Absorption Chemistry of Plasma Generated NOx Gas for Green Nitric Acid Production

1. Introduction

Ammonium nitrate fertilizer, an essential element in our global food production system is currently produced in large chemical plants [1,2]_. Ammonia (NH₃), being the base chemical, is produced in the Haber-Bosch process. Ammonia synthesis, using hydrogen gas produced from fossil methane is responsible for a significant percentage of the global carbon dioxide emissions [3,4,5]. In the subsequent Ostwald process, NH₃ is catalytically oxidized with O₂ to form NO₂, which upon reaction with water forms HNO₃. The chemistry of this industrial absorption processes is complex. The process entails about 40 entangled reaction steps with intermediates and by-products such as HNO₂, N₂O, N₂O₃ and N₂O₄ [6]. NO oxidation and NO₂ absorption are performed under pressure of up to 10 bar and at NO_x gas concentrations of around 10%. The commercial HNO₃ product is typically a concentrated aqueous solution containing 65 wt.% HNO₃ [7,8,9].

Electricity-powered warm plasma technology is an emerging alternative for nitric acid production and even has the potential to fixate N₂ into NO_x with a lower net energy consumption than the combined Haber-Bosch and Ostwald process [10]. In the plasma electrons and gas molecules are excited such that nitrogen and oxygen gas from atmospheric air react to NO _x [10,11]. Like in the Ostwald process, NO_x needs to react with water to form HNO₃. However, in a plasma process NO_x is produced at concentrations in the range of a few percentages, which is substantially lower than in the Ostwald processes (ca. 10%), leading to different absorption conditions [7,8,10].

Besides serving as a precursor for HNO₃, NO_x gas is also a known air pollutant emitted at parts-per-million levels during combustion processes [12,13]. NO_x scrubbing is an end-of-pipe solution to eliminate it from flue gas [14,15]. Under the near ambient temperature and pressure conditions, the low aqueous solubility of NO and the slow kinetics of its spontaneous oxidation to the better dissolving NO₂ often necessitate the use of oxidizing agents like H₂O₂ and O₃ [15,16,17,18,19]. The partial pressure of NO_x and the NO₂ - to - NO ratio significantly influence the absorption performance and the formation of HNO₂ versus HNO₃ [12,20,21]. Alkaline scrubbing solutions are often deemed more effective for NO_x removal compared to neutral of acidic solutions due to the acidic character of NO_x [12].

Given the current relatively small size of plasma reactors for NO_x production, a scrubber unit for producing HNO₃ from plasma-generated NO_x should be simple and operated without pressurization and added chemicals. The operation conditions of such NO_x scrubbers are in between those of the Ostwald process and NO_x emission abatement technologies. While NO_x absorption is extensively modelled for example by Patwardhan et al. [22], only limited experimental research has been performed in NO_x to HNO₃ conversion, specifically under reaction conditions relevant to plasma-NO_x. Notably, much of the experimental research in this area focuses on short-duration tests, often with recirculation times for the scrubbing solution limited to an hour [20,23].

In this work the absorption reaction network of NO_x-containing gas, simulating the gas mixture produced from ambient air in a warm plasma reactor is experimentally investigate for long term recirculation. Insight is gained on the influence of scrubbing pH on NO_x absorption kinetics, capacities and on the formation of undesired nitrous acid side product without the addition of oxidizing agents.

2. Results

2.1. NO_x Scrubbing Experiments

NO_x scrubbing experiments were performed in a column with an empty bed volume (EBV) of 780 mL with 400mL recirculation solution of which the pH was kept fixed at values of 1, 2, 3, 4 and 5 by dosing KOH (Figure 1). A gas mixture consisting of 1% NO, 20% O₂ and 79% N₂ enters the column in counter current with the solution at a flowrate of 3 L/min. In order to ensure sufficient contact area the column is filled with an inert packing material. In- and outgoing NO_x gasses are analyzed in-line using a UV detector. The HNO₃, NO₂^- and HNO₂ content of the scrubber solution was analyzed on regularly taken samples, therefore HNO₂ was deprotonated to NO₂^- by adding buffer solution of pH 7 to enable quantification with ion chromatography. The combination of NO₂^- and HNO₂ will later on be referred to as total nitrate concentration. For more information see section Methods.

