HCN production from formaldehyde during the selective catalytic reduction of NOx with NH3 over V2O5/WO3-TiO2

Abstract Raw exhaust gases may contain notable levels of formaldehyde that can negatively impact the efficiency of after-treatment systems. In the selective catalytic reduction (SCR) of NOx over V2O5/WO3-TiO2, formaldehyde was found to react with NH3 to produce HCN at concentrations above the threshold limit value set by environmental/safety organizations. Due to this side reaction, NH3 is consumed parasitically and the NOx conversion decreases by up to 15 %, even after compensating for the fraction of lost NH3. Under similar conditions, the non-reducible TiO2 support also produced HCN moderately, thereby showing that redox sites promote the reaction but are not a necessary condition. To understand the chemistry responsible for HCN formation, the roles of reaction temperature, water, and oxygen were investigated. Our results suggest a new pathway for HCN production through the direct reaction of formaldehyde and NH3, which is active at high temperature and does not proceed through the formate route previously proposed.


Introduction
The rise in anthropogenic nitrogen oxide (NO x ) emissions has impelled the development of exhaust after-treatment technologies. First introduced in the 1950′s, the selective catalytic reduction (SCR) of NO x by ammonia (NH 3 ) has enjoyed worldwide academic interest and industrial acceptance in the subsequent years. Because of its unparalleled efficiency, wide operating window, and moderate cost, SCR is the technology of choice for NOx reduction in Diesel-powered vehicles and thermal power plants [1][2][3].
More than 90 % of the NO x emissions are in the form of nitric oxide (NO), and the main reaction taking place inside the reactor is the standard SCR reaction (Eq. 1).

NO + NH 3 + O 2 → N 2 + 6 H 2 O
The presence of nitrogen dioxide (NO 2 ) in the exhaust gas invokes the occurrence of the so-called fast SCR reaction (Eq. 2), wherein the reoxidation step of the metal center occurs much faster because of the stronger oxidizing power of NO 2 than oxygen (O 2 ) [4,5].
NO + NO 2 + 2 NH 3 → 2 N 2 + 3 H 2 O Besides the abatement of carbon monoxide (CO) and unburned hydrocarbons, Diesel oxidation catalysts (DOC) also partially oxidize NO to NO 2 and favor the fast SCR reaction [6].
The inherent complexity of the exhaust gas composition and gascatalyst interactions can also promote side reactions during the SCR process. The most significant is the direct oxidation of NH 3 Although the major product of this parasitic reaction is nitrogen (N 2 ), NO can also form at high temperature, thereby negating the beneficial effect of NH 3 addition in the first place. Another undesirable side reaction is the formation of nitrous oxide (N 2 O), a very potent greenhouse gas (Eq. 4). [9][10][11] 4 NO + 4 NH 3 + 3 O 2 → 4 N 2 O + 6 H 2 O The extent of N 2 O formation is catalyst-dependent, with Cu-and Febased SCR catalysts being more susceptible than V-based ones especially below 350 °C [12,13].
The production of hydrogen cyanide (HCN), a gas considered immediately dangerous to life, is another highly undesirable SCR side reaction. It was first observed over Cu-ZSM-5 when hydrocarbons were used as the reductant instead of NH 3 [14]. The extent of HCN formation followed a volcano-shaped profile with temperature, and the activity depended on the type of hydrocarbon used. With ethene and propene, https://doi.org/10.1016/j.apcatb.2020.119462 Received 10 July 2020; Received in revised form 18 August 2020; Accepted 20 August 2020 the HCN concentration reached a maximum value at 325 and 225 °C, respectively. The presence of water in the feed generally increased the amount of HCN produced [15]. It became evident that HCN formation did not solely apply to Cu-ZSM-5, but also to other catalysts operating under hydrocarbon-SCR [16][17][18][19].
HCN was also observed during NH 3 -SCR. When ammonium formate decomposition is exploited to generate NH 3 instead of urea, it should ideally decompose according to Eq. 5 [20].
As this is an endothermic reaction (Δ r H° = 85 kJ·mol −1 ), the equilibrium shifts in favor of ammonium formate at low temperature. This can give rise to another decomposition pathway in which it undergoes subsequent dehydration steps to form formamide (Eq. 6) and eventually HCN (Eq. 7).
Both formamide and HCN were observed below 250 °C under SCRrelevant conditions, and the decrease in HCN production correlated well with the increase of SCR activity. As more NH 3 was consumed to satisfy the increased NO x reduction at higher temperatures, more ammonium formate was decomposed according to Eq. 5, thereby rendering Eq. 6 and Eq. 7 less probable. Nonetheless, this reaction path showed very low HCN yield. A very recent study proposed that a similar route was responsible for HCN production under NH 3 -SCR conditions in the presence of formaldehyde (HCHO) [21]. Formed due to the incomplete oxidation of hydrocarbons, formaldehyde is a noxious compound that is nominally present as a trace constituent in the exhaust of internal combustion engines [22][23][24][25]. This is particularly problematic for stationary natural gas engines, which usually do not operate an oxidation catalyst upstream of the SCR catalyst. The unwanted side reaction between formaldehyde and NH 3 can potentially lower NO x conversion by consuming NH 3 that is otherwise reserved for SCR. While the decrease in SCR activity is certainly detrimental, the major problem really stems from the high toxicity of the produced HCN. The European Commission set the maximum occupational exposure limit of HCN at only 0.9 ppm. Hence, the release of even minute quantities of HCN to the environment can have grave consequences for air quality and human health [26,27]. It is therefore imperative to investigate this side reaction in detail to develop potential measures for mitigation.
In this study, we show that formaldehyde reacts with NH 3 to form HCN with high yields under SCR conditions, and that the reaction does not require the formation of ammonium formate.

