore, in this work, we tried to use the ESR spectroscopy, as a promising technique, for probing the evidence of Co-SCF on Co/TiO2 catalyst. In this work, the ESR investigation Co/TiO2 catalyst prior standard reduction, and Co/TiO2 catalyst after standard reduction were performed to investigate the CoSCF complexes. The ESR spectra of supported cobalt on TiO2 with the different Ti4+/Ti3+ ratios (A, B, and D samples) prior to standard reduction was shown in Fig. 2b. based on analysis over all range of magnetic field (unit in mT), it can detect only the small peak at g value ca. 1.995 which can assign to be Ti3+ site on TiO2 support. Based on the disappearance of Ti3+ site on the surface of TiO2 support (between Fig. 2a and b) , it shows that almost Ti3+ sites were used and participated to anchor the cobalt cluster on TiO2 support resulting in an increase of cobalt dispersion as discussed in the previous section. However, it should be further mentioned that, based on result from% reducibility loss with increased Ti4+ sites, some of cobalt cluster should be participated to anchor at Ti4+ site as well. From the spectra, it can not detect any paramagnetic cobalt species as a Co(II) and Co(0) on these catalyst at g value ca. 2.16 and 1.41, respectively [29]. Based on this analysis of Co/TiO2 prior to standard reduction, although, the Co/ TiO2 was dried at 383 K for 12 h and calcined in air at 723 K for 4 h during catalyst preparation, it shows that any cobalt compound as more complex derivatives from the interaction between cobalt and TiO2 support was not present. This shows that something is missing for the formation of the Co-SCF. The ESR analysis of supported cobalt on TiO2 support with the different Ti4+/Ti3+ ratios (A, B, and D samples) after standard reduction was also performed as shown in Figs. 4c and 5a and b. These spectra were very different from the spectra of Co/TiO2 prior standard reduction.
1776
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
Fig. 2. ESR spectra obtained at 77 K without irradiation for: (a) TiO2 support (prior to cobalt loading) of samples A, B, and D; (b) Co/TiO2 before standard reduction.
These spectra could be assigned as a Co-SCF on the Co/ TiO2 catalyst. Therefore, this shows that the formation of Co-SCF can be facilitated by the presence of hydrogen dur-
ing reduction. Based on ESR theory, these spectra are the hyperfine pattern which result in multi-peak in ESR spectrum [32,33].
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
D
H2 signal (a.u.)
C
B
A
273
373
473
573
673
773
873
973
1073
Temperature (K)
Fig. 3. Temperature-programmed reduction of Co/TiO2 catalyst in the sample A, B, C, and D.
This hyperfine pattern can occur when a lone pair electron (half electron spin) contacts with at lease one nucleus (one nuclear spin) [28,34,35]. The cobalt can be classified to be a low-spin Co(II) with a d7 electron configuration or Co(0) with a d9 electron configuration [28]. Based on this electron configuration, the one lone pair electron in d orbital of both cobalt can be considered to be S = ±1/2 (haft spin of lone pair electron) [34]. Also nuclear spin, cobalt have I = ±7/2 of 59Co with abundance of 100% in the nature [36]. Based on Fig. 4a, when one lone pair electron (S = ±1/2) in the d orbital contacts with one nucleus of 59 Co with I = ±7/2, eight peaks of hyperfine pattern, which symmetrically spitted from center of one lone pair electron in d orbital of cobalt (S = ±1/2), will occur [28,34,35]. Considered on the hyperfine pattern of sample A as also seen in Fig. 4c, the well consistency between the experimental showing hyperfine pattern and ESR theory can be observed in this sample. It was showed that there are eight peaks symmetrically spitted, with equal distance of A = 26.9 mT, from the central peak at g = 1.049 (as summarized in Table 2). This center attributed to the electron spin of one lone pair electron (S = ±1/2) in d orbital of Co(0) [29]. Based on the consistency between the hyperfine pattern from experimental and theory, it can be mentioned that non-reducible compound formed after standard reduction had a form of Co(0) compound. In this hyperfine pattern, there are two patterns in hyperfine spectra (as seen in Fig. 4c), one is attributed to 59Co with a larger A value and the smaller one (A = 9.9 mT) of four sets of two sub-hyperfine patterns. Saraev et al. [28] also found the two patterns in hyperfine spectra of Co(acac)2–PBu3-complexes. One was the eight lines of a 59Co (I = ±7/2) with a larger A value and the smaller one of A value (of two sub-hyperfine peak) which was attributed to 31P (I = ±1/2). Two subhyperfine patterns came from the contacting between one lone pair electron (S = ±1/2) and one nucleus of I = ±1/2 as seen in Fig. 4b [28]. Based on the system of this Co/ TiO2 catalyst after standard reduction with hydrogen, these
1777
two sub-hyperfine patterns should come from the nucleus of 1H (I = ±1/2) with abundance of 99.98% in nature [37]. It can not come from the nucleus of 47Ti and 49Ti (abundance of 7.44% and 5.41%, consequently) because they really have the less abundance in the nature. And if it came from 47Ti and 49Ti, these sub-hyperfine pattern must have six or eight lines (with equal of distance from center) of I = ±5/2 and ±7/2, respectively [38]. These four sets of 1H can be classified to be a two symmetrical pattern with a distance from central peak equal to A = 112.5 and 154.2 mT (as summarized in the Table 2) showing a uniform structure of Co-SCF. As mentioned above, however, the occurring of two sub-hyperfine patterns must have one lone pair electron from some source contacting with one nucleus of 1H [28,37]. Considered in this system of Co/ TiO2 catalyst, this lone pair electron can come from d orbital of TiO2 support [8]. Therefore, based on the hyperfine pattern of ESR spectra, we suggested that the non-reducible compound as a Co-SCF should have the simple
Fig. 4. The consistency between the theory and experimental result of the hyperfine pattern of Co-SCF after standard reduction: (a) the theory showing the contacting between one lone pair electron (S = ±1/2) and the nucleus of 59Co (I = ±7/2); (b) the theory showing the contacting between one lone pair electron (S = ±1/2) and the nucleus of 1H (I = ±1/2); (c) The experimental results from ESR spectrum of sample A after standard reduction obtained at 77 K without irradiation.
