Hydrogenation of sunflower oil over Pt–Ni bimetallic supported catalysts: Preparation, characterization and catalytic activity

Applied Catalysis A: General 474 (2014) 78–86

Contents lists available at ScienceDirect

Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata

Hydrogenation of sunflower oil over Pt–Ni bimetallic supported catalysts: Preparation, characterization and catalytic activity Shane McArdle a,b , J.J. Leahy a,b , Teresa Curtin a,b,∗ , David Tanner a a b

Materials and Surface Science Institute, University of Limerick, Ireland Chemical and Environmental Science Department, University of Limerick, Ireland

a r t i c l e

i n f o

Article history: Received 20 March 2013 Received in revised form 16 August 2013 Accepted 17 August 2013 Available online 26 August 2013 Keywords: Pt–Ni supported on silica Hydrogenation of sunflower oil Trans fatty acids Bimetallic catalysts

a b s t r a c t It has been reported that in order to improve sunflower oil hydrogenation catalysts the geometric or electronic effects of the catalysts must be manipulated in order to achieve the desired activity or selectivity. To achieve this, the development of an active and selective bimetallic catalyst was undertaken. An innovative approach was taken to synthesize Pt–Ni catalysts supported on mesoporous silica using the surface redox reaction (Srr) technique. It was determined that the Srr preparation method resulted in better activity and lower selectivity towards trans isomer than the traditional successive and co-impregnation bimetallic impregnation techniques. The change in selectivity can be explained by a promoter electronic effect generated by the close proximity of a second metal on the catalysts surface and from a geometric effect due to the incorporation of the Ni on the surface. All the bimetallic catalysts showed a drop in the formation of trans compared to the monometallic catalyst. © 2013 Elsevier B.V. All rights reserved.

1. Introduction The field of catalytic hydrogenation of edible oil has seen a lot of advances in recent years, especially regarding efforts to reduce the formation of trans fats. These trans fats are associated with increased risk of coronary heart disease. Researchers have investigated the use of different metals, mainly noble metals, to improve the activity and selectivity of the catalyst [1–7]. Manipulation of the catalytic structure has also been studied in an effort to achieve tighter control of the reaction [8,9]. It has been found that, with some exceptions, the variations of properties achieved on different supports are not as notable as the differences among the metals themselves [10,11]. In order to further promote the selectivity toward C18:1 cis, noble metals have been modified by the addition of amines in the reaction medium and by metal additives. The influence of these additives has shown increased selectivity to cis isomers and in some cases without loss of activity for various hydrogenation reactions [5,12,13]. Rubin et al. [14] presented one of the first studies using a mixed metal system containing Ni and methyl benzoate chromium tricarbonyl for edible oil hydrogenation. They showed it was possible to retain the advantages of both catalytic metals while using them in combination. Adding Ni to Ru resulted in improved activity and

∗ Corresponding author at: Materials and Surface Science Institute, Department of Chemical and Environmental Sciences, University of Limerick, Ireland. Tel.: +353 61 202981; fax: +353 61 202568. E-mail address: [email protected] (T. Curtin). 0926-860X/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.apcata.2013.08.033

relatively low trans isomer formation. At an iodine value (I.V.) of 60 only 7% trans isomers were formed (110 ◦ C, 57.2 bar, 580 rpm). However, Ru is oil soluble and is difficult to recover at the end of hydrogenation. In addition, the operating conditions required relatively high hydrogen pressure (57.2 bar). The catalytic properties of bimetallic catalysts usually deviate from the additive properties of the metal components. This very often results in activities and selectivities that exceed those that can be achieved by using the simple monometallic catalyst. There have been many experimental and fundamental studies reported aiming to investigate the origin of this interesting effect using model surfaces [15–17]. This catalytic phenomenon can be sometimes understood in terms of the so-called ligand (electronic) effect in which unanticipated catalytic effects are the result of a change in the electron structure of the catalyst. Another important factor is the average metal–metal bond length will be different to that of the parent metals and this will result in a strain effect, modifying the electronic structure of the metal through changes in orbital overlap [18,19]. For the Pt/Ni bimetallic system, it has been shown that a Pt enriched surface exists with a subsurface of Ni and this structure is very active for low temperature hydrogenation reactions [16,20]. While electronic effects has been used to explain the influence on the selectivity and activity of bimetallic catalysts, the geometric (ensemble) effect at a surface may also play a crucial role in changing catalytic characteristics, particularly in for real catalysts. Here, the second metal may block sites on the surface of an active metal and by that means, the average size and composition of the ensemble of active sites is varied [21]. It is also recognized that a combination of electronic and geometric effects can explain