Materials
The catalyst in this study comprised of 2 wt% V 2 O 5 /WO 3 -TiO 2 that was prepared by wet impregnation. A sufficient amount of ammonium metavanadate (NH 4 VO 3 ; Fluka, > 99.0%) was dissolved in water at 60 °C, and was mixed with 10 wt% WO 3 /TiO 2 (DT-52, Cristal) for 1 h. After removal of water under reduced pressure, the sample was calcined in a muffle furnace at 550 °C for 10 h. The resulting surface area of the calcined catalyst is ∼90 m 2 g −1 , which does not vary significantly from that of the starting 10 wt% WO 3 /TiO 2 material. [28] At this relatively low loading and calcination temperature, V 2 O 5 and WO 3 are well-dispersed and only reflections from anatase could be detected by x-ray  diffraction. [28][29][30] For comparative purposes, TiO 2 (DT-51, Cristal; ∼100 m 2 g −1 ) was also used without additional modification. Cordierite honeycomb monoliths (Corning, 400 cpsi, 17 mm x 12 mm x 50 mm) were repeatedly immersed in an aqueous slurry of V 2 O 5 /WO 3 -TiO 2 or TiO 2 and then dried in air to achieve a loading of 130 g·L −1 of material. The washcoated monoliths were dried and then calcined in a muffle furnace at 500 °C for 5 h.