1778
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
Fig. 5. ESR spectra of Co/TiO2 after standard reduction obtained at 77 K without irradiation of samples: (a) B and (b) D.
chemical form (which consisted of one Co(0), four 1H, and one TiO2) as a Co0(H4TiO4). The ESR spectra of sample B and D are also presented in Fig. 5a and b, respectively. The hyperfine patterns of sample B and D still had the same pattern with that of sample A. Eight lines of Co(0) can still be observed with the four set of two sub-hyperfine patterns from 1H. When surface of TiO2 had more Ti4+, the Co-SCF occurred still had the same basic chemical form as a Co0(H4TiO4). However, when the surface of TiO2 had more Ti4+ site, the loss of symmetrical pattern can be observed indicating the occur-
ring of non-uniform structure (considered on the Fig. 5a and b and also the error of hyperfine pattern as summarized in Table 2). In case of sample B and D, the center of peak still around 1.409, however, the average spitting distance of eight line of Co(0) became more asymmetry (considered at the error) when the TiO2 had more Ti4+ sites. Considered on the distance from center of two subhyperfine patterns, it should be mentioned that the changing of these distance of 1H to be more asymmetry attributed to the occurring of the random bonding position of hydrogen atom in Co-SCF structure [39]. The observation of non-symmetric set of two sub-hyperfine patterns in the sample D, at A = 107.3 mT (left), 87.5 mT (left), 62.5 mT (left), 108.3 mT (right), and 77 mT (right), can be used to confirm the random bonding position of hydrogen atom resulting in the non-uniform structure of Co-SCF. Therefore, this proposed chemical form as a Co0(H4TiO4) is only a elementary proposed form because more asymmetry and the random peaks can be observed when the surface of TiO2 support having more Ti4+. Therefore, in this work, Co-SCF should be represented on the basic chemical form a Co0(HxTiOy), where x corresponds to the number of two sub-hyperfine patterns and y is the number of oxygen that present for balancing the total oxidation number of this compound to be a zero. Briefly, many researchers mentioned that CoMxO3+x (where x = 1, 2, . . . and M = metal element of support such as Ti, Si) should be the main form of Co-SCF (the details have been summarized in Table 3) [5,6,40–44]. Jongsomjit et al. [6] also mentioned about this chemical form of CoSCF, they reported that the form of Co-SCF seems different from the CoMxO3+x compound. Therefore, it seems that Co-SCF might have many possible chemical structures. However, in this work, based on the promising ESR technique, the simple chemical form of Co-SCF as a Co0(HxTiOy) should be one choice of a possible model for studying the formation of Co-SCF. However, there are the important question that ‘‘How can this compound occur?’’. We tried to find out this question using the proposed pathway mechanism as seen in the simplified Scheme 1. There are two possibly simplified pathways to form this compound during standard reduction. In the pathway A, it have two sub-path ways are as follows: (i) the cationic cobalt (Co(III) and Co(II)) are reduced with hydrogen to form zero-valent cobalt and then (ii) this zero-valent cobalt is facilitated by hydrogen to interact with the TiO2 support and form the Co(0) compound. In addition, based on the pathway B, the cationic cobalt migrate on Ti4+ site of the surface of TiO2 support and agglomerate to form metallic cobalt (zero-valent cobalt) coinciding with the interaction with TiO2 support using hydrogen facilitation and form the Co(0) compound as a non-reducible Co-SCF. However, based on the previous section, the pathway A contradicts with the result of the % reducibility since if it accords with this pathway, it must consume the equal amount of hydrogen, but somehow it does not. Therefore, the possible pathway should be the pathway B which competes with the
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
1779
Table 2 Parameters of the hyperfine pattern from ESR spectra of Co/TiO2 after standard reduction Sample
Centric g tensor
A B D a b c
1.409 1.392 1.401
Cobalt (I = 7/2)a
1
A tensor (mT)
A tensor (mT)b
Distance from center (mT)c
26.9 ± 0.09 21.1 ± 0.16 30.0 ± 0.11
9.9 ± 0.05 10.7 ± 0.12 9.1 ± 0.03
(i) 112.5 ± 0.00, (ii) 154.2 ± 0.05 (i) 105.2 ± 0.11, (ii) 130.5 ± 0.08 (i) 126.0 ± 0.07, (ii) 160.7 ± 0.01
H (I = 1/2)
Calculated from the average 8 peaks of 59Co (I = 7/2). Calculated from the average 4 set of two sub-hyperfine patterns of 1H (I = 1/2). Indicating mainly two symmetrical patterns of 1H (I = 1/2).
Table 3 The proposed chemical form of cobalt-support compound formation by many research works Sample
Co/Al2O3 Co–Ru/Al2O3 Co/Zr–Al2O3 Co/SiO2 Co/TiO2 Co/TiO2 a
Cobalt-support compound formation
Ref.
The proposed chemical form after standard reduction
The detection on the sample after standard reduction
Cobalt aluminate as a CoAl2O4 Cobalt aluminate as a CoAl2O4 Cobalt aluminate as a CoAl2O4 Cobalt silicate as a CoSiO3 Cobalt titanate as a CoTiO3 Co0(HxTiOy)
c.d.a c.d. c.d. c.d. c.d. Electron spin resonance spectroscopy (ESR)
[31,35] [5,32] [33] [34] [6,35] This work
c.d., cannot detect.
alyst. It revealed that the Co-SCF preferred to form on Co/ TiO2 catalyst when the most proportion surface site of TiO2 support became Ti4+. Based on ESR spectra, it was demonstrated that the Co-SCF had a simple chemical form as a Co0(HxTiOy). This compound reached to be a higher non-uniform structure when the number of Ti4+ on TiO2 surface increased. These are the promising knowledge that should be used to develop the characteristics of Co/TiO2 catalyst especially the formation of Co-SCF. The number of Ti4+ should be decreased as much as possible to prevent the formation of such the compound. It might be done treating the TiO2 surface prior to preparing the Co/TiO2 catalyst. Acknowledgements Scheme 1. The simplified mechanism of the Co(0) compound formation.
activation of cobalt (to form the active cobalt) by hydrogen reduction (as seen in the Activation Pathway in Scheme 1). It should be mentioned that the competition between activation pathway and deactivation pathway resulted in a decrease of % activity (as seen in Table 1) when the Ti4+ covering on the surface of TiO2 support increased. 4. Conclusions In this work, the effect of surface species (as a Ti4+ and Ti on TiO2 support) on the cobalt-support compound formation (Co-SCF) was investigated on the Co/TiO2 cat3+
The authors thank the Thailand Research Fund (TRF) and the Thailand-Japan Technology Transfer Project (TJTTP-JBIC) for their financial support of this project. Also thank the Catalysis Research Group in Department of Chemical Engineering and Material Science, University of California, Davis, for IR measurement. References [1] [2] [3] [4] [5]
D.J. Duvenhage, N.J. Coville, Appl. Catal. A 233 (2002) 63. J. Li, N.J. Coville, Appl. Catal. A 181 (1999) 201. N.J. Coville, J. Li, Catal. Today 71 (2002) 403. S. Sun, N. Tsubaki, K. Fujimoto, Appl. Catal. A 202 (2000) 121. Y. Zhang, D. Wei, S. Hammache, J.G. Goodwin Jr., J. Catal. 188 (1999) 281. [6] B. Jongsomjit, C. Sakdamnuson, J.G. Goodwin Jr., P. Praserthdam, Catal. Lett. 94 (2004) 209. [7] A. Kogelbauer, J.C. Weber, J.G. Goodwin Jr., Catal. Lett. 34 (1995) 269.