2.2. Scrubber Liquid Analysis

The nitrate, nitrite and total nitrite concentration, standing for the sum of NO₂^- and HNO₂ in the scrubber solution were determined at 1 h time intervals (Figure 2). The experiments at pH 4 and 5 were run over two days. No liquid samples were taken during the night. The nitrate concentration steadily increased throughout the five experiments at different pH values. After 6 h scrubbing, the nitrate concentration in the scrubber solution reached values of 12.6 mM at pH 1; 14.0 mM at pH 2; 10.4 mM at pH 2; 4.72 mM at pH 3, and 2.97 mM at pH 5 (Figure 2). The nitrate accumulation decreased with increasing pH value.

The total nitrite concentration displayed increasing concentrations at higher pH value. At pH 1, 2 and 3, the total nitrite concentration reached a plateau after 2 to 4 hours (Figure 2) at ca. 3.37 mM, 3.86 mM and 5.41 mM for pH 1, 2 and 3, respectively (Table 1). At pH 4 and 5 the total nitrite concentration in the circulating solution did not reach a plateau, even when the scrubbing was prolongated to 30 hours. The total nitrite concentrations rose to 21.6 and 38.4 mM, respectively at pH 4 and 5. At pH 5 nitrite is the main NO_x-derived compound in the scrubber liquid.

After the total nitrite concentration plateau was reached at pH 1, 2 and 3, the calculated HNO₂ concentrations reached very similar concentrations of 3.35, 3.63 and 3.32 mM, respectively (Table 1). At pH 4, the HNO₂ concentration reached 2.95 mM at the end of the 30 h testing period and did not fully reach a plateau yet, suggesting it could still increase to reach a value similar to that reached in the experiments at lower pH. At pH 5 the final HNO₂ concentration reached only 0.599 mM, which is substantially below the plateau value reached at lower pH. Under this condition, it may take a very long time to reach the plateau.

2.3. Gas Phase Analysis

The NO_x concentration in the gas exiting the scrubber under the different pH conditions is presented in Figure 3a and b. In experiments at pH 1, 2 and 3, NO_x outlet concentrations initially were around 800 ppm, and increased during the experiment until a plateau value of ca. 1,800 ppm was reached. The corresponding NO_x removal efficiency decreased from ca. 95% initially, to 82.5%, 83.9% and 83.0 % at pH 1, 2 and 3, respectively (Figure 3c).

At pH 4 and 5, the outgoing NO_x gas concentrations were significantly lower after 7 hours, and still increasing (Figure 3b). When prolonging the scrubbing to 30 hours at pH 4, a plateau value of ca. 84.2% removal efficiency was reached (Figure 3c), which is in the range of values obtained at pH 1, 2 and 3. At pH 5 even after 30 hours the removal efficiency kept on decreasing slowly. Very long times would have been needed to verify the reaching of a plateau. Gaps in data are due to a set-up malfunction leading to an artefact in the data, irrelevant for further interpretation.

3. Discussion

The reaction network of NO_x gas absorption in aqueous media is complex and involves dissolution as well as chemical reactions involving reaction intermediates and reversible and irreversible reactions in gas as well as liquid phase. The reactions considered most important in the gas phase are presented in Eqns. 1-4 [6,23,24,25].

2 NO + O₂ → 2 NO₂

(1)

NO + NO₂ ⇌ N₂O₃

(2)

NO + NO₂ + H₂O ⇌ 2 HNO₂

(3)

2 NO₂ ⇌ N₂O₄

(4)

According to Eqns. 1-3, NO gas is converted into NO₂, N₂O₃, and HNO₂, respectively, which are more water-soluble. NO₂ also dimerizes to form N₂O₄ (Eqn.4).

