Study of Composite Photocatalysts Obtained by Photodeposition of Platinum on Layered Oxide HCa₂Nb₃O₁₀ and Its Inorganic–Organic Hybrids

Abstract

In recent decades, layered perovskite-like oxides have been intensively investigated as prominent photocatalysts for water splitting as a method of hydrogen production. In many previous papers, it was shown that deposition of platinum on an oxide sample dramatically increases photocatalytic activity. Nevertheless, little research was conducted to reveal the localisation of platinum in layered oxides; either it is located on the surface or in the interlayer space. In the present work, an attempt to answer this question is made. An HCa₂Nb₃O₁₀ layered perovskite-like oxide was modified with platinum by photoreduction of H₂PtCl₆ and then was intercalated with n-alkylamines (R = Me, Bu, Oc). Moreover, another set of samples were prepared by intercalating the amines first into HCa₂Nb₃O₁₀, followed by Pt deposition. Besides conventional methods for sample characterisation, we measured the kinetics of Pt dissolution during aqua regia etching of the samples, hoping that the rate of platinum dissolution would provide some information about its localisation. It was shown that in HCa₂Nb₃O₁₀ modified with platinum, less than 20% of the platinum is located on the surface, and that in the case of HCa₂Nb₃O₁₀ intercalated with amines, an even smaller amount of platinum attaches to the surface. Moreover, Pt in HCa₂Nb₃O₁₀ intercalated with amines was found to be significantly more stable against aqua regia treatment than in HCa₂Nb₃O₁₀ decorated with platinum directly.

1. Introduction

Photocatalysis is one of the ways to generate hydrogen as a main component in the hydrogen energy paradigm. For decades, the scientific community has been conducting extensive research in this field, developing more prominent photocatalysts and enhancing their characteristics. Among the many types of photocatalysts, we would like to highlight layered perovskite-like oxides, compounds formed by alternating layers of negatively charged perovskite blocks and positively charged ions (metal, hydrogen or organic ions). These compounds are of interest because they demonstrate high levels of photocatalytic activity in the reaction of water decomposition compared to ordinary oxides [1,2] and offer a wide range of soft-chemistry reactions (intercalation, ion-exchange, etc.) and thus can be tuned in different ways [3,4,5].

As has been shown in many works [6,7,8], modification of oxides with platinum leads to an order of magnitude increase in photocatalytic activity. For layered perovskite-like oxides, this effect was also observed [9,10,11]; however, to the best of our knowledge, there is no widely accepted consensus on the localisation of platinum in layered perovskite-like oxides.

One of the most common methods for platinum deposition on oxides is photocatalytic reduction. For instance, this method was used in [12]. In this work, platinum was photocatalytically deposited onto exfoliated layers of KCa₂Nb₃O₁₀ by reduction of H₂PtCl₆ under UV light, after which the layers were restacked. It was shown that platinum forms agglomerates approximately 10 nm in size on the surface of the sheets.

Another method of obtaining Pt particles is sample impregnation with Pt-containing compounds followed by hydrogen reduction. This method was used in [13] where platinum was reduced from [Pt(NH₃)₄]Cl₂ in two different ways: onto exfoliated nanosheets of KCa₂Nb₃O₁₀ followed by restacking and onto already restacked nanosheets. In the first case, Pt formed 1 nm particles in the interlayer space and larger particles on the surface of the oxide in the second case. Similar results were obtained in [14] for HPb₂Nb₃O₁₀ and H₂PtCl₆ as a precursor. However, when an explicit comparison between methods employing [Pt(NH₃)₄]Cl₂ and H₂PtCl₆ was performed, the outcome was different. In that work [15], the higher XPS signal for platinum after H₂PtCl₆ treatment than after [Pt(NH₃)₄]Cl₂ led the authors to conclude that [Pt(NH₃)₄]²⁺ cations intercalate in the interlayer space, whereas [PtCl₆]²⁻ anions are repelled from there and are deposited on the surface.

Uncertainty about cocatalyst localisation also appears for NiO particles, which are usually formed by intercalation of Ni²⁺ ions followed by reduction by H₂ and reoxidation by O₂. As stated in [16], NiO doping of A₂La₂Ti₃O₁₀ oxides leads to the formation of NiO particles primarily on the surface. At the same time, in another study [17], NiO particles were found to be located in the interlayer space, since the XPS signal intensity from Ni in Ni-loaded K₄Nb₆O₁₇ was significantly lower than for KNbO₃, which does not have a layered structure.

Overall, one can find contradictory information about Pt localisation in layered perovskite-like oxides. This is an especially important problem to solve, as it can shed light on the mechanisms of the photocatalytic process.

It has previously been noted [18] that the intercalation of organic molecules such as amines or alcohols into the interlayer space can raise the photocatalytic activity of samples. Moreover, if one modifies such organic–inorganic hybrids with platinum, photocatalytic activity usually increases even further, which begs the question: what is accountable for the growth of photocatalytic activity—platinum, amine molecules, or some interaction between them? For example, the presence of amine molecules can indirectly influence photocatalytic activity by influencing the platinum deposition process. Since it is not yet entirely clear where two half-reactions take place—on the surface or in the interlayer space—it is not obvious how different localisation of platinum would affect photocatalytic activity.

It might be that the hydrogen reduction process takes place in the interlayer space, and then the deposition of platinum there would be favourable. In this case, intercalation of the amine before platinisation leads to an increase in interlayer distance and, on the one hand, may ease deposition of platinum there; however, on the other hand, it may also impede this process if the amine densely fills the interlayer space.

If the reduction process takes place on the surface, then the deposition of platinum there would be beneficial.

For this reason, the objective of the present work was to reveal the localisation of platinum in different intercalation products of the layered perovskite-like oxides and establish a connection between two processes: intercalation of the amine and platinum deposition. In particular, we investigated derivatives of the HCa₂Nb₃O₁₀ (HCN₃) niobate. It was of interest not only to determine where platinum is localised if one deposits it on this oxide but also to reveal how the intercalation of organic molecules (in our case, amines) influences its localisation. Moreover, the order of platinisation and intercalation steps might have affected the resulting product as well. Hence, we also investigated this factor. For these purposes, two series of samples were synthesised. The first one is the initial HCN₃ niobate with platinum deposited on it (HCN₃/Pt), which was then intercalated with amines (HCN₃/Pt×R). The second set of samples was prepared in reverse order: the HCN₃ was firstly intercalated with n-alkylamines (HCN₃×R), and then platinum was deposited on it (HCN₃×R/Pt). The synthesis sequence is schematically depicted in Figure 1.

Figure 1. Synthesis scheme of the samples mentioned in the present work.

To determine the localisation of platinum, alongside physical methods of analysis, we applied acid treatment of the samples with aqua regia because it is capable of dissolving platinum while not damaging niobate blocks. We hypothesised that if platinum dissolves slowly, then this means it is located predominantly in the interlayer space (although strictly speaking, this may be not true). The standard sample, which contains only surface platinum and against which the “slowness” of dissolving is measured, is platinised Nb₂O₅. It is worth noting that a similar approach to determine platinum localisation has already been used by other researchers [19]. In that work, K₄Nb₆O₁₇ niobate modified with platinum was treated using boiling aqua regia for one hour. Then, the samples were completely dissolved before and after aqua regia treatment and analysed using an ICP method. It was found that in the samples prepared by [Pt(NH₃)₄]²⁺ intercalation followed by hydrogen reduction, the amount of dissolved platinum was smaller than in the case of [PtCl₆]²⁻ usage. Thus, similar to other works that we mentioned above, the authors concluded that in the samples prepared with [Pt(NH₃)₄]²⁺, platinum is located in the interlayer space, whereas for [PtCl₆]²⁻, it is situated on the surface.