S. McArdle et al. / Applied Catalysis A: General 474 (2014) 78–86

the catalytic behavior of some bimetallic systems [22–24]. Other studies have reported on the effect on the reducibility of a metal in the presence of the second metal. For example, for the Ni/Pt system, the reducibility of supported NiO can be enhanced by adding a noble metal such as Pt [25–27]. In studies by Nohair et al. [4,5] the modification of Pd/SiO2 by the addition of Pb via surface redox reaction (Srr) improved the selectivity to cis C18:1 , with a small decrease in conversion. Fernández et al. [28] prepared Pd–Me/Al2 O3 catalysts where Me = molybdenum (Mo), vanadium (V) and lead (Pb). All the catalysts were prepared using successive impregnation (S-i). It was shown that the Pd–Mo/Al2 O3 and Pd–V/Al2 O3 catalysts presented the same activity as that of the respective monometallic Pd/Al2 O3 catalyst, but an increased selectivity to trans-isomers. The Pd–Pb/Al2 O3 catalyst showed a low hydrogenation activity. This drop in activity was thought to be a result of certain dimensional limitations on the space lattice of Pd for hydrogenation of double bonds due to the formation of a Pd–Pb alloy. Wright et al. explored the use of homogeneous and heterogeneous bimetallic systems for hydrogenation of vegetable oils [29,30]. They concluded that there was no advantage in combining Pd and MeBeCr(CO)3 . The homogeneous catalyst inhibited the activity of Pd and a high level of trans isomers were formed. The addition of Ni to Pd was more promising, resulting in increased catalyst activity but no significant difference in cis-selectivity was observed. It was postulated that poison adsorption by the Ni on the Ni–Pd catalyst accounted for the increase in activity. Encouraged by these reported results, this work examines the hydrogenation of sunflower oil using Pt/SiO2 catalysts promoted with Ni. The study investigates the dependence of activity and selectivity on the support characteristics, the bimetal preparation and metal precursors. The main objective is to develop a catalyst that is active while lowering the trans-isomer content in the hydrogenated sunflower oil.

2. Experimental 2.1. Catalysts preparation The mesoporous silica material, (CNS SiO2 ), was synthesized according to a procedure reported in the literature [31]. Briefly, 50 g CTAB (cetyl trimethyl ammonium bromide) was dissolved in 450 g H2 O at 30 ◦ C, after which 50 ml TEOS (tetraethoxysilane), 6.25 ml CEOS (2-cyanoethyltriethoxysilane) and 2.5 ml liquid NH4 OH (all from Sigma–Aldrich) were added. The resulting gel mixture was then heated in an autoclave at 105 ◦ C for 24 h. The product obtained was filtered, washed with water and refluxed in a methanol solution containing 1.5% HCl and 2.5% H2 O for 2 h, followed by a methanol wash. This process (washing/refluxing) was repeated 4 times, increasing the time of reflux each time by 1 h. The resulting solid was calcined at 650 ◦ C (ramp rate 1 ◦ C min−1 ) for 6 h under a stream of air (100 ml min−1 ). The supported bimetallic catalysts were prepared using the CNS SiO2 support following three different methods; (i) co-impregnation, (ii) successive impregnation and (iii) surface redox reaction. In all cases the intended loading on the support was 2 wt% Pt and 1 wt% Ni.

(i) Co-impregnation (Co-i): 0.041 g of platinum acetylacetonate (Pt(AcAc)2 ) and 0.049 g nickel nitrate hexahydrate Ni(NO3 )2 ·6H2 O (both supplied by Sigma–Aldrich), were dissolved in 50 ml ethanol and 1 g SiO2 support added. This suspension was stirred for 10 h, after which the solvent was removed under vacuum. The sample was then dried overnight at 80 ◦ C and finally calcined in air at 450 ◦ C for 5 h.

79

(ii) Successive impregnation (S-i): 0.041 g of Pt(AcAc)2 was dissolved in 50 ml ethanol and 1 g of the support (CNS SiO2 ) was then added. After stirring for 10 h, the solvent was removed under vacuum and the sample was dried and calcined as described above. The second metal, nickel, was then impregnated using Ni(NO3 )2 ·6H2 O. 0.049 g of the Ni salt, dissolved in 50 ml ethanol, was mixed with 1 g of the prepared Pt/SiO2 catalyst. This suspension was allowed to stir for 10 h before the solvent was removed and subsequently dried. Finally the solid was calcined in air at 450 ◦ C for 5 h. (iii) Surface redox reaction (Srr): this preparation method, based on a procedure reported by Nohair et al. [4], is known to induce a strong interaction between two metals. Initially, the monometallic Pt/SiO2 catalyst was synthesized by wetimpregnation, as described above. This calcined monometallic catalyst was then introduced into a Parr high pressure reactor (model 4560) under argon and was activated under hydrogen at 200 ◦ C for 4 h. A degassed solution containing 0.049 g Ni(NO3 )2 ·6H2 O dissolved in 50 ml methanol was then introduced onto the catalyst at room temperature. The suspension was stirred and maintained under a H2 flow for 90 min. The solution was then filtered and dried overnight at 80 ◦ C. 2.2. Hydrogenation of sunflower oil The hydrogenation of sunflower oil (Flora) was carried out using a 600 ml Parr pressure reactor, model 4560 (Parr Instrument Co., Moline, IL, USA) with a double agitator. 150 mg catalyst was reduced in situ by placing the pre-weighted catalyst in the reactor vessel and then passing hydrogen gas over it at 200 ◦ C for 2–3 h. It was then cooled and the system was purged with argon before the sunflower oil (Flora) was introduced (100 ml). The temperature of the system was then increased gradually to 170 ◦ C. Once the temperature was at the required level, hydrogen was introduced and this point was referred to as “time = 0 . The system was agitated using a stirring rate of 400 rpm. The hydrogen pressure was maintained constant at 3 bar throughout the reaction. Oil samples were taken from the reactor every hour and the entire reaction time was 5 h. These partially hydrogenated sunflower oil samples were analyzed for their iodine value (IV) according to EN ISO 3961 Wijs method [32]. The iodine value of an oil or fat is a measure of unsaturation and was used to monitor the activity of the catalysts. The fatty acid composition of the hydrogenated oils was determined by gas chromatography (GC) using a procedure described elsewhere [11]. All hydrogenation experiments were carried out in triplicate and an average of their results reported. 2.3. Catalyst characterization XRD patterns were obtained using a Philips X’Pert Pro X-ray ˚ diffractometer with nickel filtered K␣ Cu radiation ( = 1.542 A) between the X-ray source angles of 10◦ and 70◦ . The specific surface area measurements and the pore size distributions of the catalytic materials were determined from N2 adsorption/desorption experiments at −196 ◦ C using a Micrometrics ASAP 2010 system. The pore volume and pore size distribution were calculated using the Barrett–Joyner–Halenda theory from the adsorption isotherm. A Varian SpectrAA Atomic Absorption Spectrometer was used to measure the Pt and Ni contents of the prepared catalysts. A Jeol 2011 Transmission electron microscope was used to obtain high resolution images of the different catalysts. The imaging program, Digital MicrographTM , was used to visualize and analyze digital image data from the electron microscope. The software was calibrated against internal standards and used to calculate the average metal particle size by sampling at least 100 particles.

