High-Utilisation Nanoplatinum Catalyst (Pt@cPIM) Obtained via Vacuum Carbonisation in a Molecularly Rigid Polymer of Intrinsic Microporosity

Polymers of intrinsic microporosity (PIM or here PIM-EA-TB) offer a highly rigid host environment into which hexachloroplatinate(IV) anions are readily adsorbed and vacuum carbonised (at 500 °C) to form active embedded platinum nanoparticles. This process is characterised by electron and optical microscopy, atomic force microscopy (AFM), X-ray photoelectron spectroscopy (XPS) and electrochemical methods, which reveal that the PIM microporosity facilitates the assembly of nanoparticles of typically 1.0 to 2.5-nm diameter. It is demonstrated that the resulting carbonised “Pt@cPIM” from drop-cast films of ca. 550-nm average thickness, when prepared on tin-doped indium oxide (ITO), contain not only fully encapsulated but also fully active platinum nanoparticles in an electrically conducting hetero-carbon host. Alternatively, for thinner films (50–250 nm) prepared by spin coating, the particles become more exposed due to additional loss of the carbon host. In contrast to catalyst materials prepared by vacuum-thermolysed hexachloroplatinate(IV) precursor, the platinum nanoparticles within Pt@cPIM retain high surface area, electrochemical activity and high catalyst efficiency due to the molecular rigidity of the host. Data are presented for oxygen reduction, methanol oxidation and glucose oxidation, and in all cases, the high catalyst surface area is linked to excellent catalyst utilisation. Robust transparent platinum-coated electrodes are obtained with reactivity equivalent to bare platinum but with only 1 μg Pt cm−2 (i.e. ~100% active Pt nanoparticle surface is maintained in the carbonised microporous host). Graphical Abstract ᅟ ᅟ