2. Results and Discussion

2.1. Characterisation of the Samples

2.1.1. XRD Data

To determine the phase composition of the samples and their unit cell parameters, the powder XRD method was used. XRD patterns of the previously mentioned layered oxides are presented in Figure 2.

Figure 2. XRD patterns and unit cell parameters of initial KCN₃, HCN₃, HCN₃/Pt and amine intercalation products.

XRD patterns of KCN₃ and HCN₃ were in agreement with the existing literature, and they were also indexed with small deviations. However, in the case of HCN₃/Pt, one peak at 30° remained unidentified, and the (002) signal was identified as a back-shoulder peak; this sample probably contained two phases. All major peaks in the diffractograms of the products of amine intercalation were indexed. Turning to the data provided by these patterns, it must be stated that as amine length increases, the interlayer distance (d) (Figure 2) expands. This indicates an amine entering the interlayer space. Moreover, the a and b parameters hardly change because they are related to the Nb-O octahedron size, which does not depend on processes in the interlayer space.

As was shown in [20], HCN₃ can exist in several forms that have different amounts of intercalated water. The form of HCN₃ that was obtained in this work is a so-called α-form with a high amount of interlayer water compared to β- and γ-forms. The XRD pattern of HCN₃/Pt is similar to that of HCN₃ but has peaks from other HCN₃ forms (β or γ). In order to fix this issue, the sample was hydrated in humid air for 2 weeks to obtain a fully hydrated form, as in the case of HCN₃. However, it did not lead to the full elimination of the peaks from other phases. Nevertheless, we decided to proceed with the synthesis sequence because any subsequent reaction with HCN₃/Pt took place in aqueous media, and the hydration process would have taken place alongside any other.

If one compares XRD patterns of HCN₃×R with HCN₃×R/Pt (see Figure S1), they can see that in the case of R = Bu, Oc, the corresponding samples underwent some changes due to the photocatalytic process. In particular, the peaks related to the interlayer space distance (00n) shifted towards greater 2θ values. This may indicate that the sample lost some amount of amine. It is worth mentioning that HCN₃×Me/Pt oxide is similar to HCN₃×Me. Thus, HCN₃×Me is more stable during Pt photodeposition than HCN₃×Bu and HCN₃×Oc.

The samples obtained in a reversed order, i.e., with platinisation preceding intercalation, were found to contain multiple phases. This is proven by the doubling of the peaks at those angles where only a single peak is observed for HCN₃/Pt. Nevertheless, the sample with methylamine HCN₃/Pt×Me turned out to be monophasic and had an XRD pattern close to that of HCN₃×Me.

2.1.2. IR and Raman Spectroscopy Data

Organic presence in the samples is confirmed by IR and Raman spectra. In both of them, characteristic adsorption bands of organic bonds can be found (Figure 3). In these spectra, one can observe lattice modes of the oxide in a low-frequency area (500–1000 cm⁻¹) [21]. Around 1500 cm⁻¹, bending of C-H and N-H bonds in amine molecules can be seen [22], as well as an H-O-H band of the intercalated water. Stretching vibrations of N-H and O-H bonds are located in a high-frequency region around 3000 cm⁻¹ and above [23]. The presence of all these characteristic bands confirms that organic molecules indeed intercalated into the samples, and at least some part of them remained there even after the platinisation of HCN₃×R (see Figure S2).

Figure 3. IR spectra of HCN₃ and the products of amines intercalated into it (HCN₃×R).

Raman spectra provide the same information as infrared spectra about the presence of organic molecules in the samples (Figure 4). Nevertheless, it should be highlighted that these spectra have quite intense background noise. This can probably be attributed to the presence of Pt, because as one can notice, there is no high noise level in Raman spectra after the samples were treated with aqua regia and Pt was eliminated. Intensities of vibrational bands related to organic molecules significantly decreased after aqua regia treatment, which indicates that there are no more organic molecules inside the interlayer space, at least in the form in which they existed before the aqua regia treatment.

Figure 4. Raman spectra of platinised organic hybrids of HCN₃ before (A) and after (B) aqua regia treatment.

2.1.3. Results of CHN and TG Analyses

In order to determine the amount of intercalated amine and water molecules in the samples, CHN and TG analyses were performed. During heating of the samples even slightly above ambient temperature, layered oxides begin to lose weakly bonded water molecules. As temperature increases, interlayer water and amines escape the sample, which explains the significant mass loss observed on TG curves in the temperature range of 50–500 °C (Figure 5). The final product of all these transformations is Ca₂Nb₃O_9.5. We would also like to highlight one peculiar detail concerning sample thermal decomposition. Specifically, the samples not containing platinum start gaining mass at some point after 500 °C, whereas the samples with Pt do not behave in this way, and their mass steadily decreases. One of the possible explanations for this phenomenon (which requires more thorough research) is that at high temperatures, the oxidation proceeds incompletely, leading to the preservation of some oxidation products inside the interlayer space or on the surface as soot; meanwhile, the presence of Pt facilitates a complete oxidation process towards CO₂ formation. Nevertheless, the fact that the presence of Pt influences processes occurring in a sample during TG measurements supports the idea that Pt is located in the interlayer space.

Figure 5. TG curves of HCN₃, its platinised form and products of amines’ intercalation (HCN₃×R).

TG curves of platinised organic hybrids HCN₃×R/Pt and HCN₃/Pt×R are presented in Figure S3 and are similar for a given R except R = Oc. Mass loss in the case of HCN₃/Pt×Oc is smaller than for HCN₃×Oc/Pt, which indicates that the amount of octylamine in the sample decreased during platinisation.

Due to the insufficient precision of CHN analysis and aliasing of mass loss stages during thermogravimetric analysis, it is not feasible to draw reliable conclusions about sample compositions from these two methods if one uses them separately. However, one can use combined data from both of them to calculate x and y in the general formula HCa₂Nb₃O₁₀∙xH₂O∙yRNH₂ more precisely. The corresponding procedure with specific formulas is described in detail in [24]. The values of x and y obtained in this way are listed in Table 1 In the HCN₃×R oxides, amine molecules are intercalated in a 1:1 ratio with respect to H⁺ in HCN₃, which is in agreement with our previous results [25]. However, after the platinisation of these samples, the amount of amine decreases, which confirms the XRD data discussed before. The most significant loss is observed for HCN₃×Bu. Methylamine is the most stable under platinisation conditions among these three samples, while the loss of octylamine is lower than that of butylamine. A possible reason may be the presence of a longer alkyl chain and, as a result, stronger alkyl-alkyl interactions between octylamine molecules, as well as greater hydrophobicity. This may ultimately lead to greater stability in an aqueous environment. In general, our set of samples was found to be highly hydrated. Similar to HCN₃×R/Pt oxides, HCN₃/Pt×R compounds also contain fewer amine molecules than HCN₃×R, but, in this case, this can apparently be attributed to the modified interlayer space in HCN₃/Pt that underwent some changes during the photodeposition of platinum.

Table 1. Compositions of the samples with the general formula HCa₂Nb₃O₁₀∙xH₂O∙yRNH₂. * after aqua regia treatment for 5 min.