80

S. McArdle et al. / Applied Catalysis A: General 474 (2014) 78–86

Table 1 Support and catalyst properties. Catalyst name

Specific surface area (m2 g−1 )

Pore diameter (nm)

Pore volume (cm3 g−1 )

Actual Pt metal loading (%)

Actual Ni metal loading (%)

Particle size (nm) TEMa

SiO2 1.57%Pt/SiO2 1.16%Pt–0.41%Ni/SiO2 1.57%Pt–0.72%Ni/SiO2 0.83%Ni–1.37%Pt/SiO2 1.56%Pt–0.95%Ni/SiO2

601 541 467 431 443 518

18.9 18.4 18.4 15.6 16.1 17.9

1.53 1.41 1.30 1.08 1.11 1.37

1.57 1.16 1.57 1.37 1.56

0.41 0.72 0.83 0.95

3.0 17.8 17.1 21.7 7.9

Number-average particle size. Determined by counting a minimum of 100 free standing particles.

3. Results and discussion 3.1. Promotional effects of Ni on Pt catalysts using different preparation methods Table 1 presents the textural properties and metal loading of the SiO2 support, the monometallic Pt/SiO2 catalyst and a series of supported bimetallic Pt–Ni catalysts. The nomenclature of the catalysts includes the measured amount of metal supported onto the SiO2 and the preparation method used (co-impregnation (Coi), successive impregnation (S-i) and surface redox reaction (Srr)). The 1.57%Pt–0.72%Ni/SiO2 (S-i) catalyst was prepared by successive impregnation where the Pt metal was loaded on first and then the Ni. Correspondingly the 0.83%Ni–1.37%Pt/SiO2 (S-i) was prepared using successive impregnation but the Ni was added first. The B.E.T. analysis of the silica support showed a specific surface area of 601 m2 g−1 , while the impregnated Pt/SiO2 catalyst presented a surface area of 541 m2 g−1 . The average pore diameter of the support was found to be 18.9 nm, signifying mesoporous characteristics. The slight decrease in pore diameter and pore volume for the 1.57% Pt/SiO2 in comparison to the support may indicate that either minor sintering of the support had occurred due to the recalcination step or some of the metal particles may have blocked a fraction of the pores. The co-impregnated and successive impregnated catalysts showed the largest decrease in surface area compared to the original support. This may indicate that the metal particles were blocking a fraction of the pores on the support. In the case of the catalysts prepared by successive impregnation, the recalcination stage may have affected the textural properties such that, it may have led to sintering of the pores. Not only was the surface area lower but the average pore diameter was also found to have decreased compared to the other methods of preparation. Table 1 also presents the average particle sizes using TEM analysis. For the monometallic 1.57%Pt/SiO2 catalyst, the Pt particles averaged 3.0 nm in size. This correlates well with published data [33,34]. Large metal particles were formed when Pt and Ni were prepared by co-impregnation. The addition of the Ni with the Pt metal together may have resulted in the reduced dispersion and increase in overall particle size. Malyala et al. [35] reported that segregates of Ni particles of different sizes (21–35 nm) and shapes are formed by wet impregnation, leading to a lower degree of dispersion. The large particles present on the catalysts prepared by successive impregnation could have arisen because of the recalcination step. A decrease in the interfacial energy at the metal-oxide/support interface allows the metal particles to freely move and aggregate. This metal re-dispersion and agglomeration is well documented in the literature [23,36,37]. All the catalysts were tested for the hydrogenation of sunflower oil. The hydrogenation activity was monitored by the decay of iodine value (IV), while the GC results show the change in the fatty acid compositions over hydrogenation time. A stirring rate of 400 rpm was selected to ensure that the reduction of the IV was independent of inter-particle mass transfer. Fig. 1 shows the