Catalytic measurements
The measurements were carried out in a custom-built model gas test bench (Fig. 1). Mass flow controllers (MFCs) were used to regulate the flows of air, N 2 (99.999 vol%, Air Liquide), H 2 (99.998 vol%, Air Liquide), O 2 (99.998 vol%, Air Liquide), NO (6 vol% in N 2 , Air Liquide), and NH 3 (4 vol% in N 2 , Air Liquide). Water was generated in situ from the reaction of H 2 and O 2 through a Pt-based catalyst (H 2 O generator), which is preferred over conventional water vaporization as it allows for a pulsation-free water feed.
Formaldehyde was generated in situ through the partial oxidation of methanol over V 2 O 5 /WO 3 -TiO 2 (200−500 μm) in a separate stainless steel reactor (HCHO generator). For this purpose, N 2 was bubbled through a 50 vol% aqueous solution of methanol at ambient temperature before mixing with air and flowing through the catalyst bed heated at 250 °C to produce 150 ppm formaldehyde. Under these conditions, methanol oxidation was not complete, and the feed to the main reactor contained also 10−15 ppm methanol, ∼1 ppm formic acid, ∼350 ppm CO, and 0.1−0.2 vol% H 2 O. All gas lines were heated to 150 °C to prevent the condensation of components with low vapor pressure.
The model gas (GHSV =50,000 h −1 ) consisted of 10 vol% H 2 O and 10 vol% O 2 balanced in N 2 with variable amounts of formaldehyde, NO, and NH 3 as indicated in Table 1.
In most SCR catalytic studies, NH 3 and NO x are dosed at the same concentration across the temperature range under study. This technique is often sufficient to compare catalysts with different activities. However, for practice-relevant measurements, it is more useful to determine the NO x conversion by varying the NH 3 dosage until a defined NH 3 slip occurs (usually 10 ppm) [31]. For this purpose, the catalyst in this study was exposed to a constant feed of NO x and increasing dosages of NH 3 at a given temperature until the outlet concentration is ∼10 ppm. All of the NH 3 dosages as well as outlet concentrations of NO and NH 3 are reported in Table S The effects of O 2 and water on HCN formation were also investigated. The O 2 content was varied from 2.5 to 15 vol% in a gas feed containing 10 vol% H 2 O, 150 ppm formaldehyde, and 300 ppm NH 3 . In a similar way, the water content was changed from 0 to 12.5 vol% in a gas feed containing 10 vol% O 2 , 150 ppm formaldehyde, and 300 ppm NH 3 .
For comparison, formic acid was also dosed in the reactor instead of formaldehyde by bubbling N 2 through a 30 vol% aqueous solution of formic acid at ambient temperature. The HCHO generator was bypassed in order to prevent any subsequent oxidation of formic acid prior to meeting the catalyst in the main reactor. This resulted in 150 ppm formic acid in the feed.
The main reactor (quartz glass tube; d int = 28 mm) had three heating zones, including the preheating zone (T 1 ), which was filled with ceramic beads to avoid radial temperature gradients of the model gas feed. Thermocouples inserted in the center of the quartz tube before (T 2 ) and after the catalyst module (T 3 ) ensured uniform temperature control of the catalyst bed. The temperature of the reactor outlet was measured and controlled via the thermocouple (T 4 ) installed on the outer wall of the reactor. After being wrapped in ceramic fiber paper, the washcoated monolith piece was inserted in the rectangular opening of a cylindrical stainless steel holder that was itself wrapped with ceramic fiber paper for sealing against the reactor wall.
The outlet of the main reactor and the bypass line were connected to a gas feed pump (KNF) that was heated at 170 °C. A manual three-way valve allowed the gas from either the reactor or the bypass line to pass through a 2-m gas cell mounted in the sample compartment of a Nexus Fourier transform infrared spectrometer (Thermo-Nicolet) and heated at 180 °C. The gas components were quantified using the Omnic/ Quantpad software. The acquisition time was 8 s per measurement point (average of 4 spectra). To obtain the lowest possible detection limit (DL), a total of 10 measurement points were averaged. Typical DL values ranged from 0.2 to 1 ppm (Table S- The degree of NO x conversion was calculated according to Eq. 8. x in x out in (8) where

NO [ ] in and NO
[ ] x out are the concentration of NO in the gas feed, and the concentration sum of NO and NO 2 after the reactor, respectively.
As mentioned previously, the oxidation of methanol in the HCHO generator was not complete and left 10−15 ppm methanol in the feed to the main reactor. This unreacted fraction could potentially be oxidized once it passes through the catalyst sample. Hence, the formaldehyde conversion was calculated according to Eq. 9 to account for the unreacted methanol. are the concentrations of HCN after the reactor, formaldehyde in the gas feed, and methanol in the gas feed, respectively.
Aside from HCN, other products can also form from formaldehyde, with CO and CO 2 (CO x ) being the most notable. Hence, the selectivity of formaldehyde consumption toward the formation of HCN and CO x is given by Eq. 11 and Eq. 12.

