Kinetic Study of the Partial Hydrogenation of 1-Heptyne over Ni and Pd Supported on Alumina

© 2012 Maccarrone et al., licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Kinetic Study of the Partial Hydrogenation of 1-Heptyne over Ni and Pd Supported on Alumina


Introduction
Selective hydrogenation reactions are industrially used for the partial hydrogenation of unsaturated organic compounds in order to form more stable products or intermediate materials for different processes. The production of final organic products of high added value or intermediate compounds for the synthesis of fine chemicals is both of industrial and academic importance [1]. Alkenes are much appreciated products used in the food industry (flavours), the pharmaceutical industry (sedatives, anesthetises, vitamins) and in the perfumes industry (fragrances). They are also used for the production of biologically active compounds [2], resins, polymers and lubricants, etc.
The partial hydrogenation of acetylenic compounds using homogeneous or heterogeneous metallic catalysts provides a very viable and economically feasible way for the obtaining of these olefinic compounds. Selective catalysts and optimum operational conditions are necessary in order to avoid the complete hydrogenation of the unsaturated bond. Certain transition metals anchored on different solids have demonstrated to be very active and selective catalysts for this type of reaction. They also have the advantage that they can be operated under milder reaction conditions. It is well documented that palladium is a highly active catalyst for hydrogenation [3]. In this sense, Lindlar catalyst (Pd/CaCO3, 5 wt % of Pd modified with Pb(OAc)2) has been used since 1952 as an excellent commercial catalyst for this type of reactions [4]. The argued reasons for the differences in reactivity of Pd indicate that when the metal is electron deficient it becomes less active because alkynes are more weakly adsorbed [5].
During decades a lot of research has been carried out modifying this type of catalysts in order to increase the activity and selectivity: different supports as alumina, coal, silica [5][6][7] have been tried while modified palladium [8], or nanoparticles of Pd have also been

Temperature Programmed Reduction (TPR)
The tests were performed in an Ohkura 2002 S apparatus equipped with a thermal conductivity detector. A cold water trap was placed before the thermal detector to condense water. Before the TPR tests the samples were dried in situ at 673 K for 30 min in an oxygen flow (AGA purity 99.99%). After that the samples were cooled down up to 298 K in an argon flow. Then the temperature was increased up to 1223 K at 10 K min -1 in a H2/Ar (5% v/v) gas flow.

Hydrogen chemisorption
Hydrogen chemisorption was performed by means of the dynamic pulse method using a Micromeritics Auto Chem II apparatus equipped with a thermal conductivity detector. 0.2 g of the samples were reduced 1 h in situ at the above mentioned temperatures using a H2/Ar (5% v/v) flow. The samples were degassed in situ for 2 h under an argon flow (AGA, 99.99%) at a temperature equal to that of the corresponding reduction step and then cooled down to room temperature. In the case of the palladium catalysts they were cooled down to 373 K to make the formation of palladium hydride negligible [16]. After that the chemisorption of hydrogen was performed until total saturation of the samples.

Catalytic tests
The partial hydrogenation of 1-heptyne was carried out in a stainless steel stirred batch reactor equipped with a magnetically coupled stirrer with two blades in counter-rotation that was operated at 800 rpm. The inner wall of the reactor was completely coated with PTFE lining in order to prevent the catalytic action of steel reported by other authors [17].
The evolutions of reactant and products concentrations with reaction time were followed by Gas Chromatography using a flame ionization detector (FID) and a capillary column 30 m J&W INNOWax 19091N-213.

