1. Introduction
The expanded use of hydraulic fracturing has resulted in the venting and flaring of large volumes of hydrocarbons, especially natural gas. A lack of local pipeline capacity results in more flaring [
1] and fugitive emissions. Public sentiment is prompting government environmental regulators to force the reduction or outright banning of hydrocarbon flaring and fugitive emissions [
2] to reduce climate change by global warming. Petroleum and natural gas companies are now actively promoting their efforts to reduce their methane footprints [
3]. However, the engineering challenge of CH
4 conversion is considerable. Catalytic methods offer several conversion approaches.
Catalytic activation of CH4 is generally classified as indirect or direct. Indirect activation produces synthesis gas (primarily CO and H2) using an oxygen source by reforming (H2O—steam; CO2—dry) or partial oxidation (O2). Synthesis gas can be catalytically converted to useful products such as alcohol (usually CH3OH) or higher hydrocarbons (by Fischer–Tropsch process).
Direct activation of CH
4 uses no oxygen source. It directly breaks the very strong CH
3–H bond (4.39 × 10
5 J/mol). For example, methane dehydroaromatization (MDA) uses a Mo/HZSM-5 zeolite [
4] catalyst to form C
2H
4 and aromatics at 950–1030 K. Unfortunately, MDA is thermodynamically limited. In addition, catalyst activity drops quickly due to coke deposition.
An intermediate direct approach is the oxidative coupling of methane (OCM) that uses a very small amount of O
2 to activate the CH
4 while limiting the coke formation. The OCM catalysts are transition metal oxides on an oxide support, e.g., La
2O
3/CaO [
5] and La
2O
3/CeO
2 [
6]. Feed CH
4/O
2 molar ratios of 7–11 with temperatures ~840–1220 K have been studied. The products include C
2+H
x and synthesis gas, with the distribution depending on the catalyst and temperature. More feed O
2 favors CH
4 conversion, but lowers C
2+H
x selectivity in favor of CO
x.
Gambo et al. [
7] reviewed recent advances in OCM, including the use of catalyst nanowires, and identified avenues for further research. Using proprietary nanowire catalysts, Siluria Technologies demonstrated OCM in a continuous flow large demonstration plant [
8,
9]. The primary goal is the production of C
2H
4 for subsequent conversion to gasoline and chemicals. Siluria envisions a flexible two-stage OCM reactor in which the first stage is a packed bed reactor (PBR) feeding CH
4, O
2, and possibly C
2H
6. The second stage feeds more C
2H
6 in an endothermic pyrolysis plug flow reactor that uses the first stage exothermicity.
Considering the potential and the constraints of OCM, the reactor type becomes important. Conventional packed beds are not economically viable for OCM [
10]. Other configurations such as membrane reactors and fluidized beds should be considered. A recent study [
11] compared the packed bed reactor (PBR) and continuous stirred tank reactor (CSTR) for OCM based on calculations with a detailed reaction mechanism [
6]. Higher feed temperatures were required to achieve a “light off” of the PBR, while the CSTR required considerably lower feed temperatures to reach nearly comparable conversions. The CSTR—a fluidized bed in likely practice—favored synthesis gas production over C
2+H
x as compared to the PBR.
In a difficult experimental study, Liu et al. [
12] considered the rapid PBR “light off”. Using careful temperature control and real-time product measurement in a micro-reactor during OCM over La
2O
3 nanorod catalysts with a feed CH
4/O
2 = 3, Liu et al. observed that CO
2 is the dominant product at the lower temperatures (<853 K) when relative O
2 concentrations are high. The system transitions through a window (~873 K) to higher temperatures (>913 K) that favor C
2+H
x at the lower O
2 levels. In effect, a competition between CO
x and C
2+H
x formation occurs in this window. This suggests that reactor configuration and temperature will be critical for OCM reactor design. These observations also show that, in OCM, the production of byproduct syngas is unavoidable if C
2+H
x is the goal.