Introduction
The preparation and formulation of catalysts for applications in fuel cells [1] and in sensors [2] remain a highly active research topic. For example, the development of new types of porous substrate materials [3] offer catalysts with (i) better levels of performance [4,5], (ii) improved access into a wide range of reproducible and highly efficient catalyst composites [6], and/or (iii) access to more sustainable catalyst systems [7]. Many new inorganic [8] or carbon-based catalyst composite materials [9,10] are under development and, in particular, synthetic methods based on the carbonization of precursors to yield carbon-embedded catalysts [11,12] have proved beneficial. Novel and highly efficient types of carbon-embedded platinum catalysts have been obtained via carbonization [13]. However, directly embedding active nano-platinum catalysts without loss of activity during carbonization in a one-step procedure is uncommon and desirable.
Platinum nanoparticles are the basis for the overwhelming majority of technical catalyst systems due to the durability of platinum and due to the special place platinum takes with respect to interfacial adsorption of reaction intermediates [14]. Platinum nanoparticles are employed in PEM fuel cells [15], in methanol [16] and glucose [17] driven fuel cells, but also in many dye-sensitised solar cells [18] and in electrolysers for energy conversion [19]. The most active size range for platinum has been reported to be 2-6 nm (based on mass activity) [20]. However, the catalyst durability and corrosion characteristics are also important [21]. We have recently demonstrated that a new class of polymers with intrinsic microporosity (PIMs) can be employed to stabilise Pt nanoparticle performance for both anode and cathode activity [22,23].
Platinum nanoparticles provide very effective catalysts but suffer from high cost of the nonsustainable noble metal catalyst. Therefore, intense research efforts are directed towards either replacing platinum with cheaper and more sustainable metals (e.g. based on alloys of Fe, Co, Ni, etc. [24]) or improving utilisation. The latter involves maximising the surface area and reactivity of platinum, for example, by embedding nanoparticles into porous hosts or substrates [25]. Porous polymer materials are emerging as promising substrates (or precursors for porous substrates) for the immobilisation and growth of platinum nano-catalysts [26]. Highly porous polymer structures that can retain microporosity during carbonization are particularly interesting. For PIM materials it has been suggested that carbonization occurs via cross-linking and without breaking the of the rigid polymer backbone strands [27].
Polymers of intrinsic microporosity (or PIMs [28]) exhibit micropore diameter of typically 0.5-2.0 nm and N2-adsorption surface area values up to 1000 m 2 g -1 [29]. They were originally developed for gas storage [30] and separation [31], but recently have also proven utility for electrochemical processes such as electrocatalysis [32], ionic devices [33], or electroanalysis [34]. Most work in PIM electrochemistry to date has been carried out with PIM-EA-TB (prepared from a diaminoethanoanthracene "EA" monomer polymerised via a Tröger base "TB" coupling; see structure in Figure 1). This polymer with highly rigid molecular structure exhibits N2-adsorption surface area of 1027 m 2 g -1 and is employed here with molecular weight of 70 kDa [33].
We have demonstrated previously that relatively gentle vacuum carbonization of PIM-EA-TB at 500 °C causes conversion of the microporous starting material into a microporous conducting hetero-carbon [27]. Although the electrical conductivity of the new carbon material was limited, thin films deposited onto tin-doped indium oxide (ITO) substrates were clearly capacitive and suitable for application as high surface area carbon film electrodes. A key feature in the conversion of PIM-EA-TB to the carbonized form (cPIM) is the retention of structural porosity/integrity. There were no changes in cumulative pore volume and no apparent changes in morphology upon vacuum carbonization [27]. Therefore, this hetero-carbon formation may offer ideal conditions for the co-formation/embedding of nano-catalysts with the following advantages: (A) the rigid PIM-EA-TB does not block/affect the catalytically active Pt particle surface and (B) after carbonization, porosity and connectivity are maintained to allow electrical contact and sufficient access for reactants to the active catalyst surface.
Here, we add a well-defined amount of platinum pre-cursor (PtCl6 2-) by absorption into the porous protonated form of PIM-EA-TB and during subsequent vacuum carbonisation nanoparticulate platinum is formed in-situ and without encapsulation or blocking of the active catalyst surface (Figure 1). The microporous hetero-carbon host surrounding the platinum nano-catalyst ensures good electrical connection and therefore excellent catalyst utilisation. It is shown that PIM-EA-TB as a precursor host material can be employed to generate/maintain platinum nanoparticle size (at typically 1.0-2.5 nm) and to improve nano-catalyst performance by building up thicker layers of catalyst (with additive performance). Catalysts are prepared in a simple one-step vacuum carbonization process in which the PIM-EA-TB polymer is converted into an electrically conducting host (of similar porosity) containing the platinum nanoparticle catalyst in a way that it is accessible to the solution phase. In contrast to solution casting, spin-coating produces thinner films which, when carbonized, give highly effective platinum catalyst layers on transparent tin-doped indium oxide (ITO) electrode surfaces.

Instrumentation
Heating was performed in a TSH12 tube furnace (Elite Thermal System Ltd.). Pt nanoparticle materials were characterised by transmission electron microscopy (TEM) on a Jeol 1200EXII TEM system (with EDS and STEM). All electrochemical measurements were performed using an Ivium Compactstat potentiostat (The Netherlands) with platinum wire counter electrode and KCl-saturated calomel reference electrode (SCE, Radiometer REF401). The working electrode was prepared from ITO coated glass (tin-doped indium oxide films sputter-coated onto glass, active area 15 mm × 15 mm, resistivity ca. 15 Ω per square) obtained from Image Optics Components Ltd (Basildon, Essex, UK). XPS data were obtained on a Kratos Axis Ultra DLD system using a fixed anode monochromatic Al Kα X-ray source operating at 120 W.

Preparation of Thick Film Pt@cPIM on Tin-Doped Indium Oxide (ITO)
An aqueous platinum precursor solution was prepared by dissolving 10 mM K2PtCl6 in 0.5 M HCl (this solution was unstable as indicated by a slow colour change from yellow to orange and therefore stored in the dark at 4 o C). A tin-doped indium oxide (ITO) electrode was coated with 60 µL (1 mg/mL in chloroform) PIM-EA-TB in a drop-cast process giving approximately in average 550 nm thick deposits. Next, the ITO with PIM-EA-TB layer was dipped into the platinum precursor solution for 5 minutes (to allow sufficient time for ion exchange), then rinsed with water, and dried under ambient conditions. The colour changed from clear to yellow. Note that in the ratio of platinum atoms to nitrogen atoms in PIM-EA-TB treated with Pt(IV)Cl6 2is 0.08 (or 8%), which suggests that only partial loading of the polymer was achieved.