1780 [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26]
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
U. Diebold, Surf. Sci. Rep. 48 (2003) 53. G. Lu, A. Linsebigler, J.T. Yates Jr., J. Chem. Phys. 98 (1994) 11733. V. Henrich, R.L. Kurtz, Phys. Rev. B 23 (1981) 6280. S. Mezhenny, P. Maksymovych, T.L. Thomson, O. Diwald, D. Stahl, S.D. Walck, J.T. Yates Jr., Chem. Phy. Lett. 369 (2003) 152. J.M. Pan, B.L. Maschhoff, U. Diebold, T.E. Madey, J. Vac. Sci. Technol. A 10 (1992) 2470. I. Nakamura, N. Negishi, S. Kutsuna, T. Ihara, S. Sugihara, K. Takeuchi, J. Mol. Catal. A 161 (2000) 205. K. Suriye, P. Praserthdam, B. Jongsomjit, Ind. Eng. Chem. Res. 44 (2005) 6599. T.L. Thomson, O. Diwald, J.T. Yates Jr., J. Phys. Chem. B 107 (2003) 11700. S.H. Szczepankiewicz, A.J. Colussi, M.R. Hoffmann, J. Phys. Chem. B 104 (2000) 9842. S. Mozia, M. Tomaszewska, B. Kosowska, B. Grzmil, A.W. Morawski, K. Kalucki, Appl. Catal. B 55 (2005) 195. A.S. Vuk, R. Jese, M. Gaberscek, B. Orel, G. Drazic, Sol. Energ. Mater. Sol. Cell. 90 (2006) 452. Y. Nakaoka, Y. Nosaka, J. Photochem. Photobiol. A 110 (1997) 299. E. Serwicka, Coll. Surf. 13 (1985) 287. R.F. Howe, M. Gratzel, J. Phys. Chem. 89 (1985) 4495. S-K Lee, P.K.J. Robertson, A. Mills, D. McStay, N. Elliott, D. McPhail, Appl. Catal. B 44 (2003) 173. X.W. Xing, K.D. Schierbaum, W. Goepel, J. Solid State Chem. 119 (1995) 237. W.T. Wallace, B.K. Min, D.W. Goodman, J. Mol. Catal. A 228 (2005) 3. D.E. Resasco, G.L. Haller, Adv. Catal. 36 (1989) 173. C. Naccache, G. Coudurier, H. Praliaud, P. Meriaudeau, P. Gallezot, G.A. Martin, J.C. Vedrine, Metal-Support and Metal-Additive Effects in Catalysis, Elsevier, New York, 1982.
[27] S. Tyrlik, K. Falkowski, K. Leibler, Inorg. Chim. Acta 6 (1972) 291. [28] V.V. Saraev, Y.S. Levkovskii, A.E. Komolova, V.A. Gruznykh, M. Gomoogiin, F.K. Schmidt, Russ. Chem. Bull. 43 (1994) 149. [29] E. Hayashi, E. Iwamatsu, M.E. Biswas, Y. Sanada, S. Ahmed, H. Hamid, T. Moneda, Appl. Catal. A 179 (1999) 203. [30] F. Gendron, S.C. Garcia, E. Popota, S. Ziolkiewicz, F. Soulette, C. Julia´n, Solid State Ionics 157 (2003) 125. [31] F.K. Shmidt, V.V. Sarayev, Y.S. Levkovskii, V.G. Lipovich, V.A. Gruznykh, G.V. Ratovskii, T.V. Dmitrieva, L.O. Nindakova, React. Kinet. Catal. Lett. 10 (1979) 195. [32] A. Abragam, B. Bleaney, Electron Paramagnetic Resonance of Transition Ions, Clarendon Press, Oxford, 1970. [33] J.E. Wertz, J.E. Bolton, Electron Spin Resonance, McGraw-Hill, New York, 1972. [34] S.K. Misra, S.I. Andronenko, J. Appl. Phys. 99 (2006) 08M106. [35] M.E. Foglio, G.E. Barberis, Braz. J. Phys. 36 (2006) 40. [36] B. Dunn, J.D. Barrie, D. Vivien, J. Mater. Sci. Lett. 6 (1987) 679. [37] H. Itoh, M. Yoshikawa, I. Nashiyama, S. Misawa, H. Okumura, S. Yoshida, IEEE Trans. Nucl. Sci. 37 (1990) 1732. [38] C. Hauser, P. Cornaz, Chem. Phys. Lett. 5 (1970) 226. [39] E.L. Fasanella, W. Gordy, Proc. Natl. Acad. Sci. 64 (1969) 1. [40] B. Jongsomjit, J.G. Goodwin Jr., Catal. Today 77 (2002) 191. [41] B. Jongsomjit, J. Panpranot, J.G. Goodwin Jr., J. Catal. 204 (2001) 98. [42] B. Jongsomjit, J. Panpranot, J.G. Goodwin Jr., J. Catal. 215 (2003) 66. [43] A. Kogelbauer, J.G. Goodwin Jr., R. Oukaci, J. Catal. 160 (1996) 125. [44] M. Voß, D. Borgmann, G. Wedler, J. Catal. 212 (2002) 10.