In typical OCM, some of the limited O
2 dissociates on the catalyst surface [
13]. An adsorbed •O
s then abstracts an H• from CH
4 to form the key •CH
3 gas phase radical. Oxygen atoms on the metal oxide lattice surface might directly abstract the H• from CH
4. In this case, the gas phase O
2 replenishes the resulting surface vacancy, leaving behind an adsorbed •O
s [
7]. In either case, the •CH
3 radicals can combine to form C
2H
6. Further reactions form the desired C
2+H
x products, and the undesired CO
x and coke. An alternate CH
3-H bond activator would reduce CO
x while possibly enhancing C
2+H
x formation. Unfortunately, the desired C
2+H
x products are more susceptible to oxidation than the reactant CH
4. This can occur via gas phase O
2 or surface oxygen species [
7,
13].
An alternate or supplementary oxidant that might be less aggressive toward C
2+H
x by reducing surface oxygen species while still activating the CH
4 is the •OH gas phase radical. Gas phase hydroxyl radicals can be formed from the gas phase decomposition of co-fed H
2O
2 vapor: H
2O
2 + M = 2 •OH + M.
Section 3 below summarizes experimental literature on the use of H
2O
2 for OCM that motivates this paper.
This paper is also a sequel to the Rivera et al. [
11] study on OCM. It uses the same elementary reaction mechanism as that developed by Karakaya et al. [
6]. The paper considers vapor phase H
2O
2 as a supplemental or alternative activating oxidizer to reduce syngas production in favor of C
2+H
x. Both CSTR and PBR are considered.
2. Kinetic Mechanism and Computational Tool
The detailed reaction mechanism used in this study was developed by Karakaya et al. [
6]. It is composed of series and parallel elementary reactions [
13] in both the gas and surface phases. The gas phase portion is taken from Chen et al. [
14]. The surface portion is inspired by the work of Alexiadis et al. [
15].
In the current study, the OCM mechanism was employed in reactor simulations by Detchem
® [
16]. This program achieves material and energy balances using the mechanism, based on required reactor input data and parameters. In this study, separate adiabatic PBR (modeled as plug flow) and CSTR (modeled as perfectly mixed) runs were conducted with the Detchem
® PBED and CSTR applications, respectively. See [
11] for a listing of the governing balance equations used in each reactor simulation.
Comparative results were prepared in terms of conversions X
CH4, selectivities S
j of useful products (C
2H
x, CO, H
2) or byproducts (H
2O, CO
2), and yields Y
j:
where F
j = molar flow rates, F
j,in = molar rate at the reactor inlet, and n
j is the number of CH
4 moles needed to make one mole of product (byproduct). For example, for C
2H
4, n
j = 2; for H
2, n
j = 0.5.
In the prior study [
11] of OCM with O
2 (no H
2O
2) as the activator, twenty cases were considered in separate CSTR and PBR calculations. The process parameters considered are summarized in
Table 1. The parametric study included the molar feed CH
4/O
2 ratio (“low” = 7, “high” = 11; LR and HR, respectively), molar feed rate (“low” = 8.984 × 10
−5, “high” = 1.412 × 10
−4 mole/s; called LF and HF, respectively), and feed temperature (843–1243). At each of the five feed temperatures, four cases were considered: LR_LF, LR_HF, HR_LF, and HR_HF. These conditions were inspired by published laboratory data [
6]. Each reactor simulation assumed the same catalytic site density and total catalytic surface area, resulting in the same 2.42 × 10
−6 total moles of sites. A processing rate can be defined as the ratio of total molar feed rate to the total number of catalyst sites. The processing rate range is 37.1–58.3 s
−1. These parameters motivate the present study. Larger reactors can be scaled from these conditions.
