Historical Platinum Group Metal Use in Three-Way Catalysts

Automotive catalysts have been offsetting vehicular exhaust emissions for almost half a century. When Johnson Matthey began the design of automotive exhaust aftertreatment catalysts in the 1960s, it was platinum-based technology that steered the way, with the first-generation automotive catalyst introduced in the 1970s (1–3). More specifically, the primary gasoline exhaust aftertreatment system started out as a primitive ‘two-way’, rhodium-promoted-platinum oxidation catalyst (1–3). By the 1980s, the technology had developed into a dual component platinum-rhodium ‘three-way catalyst’, so-called for its ability to complete NOx reduction reactions simultaneously to oxidation reactions of CO and HCs with the help of engine calibration (1–4). Since the implementation of TWCs in commercial systems, there have been significant efforts in both industry and academia to explore cheaper alternatives to pgm based systems, but without widespread adoption in commercial applications. Ultimately, only three metals from the platinum group achieve the required performance, are sufficiently abundant and stand up to the harsh operating conditions of a gasoline engine: platinum, palladium and rhodium.

Research efforts have undoubtedly driven developments in TWC technology over time, but so too have external factors like pgm cost. Notably, the relatively high platinum prices of the 1990s, as seen in Figure 1(a), motivated efforts to substitute with the then cheaper palladium. However, while cost pressures drove these changes in the catalyst technology, it was ultimately external factors like improvements in fuel quality (4) that enabled them to be implemented successfully.

The late 1990s saw a move towards desulfurisation of fuel as sulfur limits were beginning to be imposed by governments around the world with the intention of reducing both sulfur-containing emissions and catalyst deactivation (5). Sulfur is a major catalyst poison for pgms to contend with in commercial systems, with sensitivity to deactivation following the order of palladium>platinum>>rhodium (5), but once sulfur was effectively removed from gasoline, the performance of palladium was better able to compete with that of platinum.

In theory, platinum and palladium are both capable of catalysing oxidation reactions to the level expected for modern three-way technology and thus should be interchangeable from this respect. However, their properties differ considerably as do their optimal operating conditions. For example, palladium’s greater sintering resistance enabled TWCs to be physically shifted closer to the engine when it largely replaced platinum as the dominant pgm by the late 1990s. This shift in position took advantage of faster catalyst warm up to reduce the slip of emissions at cold start. These two changes of desulfurisation and catalyst positioning led the way to today’s widespread TWC palladium-rhodium formula, which can realise greater than 99% emissions conversion.

Since the early 1990s, the cost of platinum has generally exceeded that of palladium, however a turning point was reached around 2018 when high demand for palladium and rhodium coincided with a stall in demand for platinum. Thorough accounts of supply and demand histories dating back to 1999 are offered through Johnson Matthey’s annual PGM Market Reports. As of 2022, the platinum price is less than half that of palladium (6). For this reason, impetus to shift the dominant pgm in commercial TWCs has returned as a key market demand, but the tides have turned with the push now to remove palladium and reincorporate platinum. As we enter a new era for the internal combustion engine, we are revisiting the origins of the technology. Can we simply shift back to the original platinum-rhodium TWC? And if not, what other options do TWCs have to make use of platinum once more?

Challenges for Platinum Incorporation in Today’s Three-Way Catalyst

Today’s TWC technology generally comprises a monolith with a catalytic coating containing palladium and rhodium supported on rare earth (RE) stabilised alumina and RE stabilised ceria-zirconia (CeZr) mixed oxides, usually with additional promotor elements like barium. In addition to catalytic activity, durability is a crucial property of TWCs with commercial systems needing to maintain high levels of performance after exposure to temperatures in the region of 950–1050°C for many hours, with regular and extreme fluctuations between reductive and oxidative atmospheres. These conditions present a greater challenge for platinum than palladium, in part due to platinum’s susceptibility to both types of sintering mechanisms experienced under TWC conditions: particle coalescence and Ostwald ripening (7).

The first of these mechanisms, particle coalescence, involves the diffusion of nanoparticles (NPs) over the support surface, causing them to collide and agglomerate into larger NPs. This process is more easily facilitated when pgm particles are reduced to their metallic form and oxygen bonds with the support are broken. Even for support materials which would usually afford pgm particles some sintering resistance through strong metal support interactions (8), the reducing conditions experienced during TWC operation can be sufficient to overcome these stabilising effects. Platinum exhibits a lower oxide decomposition temperature compared to palladium or rhodium as seen in Table I, which means that it is in its metallic form independent of the gas conditions, even when only operating at moderate TWC temperatures of ~600°C. Relative to palladium and rhodium, platinum particles therefore spend a greater proportion of time susceptible to coalescence effects.

