The H₂O Effect on Cu Speciation in Cu-CHA-Catalysts for NH₃-SCR Probed by NH₃ Titration

Abstract

The present work is focused on the effect of water on NH₃ adsorption over Cu-CHA SCR catalysts. For this purpose, samples characterized by different SAR (SiO₂/Al₂O₃) ratios and Cu loadings were studied under both dry and wet conditions. H₂O adversely affects NH₃ adsorption on Lewis acid sites (Cu ions) over all the tested catalysts, as indicated by the decreased NH₃ desorption at low temperature during TPD. Interestingly, the NH₃/Cu ratio, herein regarded as an index for the speciation of Cu cations, fell in the range of 3–4 (in the presence of gaseous NH₃) or 1–2 (no gaseous NH₃) in dry conditions, in line with the formation of different NH₃-solvated Cu species (e.g., [Cu^II(NH₃)₄]²⁺ and [Cu^II(OH)(NH₃)₃]⁺ with gaseous NH₃, [Z₂Cu^II(NH₃)₂]²⁺ and [ZCu^II(OH)(NH₃)]⁺ without gaseous NH₃). When H₂O was fed to the system, on the contrary, the NH₃/Cu ratio was always close to 3 (or 1), while the Brønsted acidity was slightly increased. These results are consistent both with competition between H₂O and NH₃ for adsorption on Lewis sites and with the hydrolysis of a fraction of Z₂Cu^II species into ZCu^IIOH.

1. Introduction

NH₃/urea SCR is recognized worldwide as an effective technology in controlling automotive emissions to meet increasingly stricter NOx emission regulations [1,2]. Amongst state-of-the-art metal-promoted zeolite catalysts used in this process, Cu-exchanged chabazite (Cu-CHA) shows excellent activity and extraordinary hydrothermal stability over a broad temperature range under SCR conditions, triggering intensive research on the relationship between its unique structure and its DeNO_x performance [3,4,5,6,7,8,9,10,11]. In this regard, the SiO₂/Al₂O₃ ratio (SAR) and the Cu loading content are important structure indexes, which could affect the nature of Cu cations existing in the catalyst framework and, based on their nature, location and redox properties, further impact the SCR activity [3,4,5,6,7,8,9]. Two Cu species are recognized to be present in the Cu-CHA catalyst: Z₂Cu^II (doubly coordinated to the framework) and ZCu^IIOH (balanced by one framework negative charge), whose relative ratio can be predicted based on Si/Al and Cu/Al ratios [5,6,11]. The proportion of such Cu species could also be affected by several factors, for example, hydrothermal aging [4,6,12] or hydrated/dehydrated conditions [5,6]. It was also proposed, based on DFT calculations, DRIFTS, XAS and NH₃-TPD tests [4,5,8,13], that these two species are able to coordinate a different number of NH₃ molecules, thus influencing, in turn, the NH₃ storage capacity, a relevant parameter for the SCR process. In line with Paolucci et al. [5], our recent paper concluded that upon full catalyst saturation, the Z₂Cu^II species could coordinate up to four NH₃ molecules, while the hydroxylated Cu^II site could bind only three molecules [8]. The NH₃ adsorption chemistry on the Lewis acid sites can be summarized by the following reactions [13]:

Z₂Cu^II _(s) + 4NH_{3 (g)} → Z₂Cu^II(NH₃)_{4 (s)}

(1)

ZCu^IIOH _(s) + 3NH_{3 (g)} → ZCu^IIOH(NH₃)_{3 (s)}

(2)

When Cu-CHA catalysts were not fully saturated and Cu ions were not able to detach from the zeolite framework, however, we observed a stoichiometry different from reaction (1) and (2) for the two copper species: Z₂Cu^II ions adsorb two NH₃ molecules while only one NH₃ is bound to ZCu^IIOH, in agreement with Luo et al. [4].

NH₃ can also be adsorbed on the Brønsted acid sites typical of the zeolite framework, according to the adsorption chemistry reported below [12]:

NH_{3 (g)} + ZH _(s) → ZNH_{4 (s)}

(3)

Herein, we carried out NH₃ adsorption + TPD tests under both dry and wet conditions over several Cu-CHA samples with different SARs and Cu loadings, which therefore contain distinct fractions of the two Cu species populations and different Brønsted sites [5,8]. The aim of the present investigation is to verify if H₂O affects the Lewis and Brønsted acidity in chabazite-based materials and the possibility to use NH₃-TPD as a probe technique for Cu speciation. In the literature, this method was implemented to characterize Cu-CHA catalysts in dry environments only in [8] or in the presence of water only in [4,12]. On the contrary, to assess the role of H₂O in NH₃ adsorption, we compared the performance of different samples under both wet and dry conditions.