3. Alternate Activator H2O2
Garibyan et al. [
17] studied OCM over Pb/aerosil, ZnO, and 10% Na
2O/ZnO catalysts. Pulses of H
2O
2 vapor into the CH
4/O
2 feed increased the C
2H
x yield while stabilizing catalytic activity during OCM at 1 atmosphere and 1023 K. With a 1% Au/5% La
2O
3/CaO catalyst, at 973–1073 K, Eskendirov et al. [
18,
19] observed that H
2O
2 increased CH
4 conversion, while enhancing C
2+ hydrocarbon yields even up to benzene. They speculated that H
2O
2 decomposition resulted in more •OH radicals for gas phase activation of the CH
4. They even observed OCM in the presence of H
2O
2 vapor without catalyst at temperatures as low as 673 K, with considerable selectivity for C
2H
x. These studies motivate this paper.
Considering the importance of making any OCM process as “green” as possible, a potentially sustainable source of H
2O
2 uses a photo-activated TiO
2-Au-Si catalyst while feeding O
2 and liquid H
2O [
20]. Spiegelman and Alvarez [
21] developed a simple yet clever technology to produce a continuous vapor stream of H
2O
2 from a liquid solution of H
2O
2 in water. Subsequent drying of the vapor stream to raise the H
2O
2 concentration runs the risk of energetic decomposition, thus posing a safety risk. In a study of the decomposition of H
2O
2 vapor on various surfaces, Satterfield and Stein [
22] generated H
2O
2 vapor concentrations of up to 0.23 atm in a 1 atm system. Therefore, in the remainder of these calculations, we used a conservative molar H
2O/H
2O
2 = 4 linkage in all cases where H
2O
2 was used.
The OCM process requires considerable preheating, so the H
2O
2 decomposition risk also calls into question the feed temperature for the H
2O/H
2O
2 vapor stream. Consider the decomposition: H
2O
2 + M → 2 •OH + M with a rate constant borrowed from the Chen et al. [
14] reaction set for the CH
4 gas phase chemistry used in the OCM mechanism [
6]. Assume a feed CH
4/H
2O
2 ratio of 11 (the HR case), with the coflowing H
2O vapor, and no O
2. A simple kinetic calculation shows that, at 1243 K, the H
2O
2 will be 100% decomposed in 10 microseconds. At 673 K, the time is a more realistic 5 s. This simple calculation suggests that preheating a combined CH
4, O
2, H
2O
2, and H
2O feed stream would be problematic, especially for a PBR. It also suggests keeping the CH
4/O
2 and H
2O/H
2O
2 vapor streams separate, with the CH
4/O
2 stream taking most or all the preheat. Separate feed streams are more easily handled with the CSTR.
Finally, the replacement of O
2 by H
2O
2 maintains the feed CH
4-to-O molar ratio, though it somewhat increases the overall H content of the feed. Consider the following overall reactions below. Though simplistic, Equations (1) and (2) show that replacing O
2 with H
2O
2 should increase the production of C
2H
4 and H
2O, while reducing CO and H
2. In addition, adiabatic reactor temperatures should be lower.
3.1. Incremental H2O2 Replacing O2 to PBR at Fixed Feed Temperature
Rivera et al. [
11] showed that, for the parametric range considered (
Table 1), the highest CH
4 conversions for the PBR were nearly 40% for the LR cases, with little impact of flow rate, for a 1243 K feed temperature. The LR_LF case showed the highest sum_C
2H
x (sum of C
2H
6, C
2H
4, and C
2H
2) selectivities and yields. First, for the same 1243 K feed temperature, H
2O
2 was incrementally substituted for O
2. It was assumed that H
2O
2 vapor will be available at 20 mole percent with the balance of H
2O vapor. The starting point was the LR_LF (feed CH
4/O
2 = 7, feed processing rate 37.1/s) case [
11] in the PBR. The PBR bed length was that used in the experiments described elsewhere [
6]. This high feed temperature did ignore the H
2O
2 stability issue. The insights gained, however, will help identify the utility of H
2O
2 as a potential CH
4 activator.