The second mechanism, Ostwald ripening, is far more detrimental to platinum dispersion. This process involves the oxygen-mediated movement of individual atoms across a support surface or through the gas phase. The loss and reincorporation of atoms between NPs leads to a net transfer of atoms from smaller particles to larger ones, resulting in sintering. The gas phase transport process is enabled through the formation of volatile gas-phase oxides when pgms are heated in an oxidising atmosphere at high temperature. Platinum in particular is known for its proclivity to form such volatile oxides to relatively high vapour pressure and well within the temperature range of TWC operation (12). Palladium, on the other hand, volatilises to an almost negligible degree in this temperature regime (13). The culmination of these properties is effectively demonstrated by comparing lanthanum-doped alumina-supported palladium and platinum catalysts that have been exposed to moderate calcination temperatures by transmission electron microscopy (TEM), shown in Figure 2. The average pgm particle size of the platinum catalyst is an order of magnitude greater than the palladium catalyst.

The result shown in Figure 2 is in agreement with other studies which show that platinum sinters in air from ~580°C (14–16). As previously mentioned, the average operating temperatures of TWCs are commonly upwards of 600°C (post light-off). Platinum therefore cannot redisperse via practical pgm redispersion routes, i.e., the oxidising conditions of the gasoline engine and air, without further measures put in place to inhibit sintering via Ostwald ripening, such as atom traps. Palladium on the other hand can undergo some redispersion in oxidising gasoline conditions below the palladium oxide decomposition temperature and following relevant TWC ageing conditions (17, 18). Sintering prevention is therefore key to preventing the permanent loss of platinum dispersion. There are many pgm stabilisation strategies, but an example which has recently received more attention in the literature is that of atom-trapping support technology (19). When applied to platinum-based catalysts, such strategies utilise the otherwise problematic issue of platinum volatility to capture and stabilise mobile molecular species on undercoordinated surface sites of a support material, which then show increased resistance to re-agglomeration (20–25). These specialised sites are maximised on high surface area materials with highly defective surface topographies, such as ceria. Studies that successfully employ atom-trap materials use mild conditions of heating at 800°C in air, i.e. diesel ageing conditions, to volatilise the platinum (20–25). However, materials such as pure ceria (which have the greatest potential for use in such a strategy) lack the stabilisers and dopants which are generally considered essential to endure TWC ageing conditions (26, 27). Preparing stable, atom trapping materials that inherently have the high energy surfaces required is therefore a paradoxically challenging research area for pgm stabilisation in TWC applications.

Another challenging consideration of note in the drive to reincorporate platinum into TWCs which is often ignored is the greater atomic mass of platinum (195 u) compared to that of palladium (106 u). Palladium sold on a weight basis contains almost twice the number of atoms as the equivalent mass of platinum. Considering the relative prices of the two metals as of 2022, the cost of palladium and platinum atoms are quite comparable. Successful palladium replacement on a weight basis therefore demands that each platinum atom be essentially twice as active as a palladium atom.

Strategies for Overcoming Platinum’s Challenging Properties

Direct Exchange of Palladium for Platinum

There are clearly challenges that must be addressed when considering platinum incorporation into modern TWCs. As a starting point, a brief exercise in swapping out palladium for platinum on representative TWC support materials was explored. Catalysts were compared through simulated activity testing of pelletised powders in a plug flow reactor under a gasoline-type rich/lean perturbed gas mix with the temperature ramped from room temperature to 500°C. Descriptions of experimental and ageing conditions can be found elsewhere (28). Where appropriate, the T₅₀ light off temperature (temperature at which 50% conversion is reached) was selected for comparisons of catalyst activity. Prior to evaluation, catalysts were hydrothermally aged at 950°C under a simulated gasoline-type rich/lean perturbed gas mix. Whether supported on RE-doped CeZr or RE-doped alumina, platinum shows a substantial loss of light-off activity compared to palladium when substituted on a 1 wt% loading basis, as shown in Figure 3. Given the aforementioned challenges platinum faces under typical gasoline operating and ageing conditions, it is perhaps unsurprising this switch results in decreased activity. More notably, this data elucidates palladium’s similar light-off activity irrespective of its support whereas platinum has a significant activity difference; the platinum catalyst exhibits a considerable detriment to light-off activity when supported on alumina. Note that the CeZr based catalysts (which have oxygen storage capacity) achieve high final conversions of >85% by buffering the perturbing gas conditions of the test.