Sample	x	y	K	Ca	Ca *
HCN3	1.7	0	0.09	1.94
HCN3/Pt	1.1	0	0.09	1.97	1.81
HCN3×Me	0.5	1	0.09	1.94
HCN3×Bu	1.3	1	0.10	1.96
HCN3×Oc	0.6	1.2	0.09	1.94
HCN3×Me/Pt	0.6	0.9	0.08	1.94	1.81
HCN3×Bu/Pt	1.6	0.6	0.10	1.97	1.86
HCN3×Oc/Pt	0.0	0.8	0.09	1.99	1.85
HCN3/Pt×Me	1.1	0.6	0.10	1.94	1.74
HCN3/Pt×Bu	1.4	0.6	0.08	2.05	1.79
HCN3/Pt×Oc	2.0	0.8	0.09	2.03	1.68

2.1.4. EDX Data

Energy-dispersive X-ray analysis was performed to verify a K:Ca:Nb ratio in all of the oxides and to try to measure Pt content in the samples (Table 1). However, it turned out that Pt produces almost no signal and was hardly detected by EDX (the maximum detected atomic fraction of Pt was equal to 0.2 at.%). From the data obtained, one can conclude that the amount of Ca in all the samples is close to the proper value of 2 per formula unit and that only around 10% of the original amount of potassium ions remained inside the interlayer space after the protonation process.

EDX analysis detected no Pt and K in the samples after processing with aqua regia. Hence, it can be concluded that Pt and K were eliminated from the sample during the 2-week etching. Moreover, we noted an interesting pattern: Ca content indicated by EDX systematically decreased for all of the samples after aqua regia treatment from a proper value of 2 to roughly 1.8 (Table 1). This result is in favour of predominant Ca elimination from the sample. It was also confirmed by the ICP-OES analysis of the solutions after the aqua regia treatment. It was found that in the first five minutes of contact with acid, only around 0.02–0.05% of the whole Nb in the samples left the oxide, whereas for Ca, this figure was around 1–3%. In subsequent days, the rate of dissolution of both Ca and Nb decreased. In particular, in the following two days, 0.5–2% of Ca and 0.01–0.06% of Nb were dissolved.

2.1.5. SEM Results

Scanning electron microscopy was utilised to obtain information about the morphology of the samples (Figure 6 and Figure S4). The SEM images of HCN₃ and its organic derivatives indicate that all of the samples represent oxide particles with a size of 0.5–1 μm. Some of the particles we found to be non-isotropic and have a scaly structure, which is caused by a layered crystal nature. The surface of the HCN₃/Pt, HCN₃/Pt×Me, HCN₃/Pt×Bu and HCN₃×Me/Pt samples also showed Pt agglomerates with a size of around 20 nm. It is worth mentioning that HCN₃/Pt×Bu visually had more dense Pt coverage than its analogues HCN₃/Pt×Me and HCN₃/Pt×Oc. This is probably due to the shorter duration of the amines’ intercalation into this sample (1 day instead of 7 days), so most Pt particles did not have time to fall off from the oxide particle at the moment of oxide particle collision, as they did in the case of HCN₃/Pt×Me and HCN₃/Pt×Oc.

Figure 6. SEM images of platinised samples of HCN₃/Pt (A), Nb₂O₅/Pt (B), HCN₃×R/Pt, and HCN₃/Pt×R (C–H) before (left) and after (right) aqua regia treatment.

Alongside platinised samples of layered oxides, SEM images (Figure 6) of platinised niobium oxide Nb₂O₅/Pt were captured. One can see from the image below that Pt agglomerates with a size of 10 nm densely cover the surface of niobium oxide. In some places, platinum formed much larger clusters with a size of up to 100 nm. In fact, Nb₂O₅/Pt had the densest platinum coverage among all the samples.

This picture is different from what is observed for platinized layered oxides. In the case of HCN₃/Pt, HCN₃/Pt×Me, HCN₃/Pt×Bu, HCN₃×Me/Pt, platinum, agglomerates can still be seen; for HCN₃/Pt×Me and HCN₃×Me/Pt, agglomerates of platinum are visually confirmed to be less dense than on the surface of Nb₂O₅/Pt. This allows us to conclude that perhaps for these three samples, less platinum is located on the surface than in Nb₂O₅/Pt. In the other three samples, HCN₃×Bu/Pt, HCN₃×Oc/Pt, and HCN₃/Pt×Oc, barely any platinum particles are visible, and there may be no surface platinum at all; however, this cannot be fully confirmed, because surface platinum can form tiny particles that are not visible via SEM. At the very least, we can confirm that there is no dense coverage with 10 nm agglomerates in the case of platinised layer oxides as seen on the surface of Nb₂O₅/Pt.

In order to convert these qualitative assumptions to quantitative ones, we used SEM images to estimate the visible Pt weight fraction. These images provide us with information about the size of oxide particles and the amount and sizes of surface Pt particles; therefore, it becomes feasible to roughly estimate the surface platinum weight fraction. The detailed procedure is described in Supplementary Material S5, and it results in values of surface Pt content equal to around 1 wt.% of total sample mass for Nb₂O₅/Pt, which aligns with the assumption that all loaded platinum was deposited on the sample surface.

Certainly, we could not miss an opportunity to obtain the results of XPS measurements, because this method is capable of detecting the very surface part of the sample (around 10 nm in depth). Thus, we were able to judge Pt content in this surface layer and to figure out whether it is distributed uniformly in the oxide particles or not. The Pt atomic fraction was determined using the 70.3 eV binding energy peak on the spectra. However, we were primarily interested in the fraction of Pt deposited on the surface relative to the total Pt content, rather than the raw atomic fraction within the pre-surface layer. To do this, we used an estimation technique, the details of which are enunciated in Supplementary Material S6. The results of these estimations are listed in Table 2. Generally speaking, during these estimations, it was assumed that XPS results provided us with information about a sample’s composition within a 10 nm thick layer. Then, the volume ratios of pre-surface layer to entire oxide particle volume were evaluated by averaging this ratio among many particles in the SEM images. Eventually, it became possible to estimate the ratio of pre-surface Pt to the loaded amount of Pt (which is equal to 1 wt.% of the entire sample’s mass).

Table 2. Amount of Pt detected by XPS in HCN₃/Pt and platinised HCN₃×R samples.

Sample	XPS Pt
	Atomic %	% of Loaded Pt (Estimate)
Nb2O5/Pt	7.50	355
HCN3	0.00	0
HCN3/Pt	1.45	68
HCN3×Me/Pt	1.55	75
HCN3×Bu/Pt	0.55	24
HCN3×Oc/Pt	0.65	31
HCN3/Pt×Me	0.60	28
HCN3/Pt×Bu	0.85	38
HCN3/Pt×Oc	0.15	8

Firstly, it is clear that within a pre-surface layer of niobium oxide, there is significantly more platinum than in the case of layered oxides. Our estimate provided a value of 3.55 wt.%, which, of course, does not seem plausible, as only 1 wt.% of platinum was loaded. However, it is worth noting that this is only an estimate, and it at least has the right order of magnitude. Even if the absolute values are not correct, they provide us with information about relative variations in pre-surface platinum content among the samples.

Since each layered oxide has a less intense XPS signal, one can conclude that there is less Pt in the pre-surface layer and that the majority of platinum is located inside the interlayer space. To be more precise, in HCN₃/Pt, less than 20% of the loaded amount of Pt is situated in the pre-surface layer. This means that the amount of surface Pt in this sample is not greater than 20% of the loaded amount.