Effect of surface sites of TiO2 support on the formation of cobalt-support compound in Co/TiO2 catalysts Kongkiat Suriye, Piyasan Praserthdam *, Bunjerd Jongsomjit Center of Excellence on Catalysis and Catalytic Reaction Engineering, Department of Chemical Engineering, Faculty of Engineering, Chulalongkorn University, Bangkok 10330, Thailand Received 25 October 2006; received in revised form 10 January 2007; accepted 12 January 2007 Available online 18 January 2007
Abstract This study has addressed the effect of surface sites (as Ti4+ and Ti3+) of TiO2 support on the formation of the cobalt-support compound on Co/TiO2 catalyst. A 20% cobalt was prepared on each TiO2 support having the different Ti4+/Ti3+ ratios covering on the surface (1.2, 1.3, 1.4, and 1.6). It can be concluded that the non-reducible compound as a Co-SCF preferred to form on the Co/TiO2 catalyst when the most proportional site of surface TiO2 became Ti4+. This is because the migration of cobalt cluster, which plays the significant role for creating Co-SCF, can be promoted by the increasing of Ti4+ sites resulting in the decrease of cobalt dispersion. The hyperfine patterns of ESR spectra demonstrated that the Co0(HxTiOy) was the simple chemical form of Co-SCF formed after standard reduction. Moreover, the structure of Co-SCF changed to be a higher non-uniform structure when the number of Ti4+ site on TiO2 surface increased. The possible mechanism to form this compound has also been discussed. Ó 2007 Elsevier B.V. All rights reserved. Keywords: Cobalt-support compound formation; Co-SCF; Ti4+; Ti3+; Supported cobalt catalyst; Titania
1. Introduction It has been known that the Co/TiO2 catalyst is considered to have a strong metal support interaction (SMSI) and shows high activity in CO hydrogenation [1,2]. This interaction is an important factor used for determining the properties of Co/TiO2 catalyst such as cobalt dispersion and reduction behavior [3]. In fact, the synthesis of highly dispersed cobalt on TiO2 support requires the strong interaction between cobalt and support. However, too strong interaction can produce the Co-support compound as a suboxide at an interface that is high resistant to reduction [4,5]. Jongsomjit et al. [6] reported that the Co-support compound formation (Co-SCF) during standard reduction resulted in a lower reducibility of Co/TiO2 catalyst. The *
Corresponding author. Tel.: +662 218 6883; fax: +662 218 6769. E-mail address: [email protected] (P. Praserthdam).
1566-7367/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.catcom.2007.01.011
compound of cobalt and TiO2 formed referred as ‘‘Co-titanate’’ was considered to be non-reducible form and it revealed that ‘‘Co-titanate’’ formed was likely to be different from CoTiO3. It was also reported that Co-SCF in SiO2 and Al2O3 can occur during standard reduction and resulted in a lower reducibility of catalyst [7]. In case of TiO2-supported cobalt, although, it has been known that the dominant surface sites of TiO2 support consist of two main sites as Ti4+ and Ti3+ [8–13], the effect of surface sites on the formation of Co-SCF has not yet been investigated as well. In our previous work, the impact of Ti3+ present in TiO2 support on catalytic properties of Co/TiO2 was discussed [14]. Therefore, in this work, the proposed research, we aim to extend the investigation to the cobalt-support compound formation on such catalyst. The effect of surface sites (as Ti4+ and Ti3+ on TiO2 support) on the cobalt-support compound formation has been investigated as the main goal. Furthermore, the simple chemical form and
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
1773
the mechanism to form the Co-SCF have also been proposed by means of electron spin resonance spectroscopy (ESR).
tivity detector was used to analyze CO and H2 in the feed and product streams. In all case, steady state was reached within 4 h.
2. Experimental section
3. Results and discussion
2.1. Materials
3.1. Effect of surface sites (Ti4+ and Ti3+) of TiO2 support on the formation of Co-SCF in Co/TiO2 catalyst
A 20 wt% of Co/TiO2 was synthesized by the incipient wetness impregnation [6]. A specific amount of cobalt nitrate [Co(NO3) Æ 6H2O] was dissolved in deionized water and then impregnated onto TiO2 which had a different Ti4+/Ti3+ ratios covering on the surface (the method for controlling the Ti4+/Ti3+ ratio was based on Ref. [14]). The resulting sample was dried at 383 K for 12 h and calcined in air at 723 K for 4 h. 2.2. Nomenclature The nomenclatures used for the Co/TiO2 catalysts in this study are following: A, B, C, D. The supported cobalt on TiO2 which had a Ti4+/Ti3+ molar ratio covering on the surface of 1.2, 1.3, 1.4, and 1.6, respectively. 2.3. Characterization CO2-temperature programmed desorption (TPD) was carried out using a 1 g of a TiO2 sample. The sample was dosed by using 1% CO2 in He for 1 h in liquid nitrogen and then desorbed at the range of temperature from 123 to 253 K by level controlling. A Gow-Mac (Series 150 thermal conductivity detector) gas chromatograph was used to analyze CO2. FT-IR spectra were collected under vacuum with a Bruker IFS 66v Spectrometer with a resolution of 2 cm 1. Measurements were performed by pressing powder between two KBr windows. Each reported spectrum is the average of 64 scans. Temperature programmed reduction (TPR), by 5% by volume H2 in Ar, was carried out using 50 mg of supported cobalt sample and a temperature ramp from 35 to 1073 K at 5 K/min. H2 chemisorption was carried out using a Micromeritics Pulse Chemisorb 2700 system. Prior to chemisorption, the supported cobalt sample was reduced at 623 K for 10 h after ramping at a rate of 1 K/min. Electron spin resonance spectroscopy (ESR) was conducted in vacuum at a 77 K and without illumination using a JEOL, JES-RE2X electron spin resonance spectrometer. CO hydrogenation was carried out at 493 K and 0.182 MPa with a flow rate of H2:CO:He = 20:2:8 ml/min in a fixed-bed flow reactor. Typically, 20 mg of a supported cobalt sample was reduced in situ in flowing H2 (30 ml/min) at 623 K for 10 h prior to the reaction. Reactor effluent samples were taken at 1 h intervals and analyzed by GC. The composition of hydrocarbons in the product stream was analyzed by a Shimadzu GC14B (VZ10) gas chromatograph equipped with a flame ionization detector. A Shimadzu GC8A (molecular sieve 5A) gas chromatography equipped with a thermal conduc-
TiO2 support having a difference of Ti4+ to Ti3+ covering ratios on the surface has been prepared following our previous work [14]. Based on CO2-TPD technique monitoring surface sites of TiO2 (original by Thomson et al. [15]), the Ti4+/Ti3+ ratios were estimated from the ratio between the CO2 desorption peak at temperature 175 K (Ti4+ site) and 200 K (Ti3+ site). The four TiO2 supports with a difference of Ti4+/Ti3+ ratios on the surface (1.2, 1.3, 1.4, and 1.6) were prepared as indicated in Table 1. The infrared spectroscopy (IR) (Fig. 1a) characterizing the OH group bonding to Ti4+ site and Ti3+ site of the three TiO2 supports of the sample A, B, and D was also used to confirm the Ti4+/Ti3+ ratios. As seen in Fig. 1b, IR spectra of these supports exhibited mainly two OH peaks at tOH = 3717 and 3673 cm 1. Based on the report of Szczepankiewicz et al. [16], the IR peak at tOH = 3717 cm 1 represented the OH bonded to Ti3+ site on the surface of TiO2 support. In case of the IR peak at tOH = 3673 cm 1, it was assigned to be the OH bonded to Ti4+ site [17,18]. Compared among the IR spectra of TiO2 supports used for preparing Co/ TiO2 in the sample A, B, and D (as seen in Fig. 1b), showing that the intensity of the band characterizing OH bonded to Ti3+ site (tOH = 3717 cm 1) decreased relatively with the number of OH bonded to Ti4+ site (tOH = 3673 cm 1). The Ti4+/Ti3+ ratios covering on the surface of each TiO2 support, estimated from the ratio between the IR peak at tOH = 3671 and tOH = 3717 cm 1, are also summarized in the Table 1. It shows that the trend of the Ti4+/Ti3+ ratios covering on TiO2 support has a good agreement with those from CO2-TPD results. The electron spin resonance spectroscopy (ESR) as seen in Fig. 2a was also used for confirming the quantitative data from CO2-TPD and IR. ESR spectra, performing at 77 K without illumination on samples, exhibited mainly one signal at the g value of 1.996 for all samples. According to the results as reported by Nakaoka et al., this peak is attributed to the Ti3+ site on TiO2 [19]. This ESR result is consistent with works of Nakamura et al. and Serwicka et al., they reported two signals occurred on the surface of TiO2 during ESR analysis without illumination including: (i) the surface Ti3+ and (ii) the bulk defect [20,13]. However, one signal of ESR result in this work agrees with the works of Howe and Gratzel [21] and Lee et al. [22]. Based on intensity of ESR spectrum as summarized in Table 1 (which corresponds to the amount of Ti3+ present in TiO2 surface), it shows that a decrease in number of Ti3+ corresponded well with an increase of the Ti4+/Ti3+ ratio obtained from CO2-TPD and IR.