Influence of HCHO on the SCR reaction
Fig . 2 shows the NO conversion obtained at 10 ppm NH 3 slip over V 2 O 5 /WO 3 -TiO 2 and the actual amount of NH 3 dosed. The performance of V 2 O 5 /WO 3 -TiO 2 in the absence of formaldehyde ( Fig. 2; SCR feed) was comparable with those reported in the literature for the same V loading and GHSV [28,30]. Regardless of the conditions applied, the trends followed a similar profile. The NO x conversion increased between 200 and 300 °C, and passed through a maximum value between 350 and 400 °C. Two factors can contribute to the decrease in the NO x conversion above 450 °C: the direct oxidation of NH 3 that becomes increasingly favorable at high temperature and the decreased amount of NH 3 on the catalyst surface due to decreased adsorption.
The presence of formaldehyde had a negative effect on the NO x conversion at the same NH 3 dosage ( Fig. 2; HCHO + SCR feed). As an illustration, nearly 95% of the NO x in the gas feed could be abated with 300 ppm NH 3 at 400 °C in the absence of formaldehyde, but this value dropped to 82% with formaldehyde. The decrease in the catalytic performance could be attributed to the depletion of the available NH 3 for SCR as it is consumed in the presence of formaldehyde.
At low temperatures, a lower NH 3 dosage was necessary to achieve 10 ppm of NH 3 slip because the catalytic activity was rather limited. Correspondingly, the NO x conversion was not affected significantly by the presence of formaldehyde, with less than 5% of catalytic performance being lost due to the introduction of formaldehyde at 250 °C. Although NH 3 could react with formaldehyde, it was still preferentially consumed by SCR. Hence, formaldehyde competed with NO x for reaction with NH 3 only when the temperature was sufficiently high (i.e., above 300 °C). These results imply that the activation energy for the side reaction is much higher than that of the SCR reaction.
To further investigate the effect of formaldehyde on the SCR reaction, the NH 3 dosage was readjusted to meet the 10 ppm slip NH 3 (i.e., compensation for the amount of NH 3 lost in the side reaction). As more NH 3 became available for SCR, the NO x conversion increased, but did not return to the original value obtained in the absence of formaldehyde. When the NH 3 dosage was adjusted from 300 to 350 ppm at 400 °C, the NO x conversion increased from 82% to 92%, but not to the initial 95% obtained in the absence of formaldehyde. This behavior implies that the side reaction consuming formaldehyde could take place at the same catalytic sites responsible for SCR. Fig. 3 shows the formaldehyde conversion over the same catalyst. Under a purely oxidative atmosphere ( Fig. 3; HCHO feed), formaldehyde conversion increased from 9% to 89% as the temperature was varied from 200 to 500 °C. CO was the major product observed, which reflects the oxidation activity of V 2 O 5 /WO 3 -TiO 2 catalysts. At 200 °C, traces of formic acid were also detected in the gas phase. This suggests that formic acid formation is the first reaction step of formaldehyde oxidation. Indeed, previous infrared spectroscopic studies have shown that surface formate species were formed almost instantaneously upon introduction of formaldehyde, and that they prevailed over surface formaldehyde species [32,33]. Nonetheless, formates do not desorb as formic acid significantly, suggesting that the formed surface species are reaction intermediates. Above 450 °C, traces of CO 2 were also observed in the gas phase.