2. Results and Discussion

An example of an NH₃ adsorption/desorption run is provided in Figure 1A, depicting the three stages of the test, namely isothermal adsorption, isothermal desorption, and TPD. The run was performed after having pre-oxidized the sample at 550 °C. The aim was not only to evaluate the NH₃ storage capacity, but more specifically to identify the different sites present in the catalyst [8].

For this purpose, we analyze the TPD phase. As shown in Figure 1A, the temperature ramp featured a release of NH₃ together with a small N₂ peak (about 25 ppm). The integral of the N₂ released (i.e., ≈0.68 µmol) is in line with the full reduction in Cu^II ions (i.e., 4.3 µmol of Cu loaded) by adsorbed NH₃ according to the following stoichiometries [13,14]:

6Cu^II _(s) + 2NH_{3 (g)} → N_{2 (g)} + 6H _(s) + 6Cu^I _(s)

(4)

6Cu^IIOH _(s) + 2NH_{3 (g)} → N_{2 (g)} + 6H₂O _(g) + 6Cu^I _(s)

(5)

Figure 1B displays the NH₃ and N₂ traces during the TPD, whereas the corresponding integral balances are reported in

Table 1. Integral balances for the NH₃ adsorption + TPD test over 1.7Cu-CHA25 shown in Figure 1.

1.7Cu-CHA25
Cutot (ICP) (µmol)	4.3
NH3 adsorbed (µmol)	23.4
NH3 physisorbed (µmol)	8.0
NH3 chemissorbed (µmol)	14.1
N2 produced (µmol)	0.7
N balance error w/o N2 (%)	5.6
N balance error with N2 (%)	0.1

. The N balance error improves significantly if the production of nitrogen is taken into account. In line with this result, Borfecchia et al. showed by means of XAS spectroscopy and UV-vis-NIR experiments that at the end of the TPD, after NH₃ adsorption, the Cu sites in their Cu-CHA catalysts were almost completely reduced [13].

In line with the literature, the TPD curve in Figure 1B showed two distinct NH₃ release peaks, ascribed to energetically different populations of acidic sites [4,8,15] (Figure 2). The physical nature of the different sites was estimated via Gaussian deconvolution of the NH₃-TPD trace, to which the contribution of N₂ according to the reactions discussed above (4) and (5) was added (green line in Figure 1B and Figure 2). The N₂ trace was not measured in our previous investigation [8]; however, it is important to include this contribution to quantify more accurately the various Cu sites taking part in the SCR chemistry.

As established in several works within the literature [4,8,12,16], the first peak in Figure 1B is related to the NH₃ adsorbed onto both Cu species and a small amount of Al extra-framework (quantified in [8] and reported in Table S1), while the second one is attributed to NH₃ desorption from Brønsted sites associated with the bare zeolite framework. It is notable that the nitrogen production in Figure 1A was well aligned with the low temperature peak of the NH₃-TPD curve, implying that mostly Lewis NH₃ is responsible for the reduction in Cu^II cations into Cu^I according to (4) and (5).

2.1. H₂O Effect with Changes in Cu Loading

Figure 3 shows the Cu loading effect on the NH₃-TPD data collected over each catalyst under both dry and wet conditions.

Figure 3A refers to the simplest case, i.e., the TPD profiles over the unpromoted chabazite catalyst with SAR 25. The curves obtained in the presence and absence of H₂O are almost overlapped, showing a single desorption peak centered at about 440 °C with a similar intensity in both tests. This peak corresponds to the desorption of NH₃ stored onto the Brønsted acid sites of the zeolite [11,17,18,19]. The small shoulder visible at 250 °C under dry conditions (black line) is linked to the extra-framework aluminum species, which confers a slight Lewis acidity to the catalyst [6,17,18,19,20,21]. This peak disappeared under wet conditions, in accordance with findings in the literature [6,20,21].

Interestingly, H₂O significantly affects the weakly bound NH₃, since H₂O molecules are stored preferentially on Lewis acid sites. A clear confirmation of this behavior was obtained comparing the NH₃ desorption profiles of the copper-exchanged samples under dry and wet conditions. Figure 3B–D show the TPD results for the zeolite catalysts with the same SAR (25) and loaded with 0.7, 1.7 and 2.1% w/w copper, respectively. The presence of H₂O considerably lowered the NH₃ storage capacity of all catalysts, affecting mostly the low-T peak, associated predominantly with Cu ions. The high-temperature Brønsted NH₃ peak was less affected by H₂O. The negative effect of H₂O on the NH₃ storage capacity of Lewis sites was confirmed by the faster dynamics during the catalyst saturation in the isothermal adsorption phase, when the samples were previously treated with H₂O (not shown): indeed, the dead time recorded during this phase was reduced from 530–860 s under dry conditions to 380–595 s under wet conditions. In addition, the integral of the TPD curves (NH₃ + 2N₂) was lower in the presence of H₂O even considering the negligible contribution of the extra-framework Al sites (Tables S1 and S2).