Catalysts characterization
The metal loadings of the catalysts determined by ICP were 3.5 and 0.35 wt % of Ni and Pd, respectively.
For the monometallic Ni/Al2O3 catalyst no hydrogen consumption was detected during the chemisorption analysis, in total accordance with previously reported results [18]. For the Pd/Al2O3 catalyst a chemisorption value of 18 μmol H2 gcat -1 was measured.
The XRD difractograms of the catalysts only present the -alumina peaks at 2θ = 37.7, 46.0 and 67.0º [24][25][26]. For this reason the difractograms are not shown. Because of the low amount of Pd or Ni on the Pd/Al2O3 or Ni/Al2O3 catalysts, the crystalline phases of palladium or nickel were undetectable. Several authors stated that high charges of nickel (>>15 wt %) are necessary to observe the peaks at 2θ = 43.3°; 63.0°; 75.5° and 79.5° of NiO [27,28]. Figure 3 shows the TPR traces of the studied catalysts. The TPR trace of the Pd monometallic catalyst had a main hydrogen consumption peak at 287 K, which can be attributed to the reduction of PdO species and to the formation of palladium hydrides [29]. This catalyst also has an inverted reduction peak at 339 K that could be related to the decomposition of the β-PdH phase [20,21,24]. As these species interact weakly with the support the palladium hydrides are completely eliminated during reduction. Figure 3 also shows the profile of reduction of Ni/Al2O3 catalyst, the principal reduction peak begins at 700 K, finishes over 900 K and presents a maximum at 802 K which corresponds to the reduction of NiO (Ni 2+ species) interacting with the alumina [30]. The second peak of minor intensity is observed at higher temperatures (>1000 K), and corresponds to the reduction of nickel aluminates [31][32]. At the reduction temperature used during the preparation of the catalysts, the obtained TPR spectra suggest the presence of species Ni 2+ and Pd 0 on Ni/Al2O3 and Pd/Al2O3 catalysts, respectively. These results are in total agreement with the XPS results.  It must be noted that the characterization techniques suggest that, after the pretreatment conditions employed during the preparation of the catalysts, only one type of active site is present on each catalytic system.

Partial hydrogenation of 1-heptyne
Before considering kinetic expressions or comparing catalyst performances it is necessary to check whether the selected reaction system proceeds in kinetic regime. The possibility of external and internal diffusional limitations during the catalytic tests was thus experimentally assessed.

Experimental verification of the absence of external and internal mass transfer limitations
In order to eliminate external diffusional limitations, experiences were carried out using different stirring speeds in the range of 180-1400 rpm. It was found that at stirring rates higher than 500 rpm, 1-heptyne conversion values remained constant, indicating that external gasliquid limitations were absent. A stirring rate of 800 rpm was therefore chosen for all the kinetic tests. On the other hand and in order to ensure that the catalytic results were not influenced by external and intraparticle mass transfer limitations, the catalyst pellets were milled to samples of different particle size: a fraction bigger than 100 mesh (<150 μm), a fraction of 60-100 mesh (250-150 μm) and pellets of 1500 μm (not milled). The obtained values of 1-heptyne conversion were the same for the two milled fractions indicating the absence of internal diffusional limitations. Then particles with sizes smaller than 250 μm were used in all tests.

Catalytic activity results
The catalytic activity results for the partial hydrogenation of 1-heptyne are shown in Figures  4 and 5, where it is represented the variation of 1-heptyne (CA) and 1-heptene (CB) concentration as a function of the reaction time for the Ni/Al2O3 and Pd/Al2O3 catalysts.
It can be clearly seen that Pd is more active than Ni, even when using one tenth of the catalyst mass of the Ni catalyst. Reasons for the differences in reactivity can be found in the literature. Most authors report that when the metal is more electron deficient it becomes less active because alkynes are more weakly adsorbed [5,33]. Nothing however is commented about the role of hydrogen. The published reports do not give a clear explanation of the effect of each metal and for this reason kinetic modelling was used in this work to shed light on these issues.

Reaction network
A series-parallel reaction network was proposed for partial hydrogenation of 1-heptyne [14], as indicated in Figure 6.a. This is composed of three hydrogenation reactions that can be a priori considered reversible. The equilibrium constant for each of the previous reactions were calculated using Joback's group contribution method [34]. The values at 323 K were calculated as K1=3.35 10 21 , K2=1.87 10 35 and K3=5.58 10 13 . These values indicate that the individual reactions in Figure 6.a can be considered as irreversible. The experimentally obtained values of total conversion of 1-heptyne confirmed this prediction. Experimentally, it is observed that while 1-heptyne is present in the reaction medium, 1heptene concentration always increases, showing higher concentration than n-heptane.
After all 1-heptyne was consumed, 1-heptene concentration begins to decrease very slowly and n-heptane concentration equally increases. These profiles are consistent with two reaction schemes: i) two parallel irreversible reactions (steps 1 and 3), and ii) series-parallel irreversible reactions, with a k1/k2 value higher than 100. The latter considerations, allows us to disregard step 2. Therefore a simplified network of parallel reactions for 1-heptyne hydrogenation can be assumed in Figure 6.b.