Though HCN₃/Pt×R samples were prepared from HCN₃/Pt without any deliberate change in Pt localisation, XPS data point out that in these three samples, the amount of Pt in a pre-surface layer dropped compared to HCN₃/Pt. We assume that this reinforces the statement that some Pt nanoparticles detached from the HCN₃/Pt oxide particles during the intercalation of amines and subsequent procedures.

HCN₃×Bu/Pt and HCN₃×Oc/Pt samples had less Pt in their pre-surface layer than HCN₃/Pt. This can indicate either that, in general, a smaller amount of Pt was deposited on those samples or that Pt managed to propagate deeper in the interlayer space, where it could not be detected by XPS.

2.1.6. TEM Data

To receive even more information about Pt inside the samples, we analysed TEM images (Figure 7), which, unlike SEM ones, can capture not only surface Pt but also interlayer Pt. It can be seen from these images that acid treatment reduced the number of Pt particles and agglomerates for all the samples (for HCN₃/Pt×Oc, this is not that pronounced due to an initially small number of visible Pt agglomerates). The HCN₃/Pt sample showed the densest distribution of Pt agglomerates among the three presented samples, which became much sparser after intercalation of octylamine for aforementioned reasons. In addition, despite the aqua regia treatment for 5 min, TEM images still indicate the presence of Pt particles sized around 3–4 nm in the samples, which begs the question: are these particles located on the surface or between the layers of oxide? Strictly speaking, it is hard to address this question based solely on TEM data. Therefore, it will be discussed later on in this paper, after the results of other methods.

Figure 7. TEM images of HCN₃/Pt (A), HCN₃/Pt×Oc (B), and HCN₃×Oc/Pt (C) before (left) and after (right) acid treatment.

Using HR-TEM images (Figure 8), one can also find the distances between atomic planes of perovskite and platinum crystal structures. A distance of 3.79 Å quite closely matches a unit cell parameter of the perovskite structure measured by XRD (Figure 2, 3.84 Å), whereas the 2.27 Å spacing between atomic planes in platinum corresponds to its (111) plane, and 15.13 Å is close to the interlayer distance of HCN₃/Pt.

Figure 8. Interlayer planar distances (marked with blue arrows) of HCN₃/Pt in HR-TEM images: perovskite block structure (A), platinum particle (B), and interlayer distance (C).

2.2. Platinum Distribution in the Samples

In this section, we will discuss the results obtained all at once and try to build a global picture of platinum’s distribution in different layered oxides. The most valuable observations can be drawn from SEM, XPS, and ICP data.

2.2.1. Platinum Dissolving During Aqua Regia Treatment

The most crucial data were received from the aqua regia treatment. This mixture can dissolve platinum at room temperature, and we assumed that the difference in the Pt dissolution rate would give us information about the depth of Pt localisation. Acid treatment was conducted at room temperature. Pt concentration in the solution of aqua regia above the sample was measured by means of ICP-OES after 5 min, as well as 2, 7, and 14 days after the start of the experiment. An additional aqua regia treatment followed by ICP-OES analysis was performed in more severe conditions (120 °C for 3 h) in order to retrieve all Pt from the samples (see Figure 9). It is worth noting that we made an assumption that after such treatment, all Pt would be removed from the samples; that said, this may be not the case, since these oxides are not completely soluble, even in highly harsh conditions.

Figure 9. Weight fractions of platinum dissolved by aqua regia in platinised derivatives of HCN₃ and HCN₃×R for different periods of time at room temperature; the residual amount of platinum was determined by subtraction from the total amount of platinum. The total amount of platinum was found by analysing the solution after aqua regia treatment at 120 °C for 3 h.

First of all, it should be mentioned that even for a compound as simple as Nb₂O₅, data discrepancy was observed. In particular, the amount of Pt detected after 3 h of aqua regia treatment at 120 °C was less than that at room temperature for 5 min. We attribute this flaw to an experimental error, which in this case happened to be around 10–15%.

One can see that the Pt deposited on Nb₂O₅ was almost eliminated in 5 min and that its total amount was close to the loaded amount. As there is no interlayer space in niobium oxide, all of this platinum was located on the surface and hence was eliminated rapidly. However, strictly speaking, it does not let us conclude for other samples that if Pt dissolves in 5 min, it is only surface Pt, because interlayer Pt can also be dissolved rapidly, given that the aqua regia propagates into the interlayer space rapidly.

Further, ICP-OES data state that in the case of HCN₃/Pt, all loaded Pt was deposited on the sample surface and that it also dissolved quite rapidly (but slower than in the case of Nb₂O₅/Pt). The set of HCN₃/Pt×R samples had a smaller total amount of Pt than HCN₃/Pt, which again confirms the assumption about the detachment of Pt nanoparticles from HCN₃/Pt during amine treatment.

In general, in the samples with amines, Pt dissolved much more slowly than in HCN₃/Pt and Nb₂O₅/Pt, so the amine provided some sort of protective effect. Since amine is located in the interlayer space, the existence of such an effect can implicitly indicate platinum localisation in the interlayer space rather than on the surface. Otherwise, such an influence is hard to explain. Moreover, the longer the amine molecule was, the stronger this protection turned out to be.

2.2.2. Samples After Aqua Regia Treatment

After two weeks of the aqua regia treatment at room temperature, the samples were subjected to further investigation with XRD (Figure S7), TG, Raman spectroscopy, SEM, and EDX methods in order to determine the changes that they underwent during this process. Some of the results have already been mentioned in this text.

Comparison of TG curves of the samples before and after (Figure S8) aqua regia treatment led us to the conclusion that mass loss for the samples after treatment is smaller than it was before; thus, organic content in the interlayer space decreased (Table 3). However, some amount of the amine remained inside because, for instance, in the case of octylamine samples, mass loss was still too large to be explained solely by interlayer water. Moreover, characteristic peaks of C-H bonds can be observed in the Raman spectra (Figure 4) of octylamine samples after the aqua regia treatment. SEM images of oxides after aqua regia etching confirm that surface Pt was eliminated during the two-week process (Figure 7).

Table 3. Relative mass losses during TG measurements of platinized samples before and after aqua regia treatment.

Sample	$∆{\mathbf{m}}_{\mathbf{b}\mathbf{e}\mathbf{f}\mathbf{o}\mathbf{r}\mathbf{e}},\mathbf{\%}$	$∆{\mathbf{m}}_{\mathbf{a}\mathbf{f}\mathbf{t}\mathbf{e}\mathbf{r}},\mathbf{\%}$
HCN3/Pt	3.68	2.17
HCN3×Me/Pt	6.86	2.70
HCN3×Bu/Pt	12.01	3.12
HCN3×Oc/Pt	21.64	7.00
HCN3/Pt×Me	6.91	1.11
HCN3/Pt×Bu	12.01	3.15
HCN3/Pt×Oc	16.75	5.79

2.2.3. Comparison of XPS and ICP Data

ICP-OES confirms that the entire amount of platinum (1 wt.%) was deposited on Nb₂O₅/Pt. This conclusion can be drawn from the fact that if one adds up the values of Pt, leaving the sample at room temperature, it will be equal to 107%. Although in another acid treatment with heating, only 88% of the loaded amount was detected, we suppose that the latter measurement underestimates the amount of platinum in Nb₂O₅/Pt due to an experimental error and, in fact, it is close to 100% of the loaded amount.

Our premise is that all this platinum is located on the surface of Nb₂O₅ oxide. We did not prove this rigorously, but the estimate from SEM data and the highest rate of Pt removal during the aqua regia treatment support this thesis.