Formaldehyde conversion
Formaldehyde conversion generally increased under SCR conditions ( Fig. 3; HCHO + SCR feed). The only exception was at 200 °C where it was slightly lower (5% vs. 9%), indicating that NH 3 probably inhibits the oxidation of formaldehyde at low temperature. As the temperature increased, more formaldehyde was consumed in presence of the SCR feed. The maximum difference of 24 % in formaldehyde conversion between the two feed conditions was recorded at 350 °C. This behavior suggests that formaldehyde reacted with NH 3 . At 500 °C, the two formaldehyde conversion levels were again very close. It is important to note that addition of 300 ppm NO in the absence of NH 3 (Table S-3) did not cause any change in formaldehyde conversion, demonstrating that NO is not involved in side reactions with formaldehyde.
When the NH 3 feed was adjusted to meet 10 ppm NH 3 slip ( Fig. 3; HCHO + adj. SCR feed), the formaldehyde conversion improved particularly in the temperature regime between 300 and 500 °C, indicating that the side reaction was further promoted by the increased NH 3 dosage. Hence, an increase in NH 3 concentration increases NO x conversion but promotes simultaneously the side reaction of NH 3 , effectively subtracting NH 3 from the main SCR reaction pathway.
Under similar conditions, TiO 2 was also able to oxidize formaldehyde to some degree, reaching 15 % and 25 % conversion at 500 °C in the absence and presence of NH 3 , respectively ( Figure S-1). Formaldehyde conversion was low likely because TiO 2 did not have appreciable redox activity in the temperature range studied.  Table 1 and Table S-2). The NH 3 dosages for SCR feed and HCHO + SCR feed are the same and therefore the respective curves perfectly overlap. Dotted lines refer to the NH 3 dosage. Fig. 3. Formaldehyde conversion over V 2 O 5 /WO 3 -TiO 2 in the presence and absence of SCR (experimental conditions listed in Table 1 and  Fig. 4 shows the concentration of HCN formed as a function of temperature in the same experiment presented in Fig. 2 and Fig. 3. In the absence of either formaldehyde or NH 3 , no HCN was observed ( Fig. 3; SCR feed and HCHO feed).

Formation of HCN
When formaldehyde and NH 3 were fed simultaneously, the HCN concentration increased from 3 to 59 ppm as the temperature was raised from 200 to 500 °C ( Fig. 4; HCHO + SCR feed). The steepest rise occurred between 250 and 300 °C, where the HCN concentration more than tripled.
When the NH 3 dosage was adjusted to meet the 10 ppm slip, even more HCN was produced ( Fig. 4; HCHO + adj. SCR feed). For example, HCN increased from 44 to 55 ppm when the NH 3 supply was changed from 300 to 350 ppm. This explains why the NO x conversion reported in Fig. 2 did not reach the same levels in the absence of formaldehyde, even if additional NH 3 was supplied. HCN production is therefore confirmed to be in direct competition with SCR, especially at high temperatures.
Even a non-reducible material such as TiO 2 showed mild HCN formation under similar conditions, the HCN concentration varying from 2 ppm at 350 °C to 11 ppm at 500 °C ( Figure S-2). While the presence of redox-active sites such as VO x increases HCN production significantly, the metal oxide support already provides a non-negligible baseline activity.
HCN formation over V 2 O 5 /WO 3 -TiO 2 was strongly temperaturedependent, but it remained constant when the O 2 content was varied from 2.5 to 15.0 vol% (Figure S-3). This suggests that excess O 2 does not cause the side reaction to stop in favor of the direct oxidation of formaldehyde. In fact, formaldehyde conversion and CO x formation were also found to be independent of O 2 content.
In contrast, water had a significant impact on HCN production. Formaldehyde conversion decreased from 99 % to 89 % as the water concentration increased from 0 to 12.5 vol% (Fig. 5). This might be due to the co-adsorption of water on the catalyst, which is also known to inhibit the SCR reaction by decreasing the adsorption sites for NH 3 , among other effects [34].
HCN selectivity also decreased similar to formaldehyde conversion, which suggests that formaldehyde requires close proximity with the active sites for HCN production. However, CO x selectivity followed an opposite trend. Without water, CO x formation was negligible, but CO x selectivity increased to 41 % at the highest water level investigated in the study. Water could transform formaldehyde into adsorbed formate species [32,33], which were already proposed to be intermediates for CO x formation in the previous sections. The positive effect of water on CO x formation further confirms this pathway. The same strong effect of water was observed also on TiO 2 . In the absence of water, 111 ppm HCN were produced from reaction of formaldehyde and NH 3 at 500 °C with practically full conversion of formaldehyde, while only 11 ppm HCN and 25 % formaldehyde conversion were observed with 10 vol% H 2 O (Figure S-4).
The increase in water concentration and the subsequent formation of formate species appeared to be detrimental for HCN formation, suggesting that it follows a different mechanism than CO x formation. To confirm that formate species were not relevant for HCN formation, formic acid was dosed together with NH 3 over V 2 O 5 /WO 3 -TiO 2 (Fig. 6). Similar to the case of formaldehyde, HCN formation from formic acid was also negatively affected by the presence of water. Under 10 vol% H 2 O -which is the usual water content in exhaust gas and the same condition applied in the measurements presented in Fig. 4 − HCN did not form in significant amounts; all of the measurements fell below the detection limit of 1 ppm. Even when the water content was decreased to 5 vol%, no HCN was observed. In fact, it is only in the absence of water  Table 1 and Table S-2). The HCN concentrations for SCR feed and HCHO feed are the same and the respective curves therefore perfectly overlap.  that formic acid formed HCN, but in rather limited amounts compare to the case of formaldehyde. This proves that HCN formation from formaldehyde cannot proceed through the formate route; instead, HCN more likely originates from the direct reaction of formaldehyde with NH 3 . Finally, it is important to note that HCN formation from formaldehyde and formic acid followed different behaviors with temperature. Whereas HCN formation increased continuously with temperature when formaldehyde was the reactant (Fig. 4), it experienced a maximum at 250 °C and then decreased rapidly when formic acid was used instead (Fig. 6). This is the same behavior observed in a previous study where ammonium formate was used as a reducing agent for SCR [20]. Hence, this confirms that two different mechanisms are responsible for HCN formation depending on the carbonyl precursor: a high-yielding route with formaldehyde that is active above 300 °C and a moderate one with formic acid that operates only below 300 °C.