Preparation of Thin Film Pt@cPIM on Tin-Doped Indium Oxide (ITO)
PIM-EA-TB was dissolved in chloroform to prepare solutions with four different concentrations (0.25%, 0.5%, 1% and 2% wt.). Spin-coating of these solutions (WS-650Mz-23NPP (Laurell Technologies) using a rotation rate of 1500 rpm for 1 minute and ITO substrates of 15 mm × 15 mm area resulted in different thicknesses of PIM-EA-TB (vide infra).
Next, the coated electrode was immersed into the platinum precursor solution (as before) for 5 minutes, then rinsed with water and dried under ambient conditions. Then the ITO electrode was transferred to vacuum carbonization under oil pump vacuum (Edwards, ca. 0.1 mTorr) in a quartz tube at 500 °C for 3 h to give Pt@cPIM thin films with platinum nanoparticles embedded.

Formation and Characterisation of Thick Film Pt@cPIM
Drop-casting was employed with 60 µL of a PIM-EA-TB solution in chloroform (1 mg/mL) applied to an ITO electrode. With the measured film density of 1.1 g cm -3 for PIM-EA-TB [36] this allows an average film thickness to be estimated as 550 nm. With platinum bound into the film (see elemental analysis results in the experimental section) a coverage of about 6 µg Pt cm -2 is anticipated. Under PtCl6 2absorption conditions employed here only approximately 8% of the nitrogen atoms in the polymer are associated with a platinum. The vacuum carbonization process when applied to a film of PIM-EA-TB on an ITO substrate is known to produce a nitrogen-containing hetero-carbon with significant electrical conductivity, the characteristic Raman signature for a graphitic structure, retained cumulative pore volume/morphology and the black coloration of carbon [27]. The N2 adsorption data for PIM-EA-TB suggests a surface area of 1027 m 2 g -1 before carbonization [33] and 242 m 2 g -1 after carbonization [27].
Here, PIM-EA-TB only is vacuum carbonized to give cPIM and with absorbed PtCl6 2in the porous structure to give Pt@cPIM. Figure 2 shows transmission electron micrographic (TEM) evidence for the presence of cPIM carbon material and for the embedded Pt nanoparticles in Pt@cPIM (contrast Figures 2A and 2B). Higher magnification ( Figure 2C and 2D) allows individual particles (black or grey) to be analysed and the average particles size to be estimated as typically 1.0-2.5 nm. This value is similar to the micropore size for the PIM-EA-TB polymer starting material (from N2 gas adsorption isotherm data [33]). Electron diffraction data ( Figure   2E and 2F) clearly confirm that the carbonization process leads to inclusion of platinum