Furthermore, one can see that both SEM images and XPS data demonstrate a lower level of platinum on the other samples than on Nb₂O₅/Pt. This means that there is less than 1 wt.% platinum on the surface of these samples. Subsequently, one can combine the ICP-OES data for the total amount of Pt and renormalised values for the pre-surface amount of Pt (Pt concentrated in a 10 nm depth pre-surface layer) (see Figure 10). A fraction of interlayer platinum was calculated as the total amount of platinum in the sample determined by ICP minus the pre-surface platinum fraction (See Appendix A).

Figure 10. Fractions of pre-surface and interlayer platinum from the loaded amount of platinum obtained by a combination of ICP-OES and XPS data.

As one can see from this chart (Figure 10), only around 20% of Pt in HCN₃/Pt is located on the surface. Moreover, after intercalation of amines, this amount decreased even further, which can be seen in SEM images as well (Figure 6). Based on these images, HCN₃/Pt×Oc and HCN₃/Pt×Me lost more surface Pt than HCN₃/Pt×Bu, and this is also confirmed by Figure 10. After intercalation of n-butylamine in HCN₃/Pt, around 10% of the loaded amount of Pt detached from the sample. This 10% is probably a part of the surface platinum in HCN₃/Pt. In the case of HCN₃/Pt×Oc, almost all pre-surface platinum (roughly 17%) left the sample. This can also be confirmed by a corresponding SEM image (Figure 6D), where no Pt particles are visible on the surface of HCN₃/Pt×Oc. Thus, we can conclude that the surface platinum fraction in HCN₃/Pt is more than 10% but less than 17% of the loaded amount.

Furthermore, during methylamine and n-octylamine intercalation in HCN₃/Pt, some amount of interlayer platinum apparently also detached (Figure 10). In the case of the n-butylamine intercalation, which took only one day, this effect was not observed.

It should be kept in mind that all the aforementioned percentages are highly approximate due to a 10–15% error in our experiment. However, ICP-OES and XPS data are generally in good agreement with each other, and it can be stated with a high degree of certainty that the surface platinum in HCN₃/Pt makes up less than 20% of the loaded amount, or equivalently, that interlayer platinum does exist for the following reasons.

-

some 60–80% of platinum in the perovskite samples is not detected by XPS;

-

coverage by Pt agglomerates in SEM images for perovskite samples is sparser than for Nb₂O₅/Pt, which has all its platinum on the surface.

-

in the case of perovskite samples, platinum dissolves more slowly in aqua regia than platinum on the surface of Nb₂O₅/Pt, and it is hard to imagine that surface platinum would be protected from exposure.

2.2.4. Arrangement of Platinum Particles

At this point, we turn our attention back to the TEM images. Since we are confident that after 5 min etching in the aqua regia, the only remaining platinum is the interlayer one, then in what form does this interlayer platinum exist? TEM images after acid treatment clearly feature Pt particles of 3–4 nm size, whereas the interlayer space of the considered oxides is around 0.15 nm. If these particles had been located between the layers, they would have caused a significant distortion of the layered structure, which should be visible in TEM images capturing the side surface of the oxide particles. In Figure 11, such fragments of TEM images that capture side facets of oxides are presented.

Figure 11. Fragments of TEM images of HCN₃×Oc /Pt (A,B) and HCN₃/Pt×Oc (C,F) after acid treatment and HCN₃/Pt (D) and HCN₃/Pt×Oc (E) before acid treatment featuring interlayer platinum.

In the images (Figure 11A,B), one can notice platinum particles sized around 1 nm which are located between layers, although not between adjacent layers of ideal crystal structure; rather, they are in the regions of defects of this layered structure, especially near the ends of some layers, forming patterns like those depicted in Figure 12.

Figure 12. Schematic representation of interlayer platinum localisation patterns.

At the same time, in Figure 11D, no significant distortion of the layered structure is observed, whereas dark patches of platinum agglomerate overlay the image. It should also be noted that those interlayer particles have a relatively small size of around 1 nm, whereas some platinum agglomerates are much larger, even after acid treatment. These clusters probably represent the surface platinum that, for some reason, managed to survive 5 min acid etching. Figure 11F serves as confirmation of this statement, since it clearly features a large agglomerate of surface platinum.

It is worth mentioning that we cannot exclude the possibility that there are particles or atoms even finer than 1 nm located in the interlayer space that were not detected in TEM images. In fact, the density of platinum agglomerates in the TEM images of HCN₃/Pt×Oc and HCN₃/Pt is drastically different (Figure 7), whereas the platinum content in both of these samples does not differ that much (Figure 9). The presence of such “invisible” platinum would explain the difference, but we did not prove its existence in this form.

As was mentioned in the Introduction, one of our objectives was to compare samples with different orders of intercalation and platinisation stages. However, it is more correct to compare HCN₃×R/Pt with HCN₃×/Pt than with HCN₃/Pt×R because the latter undergoes changes after the intercalation process and loses some amount of platinum, whereas in comparing HCN₃×R/Pt with HCN₃×/Pt, only the effect of amine presence in the interlayer space has an influence. From Figure 10, it can be seen that the HCN₃×Bu/Pt and HCN₃×Oc/Pt samples have less Pt in the pre-surface layer than HCN₃/Pt. Moreover, they have less platinum (by around 30%) in total compared to HCN₃/Pt. Perhaps intense gas emission during the platinum deposition process makes it more difficult for platinum to approach the oxide particle. Nevertheless, no dramatic difference in the proportion of pre-surface and interlayer Pt is found when one compares HCN₃×R/Pt and HCN₃/Pt. In all these cases, only a minor part of the Pt is located on the surface. Thus, the presence of amines in the interlayer space does not favour greater deposition of platinum on the surface. Moreover, as seen from Figure 9, in both series of oxides, a protective effect of platinum exists; thus, Pt is predominantly inside the samples in both cases.

One more interesting detail is that HCN₃×Me/Pt has an almost identical pattern of Pt distribution to HCN₃/Pt. The only difference is that Pt in HCN₃×Me/Pt is dissolved more slowly than in HCN₃/Pt under the influence of aqua regia.

2.3. Influence of Platinum Distribution on Photocatalytic Activity

2.3.1. Band Gap Energies

Diffuse reflectance spectra were measured to determine optical bandgap energies. Raw reflectance data served as an input to calculate the Kubelka–Munk function F(R) and to build the Tauc plot (Figure S9). From this plot, optical bandgap energy values were obtained as abscissae of the intersection points of linear sections of the graphs. The values of the bandgap energies (Table 4) happened to be quite close to each other among all the samples, at least within an error, and were located between 3.5 and 3.6 eV. Hence, it is not feasible to identify any dependence or pattern in these data.

Table 4. Light adsorption characteristics of HCN₃, HCN₃×R, and their platinized derivatives.

Sample	Eg, eV	λmax, nm
HCN3	3.50	354
HCN3/Pt	3.53	352
HCN3×Me	3.57	348
HCN3×Bu	3.59	346
HCN3×Oc	3.56	349
HCN3×Me/Pt	3.58	347
HCN3×Bu/Pt	3.59	346
HCN3×Oc/Pt	3.62	343
HCN3/Pt×Me	3.58	347
HCN3/Pt×Bu	3.61	344
HCN3/Pt×Oc	3.59	346

2.3.2. Photocatalytic Activity

Data about the rate of water splitting of almost all oxides mentioned in our work are presented in Figure 13. The activity of Nb₂O₅ and Nb₂O₅/Pt is at the level of background noise, around a 0.5 mmol/(g·h) magnitude, and is not depicted in the graph. It proves that a further increase in the photocatalytic activity for other samples is not caused by platinum on its own but by its interaction with the oxide. Pt deposition on the surface of Nb₂O₅ slightly increases the activity, but it is still far less than for the layered samples; meanwhile, platinum deposition on HCN₃ leads to a five-fold increase in the activity. Therefore, a high level of activity for the platinized layered oxides is probably caused by Pt presence in the interlayer space rather than on the external surface, since it was shown that in HCN₃/Pt, the majority of platinum is located between layers.