Figure 1 shows that the final PBR bed temperature drops with increasing H
2O
2 content. This was attributed to the reduced overall reaction exothermicity suggested by Equations (1) and (2). However, the bed temperature peaked much earlier when H
2O
2 was used in the feed oxidant. A closer examination of the post-entrance region is revealed in
Figure 2 below. The H
2O
2-containing cases showed a near-immediate rise from the 1243 K feed temperature. The very early single peak for the R = 1 (no feed O
2) case roughly corresponded to the exhaustion of the H
2O
2, which was consistent with the extremely rapid H
2O
2 decomposition described earlier at this temperature. The presence of the feed O
2 caused a second local temperature peak further downstream. These peaks (R = 0.33, 0.67 cases) and the much later single peak (R = 0) corresponded approximately to where the O
2 ran out, with no further adsorbed •O
s.
The R = 1 case (no feed O
2) showed almost no adsorbed species, suggesting that all the CH
4 conversion effectively occurred in the gas phase (i.e., non-catalytic). On the contrary, for the R = 0.33 and 0.67 cases, while H
2O
2 was still present, there was a complex parallel/series scheme ongoing with both catalytic and gas-phase reactions occurring. These observations were consistent with the claim that the CH
4 conversion is accelerated by •OH gas phase radicals produced from the H
2O
2 dissociation [
18,
19].
Figure 3 shows that increasing the H
2O
2 content improved the selectivity of sum_C
2H
x, while lowering both the selectivity and yield for syngas (H
2 + CO). There was a negligible impact on sum_C
2H
x yields. The reduction in syngas was due almost entirely to a reduction in CO. Finally, for these four cases from R = 0 to 1, the CH
4 conversions were: 39.3, 36.6, 35.1, and 33.8%, respectively. In all cases, the final O
2 and H
2O
2 conversions were 100%.
An expanded look at the long-post-entrance region provides more insight into the dramatic impact of substituting some of the feed O
2 with H
2O
2 vapor.
Figure 4 shows the selectivities for the CO, C
2H
6, C
2H
4, and CH
4 conversions for the LR_LF case, at 1243 K feed temperature, for feed ratio cases R = 0 and R = 0.33. The curves were almost unchanged after the 0.006 m bed length was reached.
Some key points can be made here. In the R = 0 case, CO selectivity exceeded C
2H
4 before the temperature peaked (see
Figure 1), but was lower than C
2H
6. After the peak temperature, consistent with experimental observations by Liu et al. [
12], C
2H
4 exceeded CO. In the R = 0.33 case, the H
2O
2 (not shown) dissociated immediately upon entry. The resulting •OH radicals abstracted •H atoms from CH
4, causing a spike in C
2H
6 formation, and a quickly rising CH
4 conversion. The C
2H
6 rapidly dehydrogenated to C
2H
4. The CO peaked at approximately where the temperature peaked. Unlike the R = 0 case, both C
2 species exceeded CO prior to the temperature peak. This all occurred much faster than for the R = 0 case. The ultimate CH
4 conversion for the R = 0.33 case was only slightly lower than for the R = 0 case, while showing a higher C
2H
4 selectivity and lower CO. Liu et al. [
12], for the R = 0 case, concluded that the selectivities of COx and C
2 depended on local O
2 concentration and temperature. Using H
2O
2 added the further complexity of gas phase chemistry to the surface reactions.
3.2. Use of H2O2 to Decrease Feed Temperature to PBR
We now discuss whether the replacement of O
2 by H
2O
2 allows for a lowering of the overall PBR feed temperature, which would be an energy and cost saving. This analysis used the R = 0 and R = 0.33 feeds with the LR_LF case, with the results shown in
Table 2.
The partial H
2O
2 substitution for O
2 produced a respectable CH
4 conversion at the lower feed temperatures where O
2 feed alone showed no OCM activity (R = 0 for 1043 and 943 K feeds). These results at lower feed temperatures were consistent with those observed experimentally [
17,
18,
19]. At the 843 K feed temperature, even the R = 0.33 case was poor.