1.7 1.6 1.1 0.9 (Ti4+–OH) and tOH = 3716 cm
1
(Ti3+–OH).
16.8 16.2 11.4 9.6 738 729 800 859 66 51 48 35
1
f
g
e
d
c
3631 3024 2898 2885 1.4
0.8 1.2
Calculated from CO2-TPD spectra. The ratio between the intensity of IR spectra at tOH = 3670 cm The intensity of ESR spectra at g value 1.996. Calculated from temperature-programmed-reduction (TPR). The temperature at the maximum point of TPR. Calculated from H2-chemisorption. Rate of CO hydrogenation after 4 h. a
Ti /Ti ratioa
1.2 1.3 1.4 1.6 A B C D
b
% Co dispersionf H2-chemisorbtion (107 molecule/g) TPR temperature (K)e % Reducibilityd Ti –OH/Ti –OH ratio (IR)b
3+ 4+ 3+ 4+
TiO2 support Sample
Table 1 Characteristics of TiO2 supports and Co/TiO2 catalysts
Ti
3+
(ESR)
c
Co/TiO2 catalyst
18 15 13 11
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
Reaction rateg (mmol g 1 s 1)
1774
A 20% cobalt was prepared on each TiO2 support which had a difference of Ti4+/Ti3+ ratios covering on the surface (1.2, 1.3, 1.4, and 1.6 based on CO2-TPD). The characteristics of each Co/TiO2 catalyst are summarized in Table 1. Considered on the% reducibility, when the most proportion of surface site on TiO2 support became Ti4+ site (from A to B, C, and D, consequently), the Co/TiO2 catalyst changed to be a harder reductive catalyst (Co3+ reduced to Co2+, and then reduced to Co0 in final step) as seen by the decrease in number of % reducibility. This is because more amount of non-reducible compound as a Co-SCF was formed on the Co/TiO2 catalyst [6,7]. The reducibility loss (based on sample A) during the standard reduction process was found to be in the range of 8% (for sample B) to 47% (for sample D). Based on the TPR profiles as shown in Fig. 3, it indicates that the reduction peaks (overlap of Co3+ > Co2+ > Co0) shifted to higher temperature when the most proportion on the surface of TiO2 support became Ti4+ site (the changing of temperature is summarized in the Table 1) confirming the existence of Co-SCF on Co/TiO2. Combined between the % reducibility loss and the moving to higher temperature of TPR profile, it can be concluded that the non-reducible compound as a Co-SCF preferred to form on the surface of Co/TiO2 catalyst, when the TiO2 support had more Ti4+ site covering on the surface. Based on the two dominant surface sites of TiO2 support as Ti4+ and Ti3+ [8–13], on the other hand, it can be referred that the Co-SCF preferred to form at the Ti4+ site, but more characterization is still needed to confirm this proposal. There are many reports showing the effect of surface site as Ti4+ and Ti3+ on the adsorption of metal and oxide, as summarized by Diebold [8]. It has been known that the bond energy for anchoring the metal is stronger at the Ti3+ site than that of Ti4+ site [23]. This is because the Ti3+ site has one lone pair electron in d orbital resulting in the higher reactive with the metal and oxide loading [8]. Based on this bond energy, Wallace et al. summarized the role of Ti3+ site on the nucleation, growth, and stability of Au cluster on such site of TiO2 support [24]. It showed that an increase of Ti3+ site strongly decreased the rate of Au cluster migration during high temperature condition due to the strongly bonding between Ti3+ site and Au cluster. The literatures showing the strong bond between Gr. VIII metals and Ti3+ have also been reported by many researchers [8,25,26]. In this work, based on this bond energy, Ti3+ site on the surface of TiO2 support should have the responsibility for fixing the cobalt cluster on the support resulting in a decrease of cobalt migration for forming a bigger size of cobalt. This suggestion can be confirmed by the changing of the cobalt dispersion on the surface of TiO2 support as indicated in Table 1. The increase of cobalt dispersion was observed when the surface of TiO2 support has more Ti3+ sites. This increase of cobalt dispersion due to the Ti3+ site can inhibit the agglomeration between cobalt clusters by the decrease of the migration rate of clusters. Zhang et al. [5] mentioned the effect of cobalt migration on the
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
a
1775
b 3717 A
B
3673
B
C
D
A
4000
3100
2200
1300
-1
Wavenumber, cm
3760
3740
3720
3700
3680
3660
3640
3620
-1
Wavenumber, cm
Fig. 1. The infrared spectroscopy: (a) characterizing the TiO2 supports (prior to cobalt loading) used for sample A, B, D; (b) considering on the OH region.