Formation and Characterisation of Thin Film Pt@cPIM
When employing spin-coating instead of drop-casting, better defined coatings of PIM-EA-TB can be produced. By changing the wt% or PIM-EA-TB in chloroform and keeping the spin coating conditions constant (see experimental) film thickness can be varied from about 50 to 250 nm (based on AFM cross-sectional data, see Figure D). These films are more uniform compared to deposits formed by drop-casting. AFM images of bare ITO and cPIM coated ITO are shown in Figure 3A and 3B consistent with a uniform coating. Perhaps surprisingly, when PtCl6 2is absorbed into the PIM deposit prior to vacuum-carbonization, the morphology is affected and only thinner deposits seem to remain (see the more apparent substrate grain boundaries in Figure 3C). This observation is confirmed by SEM imaging.   Figure 4A is formed from 1 wt% PIM in chloroform with an approximately 100 nm thin film clearly visible (containing PtCl6 2as "stain"). Films of vacuum-carbonized PIM (without Pt) also form well-defined films ( Figure   4B). However, when exploring cross-sectional images of 1 wt% and 2 wt% PIM films with PtCl6 2immobilised, it is apparent that a part of the carbon deposit is missing (removed by the vacuum carbonization process). Additional morphological images ( Figure 4G and 4H) show that platinum is present (as nanostructured deposit on top of the ITO crystals), but evidence for the amount of remaining cPIM substrate material is not clear. Next, XPS experiments were performed. Figure 5 shows survey data for PIM and PIM-Pt with the additional Pt signal clearly observed. After vacuum carbonization of a thin film of PIM-EA-TB (from 1 wt% solution), cPIM ( Figure 5C) shows features identical to those reported recently [34]. The higher oxygen signals in cPIM compared to PIM are assigned to ingress and reactivity of oxygen after vacuum carbonization when the sample is exposed to the open atmosphere. Nitrogen is present after carbonization, but the level appears lower relative to that for carbon. For the Pt@cPIM film ( Figure 5D) significant changes occur and signals from ITO are clearly evident (compared to Figure 5E). Carbon levels are higher compared to the bare ITO sample but lower when compared to the cPIM sample ( Figure   5C). Characteristic Pt4f peaks are observed (see insets) with a splitting into 7/2 and 5/2 states.
Binding energies for Pt4f in Figure 5B are consistent with the PtCl6 2precursor [38] and in Figure 5D with metallic platinum nanoparticles (see Table 1) [39].  It can be concluded that the vacuum carbonisation of thin film PIM is somewhat affected by the presence of PtCl6 2-(in part possibly due to the formation of volatile carbon halides) but that consistently platinum metal in the form of nanoparticles is formed. Even for the thin film Pt@cPIM deposits, there remains carbon to act as substrate and conductive support for the catalyst. The electrocatalytic reactivity of these materials is investigated next.

Electrocatalytic Oxygen Reduction
Platinum catalysts are employed widely in fuel cells and particularly in the oxygen reduction reaction [40], which provides an enormous challenge for sustainable production. Recent work on PIM-EA-TB stabilised platinum catalysts in oxygen reduction processes has been reported where the platinum nanoparticles and the carbon substrate were "protected" against accelerated corrosion [22]. Here, platinum nanoparticles are embedded into a heterocarbon host with sufficient porosity to allow reagent access to the catalyst and with sufficient electrical conductivity to allow effective potential control. Figure 6A shows cyclic voltammetry data for "thick film" Pt@cPIM catalyst for the reduction of oxygen saturated 0.1 M phosphate buffer pH 7 (trace iii). The reduction peak is detected at about 0.0 V vs. SCE.
When comparing with the reduction of 2 mM H2O2 in argon-degassed solution a similar reduction response is observed (trace ii) in addition to the voltammetric signal for the corresponding H2O2 oxidation. It seems likely that the reduction of oxygen under these conditions occurs predominantly with H2O as product (4-electron, see equation 1). Figure 6B shows data for the corresponding processes at a 2 mm diameter Pt disk electrode confirming the assigned mechanism.  The surface area of platinum increases systematically with Pt@cPIM film thickness as seen from the hydrogen-UPD region (see Figure 6C) and the oxide region. Similar experiments without PIM were performed by directly depositing H2PtCl6 platinum precursor onto ITO and thermal conversion without polymer ( Figure 6D). For this set of samples surface area of platinum can be seen to remain well below that observed for Pt@cPIM. Accordingly, the surface area and therefore the catalytic activity/utilisation of Pt@cPIM are improved due to the templating effect of the microporous polymer. Figure 6E shows Data are summarised in Figure 6F. The ESA value for platinum nanoparticles supported on carbonized PIM is substantially higher compared to that for platinum produced without PIM.
Data for ESA and for catalytic activity are also summarised in Table 2. Note that the ESA relative to the geometric surface area of the electrode reaches unity at about 1-2 µg cm -2 . That is a 1-2 µg cm -2 deposit offers a platinum surface similar to a bare platinum electrode. a area of platinum per geometric surface area of the electrode b anodic peak current for 1 M methanol in 0.1 M HClO4 during forward scan with 50 mV s -1 scan rate (see Figure 7).