Langmuir-Hinshelwood-Hougen-Watson (LHHW) models
Models of heterogeneous reactions were outlined using the Langmuir-Hinshelwood-Hougen-Watson formalism (LHHW models). Taking into account the previously presented characterization results of Ni/Al2O3 and Pd/Al2O3, in all the models only one type of active sites was considered to be present. Six different models with their respective basic hypotheses are presented in Table 1. The elementary steps with H2 dissociative or nondissociative adsorption reaction mechanism are presented in Table 2.

Mass balances
The following mass balances for components in the liquid phase were considered for the reaction scheme of Figure 6.b, for 1-heptyne (A), 1-heptene (B) and n-heptane (C): initial conditions were: t = 0 min, C 0 A = 0.1528 mol L -1 , C 0 B = C 0 C = 0 mol L -1 .

Numerical resolution and statistics
The system of differential equations (15)- (17) was solved numerically using the Runge-Kutta-Merson algorithm. The model parameter estimation was performed by a non-linear regression, using a Levenberg-Marquardt algorithm which minimized the objective function: are the experimental and the predicted concentration values, respectively, "i" is the chemical compound and "j" is the reaction time.
The model adequacy and the discrimination between models were determined using the model selection criterion (MSC), according to the following equation: where i C is the average relative concentration; p is the amount of parameters fitted and n is the number of experimental data. In order to compare different models, the selected one is that leading to the highest MSC value.
The Standard Deviation (S) was calculated with the following equation:

Model discrimination
The first main requisite for appropriateness of a model should be that of physical significance. A priori this means that the model parameters adopt feasible real values. A second requisite is that of adequate statistical confidence, i.e. the parameters should lie in one as small as possible confidence interval.
The practical criteria for the selection of the kinetic models were: 1. The estimated values of the parameters must be positive and different from zero. 2. The upper and lower extremes of the confidence interval (95%) must be positive. 3. The amplitude of the confidence interval must be lower than the value of the estimated parameter.
The final model is selected from the set of models complying the above 1 to 3 conditions, as the model with the lowest SCD, the summation of squares of the deviations. Another condition is that the standard deviation is smaller than the value of the parameter. If differences are not big, then the model selection criterion (MSC) should be used. Appropriate models should have a MSC value greater than 4.

Kinetic models for the reaction
Preliminary tests were performed in order to check the influence of the different variables on the reaction rate. These results will be used later in the model selection stage. The variables screened were the partial pressure of hydrogen, the initial concentration of 1heptyne and the reaction temperature. In order to analyze the influence of each variable a pseudo homogeneous reaction model was proposed in which the reaction rate was assumed to follow a potential law. The initial reaction rate should thus be written as: An Arrhenius dependent was supposed for the specific constant of reaction:  Ni/Al2O3 catalyst, high hydrogen partial pressures are beneficial for the reaction kinetics, probably both adsorption and surface reaction elementary steps could be enhanced; and b) an increase in the partial pressure of hydrogen negatively affects the reaction rate for Pd/Al2O3.
The obtained values of conversion of 1-heptyne as a function of time are plotted in Figures  9.a and 9.b. It can be seen that for both catalysts the catalytic activity is decreased when the initial concentration of 1-heptyne is increased.  The value of the reaction order in 1-heptyne (α) can be calculated along the lines described in the previous section: The graph of ln(r 0 A) vs. ln(C 0 A), Figures 10.a and 10.b, yields value of order of reaction of 1heptyne equal to -0.22 and -1.5 for Ni/Al2O3 and Pd/Al2O3, respectively. The results indicate that an increase in the initial concentration of 1-heptyne is detrimental to the reaction rates.  The experimental values of conversion of 1-heptyne as a function of time at different temperature values are plotted in Figures 11.a and 11.b. As expected the activity of the catalyst is increased while the reaction temperature is raised up. When equation (21) is linearized a value of "apparent" activation energy (EA) can be got, as indicated in Eq. (25):

Influence of the reaction temperature
The initial reaction rates of 1-heptyne were calculated as in the previous sections. The value of the apparent activation energy were obtained from the plots presented in Figure 12 of ln(r 0 A) as a function of 1/T. The calculated values were 24 and 18 KJ mol -1 for Ni/Al2O3 and Pd/Al2O3, respectively. These values have not a real physical meaning and are only apparent.