The main reactions in the liquid phase are given in Eqns. 5-8. Dissolved N₂O₃, and N₂O₄ are converted into either HNO₂ or a combination of HNO₂ and HNO₃ [22,25,26]. The reactions of Eqns. 5-8 are often assumed irreversible since the rate of the back reactions, being bimolecular, is considered negligible in the present diluted, aqueous solutions [23,27,28,29].

2 NO₂ + H₂O → HNO₂ + HNO₃

(5)

N₂O₃ + H₂O → 2 HNO₂

(6)

N₂O₄ + H₂O → HNO₂ + HNO₃

(7)

3 HNO₂ → H₂O + 2 NO + HNO₃

(8)

In liquid phase nitrous acid is prone to dismutation leading to HNO₃, NO and H₂O formation according to Eqn. 8 [22,26,30,31]. The poorly soluble NO molecules desorb from the solution and return to the gas phase, where they re-enter the reaction network [32]. The HNO₂ decomposition reaction mechanism is often assumed as described by Schwartz et al. to be 4^th order in HNO₂, but accurate rate equations remain matter of discussion [26,28,29].

The observation of a total nitrite (NO₂^- and HNO₂) concentration plateau in the scrubber liquid at fixed pH values of 1, 2 and 3 (Figure 2) can be explained as follows. Nitrous acid decomposition is a dismutation reaction involving several HNO₂ molecules and a reaction order higher than 1 (Eqn. 8) [28,29]. With increasing HNO₂ concentration, the HNO₂ dismutation reaction rate will increase until it equals the rate of HNO₂ formation (Eqns. 5-7), leading to a steady state HNO₂ concentration. At a constant pH, this also means a steady state NO₂^- concentration.

This is in agreement with what is seen in the analysis of the gas phase. The decrease of the NO_x removal efficiency with time at constant pH (Figure 3) could be attributed to this formation of NO by HNO₂ dismutation (Eqn. 8). Here as well it is remarkable that once the system reaches its steady state, NO_x removal efficiencies are very similar for pH 1, 2, 3 and 4. The similar limitation of the NO_x removal efficiency is also likely to be a consequence of the similar HNO₂ concentration (Table 1) and the equal rate of the HNO₂ decomposition reaction (Eqn. 9).

While at steady state the pH does not seem to influence the removal efficiency, in the transient period the pH does matter as observed in earlier studies [12,21,23,33]. The time after which the steady state condition is reached, is pH dependent. For tests at pH 1, 2 and 3 the HNO₂ concentrations reaches steady state after approximately four, five and six hours, respectively. At pH 4 and 5, the accumulation of HNO₂ in the scrubber solution progresses much slower (Figure 2). This can be attributed to the acid dissociation reaction. The higher the pH, the more HNO₂ directly dissociates into NO₂^- and the longer it takes for HNO₂ to accumulate enough for the dissociation reaction (Eqn. 8) to balance the formation rate. Per pH unit, ten times more HNO₂ is converted to NO₂^-, which is reflected in Table 1. The test at pH 4 nearly reached its steady state after 30 hours, but the test at pH 5 did not reach this level within the tested timespan. An important conclusion to draw is that running experiments for shorter timespans than necessary to reach steady state, could wrongly lead to the assumption that increasing the pH of the scrubbing solution also increases the removal efficiency on the long term.