Electrocatalytic Methanol Oxidation
Next, the reactivity of Pt@cPIM towards methanol oxidation was investigated. Small molecule [44,45] and in particular methanol oxidation to CO2 (equation 2) has been extensively studied as a fuel cell process [46,47]. Platinum is commonly employed as catalyst component in methanol fuel cell devices [48]. shows data for a 39 µg Pt deposit without PIM support. Figure 7B shows data for only platinum deposited from precursor onto ITO. The inset shows a plot correlating the peak current for methanol oxidation with the amount of Pt in Pt@cPIM. At substantially lower loading with Pt the Pt@cPIM catalyst performs substantially better. Figure 7E compares catalytic current per gram of platinum to give a measure of specific catalytic efficiency (mass activity based on the forward oxidation peak and the weight of platinum deposit). A loading of 2.3 µg platinum appears to give the most effective methanol oxidation catalysts. Additional experiments were performed to investigate the accelerated degradation of platinum catalyst under conditions of continuous potential cycling in aqueous 0.1 M HClO4 ( Figure 7C and 7D). Over 500 consecutive potential cycles the Pt@cPIM catalyst performs better (ca. 30 % loss of Pt activity) compared to the directly deposited Pt (ca. 70 % loss of Pt activity).

Electrocatalytic Glucose Oxidation
Glucose oxidation occurs on platinum electrode surfaces (mainly on catalytically active 100 oriented surfaces [49]) at relatively negative potentials within the hydrogen UPD region. The reaction mechanism is complex, but the main process has been suggested to be associated with a 2-electron transformation from glucose to gluconic acid or gluconolactone [50] (equation 3).
(3) Figure 8A shows cyclic voltammetry data for the oxidation of 10 mM glucose in 0.1 M phosphate buffer pH 7 as a function of scan rate. A first oxidation peak is observed at -0.4 V vs. SCE consistent with literature reports [51]. This oxidation peak seems to originate from the (111)×(100) UPD region which in turn implies that hydrogen adsorption onto the platinum surface is an intermediate reaction step in the process described by equation 3. In Figure 8B data obtained at a bare 2 mm diameter Pt disk electrode is shown and very similar features are observed. Figure 8C and 8D report the effect of changing the glucose concentration over a range from 2 to 14 mM in 0.  The oxidation of glucose on platinum is known to be very sensitive to poisons such as chloride [52,53] and even phosphate. The effect of adding chloride is demonstrated in Figure 9A and 9B for Pt@cPIM and for bare Pt, respectively. Both types of catalyst show substantial loss of activity in the presence of chloride. This needs to be taken into consideration in particular when using SCE reference electrodes, which tend to leak some chloride with time. In addition, phosphate buffer concentration effects the catalyst performance. Figure   Based on these data, it can be suggested that the microporous cPIM catalyst support for both thick film and thin film cases offers (i) sufficient electrical conductivity to allow catalysis at platinum nanoparticles, (ii) negligible surface blocking effects for embedded nanoparticles, and (iii) sufficient porosity to allow oxygen, methanol, glucose and similar small molecules access to the embedded catalyst nanoparticles.

Conclusion
It has been demonstrated that PIM-EA-TB can be employed to immobilise a PtCl6 2precursor for platinum by simple absorption from aqueous solution. Gentle vacuum carbonization then allows platinum nanoparticle catalyst of typically 1.0-2.5 nm diameter to be formed. The Therefore cPIM-Pt appears to provide maximum utilisation with all particles accessible and active in the molecularly rigid host material. At this stage, many experimental parameters remain unexplored and further study of platinum loading, carbonization conditions, and the effects introduced by different types of PIMs will be necessary. Powder samples will have to be prepared and applied as inks in order to allow more quantitative comparison to commercial standard catalysts.
We anticipate that the vacuum carbonization of PIMs with a range of different catalyst precursors will provide a general methodology for the production of high performance catalysts for various applications in electrochemistry. Instead of directly coating electrodes with films of PIM precursors, it will be beneficial to prepare bulk catalyst materials, which are then employed as an ink to apply to suitable electrode surfaces. The rigidity in the molecular backbone of PIMs ensures minimal loss of catalyst surface activity, whilst high temperature carbonization allows the metal catalyst to form and restructures the matrix to an electrically conducting but still microporous substrate. Further work is planned to explore other types of reactions and applications.