drop in IV for the indicated catalysts over time. A monometallic 1 wt% Ni/SiO2 catalyst (not shown) was tested for the hydrogenation of oil and little or no drop in iodine value was observed with hydrogenation time (the iodine value dropped to 128.5 after 5 h hydrogenation). The 1.57%Pt/SiO2 monometallic catalyst showed the highest activity of all the catalysts tested reaching an IV of 51 after 5 h. Comparatively, the bimetallic catalysts were less active following the addition of Ni to Pt. The traditional techniques of impregnation (Co and successive) produced less active catalysts, with 1.16%Pt–0.41%Ni/SiO2 (Co-i) achieving an IV of 74 after the 5 h. The 1.57%Pt–0.72%Ni/SiO2 (S-i) and 0.83%Ni–1.37%Pt/SiO2 (S-i) achieved IV’s of 71 and 69, respectively. The Srr catalyst showed the highest activity of all the bimetallic catalysts, with an IV value of 58 after the 5 h. These results indicate that the addition of Ni decreases the activity of the Pt catalysts and can be explained by the deposition of less active Ni atoms on the surface of the platinum particles, limiting the accessibility to Pt active sites. The severe inhibition of the catalyst activity by the traditional impregnation methods suggests a strong interaction between the two metals. The change in linoleate (C18:2 ), oleate (C18:1 cis and C18:1 trans) and stearate (C18:0 ) components of sunflower oil with time for the mono and bimetallic catalysts are presented in Fig. 2. For all catalysts, the linoleate is consumed, while a steady increase in the C18:1 trans and stearate content is observed with time. The vertical line indicates the time required to hydrogenate the sunflower oil to an iodine value of 70. At this value, the 1.57%Pt/SiO2 catalyst (Fig. 2a) presented a trans content of 26.5% and a stearate value of 17.8%. The four bimetallic catalysts showed varying results for trans formation. At an IV of 70 the co-impregnated bimetallic catalyst formed 23% trans (Fig. 2b). The 1.57%Pt–0.72%Ni/SiO2 (S-i) and 0.83%Ni–1.37%Pt/SiO2 (S-i) catalysts produced 23.2% and 18.5%

1.57 1.56 1.16 1.57 0.83

140 120

Iodine Value

a

(Co-i) (S-i) (S-i) (Srr)

Pt/SiO2 Pt-0.95 Pt-0.41 Pt-0.72 Ni-1.37

Ni/SiO2 (Srr) Ni/SiO2 (Co-i) Ni/SiO2 (S-i) Pt/SiO2 (S-i)

100 80 60 40

0

1

2

3

4

5

Time (h) Fig. 1. Drop in IV over hydrogenation time for indicated catalysts (170 ◦ C, 3 bar H2 , 400 rpm, 100 ml sunflower oil, 150 mg catalyst).

100

C18:0 C18:1trans C18:1cis C18:2

a)

90 80 70

Relative Composition (%)

Relative Composition (%)

S. McArdle et al. / Applied Catalysis A: General 474 (2014) 78–86

60 50 40 30 20 10 0

0

1

2

3

4

100

C18:0 C18:1trans C18:1cis C18:2

b)

90 80 70 60 50 40 30 20 10 0

5

81

0

1

2

C18:0 C18:1trans C18:1cis C18:2

c)

80 70

Relative Composition (%)

Relative Composition (%)

100

60 50 40 30 20 10 0

0

1

2

3

4

100

5

80 70 60 50 40 30 20 10 0

5

C18:0 C18:1trans C18:1cis C18:2

d)

90

0

1

2

Time (h)

Relative Composition (%)

4

Time (h)

Time (h) 90

3

3

4

5

Time (h)

100

C18:0 C18:1trans C18:1cis C18:2

e)

90 80 70 60 50 40 30 20 10 0

0

1

2

3

4

5

Time (h) Fig. 2. Change of sunflower oil composition with time using (a) 1.57%Pt/SiO2 , (b) 1.16%Pt–0.41%Ni/SiO2 (Co-i), (c) 1.57%Pt–0.72%Ni/SiO2 (S-i), (d) 0.83%Ni–1.37%Pt/SiO2 (S-i) and (e) 1.56%Pt–0.95%Ni/SiO2 (Srr) (170 ◦ C, 3 bar H2 , 400 rpm, 100 ml sunflower oil, 150 mg catalyst).

trans, respectively. It seems that the order in which the metals are added influence the trans selectivity. The amount of metals present may also affect the formation of the trans isomers. The Srr synthesized bimetallic catalyst formed the lowest concentration of trans isomers (16.8%) of the four bimetallic catalysts tested. This was a significant drop in the trans level compared to the 1.57%Pt/SiO2 monometallic catalyst.

Fig. 3 compares the % stearate and % trans at an IV of 70, for all the prepared catalysts. For the same level of conversion (IV of 70) the 1.56%Pt–0.95%Ni/SiO2 (Srr) catalyst generated the least trans-isomers of all catalysts (16.8%). The monometallic 1.57%Pt/SiO2 catalyst presented the highest level of trans but the lowest level of stearate (C18:0 ) of the tested catalysts. Overall, while the 0.83%Ni–1.37%Pt/SiO2 (S-i) and the 1.56%Pt–0.95%Ni/SiO2 (Srr)

S. McArdle et al. / Applied Catalysis A: General 474 (2014) 78–86

35

C18:0

30

C18:1 trans

25 20 15 10 5 (S rr) iO 2 i/S N 5% Pt % 56 1.