Proposed reaction pathway
Based on the discussed experimental pieces of evidence, we propose the new reaction pathway for the formation of HCN over V 2 O 5 /WO 3 -TiO 2 in Fig. 7.

Oxidation of formaldehyde
When formaldehyde adsorbs onto the catalyst, it can readily transform into formate species due to the oxidizing atmosphere of the gas feed and the catalyst surface. This oxidative transformation occurs also over TiO 2 [32,33]. Indeed, some traces of the formate species can desorb as formic acid, which was detected in trace amounts at low temperature.
Formation of formic acid might also be possible through the Cannizzaro reaction [35], a base-catalyzed process resulting in the disproportionation of formaldehyde to methanol and formic acid, according to Eq. 13. On the contrary, no methanol was observed over V 2 O 5 /WO 3 -TiO 2 , suggesting that Eq. 13 does not contribute significantly to formic acid formation in the presence of redox-active sites. Hence, the direct oxidation of formaldehyde is proposed as the major route for formic acid production in this study. Formate species are short-lived, and are quickly oxidized into CO in a subsequent step (Fig. 7). Previous studies have shown that formic acid decomposes readily in the temperature range of 180−250 °C [36][37][38]. This is in line with our catalytic measurements showing traces of formic acid only at 200 °C, the lowest temperature investigated. However, CO was the major product of formaldehyde oxidation over V 2 O 5 /WO 3 -TiO 2 and only small amounts of CO 2 were detected at 500 °C.

HCHO + H O → CH 3 OH + HCOOH
Detection of low levels of H 2 above 350 °C in the absence of water over TiO 2 confirms that formaldehyde decomposes also into CO and H 2 (Figure S-5), but not over V 2 O 5 /WO 3 -TiO 2 .

Formation of HCN through formic acid and NH 3
The formed formic acid could then react with NH 3 to yield ammonium formate, which is the reverse of Eq. 5. In this situation, HCN is formed after two successive dehydration steps (Eqs. 6 and 7). This is the same route that was previously proposed for HCN production when ammonium formate was used as the NH 3 precursor instead of urea [20].
Our catalytic measurements with formic acid revealed that this route is only possible at low temperature, reaching maximum activity at 250 °C (Fig. 6). At higher temperatures, the direct decomposition of formic acid into CO and water becomes more favorable. Under these conditions, Eq. 5 also shifts toward the formation of formic acid and NH 3 and as a result, less ammonium formate and HCN can be formed. Furthermore, this route is strongly inhibited by the presence of water (Fig. 6). Hence, this reaction pathway cannot explain the high amounts of HCN observed at high temperatures and considerable amounts of water.