Figure 13. Photocatalytic activity of initial and platinised oxides (HCN₃, HCN₃×R) as well as oxides after 5 min and 3 h heating in aqua regia. HCN₃ and HCN₃×R were not subjected to acid treatment and heating.

As we demonstrated previously, in HCN₃/Pt, around 65% of Pt is eliminated in the first 5 min of the acid treatment. From the photocatalytic activity data, it can be seen that the activity of the HCN₃/Pt sample also decreases, roughly by 60%. Acid treatment with heating is capable of completely removing all platinum, and therefore, photocatalytic activity becomes equal to that of HCN₃ without Pt.

Intercalation of amines into HCN₃ increases the activity for R = Bu, Oc but decreases in the case of R = Me. Moreover, the longer the amine, the greater photocatalytic activity, which can be explained by enhanced diffusion of water and other molecules in the wider interlayer space. Methylamine does not significantly expand the interlayer space compared to HCN₃, but its molecule is larger than a water molecule, which leads to more impeded diffusion than in HCN₃. The same behaviour and explanation were discussed in one of our previous articles [25].

Deposition of Pt on HCN₃×R raises the activity only for HCN₃×Me and HCN₃×Bu, whereas, in contrast, for HCN₃×Oc, the activity slightly decreases. As was previously discussed, HCN₃×Me/Pt and HCN₃/Pt have close patterns of Pt distribution between the surface and interlayer space. However, the activity of HCN₃×Me/Pt is lower than that of HCN₃/Pt. This is for the same reason as in HCN₃ and HCN₃×Me. Amine molecules block reactants from moving inside the interlayer reaction zone, where the majority of Pt is located. Therefore, the presence of Pt has little effect, since it is almost entirely blocked in the interlayer space. Nevertheless, the activity of HCN₃×Bu/Pt is exceptionally high and even higher than for HCN₃/Pt. Perhaps the reason for such a maximum is that this sample contains the lowest amount of amine among all the HCN₃×R/Pt samples, therefore creating additional space for water diffusion into the interlayer space. In other words, for R = Bu, the sample likely achieves the optimal balance (compared to the other two compounds) between interlayer expansion and density of amine packing, which may explain its highest photocatalytic activity.

Intercalation of the amines into HCN₃/Pt in all cases suppresses the activity to a figure even lower than that observed for HCN₃. This can probably be explained by amine molecules covering Pt in the interlayer space more densely and effectively than in the case of HCN₃×R/Pt, where Pt is deposited after intercalation of amines. In this scenario, only surface Pt takes part in the photocatalytic process; the difference in the amount of surface Pt between HCN₃/Pt×R with different R becomes clear. In particular, the activity of HCN₃/Pt×Bu is higher than that of HCN₃/Pt×Me and HCN₃/Pt×Oc due to the larger amount of surface Pt in HCN₃/Pt×Bu. In addition, the lower activity of HCN₃/Pt×R compared to HCN₃×R can be explained by the lower amount of intercalated water in HCN₃/Pt×R in general due to the higher amount of amine molecules than in HCN₃×R. Increasing the amount of intercalated amine may suppress the ability of reactants to enter the interlayer space.

In addition, it is worth mentioning that some of the Pt particles can detach during the ultrasonic treatment, which was produced with a catalyst suspension before the photocatalytic experiment. The activity of the platinised samples in this study is less than that of the analogous samples from our previous work [25], where platinum was deposited in situ without the subsequent ultrasonic treatment. At the same time, the activities of HCN₃, HCN₃×Me and HCN₃×Bu are in good agreement with our previous results.

One of the most interesting details of the results obtained is that for the samples with intercalated amines, the activity increased after the 5 min aqua regia treatment, whereas for the oxides without amines (Nb₂O₅, Nb₂O₅/Pt, HCN₃, HCN₃/Pt), the activity predictably decreased. One of the possible explanations for such behaviour is that the acid etching destroys a part of the amine molecules in the interlayer space, making reactants’ diffusion easier than in the initial oxides with amines, provided that the interlayer space does not shrink significantly after the 5 min treatment, since a part of the amine molecules remains there.

Although the activity of the samples subjected to the 3 h acid treatment with heating is lower than that of the samples treated for 5 min at room temperature, it is still considerably high and even higher than the activity of the initial samples. We assumed previously that in these samples, Pt would be absent, but it became clear that perhaps this assumption is wrong, and some amount of Pt (less than 20% of the loaded amount) is still located somewhere inside the samples and is accounted for by residual activity which is, in fact, quite substantial. Therefore, it can be concluded that the amount of platinum required to attain a certain level of photocatalytic activity can be reduced by the aqua regia treatment since only the platinum located deep inside the interlayer space is the most effective during the photocatalytic process.

3. Materials and Methods

3.1. Synthesis

The following chemicals were used for the synthesis of the compounds mentioned in this paper: Nb₂O₅ (Chemcraft Ltd., Saint Petersburg, Russia), CaO (Neva reactive Ltd., Saint Petersburg, Russia), K₂CO₃ (Vecton JSC, Saint Petersburg, Russia), MeNH₂ (Chemstore Ltd., Saint Petersburg, Russia), n-BuNH₂ (Chemstore Ltd., Saint Petersburg, Russia), n-heptane (Ecos-1 Ltd., Moscow, Russia), n-OcNH₂ (Vecton JSC, Saint Petersburg, Russia), H₂PtCl₆ 6H₂O, HNO₃ (Vecton JSC, Saint Petersburg, Russia), HCl (Vecton JSC, Saint Petersburg, Russia), MeOH (Baltpromhim Ltd., Saint Petersburg, Russia). The water used for solution preparation and laboratory glassware washing was distilled using a Liston A1110 distiller (Zhukov, Russia). Before ceramic synthesis, calcination of Nb₂O₅ (1100 °C, 7 h), CaO (1100 °C, 7 h), and K₂CO₃ (700 °C, 7 h) was performed in a Nabertherm L-011K2RN muffle furnace (Lilienthal, Germany).

3.1.1. Synthesis of KCa₂Nb₃O₁₀ (KCN₃)

The initial layered perovskite-like oxide KCa₂Nb₃O₁₀ was obtained as follows. Some 17.8645 g of Nb₂O₅, 5.0253 g of CaO and 4.0250 g (30% excess) of K₂CO₃ were mixed. The mixture of these compounds was soaked with n-heptane and ground in a Fritsch Pulverisette 7 planetary ball mill at 600 RPM for 10 min with 5 min breaks between 9 repetitions. Then, this mixture was dried at 60 °C for 2 h. It was then pelletized into 1 g tablets 1.5 cm in diameter under 50 atm pressure via an Omec PI 88.00 hydraulic press (Certaldo, Italy). The tablets were placed in corundum crucibles and heated to 800 °C within 2 h in a Nabertherm L-011K2RN muffle furnace (Lilienthal, Germany) then kept at this temperature for the following 12 h. After this, the tablets were ground in an agate mortar and pelletized again with the same method. Then tablets were again heated in corundum crucibles up to 1100 °C within 2.75 h, kept at this temperature for the following 24 h, and thoroughly ground after cooling down. Thus, the layered perovskite-like niobate with potassium ions in the interlayer space was obtained.