Nitrate, being a strong acid and assumed always in its dissociated form, accumulates over time in the scrubber solution (Figure 2). In the transient period the concentration increases more than proportional with time. This can also be attributed to the increasing rate of the HNO₂ decomposition reaction (Eqn. 8) which also forms HNO₃. Given the steady state NO₂^- and HNO₂ concentrations and the ongoing NO₃^- accumulation there is no point in expressing a nitrate selectivity in these tests as it is strongly time dependent. What can be expressed is the steady state NO₂^- amount that remains present. In practical applications of nitrate fertilizer production, nitrite and nitrous acid are undesired side products. Even though maximal allowed nitrite concentration in nitrogen fertilizer is in many countries not explicitly regulated, effects of nitrite toxicity are reported at concentrations starting from 0.11 mM [34]. It is however known that nitrite is in soil processes rapidly transformed in nitrate [35]. This study hints at a few options expected to lower the steady state HNO₂ concentration without sacrificing the NO_x removal efficiency.

A first approach to consider, is to increase the liquid volume. For a constant EBRT but larger liquid volume, a slower increase in HNO₂ concentration is expected for the same absolute amount of NO_x absorbed. This means that also the dismutation reaction is slowed down, resulting in a lower steady state HNO₂ concentration and NO formation. For experiments at pH 1 with a liquid volume of 250 mL and 400 mL, similar removal efficiencies of 83.6% and 82.5% were reached, respectively. Total nitrite concentrations stagnated after 6 hours at 4.07 and 3.37 mM for 250 mL circulation volume and 400 mL. Respectively, 4.04 mM and 3.35 mM of this were in the form of HNO₂. Increasing the volume of the scrubbing liquid 1.6 times, reduced the nitrite content with around 17%. Increasing the liquid volume as well as slowing down the dismutation reaction, also decreases the rate of NO₃^- formation. After 6 hours a nitrate concentrations of 19.6 mM and 12.6 mM were obtained for 250mL and 400mL scrubbing solution. However, this can be remedied by increasing the time over which the experiment in ran.

As shown in other experimental contexts an important factor influencing the nitrite content is the NO_x oxidation ratio [12,20]. By oxidizing NO to NO₂ and forming N₂O₄ instead of N₂O₃, reactions of Eqns. 4, 5 and 7 are favored over reactions of Eqns. 3 and 6, lowering in this way the HNO₂ formation and dismutation rate. The NO_x oxidation ratio is dependent on the composition of the plasma produced NO_x gasses but also by the oxidation reaction of NO (Eqn. 1), which is second order dependent on NO partial pressure and first order dependent on O₂ partial pressure [22,36]. As often done, pressurizing the air in or at the outlet of the plasma process will improve the NO₂ content of the NO_x and the nitrate selectivity in the scrubber as showcased by Tsonev et al. reaching an OR of 94% [37]. Another solution is to prolong the residence time of the gas in the scrubber.

A final option to further lower the nitrite content would be to store the scrubbing solution in a storage tank, after the desired nitrate concentration is reached. The dismutation reaction of HNO₂ proceeds spontaneously but is slow. Favored by a low pH, undesired nitrite and nitrous acid are expected to be eliminated over time, producing nitric acid and NO gas. The gas phase enriched with NO could be recovered from the storage tank and again injected in the scrubber to ensure no NO_x in released to the environment.

4. Materials and Methods

The experimental set-up for NO_x absorption experiments is depicted in Figure 1, see section Results. A mixture of NO, O₂ and N₂ gases delivered from gas cylinders was fed to the absorption column. Absorption liquid was circulated in such a way that gas and liquid are running in counter current. Large contact area was assured between liquid and gas by filling the column with packing material consisting of 6 mm glass Raschig rings. The column was kept at constant temperature of 293K by a cryostat. The liquid outlet of the column is connected to a round flask with openings for the gas inlet, base solution inlet, pH probe (Mettler Toledo, SevenCompact) and pH control system (Hannah instruments, HI510 and HI1006-1850).