0.

83

%

N

i-1

-0 .9

.3

72 -0 . Pt % 57 1.

(S -i) 2 iO /S

%

7%

N

Pt

i/S

iO i/S N % 41

-0 . Pt % 16

2

2

iO

iO /S Pt % 57 1. 1.

(S -i)

(C oi)

0 2

% trans/stearate at IV 70

82

Fig. 3. Percentage trans and stearate formed at an iodine value of 70 for the indicated catalysts (170 ◦ C, 3 bar H2 , 400 rpm, 100 ml sunflower oil).

catalysts generated a lower level of trans, this was offset by the higher level of stearate formed, 22% and 23.2%, respectively, at an IV 70. From these results it is clear that the combining of Ni with a Pt catalyst and the synthesis method of the bimetallic catalysts have an effect on the hydrogenation process. The results show that when Ni is deposited onto a Pt catalyst by co-impregnation or successive impregnation the activity is significantly reduced. The surface redox reaction method only slightly inhibits the activity, if at all. This catalyst also showed the best selectivity forming the lowest levels of % trans at a comparable IV. Although the 0.83%Ni–1.37%Pt/SiO2 (S-i) also formed a low level of % trans, the 1.56%Pt–0.95%Ni/SiO2 (Srr) catalyst is seen as a superior catalyst as it not only reduced the % trans, it did so without affecting the hydrogenation activity. In an attempt to further understand this phenomenon the bimetallic catalysts were characterized using different techniques. 3.2. Characterization of the surface composition of bimetallic catalysts XRD patterns of the prepared catalysts are shown in Fig. 4. Since the crystalline peaks of interest appear in the 2 range of 30–70◦ , only this section of the diffractogram is shown. The mesoporous SiO2 is amorphous and does not present a diffractive peak. The classical co-impregnation and successive impregnation techniques all show diffraction peaks primarily associated with the monometallic

Fig. 4. XRD patterns for indicated catalysts.

Fig. 5. Bright field TEM images of 1.57%Pt/SiO2 .

or monometallic oxide species. This indicates the presence of large segregated particles of Pt and NiO on the catalyst surface. Indeed, TEM analysis also shows that the Pt and Ni particles are large. A diffractive peak appearing at 35.7◦ may indicate the presence of cubic Ni0.25 O4 Pt3 (2 1 0). The appearance of this compound could be indicative of bimetallic oxide particles. The Srr synthesized catalyst and the monometallic Pt/SiO2 catalyst displayed few or no diffraction peaks. The absence of diffraction peaks indicates the presence of very small particles, which corroborates with the TEM analysis. A typical bright field image (BF) of 1.57%Pt/SiO2 is shown in Fig. 5. Good contrast is observed for the Pt particles and individual metal particles can be identified (black spots). The mesoprous material in the background appears in the overfocus condition due to the different relative height of the particles and the base material. Using the DigitalMicrographTM software it was possible to measure the Pt particle sizes on the support (3 nm). The images show that the Pt particles are well dispersed on the support. TEM micrographs were taken of each of the bimetallic catalysts and are shown in Fig. 6. Both bright field images and the corresponding dark field images are shown for all samples. Dark field imaging (DF) was used in an effort to solely identify the Pt and Ni metal particles. This technique allows for the highest contrast between amorphous and crystalline regions, the crystalline regions reflect, thus appearing extremely bright white, whilst the amorphous regions are darker. As the SiO2 support is amorphous and metals are generally crystalline, the DF images shown in Fig. 6 clearly represents the metal particles on the SiO2 . Metal particle sizes were measured from these images. The traditional impregnated catalysts (Fig. 6a–f) all showed large metallic particles present on the support surface. The particles observed, both in the BF and DF imaging, are much larger than those seen on the 1.56%Pt–0.95%Ni/SiO2 (Srr) catalyst (Fig. 6g and h). Indeed, the overall number of particles per unit area is much large in the Srr catalyst. Interestingly, the DF images show a tendency for the impregnation methods to form metallic clusters. These formations were not seen on the Srr catalyst. Sinfelt et al. [38] postulated that at sufficiently high temperatures the two metals will be completely miscible in order to explain the presence of these clusters. When the bimetallic co and successive – impregnated catalysts were recalcined at a temperature of 450 ◦ C for 5 h, this temperature may have been high enough to promote the formation of Pt–Ni clusters. The Srr reaction did not involve a recalcination step therefore it is more likely that the individual phase will exist as separate particles.

S. McArdle et al. / Applied Catalysis A: General 474 (2014) 78–86

83

Fig. 6. TEM images of 1.16%Pt–0.41%Ni/SiO2 (Co-i) (a) (BF), (b) (DF); 1.57%Pt–0.72%Ni/SiO2 (S-i) (c) (BF), (d) (DF); 0.83%Ni–1.37%Pt/SiO2 (S-i) (e) (BF), (f) (DF) and 1.56%Pt–0.95%Ni/SiO2 (Srr) (g) (BF), (h) (DF).