formation of Co-SCF, they investigated the reducibility of Co–Ru/c-Al2O3 during standard reduction in the presence of added water vapor. They reported that the migration of cobalt played the significant role for the forming of Co-SCF. In this work, the results showing the existence of Co-SCF with the observing the decrease of cobalt dispersion corresponded well with the conclusion of Zhang et al. [5]. Therefore, it can be suggested that the nonreducible compound, as the Co-SCF, preferred to form on Co/TiO2 catalyst having more Ti4+ site on the support due to the increasing of cobalt migration rate. 3.2. Determination of Co-SCF characteristics Based on the previous work in our laboratory, Jongsomjit et al. also investigated to identify the nature of Co-SCF on Co/TiO2 catalyst using several characterization techniques including SEM-EDX, XRD, and Raman spectroscopy [6]. In case of SEM-EDX and XRD, it cannot detect a Co-SCF on the Co/TiO2 because this compound has highly dispersed form. However, Raman spectroscopy revealed that this highly dispersed Co-SCF formed was likely to be different from CoTiO3 and present as the more complex derivatives on Co/TiO2 catalyst. It should be noted that, however, the Co-SCF has still never been directly detected by any techniques. In order to determine the nature of cobalt compound as more complex derivatives from the interaction between cobalt and the other metal compound, the electron spin resonance has been used by many researchers [27–31]. Schmidt et al. [31] investigated the nature of cobalt complexes in Ziegler type catalytic system using ESR and reported that the reduction of cationic cobalt to the zero-valent state can facilitate the formation of paramagnetic Co(II) and Co(0) complexes. However, in case of Co/TiO2 catalyst used for hydrogenation reaction, the evidence of Co-SCF by moni-
toring of ESR has never been done so far. Therefore, in this work, we tried to use the ESR spectroscopy, as a promising technique, for probing the evidence of Co-SCF on Co/TiO2 catalyst. In this work, the ESR investigation Co/TiO2 catalyst prior standard reduction, and Co/TiO2 catalyst after standard reduction were performed to investigate the CoSCF complexes. The ESR spectra of supported cobalt on TiO2 with the different Ti4+/Ti3+ ratios (A, B, and D samples) prior to standard reduction was shown in Fig. 2b. based on analysis over all range of magnetic field (unit in mT), it can detect only the small peak at g value ca. 1.995 which can assign to be Ti3+ site on TiO2 support. Based on the disappearance of Ti3+ site on the surface of TiO2 support (between Fig. 2a and b) , it shows that almost Ti3+ sites were used and participated to anchor the cobalt cluster on TiO2 support resulting in an increase of cobalt dispersion as discussed in the previous section. However, it should be further mentioned that, based on result from% reducibility loss with increased Ti4+ sites, some of cobalt cluster should be participated to anchor at Ti4+ site as well. From the spectra, it can not detect any paramagnetic cobalt species as a Co(II) and Co(0) on these catalyst at g value ca. 2.16 and 1.41, respectively [29]. Based on this analysis of Co/TiO2 prior to standard reduction, although, the Co/ TiO2 was dried at 383 K for 12 h and calcined in air at 723 K for 4 h during catalyst preparation, it shows that any cobalt compound as more complex derivatives from the interaction between cobalt and TiO2 support was not present. This shows that something is missing for the formation of the Co-SCF. The ESR analysis of supported cobalt on TiO2 support with the different Ti4+/Ti3+ ratios (A, B, and D samples) after standard reduction was also performed as shown in Figs. 4c and 5a and b. These spectra were very different from the spectra of Co/TiO2 prior standard reduction.
1776
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
Fig. 2. ESR spectra obtained at 77 K without irradiation for: (a) TiO2 support (prior to cobalt loading) of samples A, B, and D; (b) Co/TiO2 before standard reduction.
These spectra could be assigned as a Co-SCF on the Co/ TiO2 catalyst. Therefore, this shows that the formation of Co-SCF can be facilitated by the presence of hydrogen dur-
ing reduction. Based on ESR theory, these spectra are the hyperfine pattern which result in multi-peak in ESR spectrum [32,33].
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
D
H2 signal (a.u.)
C
B
A
273
373
473
573
673
773
873
973
1073
Temperature (K)
Fig. 3. Temperature-programmed reduction of Co/TiO2 catalyst in the sample A, B, C, and D.
This hyperfine pattern can occur when a lone pair electron (half electron spin) contacts with at lease one nucleus (one nuclear spin) [28,34,35]. The cobalt can be classified to be a low-spin Co(II) with a d7 electron configuration or Co(0) with a d9 electron configuration [28]. Based on this electron configuration, the one lone pair electron in d orbital of both cobalt can be considered to be S = ±1/2 (haft spin of lone pair electron) [34]. Also nuclear spin, cobalt have I = ±7/2 of 59Co with abundance of 100% in the nature [36]. Based on Fig. 4a, when one lone pair electron (S = ±1/2) in the d orbital contacts with one nucleus of 59 Co with I = ±7/2, eight peaks of hyperfine pattern, which symmetrically spitted from center of one lone pair electron in d orbital of cobalt (S = ±1/2), will occur [28,34,35]. Considered on the hyperfine pattern of sample A as also seen in Fig. 4c, the well consistency between the experimental showing hyperfine pattern and ESR theory can be observed in this sample. It was showed that there are eight peaks symmetrically spitted, with equal distance of A = 26.9 mT, from the central peak at g = 1.049 (as summarized in Table 2). This center attributed to the electron spin of one lone pair electron (S = ±1/2) in d orbital of Co(0) [29]. Based on the consistency between the hyperfine pattern from experimental and theory, it can be mentioned that non-reducible compound formed after standard reduction had a form of Co(0) compound. In this hyperfine pattern, there are two patterns in hyperfine spectra (as seen in Fig. 4c), one is attributed to 59Co with a larger A value and the smaller one (A = 9.9 mT) of four sets of two sub-hyperfine patterns. Saraev et al. [28] also found the two patterns in hyperfine spectra of Co(acac)2–PBu3-complexes. One was the eight lines of a 59Co (I = ±7/2) with a larger A value and the smaller one of A value (of two sub-hyperfine peak) which was attributed to 31P (I = ±1/2). Two subhyperfine patterns came from the contacting between one lone pair electron (S = ±1/2) and one nucleus of I = ±1/2 as seen in Fig. 4b [28]. Based on the system of this Co/ TiO2 catalyst after standard reduction with hydrogen, these
1777
two sub-hyperfine patterns should come from the nucleus of 1H (I = ±1/2) with abundance of 99.98% in nature [37]. It can not come from the nucleus of 47Ti and 49Ti (abundance of 7.44% and 5.41%, consequently) because they really have the less abundance in the nature. And if it came from 47Ti and 49Ti, these sub-hyperfine pattern must have six or eight lines (with equal of distance from center) of I = ±5/2 and ±7/2, respectively [38]. These four sets of 1H can be classified to be a two symmetrical pattern with a distance from central peak equal to A = 112.5 and 154.2 mT (as summarized in the Table 2) showing a uniform structure of Co-SCF. As mentioned above, however, the occurring of two sub-hyperfine patterns must have one lone pair electron from some source contacting with one nucleus of 1H [28,37]. Considered in this system of Co/ TiO2 catalyst, this lone pair electron can come from d orbital of TiO2 support [8]. Therefore, based on the hyperfine pattern of ESR spectra, we suggested that the non-reducible compound as a Co-SCF should have the simple
Fig. 4. The consistency between the theory and experimental result of the hyperfine pattern of Co-SCF after standard reduction: (a) the theory showing the contacting between one lone pair electron (S = ±1/2) and the nucleus of 59Co (I = ±7/2); (b) the theory showing the contacting between one lone pair electron (S = ±1/2) and the nucleus of 1H (I = ±1/2); (c) The experimental results from ESR spectrum of sample A after standard reduction obtained at 77 K without irradiation.