Optimising the Ceria-to-Zirconia Ratio for Platinum Three-Way Catalysts

The above results highlighted a potential pathway by which platinum’s activity can be improved and prompted a study to examine 1 wt% platinum and palladium catalysts supported on a range of CeZr mixed oxides of differing ceria-to-zirconia ratios. All catalysts were hydrothermally redox aged at 950°C. Descriptions of experimental and ageing conditions can be found elsewhere (28). As seen in Figure 4, there is a strong dependence of the CO T₅₀ value on the ceria-to-zirconia ratio for the platinum catalysts, whereas evaluation of the palladium catalysts reveals no relationship. This observation emphasises the very different catalytic properties of these two pgms which must be considered when trying to use them interchangeably for TWC applications. At low temperatures, and hence during catalyst light-off, platinum particles can be inhibited by rich gas conditions, i.e. platinum is temporarily poisoned by CO (29, 30). The initial light-off ‘hump’ seen in Figure 3 for the platinum-alumina catalyst illustrates this inhibitory behaviour. Interactions between platinum and ceria appear to reduce these inhibition effects. Many studies have explored the phenomenon of reverse oxygen spillover across the support-metal boundary from ceria to platinum at low temperatures. It is suggested that this ‘activated oxygen’ initiates lower temperature CO oxidation and keeps catalytic sites clear at the pgm-support interface, thus alleviating some inhibition effects (29, 30). Whilst there are certainly activity and durability benefits to be gained from optimising the ceria content to help overcome some of platinum’s inherent challenges, an alternative strategy discussed in the final section of this paper actually seeks to turn these challenges into an advantage for promotion of overall TWC performance.

Partial Platinum Substitution: Rhodium-Platinum Alloys

As clearly laid out in previous sections of this paper, platinum has a number of inherent properties which make wholescale replacement of palladium in TWCs a challenging prospect. One alternative strategy for achieving this goal is designing ‘trimetal’ TWCs by partially replacing some of the more expensive palladium or rhodium with platinum; specifically through co-location of platinum with palladium or rhodium on a support material. Given the relatively high cost of rhodium as of 2022 as well as general price volatility, illustrated in Figure 1(b), partial replacement of rhodium with platinum has been chosen as a case study to illustrate this strategy considering the dual potential benefits of platinum incorporation and rhodium thrifting.

Rhodium catalysts partially replaced with platinum were prepared at a pgm loading of 0.2 wt% and supported on a RE-doped CeZr with ~38 mol% ceria. In addition to the monometallic platinum-only and rhodium-only reference catalysts, bimetallic rhodium-platinum catalysts with rhodium-platinum weight ratios of 5:1, 2:1 and 1:1 were prepared. These ratios were chosen to maintain an overall rhodium character and activity. All catalysts were hydrothermally redox aged at 950°C. Descriptions of experimental and ageing conditions can be found elsewhere (28). In order to emphasise the themes of rhodium thrifting and stabilisation alongside platinum substitution, an oxidising cooldown step was chosen for the ageing procedure. Such shutdown conditions are particularly stressful for rhodium performance given its susceptibility to form inactive oxides under oxidising conditions until exposed to sufficiently reducing conditions to recover the metallic state. Catalysts were evaluated through simulated activity testing as described previously, with the nitric oxide (NO) T₄₀ value chosen for comparison as the rhodium component of a TWC is especially important for controlling NO emissions. T₄₀ (and not T₅₀) values were chosen to enable comparison to the two ≤0.13% rhodium-only references which had not reached 50% conversion by the end of the test at 500°C. The results presented in Figure 5 reveal that activity is maintained upon replacing rhodium with platinum up to the maximum substitution of 50%. Comparison of the activities of the 0.1 wt% and 0.2 wt% rhodium-only references with the 1:1 rhodium-platinum catalyst reveals the strong interaction that low rhodium loadings can have with a support following ageing (31, 32). A synergistic interaction between rhodium and platinum is evidently able to limit this deactivation mechanism.

Physical characterisation methods were employed to probe the synergistic relationship apparently underpinning the enhanced catalytic behaviour, with the pgm loading increased to 3 wt% to emphasise observable effects. Both X-ray diffraction (XRD) and energy-dispersive X-ray spectroscopy analysis via TEM (EDX TEM) results clearly evidence the presence of bulk rhodium-platinum alloy particles as seen in Figure 6. This observation aligns with expectations from literature based on the phase diagrams (33) of rhodium-platinum (and palladium-platinum), which indicate the ageing temperature employed, and thus the extreme operating temperatures of TWCs, are sufficient to induce rhodium-platinum (and palladium-platinum) alloy formation.

