Promotion of reduction in Ce0.5Zr0.5O2: the pyrochlore structure as effect rather than cause?

Evidence is presented that the well-established phenomenon of promotion of Ce0.5Zr0.5O2 reduction by suitable treatment procedures may be controlled by adjusting the conditions; and that such reduction leads to the formation of the pyrochlore structure at extremely low temperature.

The redox properties of ceria–zirconia mixed oxides (CZMOs) have been extensively investigated due to their use in automotive three-way catalysts (TWCs), in which CZMOs are included as the oxygen storage component.1

Correlation between structure and ease of reduction, which has received particular interest, remains an open question.

The Ce1–xZrxO2 phase diagram2 may be summarised as follows: for x ≤ 0.15 a cubic, fluorite-type phase is formed, while for x ≥ 0.85, a monoclinic phase is present.

At intermediate compositions, various phases (t, t′, t″, κ and t*) have been identified.2–4

Oxygen deficient structures, such as the pyrochlore structure (A2B2X7), may also be formed.5

This is an ordered defect structure, related to that of fluorite, with ⅛ of the anions missing.

In practice, pyrochlores are often not stoichiometric and are more correctly represented as A2B2X7+y.

Current interest in the pyrochlore structure in CZMOs is strong as its formation has been suggested to enhance oxygen storage behaviour.3,4,6,7

For CZMOs, they are generally reported to form under severe reducing conditions.5

Here, we report evidence for the formation of a pyrochlore for Ce0.5Zr0.5O2 during low temperature reduction of the sample.

Ce0.5Zr0.5O2, BET surface area 40 m2 g–1 after calcination at 773 K (5 h), was prepared by a previously reported citrate route.8

An in-situ cleaning procedure (CP) was applied before each set of experiments.

Temperature programmed reduction (TPR) profiles (25 mg of sample) were recorded in a system equipped with a thermal conductivity detector: CP-O2 at 773 K (1 h), then cool, first to 423 K in O2, then to room temperature (rt) in Ar; TPR-H2 (10 or 1%) in Ar at rt (30 min), ramp to 1173 or 1273 K, hold 15 min, switch to Ar (15 min).

TPR profiles are designated as severe reduction (SR) procedures; after TPR, cool to 700 K and oxidise (mild oxidation—MO) in O2 (1 h), or oxidise at 1173 or 1273 K (severe oxidation—SO).

O2 was pulsed in all cases (100 µl, every 75 s).

Flow and ramp rates: 25 ml min–1 and 10 K min–1 respectively.

Powder XRD spectra were collected on a Siemens Kristalloflex Mod.F (Ni-filtered Cu Kα).

TPR-Raman spectra were obtained on a Renishaw Raman System 1000, with an Ar laser (514 nm, 25 mW) fitted with an in-situ cell that allows control of temperature and gas flow.

Reduction was performed by introducing the reducing gas (1 or 10% H2 in Ar), heating in a stepwise manner (50 K intervals) and recording the spectra during the isothermal steps (15 min).

Oxidation was performed in the same cell at the same temperatures as for TPR experiments.

It is known that the temperature at which CZMOs release oxygen under reducing conditions may be decreased or increased upon application of SR/MO or SO treatment regimes.3,4,8,9

This effect is reported in Fig. 1, which shows a series of TPR profiles obtained using 10 and 1% H2.

SR and SO were conducted at both 1173 K (A) and 1273 K (B).

Due to the different response to 1 and 10% H2, the profiles shown have been normalised to the total area detected.

Profiles A1, A2, B1 and B2 were obtained after: SR–MO–SR–SO.

All show H2 uptake at relatively high temperature, the maximum of which depends on H2 concentration.

The latter is an established TPR phenomenon.10

After intermediate MO procedures, further TPR profiles (A1′, A2′, B1′ and B2′) were recorded.

The most striking feature of this data set is that, for SR and SO at 1173 K (A), the procedure followed results in low temperature reduction with 10% H2 but not with 1% H2.

It should be noted that the profile always remains on-scale with 1% H2.

Thus, reduction is not limited by H2 concentration.

For SR and SO at 1273 K (B), low temperature reduction is observed with both H2 concentrations, but the effect is more pronounced with 10% H2, in the sense that the relative intensities of the two peaks observed indicate more uptake in the low temperature feature.

Cycling between the related profiles in Fig. 1B may be achieved by further SO and SR–MO treatments.

Thus, aspects of the behaviour reported in Fig. 1 are consistent with previous observations.8,9

However, here we show that the effect is also strongly dependent on H2 concentration and temperature, in addition to the known dependence on the oxidation conditions (SO vs. MO).9

Rietveld analysis of the XRD profile after SO at 1173 and 1273 K initially indicated that the structure may be fitted equally well using cubic (Fm3m) or tetragonal (P42/nmc) models.

Careful collection of XRD data revealed the presence of a low intensity characteristic t′ reflection (2θ = 42°).

Similar analysis after subsequent SR (corresponding to 1A2 and 1B1)–MO showed only a small cell volume decrease from the SO-treated samples, suggesting that the procedures do not induce significant changes in the cation sub-lattice, at least on the XRD scale of detection (see ESI).