1778
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
Fig. 5. ESR spectra of Co/TiO2 after standard reduction obtained at 77 K without irradiation of samples: (a) B and (b) D.
chemical form (which consisted of one Co(0), four 1H, and one TiO2) as a Co0(H4TiO4). The ESR spectra of sample B and D are also presented in Fig. 5a and b, respectively. The hyperfine patterns of sample B and D still had the same pattern with that of sample A. Eight lines of Co(0) can still be observed with the four set of two sub-hyperfine patterns from 1H. When surface of TiO2 had more Ti4+, the Co-SCF occurred still had the same basic chemical form as a Co0(H4TiO4). However, when the surface of TiO2 had more Ti4+ site, the loss of symmetrical pattern can be observed indicating the occur-
ring of non-uniform structure (considered on the Fig. 5a and b and also the error of hyperfine pattern as summarized in Table 2). In case of sample B and D, the center of peak still around 1.409, however, the average spitting distance of eight line of Co(0) became more asymmetry (considered at the error) when the TiO2 had more Ti4+ sites. Considered on the distance from center of two subhyperfine patterns, it should be mentioned that the changing of these distance of 1H to be more asymmetry attributed to the occurring of the random bonding position of hydrogen atom in Co-SCF structure [39]. The observation of non-symmetric set of two sub-hyperfine patterns in the sample D, at A = 107.3 mT (left), 87.5 mT (left), 62.5 mT (left), 108.3 mT (right), and 77 mT (right), can be used to confirm the random bonding position of hydrogen atom resulting in the non-uniform structure of Co-SCF. Therefore, this proposed chemical form as a Co0(H4TiO4) is only a elementary proposed form because more asymmetry and the random peaks can be observed when the surface of TiO2 support having more Ti4+. Therefore, in this work, Co-SCF should be represented on the basic chemical form a Co0(HxTiOy), where x corresponds to the number of two sub-hyperfine patterns and y is the number of oxygen that present for balancing the total oxidation number of this compound to be a zero. Briefly, many researchers mentioned that CoMxO3+x (where x = 1, 2, . . . and M = metal element of support such as Ti, Si) should be the main form of Co-SCF (the details have been summarized in Table 3) [5,6,40–44]. Jongsomjit et al. [6] also mentioned about this chemical form of CoSCF, they reported that the form of Co-SCF seems different from the CoMxO3+x compound. Therefore, it seems that Co-SCF might have many possible chemical structures. However, in this work, based on the promising ESR technique, the simple chemical form of Co-SCF as a Co0(HxTiOy) should be one choice of a possible model for studying the formation of Co-SCF. However, there are the important question that ‘‘How can this compound occur?’’. We tried to find out this question using the proposed pathway mechanism as seen in the simplified Scheme 1. There are two possibly simplified pathways to form this compound during standard reduction. In the pathway A, it have two sub-path ways are as follows: (i) the cationic cobalt (Co(III) and Co(II)) are reduced with hydrogen to form zero-valent cobalt and then (ii) this zero-valent cobalt is facilitated by hydrogen to interact with the TiO2 support and form the Co(0) compound. In addition, based on the pathway B, the cationic cobalt migrate on Ti4+ site of the surface of TiO2 support and agglomerate to form metallic cobalt (zero-valent cobalt) coinciding with the interaction with TiO2 support using hydrogen facilitation and form the Co(0) compound as a non-reducible Co-SCF. However, based on the previous section, the pathway A contradicts with the result of the % reducibility since if it accords with this pathway, it must consume the equal amount of hydrogen, but somehow it does not. Therefore, the possible pathway should be the pathway B which competes with the
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
1779
Table 2 Parameters of the hyperfine pattern from ESR spectra of Co/TiO2 after standard reduction Sample
Centric g tensor
A B D a b c
1.409 1.392 1.401
Cobalt (I = 7/2)a
1
A tensor (mT)
A tensor (mT)b
Distance from center (mT)c
26.9 ± 0.09 21.1 ± 0.16 30.0 ± 0.11
9.9 ± 0.05 10.7 ± 0.12 9.1 ± 0.03
(i) 112.5 ± 0.00, (ii) 154.2 ± 0.05 (i) 105.2 ± 0.11, (ii) 130.5 ± 0.08 (i) 126.0 ± 0.07, (ii) 160.7 ± 0.01
H (I = 1/2)
Calculated from the average 8 peaks of 59Co (I = 7/2). Calculated from the average 4 set of two sub-hyperfine patterns of 1H (I = 1/2). Indicating mainly two symmetrical patterns of 1H (I = 1/2).
Table 3 The proposed chemical form of cobalt-support compound formation by many research works Sample
Co/Al2O3 Co–Ru/Al2O3 Co/Zr–Al2O3 Co/SiO2 Co/TiO2 Co/TiO2 a
Cobalt-support compound formation
Ref.
The proposed chemical form after standard reduction
The detection on the sample after standard reduction
Cobalt aluminate as a CoAl2O4 Cobalt aluminate as a CoAl2O4 Cobalt aluminate as a CoAl2O4 Cobalt silicate as a CoSiO3 Cobalt titanate as a CoTiO3 Co0(HxTiOy)
c.d.a c.d. c.d. c.d. c.d. Electron spin resonance spectroscopy (ESR)
[31,35] [5,32] [33] [34] [6,35] This work
c.d., cannot detect.
alyst. It revealed that the Co-SCF preferred to form on Co/ TiO2 catalyst when the most proportion surface site of TiO2 support became Ti4+. Based on ESR spectra, it was demonstrated that the Co-SCF had a simple chemical form as a Co0(HxTiOy). This compound reached to be a higher non-uniform structure when the number of Ti4+ on TiO2 surface increased. These are the promising knowledge that should be used to develop the characteristics of Co/TiO2 catalyst especially the formation of Co-SCF. The number of Ti4+ should be decreased as much as possible to prevent the formation of such the compound. It might be done treating the TiO2 surface prior to preparing the Co/TiO2 catalyst. Acknowledgements Scheme 1. The simplified mechanism of the Co(0) compound formation.