CO probe Fourier transform infrared spectroscopy (CO-FTIR) was employed to examine the nature of the catalytically active rhodium-platinum surface. Figure 7 shows the first scan data taken immediately following the introduction of CO at room temperature and compares rhodium-only and platinum-only references with the rhodium-platinum 1:1 alloyed catalyst at 0.2 wt% pgm loading. Prior to CO introduction, self-supporting pellets of catalyst of 13 mm diameter were subjected to a 5% hydrogen pretreatment for 10 min at 250°C. CO-probe data was collected at 30°C in a flow of 500 ppm CO in helium. Difference spectra were obtained by using t = 0 as the background for subsequent spectra under CO flow. Difference spectra were normalised by the weight of the pellet used. The difference spectra shown here are at t = 5 min. Note that the negative band at ~2127 cm^–1 is ascribed to Ce³⁺ defects induced by the hydrogen pretreatment. As CO is introduced over the catalysts, this band is steadily lost due to direct interaction of CO with the defect (34–37). The rhodium and rhodium-platinum catalysts present major bands at 2003 cm^–1 and 2073 cm^–1, assigned to the symmetric and antisymmetric stretching modes of gem-dicarbonyl Rh⁺(CO)₂ (38, 39), i.e. CO adsorbed to dispersed, oxidic rhodium species. It is apparent that the spectrum of the rhodium-platinum catalyst closely resembles that of the rhodium-only reference, and in fact, reveals similar intensity in the Rh⁺(CO)₂ bands, indicating similar rhodium surface area despite there being half the amount of rhodium present in the catalyst overall. Platinum, therefore, is able to promote a similar accessible rhodium surface area (compared to the addition of the same weight percentage of rhodium as seen in the rhodium-only reference) during exposure to the excess-oxygen conditions at high temperatures. This accounts for the equivalent light-off activity seen for the rhodium-platinum catalysts.

As previously discussed, platinum is easily sintered by TWC conditions. Moreover, it does not recover a bulk reoxidised form following sintering of its metallic state. It reforms a passivating oxygen layer at its surface only (40) and by this definition, is the most noble of the three pgms used for TWC applications. It is proposed that when alloyed with rhodium or palladium, platinum lends some of its nobility to form more metallic particles that show greater resistance to severe deactivation with the support during oxidising conditions, enabling the use of, in the case examined here, lower rhodium loadings. It is further suggested that the oxidic rhodium seen via the CO-FTIR results, i.e. Rh⁺(CO)₂, resides at the surface of stabilised bulk rhodium-platinum alloyed particles, demonstrated by the XRD and EDX TEM measurements. This proposed structure is illustrated in Figure 8.

This rhodium-platinum alloy technology demonstrates not only a route by which platinum can be incorporated into TWCs, but also one where the challenging properties of platinum can be repurposed towards promoting the activity and durability of one of the most important, and costly, TWC components. As previously stated, the first generation TWCs were rhodium-promoted platinum oxidation catalysts. This study has shown that platinum has the potential to return the favour by promoting today’s rhodium TWCs to mitigate rhodium’s less-desirable properties.

Conclusions

The requirements of the TWC over the past ~25 years have evolved far beyond the capabilities of the original platinum-rhodium formulations. Today’s palladium-rhodium TWC has been fine-tuned to achieve current emission limits, including optimisation of supports and promotors. While pgm price fluctuations take place over a relatively fast timescale, prompting the requirement for pgm flexibility, similar platinum TWC optimisation efforts should be expected in order to see platinum reinstated in palladium’s place. Non-equivalent performance from straight exchange to the otherwise unchanged TWC technology is the beginning of the story, not the end! It is evident that there are options and furthermore, opportunities for platinum in returning to the TWC. Results shown here demonstrate that platinum activity and stability are enhanced by increasing the ceria-to-zirconia ratio, with further optimisation opportunities to hone platinum’s ultimate TWC performance. Moreover, it was shown that platinum can be effectively substituted for a portion of the other pgms in the system due to the synergistic nature of the alloys formed. And beyond this substitution achievement, it was demonstrated that mixing the properties of platinum with those of rhodium can positively impact the TWC performance of these existing components.