2.2. Water Effect with Changes in SAR

Results were also collected over Cu-CHA catalysts with fixed Cu loading (1.7% w/w) and different SARs (10, 13, 22, 25): the corresponding TPD curves are shown in Figure 4. Under dry conditions, the differences in the relative intensities of the low and high temperature peaks indicate that on increasing the SAR, the overall stability of the adsorbed NH₃ molecules decreased [8]. In fact, in line with our previous studies, intensities of the low-T peaks were lower than those of the high-T peaks for high SARs (22, 25), and the opposite trend was seen at low SARs (10, 13), where the intensities of the low-T peaks were higher [8]. In all cases, the overall amount of chemisorbed NH₃ calculated from the integration of the TPD curves was found fairly constant upon changing the SAR, in the range of 950–1120 μmol/g.

A similar effect was observed in presence of H₂O (light-blue lines). Comparing the TPD under wet and dry conditions, it is apparent that H₂O reduced ammonia storage (Figure 4). Similarly to the data presented in the previous section, the low-temperature desorption feature diminished as a result of H₂O addition. Furthermore, the overall chemisorbed NH₃ decreased to 830–880 μmol/g for all the catalysts. Accordingly, the deadtime recorded during the adsorption phase was affected by the presence of water (not shown), decreasing from about 800 to 650 s.

2.3. H₂O Effect on Cu Speciation

At this point, we used Gaussian deconvolution of the bimodal TPD profiles to compare the amount of NH₃ to Cu atoms and to Brønsted acid sites in wet and dry atmosphere. Notice that the NH₃ legated to the Al extra-framework was completely lost in the presence of water, as apparent in Figure 3A for H-CHA (SAR 25) and in Figure S1 for H-CHA (SAR 13).

The results in Figure 5 and in Table S1 indicate that by cofeeding water: (i) the quantity of NH₃ stored on Brønsted sites slightly increased, while (ii) the Lewis NH₃ decreased. This trend could be related to the hydrolysis of Z₂Cu^II species, reaction (6) [6,22]:

Z₂Cu^II _(s) + H₂O _(g) → ZCu^IIOH _(s) + ZH _(s)

(6)

In fact, according to this reaction, the transformation of Z₂Cu^II into ZCu^IIOH implies a loss in Cu-NH₃ ligands. Indeed, the Z₂Cu^II species can bind up to four NH₃ molecules, while the hydroxylated Cu ions can coordinate only three molecules [5,8,23].

According to our deconvolution results, however, the increment in Brønsted NH₃ in wet conditions was less than the decrease in Lewis NH₃ for some samples. This behavior could be explained by a further loss of Lewis NH₃ due to the competitive adsorption of water on the Cu sites [20], as indicated by the decrease in the overall NH₃ storage capacity (i.e., the sum of Brønsted and Lewis NH₃ in Figure 5).

The fraction of Z₂Cu^II transformed into ZCu^IIOH was estimated by the following equation:

$$\frac{\mathrm{Amount} \mathrm{of} {\mathrm{Z}}_2{\mathrm{Cu}}^{\mathrm{II}} \mathrm{hydrolised} }{\mathrm{initial} \mathrm{amount} \mathrm{of} {\mathrm{Z}}_2{\mathrm{Cu}}^{\mathrm{II}}\left(\mathrm{Table} 2\right)} = \frac{N{H}_3 \mathrm{Brønsted} \mathrm{wet} - N{H}_3 \mathrm{Brønsted} \mathrm{dry}}{\mathrm{initial} \mathrm{amount} \mathrm{of} {\mathrm{Z}}_2{\mathrm{Cu}}^{\mathrm{II}}\left(\mathrm{Table} 2\right)}\%$$

(7)

From the results in

Table 2. Estimates of the fraction of Z₂Cu^II hydrolyzed according to reaction (3).