activation of cobalt (to form the active cobalt) by hydrogen reduction (as seen in the Activation Pathway in Scheme 1). It should be mentioned that the competition between activation pathway and deactivation pathway resulted in a decrease of % activity (as seen in Table 1) when the Ti4+ covering on the surface of TiO2 support increased. 4. Conclusions In this work, the effect of surface species (as a Ti4+ and Ti on TiO2 support) on the cobalt-support compound formation (Co-SCF) was investigated on the Co/TiO2 cat3+
The authors thank the Thailand Research Fund (TRF) and the Thailand-Japan Technology Transfer Project (TJTTP-JBIC) for their financial support of this project. Also thank the Catalysis Research Group in Department of Chemical Engineering and Material Science, University of California, Davis, for IR measurement. References [1] [2] [3] [4] [5]
D.J. Duvenhage, N.J. Coville, Appl. Catal. A 233 (2002) 63. J. Li, N.J. Coville, Appl. Catal. A 181 (1999) 201. N.J. Coville, J. Li, Catal. Today 71 (2002) 403. S. Sun, N. Tsubaki, K. Fujimoto, Appl. Catal. A 202 (2000) 121. Y. Zhang, D. Wei, S. Hammache, J.G. Goodwin Jr., J. Catal. 188 (1999) 281. [6] B. Jongsomjit, C. Sakdamnuson, J.G. Goodwin Jr., P. Praserthdam, Catal. Lett. 94 (2004) 209. [7] A. Kogelbauer, J.C. Weber, J.G. Goodwin Jr., Catal. Lett. 34 (1995) 269.
1780 [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26]
K. Suriye et al. / Catalysis Communications 8 (2007) 1772–1780
U. Diebold, Surf. Sci. Rep. 48 (2003) 53. G. Lu, A. Linsebigler, J.T. Yates Jr., J. Chem. Phys. 98 (1994) 11733. V. Henrich, R.L. Kurtz, Phys. Rev. B 23 (1981) 6280. S. Mezhenny, P. Maksymovych, T.L. Thomson, O. Diwald, D. Stahl, S.D. Walck, J.T. Yates Jr., Chem. Phy. Lett. 369 (2003) 152. J.M. Pan, B.L. Maschhoff, U. Diebold, T.E. Madey, J. Vac. Sci. Technol. A 10 (1992) 2470. I. Nakamura, N. Negishi, S. Kutsuna, T. Ihara, S. Sugihara, K. Takeuchi, J. Mol. Catal. A 161 (2000) 205. K. Suriye, P. Praserthdam, B. Jongsomjit, Ind. Eng. Chem. Res. 44 (2005) 6599. T.L. Thomson, O. Diwald, J.T. Yates Jr., J. Phys. Chem. B 107 (2003) 11700. S.H. Szczepankiewicz, A.J. Colussi, M.R. Hoffmann, J. Phys. Chem. B 104 (2000) 9842. S. Mozia, M. Tomaszewska, B. Kosowska, B. Grzmil, A.W. Morawski, K. Kalucki, Appl. Catal. B 55 (2005) 195. A.S. Vuk, R. Jese, M. Gaberscek, B. Orel, G. Drazic, Sol. Energ. Mater. Sol. Cell. 90 (2006) 452. Y. Nakaoka, Y. Nosaka, J. Photochem. Photobiol. A 110 (1997) 299. E. Serwicka, Coll. Surf. 13 (1985) 287. R.F. Howe, M. Gratzel, J. Phys. Chem. 89 (1985) 4495. S-K Lee, P.K.J. Robertson, A. Mills, D. McStay, N. Elliott, D. McPhail, Appl. Catal. B 44 (2003) 173. X.W. Xing, K.D. Schierbaum, W. Goepel, J. Solid State Chem. 119 (1995) 237. W.T. Wallace, B.K. Min, D.W. Goodman, J. Mol. Catal. A 228 (2005) 3. D.E. Resasco, G.L. Haller, Adv. Catal. 36 (1989) 173. C. Naccache, G. Coudurier, H. Praliaud, P. Meriaudeau, P. Gallezot, G.A. Martin, J.C. Vedrine, Metal-Support and Metal-Additive Effects in Catalysis, Elsevier, New York, 1982.
[27] S. Tyrlik, K. Falkowski, K. Leibler, Inorg. Chim. Acta 6 (1972) 291. [28] V.V. Saraev, Y.S. Levkovskii, A.E. Komolova, V.A. Gruznykh, M. Gomoogiin, F.K. Schmidt, Russ. Chem. Bull. 43 (1994) 149. [29] E. Hayashi, E. Iwamatsu, M.E. Biswas, Y. Sanada, S. Ahmed, H. Hamid, T. Moneda, Appl. Catal. A 179 (1999) 203. [30] F. Gendron, S.C. Garcia, E. Popota, S. Ziolkiewicz, F. Soulette, C. Julia´n, Solid State Ionics 157 (2003) 125. [31] F.K. Shmidt, V.V. Sarayev, Y.S. Levkovskii, V.G. Lipovich, V.A. Gruznykh, G.V. Ratovskii, T.V. Dmitrieva, L.O. Nindakova, React. Kinet. Catal. Lett. 10 (1979) 195. [32] A. Abragam, B. Bleaney, Electron Paramagnetic Resonance of Transition Ions, Clarendon Press, Oxford, 1970. [33] J.E. Wertz, J.E. Bolton, Electron Spin Resonance, McGraw-Hill, New York, 1972. [34] S.K. Misra, S.I. Andronenko, J. Appl. Phys. 99 (2006) 08M106. [35] M.E. Foglio, G.E. Barberis, Braz. J. Phys. 36 (2006) 40. [36] B. Dunn, J.D. Barrie, D. Vivien, J. Mater. Sci. Lett. 6 (1987) 679. [37] H. Itoh, M. Yoshikawa, I. Nashiyama, S. Misawa, H. Okumura, S. Yoshida, IEEE Trans. Nucl. Sci. 37 (1990) 1732. [38] C. Hauser, P. Cornaz, Chem. Phys. Lett. 5 (1970) 226. [39] E.L. Fasanella, W. Gordy, Proc. Natl. Acad. Sci. 64 (1969) 1. [40] B. Jongsomjit, J.G. Goodwin Jr., Catal. Today 77 (2002) 191. [41] B. Jongsomjit, J. Panpranot, J.G. Goodwin Jr., J. Catal. 204 (2001) 98. [42] B. Jongsomjit, J. Panpranot, J.G. Goodwin Jr., J. Catal. 215 (2003) 66. [43] A. Kogelbauer, J.G. Goodwin Jr., R. Oukaci, J. Catal. 160 (1996) 125. [44] M. Voß, D. Borgmann, G. Wedler, J. Catal. 212 (2002) 10.