Formation of HCN through the direct reaction of formaldehyde and NH 3
Based on the above considerations, we propose that under the   [39], which is analogous to the reaction of aldehydes with water to form a geminal-diol or with an alcohol to form a hemiacetal [40,41].
Methanolamine can undergo two dehydration pathways that both ultimately lead to the production of HCN. The first one is oxidative dehydration, which requires at least a mild oxidation catalyst. Because of the excess O 2 in the feed and the fact that VO x supported on TiO 2 possesses oxidation activity [42], methanolamine can easily transform into formamide (Equation 15).
Formamide dehydrates to form HCN in an endothermic reaction (Δ r H° = 140 kJ·mol −1 ) according to Eq. 16. This reaction pathway is in agreement with the high levels of HCN observed as the temperature was increased.
The two intermediates, methanolamine and formamide, were not observed in this study and are thus supposed to be formed and consumed at the surface of the catalyst. It is also important to mention that this final dehydration step is shared with the reaction pathway starting from formic acid (Eq. 6 and Eq. 7). However, formic acid is very easily oxidized to CO, thus formamide is not formed in the first place.
The second possible dehydration pathway follows the Strecker synthesis of amino acids [43]. According to this reaction, methanolamine decomposes into methanimine and water, as confirmed by calculations (Eq. 17) [44].
Methanimine is a highly unstable molecule [45,46] and is therefore not detected as a product even at low temperatures. Instead, it can oxidize rapidly on the catalyst surface to yield HCN and water (Eq. 18).

Practical relevance
HCN forms under SCR conditions in the temperature range of 200−500 °C. This side reaction is highly undesired because of two reasons. First, HCN is a poisonous gas that harms human health and the environment. Second, the side reaction leading to HCN formation not only consumes NH 3 parasitically, but also retards SCR activity by up to 20 % (Figure S-6). This means that a larger catalyst bed or monolith brick would be needed to achieve the same reactivity in the absence of formaldehyde.
Formaldehyde and other potential carbonyl species in the gas feed result from the partial oxidation of hydrocarbons. Hence, these species must be fully oxidized to avoid potential side reactions before passing through the SCR catalyst compartment. For mobile systems, this is not really an issue because oxidation catalysts and NH 3 slip catalysts are typically installed before and after SCR catalysts, respectively. These materials are usually based on Pt and possess high activity for formaldehyde oxidation [47,48]. Even if some formaldehyde escaped oxidation and eventually formed HCN, the NH 3 slip catalyst could oxidize it prior to release to the environment. We have shown that Pdand Pt-based catalysts abate HCN effectively above 300 °C [49], which is the temperature region where maximum levels of HCN were detected. Nonetheless, both formaldehyde and HCN can resist oxidation at lower temperatures, and therefore still represent a potential issue.
While mobile exhaust treatment systems are in principle already well designed to prevent HCN formation or emission, not all stationary systems are. Many stationary SCR installations function as standalone units directly downstream of the engine. Where operation conditions of the engine allow formation of formaldehyde in the exhaust gas, a catalytic oxidation system must be installed upstream of the SCR unit to avoid HCN formation.
Currently, HCN is not monitored in the exhaust of vehicles and stationary plants. However, legislations are constantly evolving [50], and it is recommendable to include HCN in the lists of regulated emissions due to its hazardous effects.

Conclusions
HCN can form in significant amounts during the SCR treatment of formaldehyde-containing exhaust gases over V 2 O 5 /WO 3 -TiO 2 . This undesired reaction not only lowers the SCR activity and consumes NH 3 parasitically, but also results in the production of a highly poisonous substance. While the detailed mechanism is not yet conclusive, we provide evidence that HCN production occurs from the direct reaction of formaldehyde and NH 3 , and not through the formation of formic acid or ammonium formate. These results should serve as a basis for system designs and/or environmental legislation to prevent HCN emissions.

Declaration of Competing Interest
There are no conflicts of interest to declare.