Dehydrogenation of Formic Acid over a Homogeneous Ru-TPPTS Catalyst: Unwanted CO Production and Its Successful Removal by PROX

Abstract

Formic acid (FA) is considered as a potential durable energy carrier. It contains ~4.4 wt % of hydrogen (or 53 g/L) which can be catalytically released and converted to electricity using a proton exchange membrane (PEM) fuel cell. Although various catalysts have been reported to be very selective towards FA dehydrogenation (resulting in H₂ and CO₂), a side-production of CO and H₂O (FA dehydration) should also be considered, because most PEM hydrogen fuel cells are poisoned by CO. In this research, a highly active aqueous catalytic system containing Ru(III) chloride and meta-trisulfonated triphenylphosphine (mTPPTS) as a ligand was applied for FA dehydrogenation in a continuous mode. CO concentration (8–70 ppm) in the resulting H₂ + CO₂ gas stream was measured using a wide range of reactor operating conditions. The CO concentration was found to be independent on the reactor temperature but increased with increasing FA feed. It was concluded that unwanted CO concentration in the H₂ + CO₂ gas stream was dependent on the current FA concentration in the reactor which was in turn dependent on the reaction design. Next, preferential oxidation (PROX) on a Pt/Al₂O₃ catalyst was applied to remove CO traces from the H₂ + CO₂ stream. It was demonstrated that CO concentration in the stream could be reduced to a level tolerable for PEM fuel cells (~3 ppm).

1. Introduction

Currently, many methods to store and/or to transport energy in a green and durable way are being investigated in order to transition to an environmentally friendly economy. Examples include mechanical storage, batteries, superconductors, chemical storage (hydrogen and other high energy molecules) [1,2]. Formic acid (FA) containing ~4.4 wt % of H₂ has been considered a potential energy carrier since 1978 [3]. Essentially, H₂ could be chemically combined with carbon dioxide (CO₂) to form liquid FA [4], which is more easily stored and transported than gaseous H₂ considering practical and safety issues. FA can then be catalytically decomposed back to H₂ and CO₂ in a 1:1 ratio, closing a loop for CO₂ and providing H₂ for fuel cell technology to generate electricity. Thus, FA would simply function as an intermediate to store and to transport energy in a clean way (Figure 1).

Recently, FA has sparked interest again, because catalysts highly effective in FA dehydrogenation in reasonable conditions have been found [5,6,7,8,9]. This fact in combination with easy and safe handling opens a route of using FA in applications such as energy storage and transport [10].

Although very high catalyst selectivities have been reported in literature [7,9,11], a side-reaction of FA dehydration takes usually place, next to FA dehydrogenation, due to a still insufficient catalyst selectivity and/or thermal decomposition of FA:

HCOOH → H₂ + CO₂,

(1)

HCOOH → H₂O + CO.

(2)

It is very important to consider this side-reaction, because a CO content, even at ppm-levels, is poisonous for many fuel cells [12,13]. FA thermal decomposition was investigated both computationally and experimentally. However, most of the experiments were conducted in supercritical water [14,15,16,17].

Mitigating strategies such as improving CO-tolerance of fuel cells showed that Pt–Ru alloying, for example, can weaken the binding of CO to Pt catalysts, decreasing the poisoning effect [18,19]. Air-bleed—injecting a small amount of air or oxygen into the fuel feed stream—is also applied to lessen CO poisoning. Here, CO oxidation takes place directly on a fuel cell membrane as an oxidation catalyst. The method can recover up to 90% of the fuel cell performance at 200 ppm CO [20,21]. However, air-bleed was shown to cause long-term degradation of the membrane where O₂ is reduced to H₂O₂ which corrodes the catalyst [22]. To avoid this problem, preferential oxidation (PROX) could be used to decrease CO concentration. In this approach, CO is removed from the H₂ stream through injecting a small amount of O₂ and selective CO oxidation over a heterogeneous catalyst prior to entering the fuel cell. Nowadays, PROX is mainly applied to bulk production of H₂ where a significant CO amount (0.5–1%) is produced through steam reforming and water gas shift reactions [23]. Several catalysts such as noble metals supported on alumina or ceria as well as transition metals on similar supports are available [24]. The catalysts are designed to work in gas streams containing at least 85% of H₂. In H₂ rich gas streams, Equations (3) and (4) are of importance

2CO + O₂ → 2CO₂

(3)

2H₂ + O₂ → 2H₂O

(4)

To minimize H₂ loss and H₂O emittance (Equation (4)), the catalyst selectivity towards CO oxidation needs to be as high as possible. An important factor is the amount of supplied O₂, because selective catalysts start converting H₂ after complete CO conversion [24]. Moreover, FA dehydrogenation gas mixtures contain much more CO₂ (~50%) which could possibly effect the conversion of CO.