3.3. Incremental H2O2 Replacing O2 to CSTR
As mentioned above, the CH
4/O
2 and H
2O/H
2O
2 vapor streams would likely be fed separately into the CSTR due to safety concerns about preheating a vapor stream containing H
2O
2. For example, for the R = 0.33 and HR_HF case, to achieve an effective (hypothetical) 843 K feed temperature while holding the H
2O/H
2O
2 stream at 373 K, the CH
4/O
2 stream would be preheated to about 883 K. The HR_HF case was chosen for this CSTR analysis because it showed the best yield and selectivity of sum_C
2H
x at the lowest feed temperatures in the earlier study [
11].
For the CSTR calculations, the volume was the same as the open (gas) volume of the packed bed, with the same catalyst surface area. The CSTR might be a single-phase ideal fluidized bed, or a perfectly mixed (e.g., jet-stirred) reactor with catalyst on the walls.
Substitution of all or some of the feed O
2 content with H
2O
2 had a marked impact on the CSTR performance.
Figure 5 (left) shows that substituting H
2O
2 for O
2 reduced the exit temperature somewhat (~65–110 K), as might be expected from the lower exothermicity (see Equations (2) and (3)). However,
Figure 5 (right) shows a complex story for the impact on CH
4 conversion. For effective feed temperatures of 843 and 943 K, switching from O
2 to H
2O
2 reduced CH
4 conversion by only ~3 percentage points. With a 1043 K feed temperature, there was little impact on conversion. At 1143 and 1243 K, switching to H
2O
2 actually increased CH
4 conversion. While literature experiments [
17,
18,
19] used a PBR, the results here were still found to be consistent with those observations in terms of the activity of H
2O
2.
Figure 6 shows that the yield of sum_C
2H
x was notably higher than the CO + H
2 yield for all R cases, while increasing H
2O
2 had a greater impact on yields at the higher feed temperatures. These results were consistent with those revealed in
Figure 3 for the PBR. Selectivities of sum_C
2H
x remained higher than CO + H
2, especially at the lower feed temperatures, as seen in
Figure 7.
Equations (2) and (3) suggest that replacing O
2 by H
2O
2 will increase the production of C
2H
4 and H
2O, while reducing CO and H
2. Consider the HR_HF case, with the effective feed temperature into the CSTR of 1243 K, with the results shown in
Table 3. As the H
2O
2 fraction in the oxidant increased (i.e., higher R value), the CSTR exit temperature fell, but the CH
4 conversion increased. The C
2H
4 and H
2O production increased, while CO and H
2 dropped. Finally, the fraction of catalytic sites occupied by adsorbed O atoms (O
s) decreased as R increased. Since •H abstraction by •O
s is the primary catalytic step for CH
4 activation [
6] by O
2, the drop in •O
s fraction was consistent with a shift from heterogeneous catalyzed to homogeneous non-catalyzed conversion pathways at higher R.
3.4. Brief Reactor Comparison Summary
A simple comparison between the PBR and CSTR for OCM with and without H
2O
2 as an activating oxidant is shown in
Table 4. This illustration is based on the LR_LF case (see
Table 1). The results will vary somewhat for other cases, but the observations will be similar. This summary considers both the current results and those published earlier with just O
2 as an activator [
11]. While the claims are based on calculations using the Karakaya et al. [
6] mechanism for La
2O
3/CeO
2 catalyst, it is anticipated other OCM catalysts would give rise to similar claims.
Table 4 shows several points. The lowest practical feed temperature is the value below which there is no appreciable CH
4 conversion. All remaining values in each column correspond to those temperatures. Replacing a portion of the feed O
2 with H
2O
2 vapor allows the CSTR to achieve good CH
4 conversions at the lowest feed temperature. It also allows the PBR to run with a reduced feed temperature. Even at this low feed temperature, the CSTR has a sum_C
2H
x selectivity that exceeds the PBR at a much higher temperature. The CSTR also shows reduced syngas (CO + H
2) and improved sum_C
2H
x selectivity when using the H
2O
2.