The inlet gas mixture was composed of 1% NO, 20% O₂ and 79% N₂, and fed from gas cylinders using mass flow controllers (Bronkhorst, EL-FLOW). The flow rate was 3 L/min. NO and NO₂ concentrations in the gas mixture were measured with an UV-analyzer (ABB AO2020 Limas11HW). Before starting the absorption, the feed gas was run during 30 minutes in bypass mode and sent directly to the analyzer without passing through the column to allow the gas composition to stabilize and to be verified. The NO and NO₂ removal efficiency (RE) was estimated as the difference between the in- and outlet concentration divided by the inlet concentration (Eqn. 9):

$$RE\left(N{O}_x\right)\%=1-{\displaystyle \frac{{{\displaystyle \int }}_{t_{end-5min}}^{t_{end}}\left(C_{N{O}_{xout}}\right)}{{{\displaystyle \int }}_{t_{end-5min}}^{t_{end}}\left(C_{N{O}_{xbypass}}\right)}}\ast 100$$

(9)

In the bypass phase of each experiment some NO oxidation already occurred in the gas lines, resulting in a mixture of NO and NO₂ to reach the scrubber column. The oxidation ratio (OR) defined as the ratio of the concentration of NO₂ over total NO_x is determined according to Eqn. 10. Oxidation ratios were very similar for all tests with OR values of 0.56 ± 0.008.

$$OR={\displaystyle \frac{C_{N{O}_{2in}}}{C_{N{O}_{2in}}+C_{N{O}_{in}}}}$$

(10)

The pH of the scrubbing solution was kept constant as follows. The round flask was filled with acidified aqueous solution of H₂SO₄ (Acros Organics, 96 wt%) or H₃PO₄ (Acros Organics, 85 wt%) at the intended pH, in the range from pH 1 to pH 5. During operation of the absorption column, the pH was monitored and kept constant by dosing 0.2 M KOH (Sigma-Aldrich, ≥85%). Liquid samples were taken every hour. Liquid samples taken from the round flask were diluted 100 times and buffered at pH 7 using an equimolar mixture of 0.005 M NaH₂PO₄ (Merck, > 99%) and Na₂HPO₄ (Merck, 98-100%). Nitrous acid is a weak acid being only partially deprotonated under acidic conditions. Protonation-deprotonation equilibrium of nitrous acid was considered to be established in the scrubber solution under all conditions. Estimates of the actual concentration of nitrous acid based on total nitrite analysis were made assuming a nitrous acid pK_a value of 3.20 (Eqn. 11) [38].

$$\left[HN{O}_2\right]={\displaystyle \frac{\left[N{O}_2^{-}\right]\ast {10}^{-pH}}{{10}^{-p{K}_a}}}$$

(11)

At this pH value nitrous acid is deprotonated which is needed for quantification using anion chromatography. The nitrate and nitrite content of the liquid samples were analyzed using ion chromatography (Metrohm Eco IC, A Supp 5 column, 850 IC conductivity detector).

5. Conclusions

This study provides experimental data on the NO_x absorption chemistry in aqueous solution under conditions relevant for small scale fossil carbon-free nitric acid production based on thermal plasma technology producing NO_x gas at concentrations of around 1%. An aqueous solution with constant pH in the range 1 to 5 was recirculated over a packed scrubber column ensuring large contact area.

The NO_x removal efficiency is observed to be higher for increased pH at the onset of the experiment, yet decreases over the course of the experiment to a plateau value of 83%, independent of pH. The explanation lies in a corresponding increase in the HNO₂ concentration until a plateau is reached, and a negative correlation between HNO₂ concentration and removal efficiency. The steady state HNO₂ concentration and, therefore, the HNO₂ decomposition rate are observed to be very similar for all pH values. It is thus concluded that pH does not influence the removal efficiency when running experiments sufficiently long for steady state to be reached. However, pH does remain an important parameter concerning the NO₂^- concentration as it is the determining factor in the NO₂^- ↔ HNO₂ equilibrium.

This paper indicates the importance of HNO₂ as a key compound in the systems transient period as well as in the reaction steps which contribute to ultimately reaching a steady state removal efficiency.