Other methods to remove CO from gas streams are adsorption and methanation. However, it is hard to implement pressure swing CO adsorption in compact applications [25]. Methanation is usually mentioned as an option next to PROX:

CO + 3H₂ → CH₄ + H₂O,

(5)

CO₂ + 4H₂ → CH₄ + 2H₂O.

(6)

Supported Ni and Ru methanation catalysts are well investigated [26]. Fuel cells are tolerant to methane [12]. No gas needs to be added to the stream. However, significant amount of H₂ could be lost in this process depending on the CO concentration and catalyst selectivity towards CO methanation. Next to that, most of the methanation catalysts become active only at temperatures over 250 °C.

In the present study, a H₂ + CO₂ gas mixture (reformate) resulted from catalytic FA dehydrogenation carried out in a continuous-flow gas–liquid bed reactor was analyzed. The liquid reaction phase was comprised of an aqueous solution of Ru(III) chloride and mTPPTS used as catalyst precursors. mTPPTS was chosen as a phosphine-ligand because of its high stability and water solubility of the resulting catalytically active complex [7]. Previously, it was reported that this catalyst was stable over more than one year of intermittent use, while being kept in air [27]. The turnover frequencies (TOF’s) reached in a continuous mode were 210 h⁻¹ and 670 h⁻¹ at 100 °C and 120 °C, respectively [27]. CO content in the resulting gas mixture was not measured in a continuous mode, but batch experiments performed using the same reactant concentrations showed no CO traces (detection limit ~ 3 ppm) [7].

In the present study, FA was fed to the reactor continuously. The CO concentration (ppm level) in the H₂ + CO₂ gas streams was measured as a function of the reaction temperature and FA feed flow. Quantification of such low CO concentrations (especially in CO₂-rich gas mixtures) is an extremely difficult task. In this research, gas chromatography (GC) coupled with sample methanation was used to achieve a good CO/CO₂ separation and sensibility of the measurements, while it was also possible to measure a H₂ concentration.

Next, CO removal from H₂ + CO₂ gas streams by PROX was tested using a commercially available Pt/Al₂O₃ catalyst placed downstream from the FA dehydrogenation reactor. The PROX catalyst was not heated. A controlled amount of O₂ (in air) was injected into the gas stream before the PROX catalyst. The considered parameters were the CO concentration in the gas stream and the amount of added O₂. A combination of these numbers can be expressed through the O₂ excess parameter $\lambda$ (Equation (7)), representing the O₂ excess, relatively to its amount required for total CO oxidation. At $\lambda =1$, the O₂ concentration in a gas mixture is exactly enough to convert all CO (catalyst selectivity ~ 100%)

$$\lambda =\frac{2C_{{\mathrm{O}}_2}}{C_{\mathrm{CO}}}=\frac{2p_{{\mathrm{O}}_2}}{p_{\mathrm{CO}}}$$

(7)

2. Results and Discussion

2.1. CO Side-Production in FA Dehydrogenation

The CO concentration in H₂ + CO₂ gas mixtures was measured in the reaction temperature range of 60–80 °C. Below 60 °C, the Pt/Al₂O₃ catalyst activity was too low for the application, while at 80 °C and higher, water evaporation became quite intensive constantly reducing the catalyst solution volume. At a steady-state, the total gas flow rate was found to be the same for the all measured temperatures (

Table 1. Total H₂ + CO₂ gas flow as a function of FA feed measured at 60, 70 and 80 °C.

FA Feed (mL/min)	Total Gas Flow (L/min)
0.3	0.43
0.6	0.86
0.9	1.29

) indicating that in this temperature range the reaction rate was limited by the FA feed flow rate.

As can be seen in Figure 2, the CO concentration in the gas mixture was below 10 ppm at the FA feed flow rate of 0.3 mL/min but increased with increasing FA feed. It was supposed that, due to an imperfect reactant mixing in the reactor, at the high FA feed flow rates the size of injected FA droplets is bigger and hence their lifetime is longer. At higher temperatures, it leads to more pronounced side-reaction of FA thermal dehydration and higher CO content in the reformate.

