Article pubs.acs.org/JPCA
Experimental and ab Initio Investigations of H2S‑Assisted Propane Oxidative Dehydrogenation Reactions Zahra A. Premji,†,‡ John M. H. Lo,*,† and Peter D. Clark†,‡ †
Alberta Sulphur Research Ltd., University Research Center, University of Calgary, Unit 6-3535 Research Road NW, Calgary, Alberta, Canada T2L 2K8 ‡ Department of Chemistry, University of Calgary, 2500 University Drive NW, Calgary, Alberta, Canada T2N 1N4 ABSTRACT: The oxidative dehydrogenation (ODH) reaction of propane was investigated at temperatures between 923 and 1023 K using either O2 or O2/H2S mixture as oxidant. GC analysis of the product mixtures showed that ethylene was the major olefin product in the conventional ODH reaction whereas propylene became dominant when H2S was included in the feed gas. With an oxygen-rich feed (4:2:2 C3H8:O2:H2S), ∼70% propane conversion, and ∼50% propylene selectivity could be achieved at 1023 K, a level of performance comparable to that for the ODH reaction employing reducible solid oxide catalysts. Theoretical calculations utilizing CBS-QB3 method were also conducted to explore the causes of the enhanced propylene yield and selectivity of the H2S-assisted ODH reaction. It was found that the increased propane conversion was due to a large enthalpy gain from the in situ formation of S2 that compensated for the high energy cost of hydrogen abstraction by SH and S2H. Also, the promoted propylene selectivity was attributed to the instability of the sulfur-containing products, which made the reaction route to propylene the most thermodynamically favored.
1. INTRODUCTION Light unsaturated hydrocarbons, such as ethylene and propylene, are important feedstocks used in the manufacturing of chemicals and polymers. More specifically, propylene is used in the large-scale manufacture of polypropylene, acrylonitrile, and propylene oxide. Much of the world’s propylene is produced via cracking of hydrocarbons, specifically by fluid catalytic cracking (FCC) of gas oils in refineries and steam cracking of small alkanes1 in which propylene is a coproduct. The growth in supply of propylene from cracking processes is, however, unable to keep up with its continually growing demand; hence, producers are always interested in developing on-demand technologies that produce propylene specifically. Olefins can also be produced from dehydrogenation reactions of alkane feedstocks; however, simple dehydrogenation reactions suffer from numerous limitations that restrict their extensive applications. The limitations include possible side reactions, thermodynamic constraints on selectivity and conversion, the endothermic nature of the process, which necessitates a high energy input, and coke formation resulting in catalytic deactivation.2 One modification that may overcome many of these problems is oxidative dehydrogenation (ODH), in which an oxidant is added to the reaction feed of a thermal dehydrogenation unit, thereby initiating the reaction via the oxidant rather than high temperature alone.2,3 Oxygen is a commonly used oxidant because of its low cost, low environmental impact, and the formation of H2O, which is highly exothermic, providing sufficient energy to compensate for the energy cost of dehydrogenation of alkane. © 2014 American Chemical Society
Despite the desirable exothermic nature and the limited coke formation, ODH based on oxygen is confronted with the challenge of over oxidation of olefins to CO and CO2, as well as the possibility of reaction runaway.4 To suppress the complete oxidation of alkane to COx to ensure a reasonable selectivity while maintaining good conversion of the ODH reactions, application of catalysts has been investigated, and many catalytic systems were found to be active for the process.5 Supported vanadium-based oxides and mixed-metal oxides are among the most extensively studied because of their high thermal stability, large surface areas, and the favored reducibility and basicity that promote the ODH activity and selectivity.6−10 Nevertheless, to date, none of these systems has shown product yields comparable to those obtained by the current industrial technologies of olefin production. Another approach of reducing the overoxdiation of alkane and olefins in the ODH reactions is the use of an additive or a milder oxidant. Species such as halogens,11 sulfur-containing compounds,12−14 N2O,15 and CO216 have been tested respectively. CO2 with ceria-based catalyst increased the ethylene selectivity in the ethane ODH reaction to about 60% but a much higher reaction temperature was required because of the low activity of CO2.16 On the other hand, a N2O/O2 oxidant mixture could enhance the propylene selectivity by 25% for the vanadia-catalyzed propane ODH Received: October 31, 2013 Revised: February 11, 2014 Published: February 13, 2014 1541
dx.doi.org/10.1021/jp410750c | J. Phys. Chem. A 2014, 118, 1541−1556
The Journal of Physical Chemistry A
Article
Figure 1. Experimental setup employed in this work and the reactor dimensions (shown in inlet).
summarized. With an aid of these results, the experimental product distributions for the C3H8/O2 and C3H8/O2/H2S reactions will be analyzed in detail.