Model discrimination for Ni/Al2O3 catalyst
Models II, III, V and VI of Table 1 were discarded because they could not explain the negative and positive orders in 1-heptyne and hydrogen obtained experimentally. The parameters estimated for the models I and IV are indicated in Table 3. A statistical analysis was performed to discriminate between the different models, by means of the selection criteria described in Section 4.5. The results of this analysis are detailed in Table 3. It can be concluded that the best fit is achieved with model I-B. In this model the value of P2 is equal to zero, indicating that n-heptane is not adsorbed. The model IV-D also shows a good fit of the experimental data, but a pseudo homogeneous kinetic expression is obtained with different reaction orders than those previously calculated.  Table 3. Estimated parameters and model discrimination for Ni/Al2O3. Figure 13 contains experimental values of the concentration of 1-heptyne, 1-heptene and nheptane along with theoretical values (solid line) estimated with model I-B, as a function of time. A good fit between the two sets of values can be seen. The same regression with model I-B was done with experimental data obtained at other reaction temperatures in the 293-323 K range. In all cases and as a consequence of the fit, parameters different from zero were obtained for a confidence interval of 95% and with values of the MSC parameter greater than 4.0. The thermodynamic consistency of the P1 and P3 parameters was graphically evaluated by plotting lnP1 and lnP3 as a function of 1/T. In both cases a straight line was obtained ( Figure 14) indicating that the constants have an Arrhenius dependence on temperature.
In equations (26) and (27) Considering that B H  ≈ 0, in accord with the experimental results, a value of the enthalpy of adsorption for 1-heptyne can be obtained from equation (28): Introducing this value in equation (29) a value of EH2 of 22.2 KJ mol -1 can be estimated.
From the results it could be concluded that: 1. Model I-B that supposes dissociative adsorption of hydrogen as the rate-controlling step of reaction and a single type of active sites with total coverage, is the one that best fits experimental data with statistical and thermodynamic consistency. 2. The model does not allow to directly obtain the enthalpies of adsorption of 1-heptyne and 1-heptene and the activation energy for the adsorption of hydrogen. 3. From equation (28) it can be inferred that the enthalpy of adsorption of 1-heptyne is greater than that of 1-heptene, in agreement with the information available in the literature on the partial hydrogenation of alkynes [34]. 4. This model also supposes that the alkane is not adsorbed. If we additionally suppose that the enthalpy of adsorption of 1-heptene is negligible, in accordance with experimental results, a value of the enthalpy of adsorption of 1-heptyne and activation energy for the H2 adsorption can be obtained from eqs. (28) and (29).
Then, it can be concluded that the dissociative adsorption of H2 is the rate limiting step for the Ni/Al2O3 catalyst, and that the active sites are preferentially occupied by 1-heptyne.