Fig. 2 shows two series of in-situ TPR-Raman spectra obtained during reduction of Ce0.5Zr0.5O2 using 1% H2 (A) and 10% H2 (B).

These Raman spectra offer a representative study of the TPR profiles A2′ and B1′ shown in Fig. 1.

In both cases, there was negligible change in the position and a small decrease in intensity of the peaks between rt and 473 K (first spectra shown).

It is important to underline the possibility of superimposed luminescence signals, derived from rare earth impurities at ppm levels, in these spectra.11

The intensity of such peaks strongly depends on factors such as temperature and environment, so great care must be taken when comparing spectra for structural analysis.

For the sample treated to 1173 K in 1% H2 (Fig. 2A), the first spectrum shows peaks at 225, 306, 470, and 623 cm–1 which are consistent with the presence of a t′ phase.12

As the reduction temperature increases, the peak intensities decrease, until, at temperatures corresponding to complete reduction of the sample (Fig. 1A2′), a featureless spectrum is observed.

No Raman features could be observed at 1173 K and above due to the blackbody radiation, which overwhelms the Raman signal.

In the case of the sample treated to 1273 K in 10% H2, the first spectrum shows peaks at 184, 301, 470, 540 and 623 cm–1, which are also consistent with the presence of a t′ phase.

Between 523 and 573 K, after commencement of sample reduction (Fig. 1B1′), peak intensities decrease strongly.

At 623 K a weak peak appears at 295 cm–1, which becomes more intense and shifts to lower wavenumber as the temperature increases, while others appear at 380 and 495 cm–1.

These peaks are consistent with the formation of a pyrochlore phase.3

It is clear that the temperature at which the pyrochlore structure becomes visible (673 K, Fig. 2) is well above the start of the reduction profile (<500 K, Fig. 1B1′).

The difference in experimental procedure (ramped vs. stepped temperature increase) should mean that, at a given temperature, the sample is more reduced during TPR-Raman than during TPR.

Finally, after a TPR-Raman series corresponding to Fig 1B1, the sample was cooled to 700 K and spectra were measured before and after oxidation.

No evidence of the pyrochlore or related structures was found, indicating that the pyrochlore was not formed at high temperature during the previous TPR.

The metastable nature of CZMOs with intermediate compositions has created difficulty in establishing certainties about the structure of such materials; which has generally complicated structural analysis of TPR behaviour.

Despite this, the observation of promoted TPR profiles has recently been linked to a newly reported κ phase, formed by oxidation (873 K) of Ce0.5Zr0.5O2 with the pyrochlore structure and destroyed by high temperature oxidation (1323 or 1423 K) to yield another new phase (t*).3,4

The pyrochlore was initially formed by high temperature reduction of a t′ phase.

Here we present evidence of formation of the pyrochlore phase during low temperature reduction.

Clearly, the preceding treatments have already conferred upon the sample the potential to undergo low temperature reduction, which in turn leads to very low temperature pyrochlore formation.

Neither effect is observed with 1% H2, SR and SO 1173 K, thereby implying a link between promoted reduction and low temperature pyrochlore formation.

Table 1 reports the degree of reduction for the eight TPR profiles of Fig. 1, as measured by O2 uptake during MO.

Comparable and high degrees of reduction were achieved in all experiments suggesting that formation of pyrochlore should be equally favoured.

Despite this, pyrochloric structure was observed only in Fig. 2B, corresponding to TPR profile B1′, thereby indicating that extent of reduction is not directly responsible for pyrochlore formation.

Finally, the Raman spectra recorded after the CP, in Fig. 2 and at the start of TPR profile B1 (not shown) may all be assigned to t′.

We find no clear correlation between the appearance of the Raman spectra and reduction profile (or formation of pyrochlore), although it must be reiterated that the possible presence of impurity peaks makes identification of any such correlation difficult.

Instead, we believe that a more complex situation than a simple structure–TPR relationship may be inferred for the TPR behaviour of this sample.

The formation of surface domains which promote reduction of the sample could instead offer an explanation.8

The kinetics of the pyrochlore formation at high temperature has not been carefully studied.

Therefore, as suggested by an anonymous referee, it cannot be excluded that the pyrochlore nucleation process occurs in high temperature conditions (its rate being increased by increasing hydrogen pressure), while the growth process could occur with a significant rate at moderate temperatures.

Once formed superficially at high temperature, a very thin pyrochlore layer (the nuclei) could be transformed into the κ phase under mild conditions, which in turn gives the pyrochlore structure (as in Fig. 2b) at low temperature, first on the surface then inwards of the particle (growth process).

Such a process would be difficult to observe given the detection limits of the techniques employed here.

In addition, the existence of antiphase domain boundaries may further weaken the XRD peaks due to the κ phase.

On the other hand while the Raman technique proved to be sensitive in detecting the κ phase,13 no evidence for its presence was observed here.

Further studies are in progress in order to ascertain these aspects.

Consistent with the suggestion of the importance of surface modifications, studies carried out on the present sample indicated that the scrambling capacity between D2 and surface hydroxyl groups is affected by the redox treatments.8

In summary, formation of a bulk pyrochloric structure, i.e. an ordered system, is not a necessary prerequisite to achieve promoted redox properties in Ce0.5Zr0.5O2.