3.1.2. Synthesis of HCa₂Nb₃O₁₀∙xH₂O (HCN₃)

In order to exchange potassium cations with protons, a slurry of the previously obtained KCa₂Nb₃O₁₀ was prepared by adding 5 g of this compound into 100 mL of 10 M HNO₃. The water solution of HNO₃ was prepared by mixing 207 mL of concentrated (65 wt.%) HNO₃ and 93 mL of water. The slurry was stirred with a magnetic anchor for 24 h. The product was separated by three repetitions of centrifugation with water rinsing (10 min at 2300 RCF using an ELMI CM 6MT centrifuge) and a decantation cycle. The sample was dried in a desiccator with zeolite for two weeks and then stored in a humid atmosphere to obtain a hydrated form of HCa₂Nb₃O₁₀.

3.1.3. Synthesis of HCa₂Nb₃O₁₀/Pt∙xH₂O (HCN₃/Pt) and Nb₂O₅/Pt

The platinized samples were obtained by the reduction of hexachloroplatinic acid under photocatalytic conditions. For this purpose, the custom photocatalytic equipment was used, which was thoroughly described in our previous work [26]. A total of 125 mg of HCN₃ or Nb₂O₅ was mixed with 50 mL of 1 mol. % aqueous MeOH and exposed to 10 min ultrasonic treatment in an Elmasonic S10H ultrasound bath (Elma, Singen, Germany), followed by the addition of 0.67 mL of a 5 g/L H₂PtCl₆∙6H₂O aqueous solution so that the mass of platinum in a flask was equal to 1 wt.% of the sample mass. The obtained slurry was pumped into the photocatalytic device and exposed to UV irradiation from a medium-pressure mercury lamp DRT-125 (125 W) for 30 min. Before reaching the oxide slurry, light passed through a light filter (KCl + NaBr aqueous solution, 6 g/L of each salt, 2 cm optical path, 15 °C temperature maintained by water thermostat). Then, passing through the photocatalytic slurry, light initiated the separation of holes and electrons in oxide particles. The reaction of the former with platinum ions led to its transformation to metallic platinum on the surface of the sample or in the interlayer space. After this, the slurry was drained from the device, and the sample was separated by an analogous centrifugation method as described previously.

3.1.4. Synthesis of HCa₂Nb₃O₁₀∙xH₂O∙yRNH₂ (HCN₃×R)

Intercalation of the n-alkylamines into the interlayer space of the niobates was performed in accordance with the conditions specified in Table 5. A slurry of HCN₃ in an amine solution with a particular oxide concentration was prepared and stirred via a rotating shaker at constant temperature and for a specified period of time. Separation and drying of the samples were performed in accordance with previously described procedures.

Table 5. Conditions of HCa₂Nb₃O₁₀∙xH₂O∙yRNH₂ synthesis.

Sample	Oxide Loading, g/L	Reaction Media	Duration, Days	Temperature, °C
HCN3×Me	25	38 wt.% MeNH2 in water	7	25
HCN3×Bu	25	90 wt.% n-BuNH2 in water	1	25
HCN3×Oc	75	30 wt.% n-OcNH2 in n-heptane	7	60

3.1.5. Synthesis of HCa₂Nb₃O₁₀∙xH₂O∙yRNH₂/Pt (HCN₃×R/Pt)

This set of samples was obtained analogously (see Section 3.1.3) to the HCN₃/Pt, with the only difference being that HCN₃×R was used instead of HCN₃.

3.1.6. Synthesis of HCa₂Nb₃O₁₀/Pt∙xH₂O∙yRNH₂ (HCN₃/Pt×R)

This set of samples was obtained analogously (see Section 3.1.4) to the HCN₃×R, with the only difference being that HCN₃/Pt was used instead of HCN₃.

3.1.7. Aqua Regia Treatment

Agua regia treatment with subsequent ICP-OES analysis of the solutions was conducted in order to measure the kinetics of platinum dissolving, which can help to determine its localisation, as previously discussed. In a typical experiment, 30–60 mg of a sample was placed in a 2 mL polypropylene Eppendorf tube, which was then filled with 0.5 mL of aqua regia. Aqua regia was prepared by mixing 12 mL of concentrated (35 wt.%) HCl and 4 mL of concentrated (65 wt.%) HNO₃. Then, the Eppendorf tube was closed and stirred using a Biosan MSV-3500 vortex mixer (Riga, Latvia) at 1500 RPM for 1 min, after which the sample was centrifuged at 15,000 RCF for 4 min using an ELMI CM-50M centrifuge mixer (Riga, Latvia). After that, 0.2 mL of the solution was transferred from the Eppendorf tube into a test tube and diluted by adding 20 mL of water. After 2, 7, and 14 days starting from the very beginning of the experiment, the cycles of vortex stirring, centrifugation, and solution exchange were repeated. A further 14 days after the start of the acid treatment, a new portion of aqua regia was not added, but the solution was drained from the Eppendorf tube. Subsequently, 1.5 mL of water was added instead of aqua regia, followed by 1 min vortex stirring at 1500 RPM and 4 min centrifugation at 15,000 RCF. Then, water was drained, and the rinsing cycle was repeated two more times. After the last water draining, the samples were placed in a laboratory oven Memmert UF55 (Schwabach, Germany) at 60 °C for two days. After drying, the samples were stored in a desiccator with zeolite.

Alongside measuring the Pt concentration change during the acid treatment, the total amount of Pt in the samples was determined as well. For this purpose, 10–20 mg of a sample was placed in a 2 mL glass vial, and then 1 mL of aqua regia was added. The vial was closed with a lid featuring a polypropylene septum and placed into a laboratory autoclave with a Teflon inner part. The autoclave was heated at 120 °C in a laboratory oven, Memmert UF55 (Schwabach, Germany), for 1 h. After cooling the vials down, a 0.5 mL probe was taken from each of the vials and placed into an Eppendorf tube, which was then centrifuged at 15,000 RCF for 10 min. After centrifugation, 0.2 mL of the solution was sampled from the Eppendorf tube and diluted by mixing with 20 mL of water.

3.2. Characterisation Methods

3.2.1. X-Ray Diffraction Analysis

Powder X-ray diffraction (XRD) patterns were measured on a Rigaku Miniflex II desktop Röntgen diffractometer (Tokyo, Japan) (CuKα radiation, hν = 8.0478 keV, angle range 2θ = 3–60°, scanning rate 10°/min, step 0.02°). Lattice parameters and estimated space groups were calculated in the tetragonal system based on all the reflections observed using DiffracPlus Topas 4.2 software (Bruker, Karlsruhe, Germany).

3.2.2. Raman Spectroscopy

Raman spectra were obtained using a Bruker Senterra spectrometer (Billerica, MA, USA) (wavenumber range 100–4000 cm⁻¹, incident laser 532 nm). For the samples before aqua regia treatment, 5 mW power was used with an accumulation time equal to 30 s; for samples HCN₃×Oc/Pt, HCN₃/Pt×Bu, 2 mW power was used. For all samples after aqua regia treatment, a 2 mW source was used, whereas accumulation time varied: 80 s for Nb₂O₅/Pt, 100 s for HCN₃/Pt, 140 s for HCN₃/Pt×Me, 150 s for HCN₃/Pt×Bu, 300 s for HCN₃/Pt×Oc, and 200 s for HCN₃×Me/Pt, HCN₃×Bu/Pt, HCN₃×Oc/Pt.