Model discrimination for the Pd/Al2O3 catalyst
Models I, II, IV and V of Table 1 were discarded because they could not explain the negative orders in 1-heptyne and hydrogen obtained experimentally. The parameters estimated for the models III and VI are indicated in Table 4. A statistical analysis was performed to discriminate between the different models, by means of the selection criteria described in Section 4.5. The results of this analysis are detailed in Table 4; these results indicate that model III-C gives the best fit of the experimental data. In this model the value of the parameters P4 and P5 are equal to zero. Therefore the only species being adsorbed on the Pd/Al2O3 catalyst are 1-heptyne and hydrogen.   The heterogeneous model VI-E also produces a good fit of the experimental data but its parameters KB, KC, P13 and P15 are equal to zero, thus transforming into a pseudo homogeneous reaction rate expression in which the reaction orders in 1-heptyne and hydrogen are positive. This is not in agreement with the observed results, so this model was discarded.  The same fit with model III-C was repeated with experimental data at other reaction temperatures in the 293-323 K range. In all cases and as a consequence of the fit, parameters were obtained with values different from zero in a confidence interval of 95% and with values of the MSC parameter greater than 4.0. The thermodynamic consistence of parameters P7, P10 and P11 was evaluated by plotting lnP7, lnP10 and lnP11 as a function of 1/T. In all cases straight lines were obtained confirming the hypotheses of Arrhenius dependence with respect to temperature ( Figure 16).
The slopes of the straight lines obtained correspond to the enthalpies of adsorption of 1heptyne and H2 and the energies of activation of the surface reactions of hydrogenation. Taking into account the definition of the parameters P7, P10 and P11 ( From the results it can be concluded that: 5. Model III-C that poses the surface chemical reaction as the limiting step is the one that best fits the experimental data. The model also poses the dissociative adsorption of H2 and the competition with 1-heptyne for the active sites. The model presents statistical and thermodynamic consistency. 6. The model enables obtaining directly the values of the enthalpies of adsorption of 1heptyne and the activation energy for the hydrogenation of 1-heptyne to 1-heptene (E1). Neither the calculation of the activation energy for the hydrogenation of 1-heptyne to nheptane (E3). 7. The model indicates that 1-heptyne and H2 are the only species adsorbed on the active sites. The enthalpy of adsorption of 1-heptyne over Pd (-19.64 KJ mol -1 ) is not much different from the value reported by Semagina et al [36] for the hydrogenation of 1hexyne over Pd nanoparticles. 8. Equation (34) shows that the enthalpy of adsorption of hydrogen is higher than that of 1-heptyne over the Pd/Al2O3 catalyst. This suggests that there are not thermodynamic limitations for the adsorption of H2. This was confirmed by the tests of hydrogen chemisorptions as Pd is able to chemisorb an important amount of H2 at room temperature, suggesting that there is not a kinetic impediment as that observed for Ni. Consequently, the dissociative adsorption of hydrogen is fast and then the controlling step is the surface chemical reaction. 9. The value obtained for the activation energy of the hydrogenation reaction of 1-heptyne to 1-heptene (E1) turned out to be quite low (18.58 KJ mol -1 ). This coincides with the fact that the reaction can proceed at low temperatures. 10. The kinetic modelling of the reactions over the Ni and Pd catalysts gives an explanation of the different reactivity of the catalysts. P d i s m o r e a c t i v e t h a n N i f o r p a r t i a l hydrogenation reactions because there is a kinetic limitation for the adsorption of hydrogen on Ni. Hydrogen is more strongly chemisorbed on Pd, so there is a great coverage of the surface by H2, therefore making the surface reaction step as the ratecontrolling one.

Conclusions
Pd/Al2O3 was more active and selective than Ni/Al2O3 for the partial hydrogenation of 1heptyne to 1-heptene.
In order to analyze the influence of the different variables (hydrogen partial pressure, initial concentration of 1-heptyne and reaction temperature) on the reaction rate a pseudo homogeneous model was proposed based on power law kinetics. Reaction orders for hydrogen and 1-heptyne were obtained as well as the apparent activation energy. For the Ni/Al2O3 catalyst, reaction orders of 1.3 in hydrogen and -0.22 in 1-heptyne, and apparent activation energy of 24 KJ mol -1 were obtained. For the Pd/Al2O3 catalyst, reaction orders of -2.6 and -1.5 in hydrogen and 1-heptyne, respectively, and apparent activation energy of 18 KJ mol -1 were obtained.
In order to elucidate the role of Ni and Pd on the reaction rate, the kinetic data were fitted with six different heterogeneous LHHW models. The results obtained indicate that for the Ni/Al2O3 catalyst the controlling step is the dissociative adsorption of hydrogen over the metal active sites and the reaction rate can be expressed by: If it is assumed that the adsorption enthalpy of 1-heptene can be considered negligible, a value of -17.91 KJ mol -1 is obtained for the adsorption enthalpy of 1-heptyne. In the same way, the value of activation energy for the hydrogen adsorption is 22.2 KJ mol -1 .
For the Pd/Al2O3 catalyst the controlling steps are the surface hydrogenation reactions (1-heptyne to 1-heptene and 1-heptyne to n-heptane). The corresponding reaction rates are:   Besides, the value obtained for the activation energy for the hydrogenation reaction of 1heptyne to 1-heptene (E1) was 18.58 KJ mol -1 . This coincides with the fact that the reaction can proceed at low temperatures.
The different activity levels of the Pd/Al2O3 and Ni/Al2O3 catalysts are due to a kinetic limitation for the adsorption of hydrogen on Ni. In the case of Pd this limitation does not exist.