In Fig. 6(g) large dark patches are identifiable for the 1.56%Pt–0.95%Ni/SiO2 (Srr) catalyst, which would indicate some metal sintering during synthesis. This sintering phenomenon was observed by other researchers under the same conditions;

hydrogen bubbling through an aqueous solution, over a Pt/SiO2 catalyst [4,5,39,40]. The authors explained the effect by the instability of the metal particles (weaker metal-support interaction), due to a Me–Hads binding energy that prevails over the metal–SiO2

84

S. McArdle et al. / Applied Catalysis A: General 474 (2014) 78–86

Fig. 7. TEM images of 1.56%Pt–0.95%Ni/SiO2 (Srr); (a) bright field image, (b) close-up magnification of surrounded area, and (c) showing crystal lattice configuration.

interaction. This allows for the migration of the unstable metal particles across the support causing agglomeration. Fig. 7(b) and (c) shows high magnification images of one of the relatively large metal particles typically found on the 1.56%Pt–0.95%Ni/SiO2 (Srr) catalyst. Particles were composed of small isolated crystallites or as aggregates on the support, suggesting that while migration of the particles does occur, the particles maintain their individual small character rather than agglomerating to form single large particles. A possible explanation for this may be that the Pt in a reduced state provides an electrostatic stabilization against full particle agglomeration and since there is not a recalcination step the particles are not miscible. Scanning transmission electron microscopy (STEM), implemented on a conventional transmission electron microscope with STEM-attachment was used to analyze the bimetallic catalysts. STEM-EDX microanalysis (138 eV resolution) on a metal particleby-particle basis, showed (on all four bimetallic catalysts) the coincident presence of Ni and Pt in some of the particles indicating the presence of bimetallic particles. Fig. 8 presents the EDX analysis of metal particles on the 1.57%Pt–0.72%Ni/SiO2 (S-i) catalyst. The presence of copper is due to the copper grid used to hold the sample. The composition of the particles ranged from Ni-rich to Pt-rich. This supports what was seen in XRD diffraction patterns, the presence of the Ni0.25 O4 Pt3 (2 1 0) bimetallic peak and the presence of the monometallic species. Pt-only and Ni-only particles were detected

in the Co-impregnated and successive impregnated catalysts, again corresponding with what was seen from XRD. Predominantly, the Ni-only particles were found to be larger (averaging 32 nm) than the Pt-only particles (7–12 nm). The bimetallic particles identified were found in clusters and usually fell within the 15–25 nm range. The 1.56%Pt–0.95%Ni/SiO2 (Srr) catalyst did not show any individual metal particles, all particles examined showed a bimetallic composition, indicating that Pt and Ni were deposited into close contact and little or no monometallic particles were present on the support. Wright et al. [30] reported no significant advantage in adding nickel to palladium catalysts for the hydrogenation of canola oil. They observed little differences in trans selectivity over various catalysts. The only benefit detected was an increase in catalytic activity, which occurred due to poison adsorption by the Ni. On the other hand, the results presented in Fig. 3 indicate very obvious difference in selectivity toward C18:1 trans. A noticeable drop in activity was also evident following the addition of Ni to the Pt catalysts by co and successive impregnation. The activity was much less affected when Ni was deposited by surface redox reaction. In general, the behavior of the bimetallic catalysts resembled that of the monometallic Pt/SiO2 catalyst, rather than the less active Ni catalyst. Results have shown that the surfaces of these bimetallic catalysts were composed of both Pt and Ni atoms as bimetallic entities and as individual monometallic particles. While it is known

S. McArdle et al. / Applied Catalysis A: General 474 (2014) 78–86

85

Fig. 8. (a) TEM (bright field) of 1.57%Pt–0.72%Ni/SiO2 (S-i); EDX analyze on metal particles on the this catalyst shows the presence of a (b) bimetallic particle and (c) a Ni-only particle.

that the surface composition of alloys differ from the bulk, surface enrichment in the presence of adsorbed gas molecules has been documented for bulk alloys and bimetallic systems. It is generally accepted that the surface of the Pt–Ni alloys become enriched in Pt when the alloy is submitted to thermal treatments in H2 , hence the resemblance of the bimetallic systems to the Pt monometallic catalyst [16,20,41–44]. The nickel is found as a subsurface layer which alters the electronic and chemical properties of the bimetallic catalyst. This results in a marked decrease in the adsorption strength of H2 and hydrocarbons on the catalyst surface for a number of different reactions, including cyclohexanene, ethylene and benzene hydrogenation [13,16,18,20,33–38]. This consequence may be considered in terms of two effects, one being an electronic interaction and the other a geometric effect between the Ni and Pt. The electronic effect would change the strength of bonding of H2 while the geometric effect would affect the number and size of active sites available for adsorption. In the present study the change in activity on the bimetallic surfaces probably arises due to the geometric effect of the less active Ni blocking the Pt active sites, thus preventing the adsorption of C18:2 molecules and their subsequent hydrogenation. Dilution of the Pt metal by the less active Ni is another geometric effect observed. Bertolini [45] was the first to explain the influence of strain in the electronic properties of the surface atoms as the most probable reason for the modification of the chemical reactivity of Pt–Ni alloys. Boitiaux et al. [44] studied the effects of additives on adsorption coefficients. Co-adsorption, on the catalyst surface, of electron-attracting compounds is inclined to increase the adsorption strength of a hydrocarbon. As already noted the addition of Ni to Pt reduces the adsorption strength of H2 and hydrocarbons, suggesting an electron donor effect. Nohair et al. [4,5] reported that this electron donor effect would increase the electron density of the metal particles, and thus weaken the