2.2. PROX

Figure 3 shows CO concentrations in H₂ + CO₂ gas mixtures treated on the PROX catalyst. Varying the FA feed flow at constant reaction temperature, both the total gas flow rate and CO concentration were varied. It was observed during the experiments that the catalyst container became warm (50–60 °C) due to the exothermic oxidation of CO and H₂. It was found for all reformate flow rates that at high concentrations of injected O₂ ($\lambda$ = 3–14), the PROX treatment led to a drop in the CO concentrations to ~3 ppm. At $\lambda$ = 1–2 a small increase of the resulting CO concentration up to 4–6 ppm was observed. Surprisingly, at $\lambda$ < 1, a high CO conversion was observed as well (Figure 3). This fact was explained by the presence of a considerable amount of O₂ adsorbed on the catalyst surface. After an N₂ purge of the PROX catalyst for three days, the CO concentration decreased under the same conditions ($\lambda$ < 1) from initial 35 to 16 ppm indicating O₂ strongly adsorbed on the catalyst that could not be removed by a simple purge.

It is worth noticing that at the total H₂ + CO₂ gas flow rate of 0.86 L/min and high O₂ concentrations ($\lambda$ > 5), water condensation in the PROX catalyst container and drop of the CO conversion were detected (Figure 3) indicating a blockage of the catalyst surface by water.

This effect was previously reported for PROX catalysts [24]. For the lower gas flow rates, water clogging was observed only after a long period of catalyst exposure. The activity of the catalyst deactivated by the water clogging could be completely recovered by an air or nitrogen purge at room temperature.

3. Materials and Methods

A homogeneous Ru-mTPPTS catalyst (mTPPTS: meta-trisulfonated triphenylphosphine, Figure 4) solution (33 mM, 700 mL) was prepared by combining an aqueous solution of RuCl₃ (5.6 g) and mTPPTS (2.8 equivalent) with FA (128 g) and Na formate (30 g). The catalyst was activated prior to its placing into the reactor by steering the solution at 60 °C for 4 h under a N₂ flow.

FA was dehydrogenated in a 1 L continuous-flow gas–liquid bed reactor (Figure 5). A cylindrical, non-stirred reactor vessel of a 7.0 cm inner diameter and 40 cm length was heated by means of an internal heating element connecting to an oil circulator (Huber Pilot One). The temperature inside the reactor was varied in the range of 60–80 °C. FA was injected into the reactor (0.3–0.9 mL/min) at the reactor bottom through a HPLC pump (HPLC Pump 422, Kontron Instruments, Montigny Le Bretonneux, France). The gas bubbles produced due to the reaction rose, mixing the reaction solution. Outside the reactor, a H₂ + CO₂ gas mixture was passed through a condenser and an adsorber (activated carbon) to remove FA and water vapors. The total gas flow was measured using a bubble flowmeter. In separate experiments, a H₂ + CO₂ gas mixture was passed through a container of 2.8 cm inner diameter, containing a PROX catalyst (50 g). A commercial 0.5 wt % Pt supported on alumina (spheres of 1.8–3.5 mm diameter, Johnson Matthey, Royston, UK) was used as a PROX catalyst. An air flow controlled by a mass flow controller was injected into the gas stream before the PROX catalyst.

Gas samples were collected at a steady state before and after the PROX catalyst using a rubber balloon flushed with nitrogen and vacuumed. The collected gas samples were analyzed by a GC method using an Agilent 7890B gas chromatograph (Agilent, Santa Clara, CA, USA) equipped with a CarboPlot P7 column (25 m × 0.53 mm × 25 µm), a Nickel Catalyst Kit (to convert CO (CO₂) to CH₄), flame ionization (FID), and thermal conductivity (TCD) detectors. At least three samples were taken for every point. At least three GC measurements were done for every gas sample. The GC was calibrated using standard (50% H₂ + 40 ppm CO + CO₂) and (50% CO₂ + 100 ppm CO + N₂) mixtures (Carbagas, Lausanne, Switzerland). The detection limit of the GC method was ~3 ppm.

4. Conclusions

• It was found that unwanted CO concentrations in H₂ + CO₂ gas streams catalytically produced through FA dehydrogenation over a homogeneous Ru-mTPPTS catalyst in a continuous-flow gas–liquid bed reactor (60–80 °C) was too high (8–70 ppm) to be used directly in a fuel cell. The CO concentration was dependent on the FA feed flow rate and reaction design (mixing).
• It was shown that the Pt/Al₂O₃ catalyst can be easily implemented to clean H₂ + CO₂ gas streams from CO by PROX. It was shown that the O₂ supply should be fine-tuned well to prevent both a water clogging of the PROX catalyst and a H₂ loss due to its oxidation. Experimentally, CO concentration was confidently decreased by PROX at O₂ concentrations close to stoichiometry (0.008–0.065 v % O₂) to the level acceptable for fuel cell applications (<5 ppm). To our knowledge, this is the first example of successful CO removal from a FA reformate by PROX.