3.2.3. Infrared Spectroscopy

Infrared absorption spectra were obtained using a Shimadzu IRAffinity-1 spectrometer (Kyoto, Japan) (spectral range 400–4000 cm⁻¹, KBr used as tablet base) with Fourier transformation.

3.2.4. Thermogravimetric Analysis

For thermogravimetric (TG) analysis, approximately 10 mg of the sample was heated up in a combined flow of oxygen (3 mL/min) and argon (3 mL/min + protective Ar flow 3 mL/min) from 30 to 950 °C with a 10 °C/min heating rate using a Netzsch TG 209 F1 Libra thermobalance (Selb, Germany).

3.2.5. CHN Analysis

A Euro EA3028-HT analyser (EuroVector, Pavia, Italy) was used to determine weight fractions of hydrogen, carbon, and nitrogen in the samples. Roughly 1 mg of the sample was placed in an aluminium foil container and subjected to flash pyrolysis at 980 °C. The produced gases were analysed chromatographically.

3.2.6. SEM

Scanning electron microscopy (SEM) was performed on a Zeiss Merlin electron microscope (Oberkochen, Germany). A field emission cathode, GEMINI-II electron optics column, and oil-free vacuum system were used during the measurements.

3.2.7. TEM

Transmission electron microscopy (TEM) measurements were carried out to obtain images of interlayer platinum, in contrast to SEM, which allows observing only the surface of the sample. The particle slurry in isopropanol was prepared by sonication and then transferred with a pipettor onto a carbon membrane with copper support. TEM measurements were performed using a Carl Zeiss Libra 200FE microscope (Oberkochen, Germany) under 200 kV accelerating voltage and Ω energy filtering. Light-field images were made with a CCD camera, whereas dark-field images were obtained using a HAADF detector in a STEM mode.

3.2.8. EDX

X-ray radiation, which was emitted during SEM measurements, was detected by an Oxford Instruments INCA X-max 80 energy-dispersive X-ray (EDX) microanalyser (Abingdon, UK). The atomic composition of the samples was determined using calibration dependences. For each sample, 10 data points were obtained and then averaged.

3.2.9. Diffuse Reflectance Spectroscopy

A Shimadzu UV-2550 spectrophotometer (Kyoto, Japan) with the ISR-2200 (Shimadzu, Kyoto, Japan) integrating sphere attachment (operating in the range of 200–2500 nm) was utilised to receive diffuse reflectance spectra (DRS) after sample deposition on a barium sulfate substrate.

3.2.10. XPS

X-ray photoelectron spectroscopy measurements were performed on a Thermo Fisher Scientific Escalab 250Xi spectrometer (Waltham, MA, USA) equipped with an AlKα monochromatic radiation source (photon energy 1486.6 eV). The spectrometer was calibrated against the Au 4 f7/2 line (binding energy 84.0 eV). The spectra were recorded in the constant transmission energy mode at 50 eV with an XPS spot size of 650 μm. The total energy resolution of the experiment was about 0.3 eV. The studies were carried out at room temperature in an ultra-high vacuum of the order of 10⁻⁹ mbar. To remove the sample charge, a combined ion–electronic charge compensation system was used.

3.2.11. ICP-OES

Solutions that were prepared by diluting the solutions from Eppendorf tubes were analysed for Pt, Ca, K and Nb content via the ICP-OES method. For this purpose, a Shimadzu ICPE-9000 spectrometer (Kyoto, Japan) was used. Standard solutions of Pt and Ca were prepared from a multicomponent standard Merck solution in 0.1 N HNO₃. Standard solutions of Nb were prepared from a Nb salt in 0.1 N HNO₃. Standard solutions had concentrations ranging from 0.001 to 10 mg/L. Spectral analysis of the samples was performed in axial mode with a mini-burner without additional dilution. An absolute error was estimated on the basis of three parallel measurements. The following emission peaks were used to determine concentrations of the elements: 203.646 nm for Pt, 422.673 nm for Ca, 769.896 nm for K, and 309.418 nm for Nb.

3.2.12. BET Measurements

In order to determine the specific area of Nb₂O₅ and HCN₃ samples’ adsorption, isotherms were measured using QuadraSorb Station 1 (Boynton Beach, FL, USA) at 77.35 K temperature and vacuum degassing at 293 K for 3 h. Nitrogen was used us an analysis gas.

3.2.13. Measurements of Photocatalytic Activity in the Reaction of Hydrogen Production from Methanol Solution

In order to evaluate the influence of the Pt deposition method and aqua regia treatment on photocatalytic activity, this property was investigated for almost all of the mentioned samples. We used the same method as described in our previous papers [26] with the only exception that instead of 1 mol.% methanol, a 10 mol.% solution was used, and platinum was not deposited in situ on any of the samples.

In general, this procedure was carried out as follows. Approximately 0.005 g of the oxide was placed in a flask containing 50 mL of a 10 mol.% methanol solution. The flask was then subjected to 10 min of ultrasonic treatment to disperse the particles. After that, the slurry was pumped into a glass cell connected to a chromatographic unit. The glass cell was irradiated with light passing through a filter solution (aqueous solution containing 6 g/L NaCl and 6 g/L KBr) that cuts off high-frequency waves (λ > 400 nm). Hydrogen formed in the cell was intermittently withdrawn for analysis. Then, using several data points, a kinetic curve was plotted, depicting the amount of hydrogen versus time, and the activity of the sample was determined as the tangent of the angle formed by this curve and the x-axis.

4. Conclusions

In the aftermath of this study, we can make the following conclusions. Firstly, during the platinisation of HCN₃ and its derivatives, loaded platinum is deposited in both the interlayer space and on the surface of the samples. In HCN₃/Pt, only a minor part of Pt (between 10% and 17%) is located on the surface of the oxide; the majority of it is interlayer Pt. If one intercalates amines in HCN₃ before platinum deposition, it will reduce the amount of Pt deposited on the surface and favour its deposition in the interlayer space. The observed changes in the photocatalytic activity of the samples obtained in this way suggest the possibility of a synergistic effect between the intercalation of organics and the photodeposition of platinum (particularly in the case of butyl and octylamine compounds). Intercalation of amines in pre-modified with platinum compound HCN₃/Pt leads to a significant drop in photocatalytic activity, compared to the initial platinated compound. This could be partially explained by a partial loss of platinum during the intercalation process, although the main reason for this is apparently that the amine molecules introduced into the pre-platinised sample effectively “screen” the interlayer platinum particles and prevent them from participating in the photocatalytic process.

The results of analysing the photocatalytic activity of samples after etching platinum particles with aqua regia suggest that the sample activity is maintained even in the presence of platinum particles located only in the interlayer space (approximately 10–20 per cent of the total deposited platinum), while surface and near-surface clusters exert a significantly smaller influence. This fact allows for effective platinum conservation during activation of such catalysts. In particular, it is possible to use amine intercalation for the preferential introduction of platinum or another cocatalyst into the interlayer space (where it is better protected from mechanical and chemical influences but retains its activity), with subsequent removal of excess amount of noble metal from the surface by dissolution, after which most of the catalyst activity is still retained. Moreover, such data may indicate the nature of charge carrier localisation. In particular, the preservation of high activity for samples without surface platinum may indicate the preferential occurrence of the reduction half-reaction in the interlayer space.

Overall, the obtained results indicate the potential for the targeted production of highly efficient photocatalysts based on layered oxides and may help to expand the range of available methods for activating catalysts and photocatalysts of this type.