Sample	Cu Loaded, µmol	Z2CuII from NO2 Adsorption, µmol [8]	NH3 Brønsted dry, µmol	NH3 Brønsted wet, µmol	Z2CuII Hydrolised/Initial Z2CuII, %	Z2CuII Hydrolised/Cutot, %
0.7% w/w Cu, SAR = 25	1.8	0.8	8.02	8.38	45	20
1.7% w/w Cu, SAR = 25	4.3	1.8	7.9	8.67	43	19
2.1% w/w Cu, SAR = 25	5.2	1.7	7.07	8.07	59	19
1.7% w/w Cu, SAR = 10	4.3	2.4	6.20	7.54	56	31
1.7% w/w Cu, SAR = 13	4.3	3.1	8.30	9.50	39	28
1.7% w/w Cu, SAR = 22	4.3	1.8	8.00	9.28	71	30

it can be noticed that only a fraction of Z₂Cu^II (40–70%) was hydrolyzed after the wet NH₃ adsorption. Furthermore, even if 1.7Cu-SAR25 and 1.7Cu-SAR22 are characterized by similar speciation and Cu/Al ratio, they display different behaviors in a wet atmosphere, more specifically the 1.7Cu-SAR22 shows a more enhanced hydrolysis of the Z₂Cu^II sites. One possible parameter affecting this reaction could be the distribution of the Al atoms in the zeolite framework. In addition, two different types of the Z₂Cu^II sites are viable, with para and metal configuration, respectively, which show distinct features in EPR spectra and distinct reactivity under Standard SCR conditions [24]. In principle, these species could be associated with a different response to the hydrolysis reaction. A dedicated investigation of the hydrolysis reaction is ongoing.

At this point we calculated the NH₃/Cu ratio, taking into account the area under the low-T TPD peak, proportional to the NH₃ stored only on Cu cations, and the number of copper atoms known from the weight fraction of Cu in each sample (

Table 3. Characterization of the tested catalyst samples [8].

Sample	Cu Loaded, µmol	Cu/Al Ratio, -	NO2 Adsorption Experiment, µmol	ZCuIIOH, %	Z2CuII, %	NOx Conversion at 225 °C, %
0.7% w/w Cu, SAR = 25	1.8	0.11	1.0	55	45	16
1.7% w/w Cu, SAR = 25	4.3	0.24	2.5	58	42	91
2.1% w/w Cu, SAR = 25	5.2	0.29	3.5	67	33	96
1.7% w/w Cu, SAR = 10	4.3	0.12	1.9	44	56	58
1.7% w/w Cu, SAR = 13	4.3	0.17	1.2	28	72	67
1.7% w/w Cu, SAR = 22	4.3	0.22	2.5	58	41	83

). Figure 6 illustrates how this ratio varies with the Cu/Al ratio.

Under dry conditions, the experimental data were in accordance with the literature [8,23]: the ratio was in the range 1–2 when only the Cu-NH₃ desorbed during the TPD was considered, and in the 3–4 range when the physisorbed NH₃ was also considered (area highlighted in orange, Figure 1A). In particular, the increase in Cu/Al lead to a decrease in the NH₃/Cu ratio, from a value close to 4 (or 2 considering only the TPD phase) for the catalysts with the lowest SAR and Cu loading, to a value closer to 3 (or 1 considering only the TPD phase) for the other tested samples (Figure 6). When water was added to the system, we observed a shift of all the NH₃/Cu ratios evaluated from the TPD to values closer to 3 (or 1 considering only the TPD phase). These results are consistent both with competition between H₂O and NH₃ adsorption onto Lewis sites and with the hydrolysis of a fraction of Z₂Cu^II species into ZCu^IIOH according to reaction (6), as discussed above.

At this stage of the discussion, our opinion is that analysis of the NH₃-TPD data is not the best method to probe the Cu speciation. First, quantification of the Lewis and Brønsted acid sites requires deconvolution of the NH₃-TPD curves. Second, during the temperature ramp the Cu atoms are reduced, producing N₂; hence, it would be important to include its trace in the TPD plot to accurately analyze the experimental data. Finally, under wet conditions the Cu species are modified by the hydrolysis reaction. The occurrence of these phenomena complicates the interpretation of experimental TPD traces, thus it is suggested not to solely use this method to quantify the fraction of Z₂Cu^II and ZCu^IIOH species in the catalysts. The Cu speciation could be quantified by means of simpler techniques such as, for example, NO₂ adsorption + TPD and H₂ TPR, as discussed in [8,11,25]. However, NH₃-TPD runs under wet conditions provide useful experimental evidence of the hydrolysis reaction over Cu-CHA materials.