adsorption of the C18:1 and the C18:2 molecules on the catalysts surface. The mechanism often used to describe the formation of trans fats is the half-hydrogenation mechanism. The cis double bonds of the C18:2 (or the C18:1 ) molecules are adsorbed onto the metal surface. It is recognized that the C18:2 molecule bonds are more strongly adsorbed and are, in effect, hydrogenated first. For this reason there is a clear fall in the level of the linoleate with hydrogenation time (see Fig. 2). Following adsorption, one of the double bonds are half hydrogenated by the addition of one hydrogen from the metal surface. Complete hydrogenation to form a saturated C C bond requires the addition of a second H atom. However, very often in the absence of this second hydrogen, the half hydrogenated species will loose the hydrogen and reform the double bond, forming either a cis or trans double bond [46]. The trans double bond it the most thermodynamically favored of the two isomers. Since hydrogenation is accompanied by an isomerization process, it can be proposed that the electron donor characteristic of the bimetallic catalyst would also affect this reaction. If the chemisorbed intermediate is removed quickly enough it may not have time to isomerize to trans or in the case of the linoleate, when hydrogenation of one of the double bonds is complete the fat molecule is released before hydrogenation of the second double bond can occur. This modification of the electronic properties and the presence of the geometric effect both may be used to explain the limited formation of C18:1 trans compared to that of the monometallic Pt catalyst. The beneficial effect of introducing the bimetallic Pt Ni system for the hydrogenation of sunflower oil is observed for all of the catalysts tested in this work when considering the reduced formation of trans isomers. However, the greatest benefit has been observed with the Srr prepared catalyst. This would indicate that this system of preparation is the best in forming a catalyst with good contact between the different metals in the catalytic system.

86

S. McArdle et al. / Applied Catalysis A: General 474 (2014) 78–86

It has been reported that classical impregnation techniques like, co-impregnation and successive impregnation often prove to be unsatisfactory for bimetallic catalysts preparation. The main problem associated with these techniques is the failure to create an intimate contact between the two metals to obtain bimetallic entities. Lafaye et al. reported that the high temperature decomposition of the precursor salts in oxidative or reductive atmospheres does not lead necessarily to the formation of bimetallic species [39]. This was also observed in the work presented here where XRD and STEM showed evidence of the monometallic species. Surface redox reaction proved to be a means of optimizing the interaction between the two metals. 4. Conclusions From the results it is clear that the addition of Ni to the Pt/SiO2 catalyst has a beneficial effect on the selectivity of the hydrogenation reaction. The 1.57%Pt/SiO2 catalyst forms 26.5% trans fats at an IV of 70. Using the classic method of synthesizing bimetallic catalysts, co-impregnation, it can be seen that the addition of Ni lowers the trans fat content to 23% at the same IV. A further reduction was observed when the catalysts were prepared by surface redox reaction (Srr), whereby the trans selectivity fell to 16.8%. This change in selectivity seems to be related to the weakened adsorption of the C18:1 molecules on the catalysts surface. The method in which the bimetallic catalysts are prepared is also important for the overall activity. The classical preparation methods of co-impregnation and successive impregnation seem to have a negative effect on the activity of the reaction. With these impregnation methods the relatively inactive Ni may block the highly active Pt particles, thus resulting in a drop in activity. Using the Srr method prevents this ‘blocking’ effect. References [1] M.W. Balakos, E.E. Hernandez, Catal. Today 35 (1997) 415–425. [2] H.P. Choo, K.Y. Liew, H.F. Liu, C.E. Seng, J. Mol. Catal. A: Chem. 165 (2001) 127–134. [3] H.P. Choo, K.Y. Liew, H. Liu, C.E. Seng, W. Mahmood, A. Kamil, M. Bettahar, J. Mol. Catal. A: Chem. 191 (2003) 113–121. [4] B. Nohair, C. Especel, P. Marecot, C. Montassier, L.C. Hoang, J. Barbier, C. R. Chim. 7 (2004) 113–118. [5] B. Nohair, C. Especel, G. Lafaye, P. Marecot, L.C. Hoang, J. Barbier, J. Mol. Catal. A: Chem. 229 (2005) 117–126. [6] C. Okkerse, A. de Jonge, J.W.E. Coenen, A. Rozendaal, J. Am. Oil Chem. Soc. 44 (1967) 152–158. [7] M.B. Fernandez, J.F. Sanchez, F. Jhon, G.M. Tonetto, D.E. Damiani, Chem. Eng. J. 155 (2009) 941–949.