reactions, but this increased selectivity could be obtained only at low propane conversion (1−10%).15 Some studies have also been performed on the effect of H2S or other sulfur species on the ODH reactions of light alkanes.13,14 Initial works by Tischler and Wing17 and Porchey and Royer18 have demonstrated the merits of adding H2S to an uncatalyzed system for the dehydrogenation of propane. The promoting properties of H2S were further explored by Clark et al.,19 and they observed that the conversion and propylene selectivity of the propane ODH reaction over V2O5/Al2O3 were enhanced to 53.7 and 56.5%, respectively. Subsequent investigation employing elemental sulfur as the oxidant revealed that the ethylene selectivity of up to 89% with 84% overall conversion could be achieved in the gas-phase ODH reaction of ethane at 1123 K.20 It was postulated in these studies that the enhanced olefin selectivity is attributed to the S2 intermediate, formed from reacting H2S with O2, that participates in the hydrogen abstraction of alkane, producing the olefin product and H2S, which is recycled in the reaction.19,20 Therefore, the present work aims at acquiring a better understanding of the effect of H2S in the alkane ODH reaction. In particular, emphasis will be placed on how H2S dictates the product selectivity, that is, the ratio of olefin to other products, in the gas-phase propane ODH reaction. The following discussion can be divided into two sections. The first part will describe the details of the ODH experiments involving C3H8/O2 and C3H8/O2/H2S feed mixtures, respectively, performed in this work. Data analysis of conversion and product selectivity by means of gas chromatography will be presented. The influence of feed gas composition, reaction temperatures, and contact time on the experimental outcomes will also be discussed. In the second part of the discussion, the data obtained from the high-level ab initio calculations for the elementary steps in the propyl + O2/S2 reactions based on the mechanisms as described by Klippenstein et al.21 will be
2. METHODS 2.1. Experimental Details. 2.1.1. Safety Notes. As this work requires the use of H2S, the entire experimental setup including H2S gas cylinders was contained in ventilated cabinets equipped with electronic gas detectors and automatic shut down systems designed to operate at low level alert and shut down mode (2−10 ppmv H2S), respectively. A hazards opeartions review of the experimental procedures was also conducted before experiments. Further details of safety operations will be described in the following sections. 2.1.2. Overall Setup and Operation Conditions. A tubular quartz tube reactor with plug flow characteristics, housed in a heated furnace, was used to carry out the experiments. Gases (nitrogen (ultra high purity), O2, H2S and propane) were purchased from Praxair and metered to the reactor using mass flow controllers and stainless steel tubing (Figure 1). Check valves and pressure relief systems were placed at various points along with a pressure gauge to monitor any back-pressure caused by solid sulfur accumulation downstream of the reactor. A relief valve, set to 10 psi, was attached to the gauge to allow venting of gases to an aqueous KOH scrubber system. The tube reactor was made of quartz, which was shown to have a minimal catalytic effect on the reactions under study. It was designed with a wider front and back end and a narrower middle portion. The outer diameter (OD) and length of the front end was 12.7 mm and 27.5 cm, respectively. The back end of the reactor had the same OD but had a length of 20 cm. The OD and length of the middle portion was 8.1 mm and 22.5 cm, respectively. The front end of the reactor was designed to be longer to allow the catalyst bed to be positioned a few centimeters into the hot zone of the furnace thus avoiding the 1542
dx.doi.org/10.1021/jp410750c | J. Phys. Chem. A 2014, 118, 1541−1556
The Journal of Physical Chemistry A
Article
interval. The results of the three samples were then averaged to obtain the final numbers reported in this work. 2.1.3. Error Estimation and Data Processing. To test for analytical accuracy, samples were analyzed multiple times using the same experimental and analysis system that was subsequently used for the reaction studies. The experimental system was setup to flow a mixture of gases of known composition through an empty reactor. The system was kept at room temperature to prevent any reaction that changes the gas compositions. The setup was left flowing throughout the sample collection and analysis period to prevent any external factors from affecting the results. Samples were drawn from the sample port using a syringe and injected into the GC, similar to the way it was done in experiments. Three samples, 30 min apart, were taken for different gas mixtures. The average and standard deviation of the GC data were calculated and the standard deviation was further converted to relative standard deviation in percentage. The observed experimental errors were indeed a sum of several errors, including the errors due to user sampling, GC instrument errors as well as fluctuations within the mass flow controllers over the specific time period. Experimental errors ranging from 0.13% for components in the highest amounts (60 vol %) to 3.6% for the components in the lowest amounts (5 vol %) were obtained. Accordingly, it is prudent to include an experimental error of up to 5% of the amount of reactant for this system in data analysis. The residence time of a gas mixture tr (s) was deduced by the following expression:
temperature gradient that can exist at the boundaries of the hot zone in such furnaces. The reactor design was chosen to allow shorter contact times (10−200 ms range) for the flow rates available from the mass flow controllers. The catalyst bed was housed at the front end of the middle portion of the reactor, within the hot zone, and secured between plugs of quartz wool. A thermocouple was inserted into the front end of the catalyst bed from the front end of the reactor to record catalyst bed temperatures. A sampling tube, inserted from the back end of the reactor, allowed samples to be drawn out from the back end of the catalyst bed. The sampling tube used had a thinner diameter, hence reducing the amount of time spent in the hot zone after the catalytic reaction. The reactor was connected to a sulfur trap operated at ambient temperature followed by a dry trap, a pair of KOH scrubbers, and carbon trap to remove moisture, sulfur gases, and trace mercaptans, respectively. The entire reactor system was housed in a ventilated cabinet equipped with H2S, SO2, and CO detectors connected to an alarm system panel. The detectors were connected to automatic shutdown valves that were set to the gas exposure limits as set out by the Alberta Environment for an 8 h working day. The lines connecting the H2S and propane gas tanks to the mass flow controllers were also connected via an air-operated valve (AOV) which, for safety reasons, had to be activated prior to each experiment and which was shut off at the end of the experiment. Crystalline silica (Aldrich, grade 40 silica gel) crushed to 16− 30 mesh size was used to fill the bed for the inert filled-bed investigations. Amounts in the range 0.2−2 g of crystalline silica and total flow rates of 300−1500 mL/min were used depending on the desired contact time. Nitrogen was used to dilute the feed gas mixture and as the internal standard to correct for condensable products. Fresh silica was loaded into the reactor before the start of each experiment. The experimental setup was then connected and sealed, followed by a leak check. The mass flow controllers were then setup to flow the feed gas at room temperature to verify feed gas concentrations, after which the reactant gases were shut off and only nitrogen was left to flow during the warm-up period of the furnace. Once all checks were complete, the furnace and heating tapes were turned on and set for the desired temperature and allowed to equilibrate for 1−2 h. The experiment was then started by initiating reactant gas flow and after 20−30 min of equilibration time, sampling began. Once all samples at 923 K were completed, the furnace temperature was increased to 1023 K and after a sufficient equilibration period, samples were taken for analysis. Gas samples were obtained by syringe, with a P2O5 trap placed in front of the sampling point to dehydrate and remove elemental sulfur from the samples. Analyses were carried out using gas chromatography. A Varian instrument, equipped with 2 columns (a 1.5 mm molecular sieve and an HP Q-PLOT), both of which were connected to thermal conductivity detectors (TCD) was used. A pulsed flame photometric detector (PFPD) in series with the TCD for the second column was employed for detection of sulfur compounds at ppmv levels. An SRI GC equipped with a molecular sieve column for hydrogen analysis and a T-PLOT column for hydrocarbons determination was also used. The gas samples were first analyzed as taken, and then diluted with argon (5× dilution) to obtain better resolution of the propane/propylene peaks. For each set of conditions, the outlet gas was sampled three times, each sample being taken after a 30 min time
tr (s) =
Vreactor (mL) × 298 (K) × 60 (s/min) Ftotal (mL/min) × T (K)
where Vreactor is the volume of an empty reactor, T the reaction temperature, and Ftotal the flow rate of all gas components. The data obtained from the GC were used to calculate the conversion, selectivity, and yield for each component according to the following formulas: conversion = moles of reactant i in feed − moles of reactant i in product × 100% moles of reactant i in feed
selectivity =
moles of reactant converted to product j × 100% total moles of reactant converted
yield = conversion × selectivity
in which i and j refer to reactants and products respectively. The amounts of all gaseous components were obtained from gas chromatography whereas those of liquid and solid components were calculated from the remaining mass balances. Specifically, the quantities of H2O and elemental sulfur were deduced from the mass balance of O2 and sulfur components as they could not be measured directly. In the analysis of sulfur species, only H2S and elemental sulfur were expected to be present in the product mixture because the reducing equivalent (C3H8) in the feed shifts the final H2S:SO2 ratio to very high values (i.e., amount of SO2 below the detection limit) by converting SO2 back to H2S: SO2 + 3H 2 → H 2S + 2H 2O
Finally, residual solid carbon, which contains carbonaceous materials and any other carbon-containing products not 1543
dx.doi.org/10.1021/jp410750c | J. Phys. Chem. A 2014, 118, 1541−1556
The Journal of Physical Chemistry A
Article
Table 1. Comparison of Product Distributions between the ODH of Propane with and without H2S at 20 ms Residence Time Using an Unfilled Reactora conversion (%) feed ratio
a
products (mol/100 mol of feed) [selectivity (%)]
temp (K)
C3H8
O2
H2S
C3H6
C2H4
C2H6
CH4
CO
CO2
C
S2
H2O
H2
4:1
923
19
23
0
80
59
58
0
0.0
11.7
2.9
4:2:1
973
50
85
58
5.6
14.6
3.2
4:1
1023
50
83
0
0.0
13.9
5.7
4:2:1
1023
62
92
60
18.6 [63] 12.4 [29] 22.9 [39] 13.5 [24] 25.0 [32] 14.2 [20]
1.9
37
0.5 [2] 0.0 [0] 0.8 [1] 0.0 [0] 1.5 [2] 0.1 [