3. Material and Methods

For this study, we investigated two sets of Cu-exchanged chabazite (Cu-CHA), research catalysts supplied in the form of powders by Johnson Matthey. One set was purposely characterized by different SiO₂/Al₂O₃ (SAR) ratios (10, 13, 22, 25), with fixed Cu loading = 1.7% w/w. The other had different Cu loadings (0, 0.7, 1.7, 2.1% w/w, with fixed SAR = 25). The textural properties of some samples, comprising the BET surface area and the micropore volume, are reported in Table S3 and are typical of commercial CHA based materials [26,27]. The samples were conditioned by heating them up to 600 °C at 5 °C/min and holding the maximum temperature for 5 h, while 10% O₂ and 10% H₂O were fed in to remove all the impurities and the species that may be formed on the sample due to contact with atmosphere.

Sixteen milligrams of catalyst powder, sieved to obtain an average particle size of 90 micron and diluted up to 130 mg with cordierite, was loaded into a quartz flow reactor (6 mm ID) inserted in a vertical electric furnace. The temperature was controlled by a K-type thermocouple directly immersed in the catalyst bed.

Helium was used as balance gas in all the micro-reactor runs. Gases were controlled using mass flow controllers (Brooks Instruments, Hatfield, PA, USA). Water vapor was fed in by means of a saturator and the water concentration was controlled by adjusting the saturator’s temperature according to Antoine’s Law. NH₃ was fed in using a 6-way valve, enabling step changes in the reactant concentrations. Temporal evolution of the species at the reactor outlet was followed by a mass spectrometer (QGA Hiden Analytical, Warrington, UK) and a UV analyzer (ABB LIMAS 11 HW, Zurich, Switzerland) arranged in a parallel configuration, which allowed the simultaneous measurement of all the species involved.

The interaction between NH₃ molecules and the catalyst was studied by performing NH₃ adsorption + temperature programmed desorption (TPD) runs. The catalysts were saturated using 500 ppm of NH₃ and balance He (GHSV = 266,250 cm³/(g_cat*h) STP), both under dry and wet conditions (5% w/w H₂O), at 150 °C. After that, the catalysts were purged with He (or He + H₂O) for 1 h to desorb weakly bound ammonia, and then heated in He only (or He + H₂O) from 150 °C up to 550 °C at a rate of 15 °C/min to release the strongly adsorbed molecules. Prior to each experiment, in order to control the oxidation state of the catalysts, the samples were pre-treated at 550 °C, feeding 8% O₂ in He for 1 h, then the adsorption temperature was reached by cooling down under the same gas feed, with a final purge in He for 15 min.

The catalysts were characterized in our previous work [8], the amount of Cu loaded and the Cu speciation is reported in Table 3. In particular, NO + NH₃ TPR and ICP-MS analysis were used to quantify the number of reducible Cu^II sites. Meanwhile, the NO₂ adsorption -TPD protocol was implemented to assess the amount of ZCu^IIOH species [8]. The samples show the typical deNOx activity of CHA based materials as apparent from the NO conversions in Standard SCR conditions reported in Table 3 (i.e., GHSV = 937,500 cm³/h/g_cat (STP), g_cat = 0.016 g, NH₃ = 500 ppm, NO = 500 ppm, H₂O = 5%, O₂ = 8%, T = 225 °C).

4. Conclusions

Herein, we have studied the effect of H₂O on the NH₃ storage onto different Cu-CHA catalysts, using standard NH₃ adsorption/TPD experiments. According to our preliminary results, during the temperature ramp of the TPD run the adsorbed NH₃ fully reduces the initially oxidized Cu sites, releasing a stoichiometric amount of N₂. This effect is often overlooked in the literature but impacts the quantification of Cu-related and Brønsted NH₃ adsorption sites.

On increasing both the SAR and Cu loading, we found the adsorbed NH₃/Cu ratio decreased from 4 to 3 when H₂O was absent from the gas mixture (and from 2 to 1 in the absence of gaseous NH₃), which correlates nicely with the corresponding expected increase in the proportion of ZCu^IIOH over Z₂Cu^II species. Upon addition of water to the feed stream, all the tested catalysts showed NH₃/Cu ratios closer to 3 (or to 1 in the absence of gaseous NH₃). These results are consistent both with the hydrolysis of a fraction of Z₂Cu^II species to form ZCu^IIOH, even though competition between H₂O and NH₃ for adsorption onto Lewis sites cannot be ruled out.

This work is novel because it includes both the Cu reduction and the hydrolysis reaction in the analysis of the NH₃-TPD data. However, the occurrence of these phenomena complicates the interpretation of the experimental TPD runs, thus it is suggested not to solely use this method to probe the Cu speciation in Cu-CHA catalysts [8].