[8] A. Bernas, N. Kumar, P. Maki-Arvela, N.V. Kul’kova, B. Holmbom, T. Salmi, D.Y. Murzin, Appl. Catal. A: Gen. 245 (2003) 257–275. [9] M. Plourde, K. Belkacemi, J. Arul, Ind. Eng. Chem. Res. 43 (2004) 2382–2390. [10] S. McArdle, S. Girish, J.J. Leahy, T. Curtin, J. Mol. Catal. A: Chem. 351 (2011) 179–187. [11] S. McArdle, T. Curtin, J.J. Leahy, Appl. Catal. A: Gen. 382 (2010) 332–338. [12] M. Arai, Y. Takada, Y. Nishiyama, J. Phys. Chem. B 102 (1998) 1968–1973. [13] M. Arai, T. Ebina, M. Shirai, Appl. Surf. Sci. 148 (1999) 155–163. [14] L.J. Rubin, S.S. Koseoglu, L.L. Diosady, W.F. Graydon, J. Am. Oil Chem. Soc. 63 (1986) 1551–1557. [15] J. Massardier, J.C. Bertolini, J. Catal. 90 (1984) 358–361. [16] M.P. Humbert, J.G. Chen, J. Catal. 257 (2008) 297–306. [17] W. Yu, M.D. Porosoff, J.G. Chen, Chem. Rev. 112 (2012) 5780–5817. [18] Y. Shu, L.E. Murillo, J.P. Bosco, W. Huang, A.I. Frenkel, J.G. Chen, Appl. Catal. A: Gen. 339 (2008) 179. [19] J.R. Kitchin, J.K. Norskov, M.A. Barteau, J.G. Chen, Phys. Rev. Lett. 93 (2004) 156801. [20] N.A. Khan, M.B. Zeller, L.E. Murillo, J.C. Chen, Catal. Lett. 95 (2004) 1–6. [21] J. Arenas-Alatorre, A. Gómez-Cortés, M. Avalos-Borja, G. Diaz, J. Phys. Chem. B 109 (2005) 2371–2376. [22] J.H. Sinfelt, J.L. Carter, D.J.C. Yates, J. Catal. 24 (1972) 283–296. [23] J. Arenas-Alatorre, M. Avalos-Borja, G. Diaz, Appl. Surf. Sci. 189 (2002) 7–17. [24] V. Ponec, G.C. Bond, Catalysis by Metals and Alloys, Elsevier, Amsterdam, 1995. [25] N.H.H. Abu Baker, M.M. Bettahar, M. Abu Baker, S. Monteverdi, J. Ismail, J. Mol. Catal. A: Chem. 333 (2010) 11–19. [26] K. Yoshida, N. Begum, S. Ito, K. Tomishige, Appl. Catal. A: Gen. 358 (2009) 186–192. [27] S. Loiha, W. Klysubn, P. Khemthong, S. Prayoonpokarach, J. Wittayakun, J. Taiwan Inst. Chem. Eng. 42 (2011) 527–532. [28] M.B. Fernández, C.M. Piqueras, G.M. Tonetto, G. Crapiste, D.E. Damiani, J. Mol. Catal. A: Chem. 233 (2005) 133–139. [29] A.J. Wright, A.L. Mihele, L.L. Diosady, Food Res. Int. 36 (2003) 797–804. [30] A.J. Wright, A. Wong, L.L. Diosady, Food Res. Int. 36 (2003) 1069–1072. [31] A. Goradia, J. Cooney, B.K. Hodnett, E. Magner, J. Mol. Catal. B: Enzymatic 32 (2005) 231–239. [32] ISO, Animal and vegetable fats and oils – determination of iodine value, in: British Standard 3961:1999, 1999. [33] A. Dolev, G.E. Shter, G.S. Grader, J. Catal. 214 (2003) 146–152. [34] A. Douidah, P. Marécot, J. Barbier, Appl. Catal. A: Gen. 225 (2002) 11–19. [35] R.V. Malyala, C.V. Rode, M. Arai, S.G. Hegde, R.V. Chaudhari, Appl. Catal. A: Gen. 193 (2000) 71–86. [36] J. Okal, H. Kubicka, L. Képifiski, L. Krajczyk, Appl. Catal. A: Gen. 162 (1997) 161–169. [37] T. Ueckert, R. Lamber, N.I. Jaeger, U. Schubert, Appl. Catal. A: Gen. 155 (1997) 75–85. [38] J.H. Sinfelt, Acc. Chem. Res. 10 (1977) 15–20. [39] G. Lafaye, C. Micheaud-Especel, C. Montassier, P. Marecot, Appl. Catal. A: Gen. 230 (2002) 19–30. [40] A. Douidah, P. Marécot, S. Labruquère, J. Barbier, Appl. Catal. A: Gen. 210 (2001) 111–120. [41] J. Sedlácek, L. Hilaire, P. Légaré, G. Maire, Surf. Sci. 115 (1982) 541–552. [42] C.A. Menning, J.G. Chen, J. Chem. Phys. 130 (2010) 239–250. [43] G.F. Cabeza, N.J. Castellani, P. Legare, Comput. Mater. Sci. 17 (2000) 255–259. [44] J.P. Boitiaux, J. Cosyns, E. Robert, Appl. Catal. 49 (1989) 235–246. [45] J.-C. Bertolini, Appl. Catal. A: Gen. 191 (2000) 15–21. [46] I.M. Campbell, Catalysis at Surfaces, Chapman and Hall, New York, 1988.