X-ray diffraction study of nickel oxide reduction by hydrogen

Abstract

Hydrogen reduction of porous bulk NiO particles has been studied with in situ hot-stage X-ray diffraction (XRD) in the temperature range 175–300 °C. This technique has the ability to measure NiO disappearance and Ni appearance simultaneously, together with the crystallite size of each. Since the sample was a very thin, 50-μm slab of dispersed 20-μm diameter grains, textural and morphological features normally encountered during studies with fixed beds of NiO particles were absent and measurements reflected only the chemical mechanism and kinetics.

The results indicated that reduction in the absence of water added to the reducing gas followed a series of steps: (1) an induction period associated with the initial reduction of NiO and the appearance of Ni metal clusters; (2) acceleration of the reduction rate as the size of the clusters increase; and (3) a pseudo-first-order (excess H₂) process in which NiO disappeared and Ni appeared in concert until reduction slowed at a fractional conversion of about 0.8. Crystallite size measurements showed NiO crystallites of about 3 nm in size were transformed into Ni crystallites of more than 20 nm, implying that Ni⁰ ion transport following reduction was very fast due to the close proximity of the NiO crystallites being reduced.

When 2.2×10⁻² atm of H₂O was added to the reducing gas, induction times increased by approximately a factor of two and reduction rates decreased (increasingly at lower temperatures) with an apparent activation energy of 126±27 kJ mol⁻¹ compared to 85±6 kJ mol⁻¹ without added water. The lag between NiO reduction and Ni growth observed in previous studies was not seen, indicating that textural and morphological factors are very important in establishing the role of water vapor in the reduction process.

Introduction

Reduction of nickel oxide by H₂

$$\text{NiO}\text{+}\text{H}{}_2\text{→}\text{Ni}\text{+}\text{H}{}_2\text{O}$$

is irreversible, since the equilibrium constant, K_p, varies from 10³ to 10² in the temperature range 0–1000 °C, and the reaction is only slightly exothermic. Extensive studies on reaction (1) have appeared in the literature, due to its practical importance in the reduction of nickel ores [1] and catalysts [2]. It is also a model reaction for oxide reduction mechanisms, since only two solid phases are involved (as contrasted to iron oxide [3]) and no excessive heat effects are present (as with CuO [4]).

The first reported systematic measurements of bulk NiO reduction was by Benton and Emmett in 1924 [5]. These authors measured water formation as an indication of the extent of reaction, and their results are reproduced in Fig. 1 for a sample of NiO made by heating nickel nitrate at 400 °C. Their seminal research introduced many of the conclusions regarding NiO reduction that are accepted today. These are: (1) reduction occurs at the interface between NiO and previously reduced Ni; (2) there is an “autocatalytic” effect, i.e. reduction rate is proportional to the interfacial area; (3) there is an induction (i.e. nucleation) period that depends on the nature of the sample and temperature; and (4) added water reduces the reduction rate and increases the induction period. This work was confirmed in 1930 by Taylor and Starkweather [6], who measured hydrogen consumption during reduction and demonstrated the catalytic effect by adding metallic nickel to the oxide.

These early experiments were followed by numerous studies that revealed the complexity of the reduction process, and results up to 1979 were reviewed by Boldyrev et al. [7]. Among features that influence the induction period are the presence of defects or altervalent ions in the outer surface of NiO grains [8], [9], [10], [11], [12], [13], thereby emphasizing the sensitivity of the initiation or nucleation process to impurities, preparational parameters and pretreatment. Even the thermomagnetic phase transition at 260 °C (the Neel temperature of antiferromagnetic NiO) produces an anomaly in reduction rate versus temperature curves [14].

Although emphasizing why experimental findings often differ from one study to another, these results also gave insight into the prevailing mechanism for chemical reduction. This is: (1) dissociation of H₂ (initially by NiO during the induction period, then by previously formed Ni metal at the growing interface); (2) surface diffusion of hydrogen atoms to a reactive center; (3) rupture of Ni–O bonds to produce Ni⁰ atoms; (4) nucleation of Ni⁰ atoms into metallic clusters; and (5) growth of nickel metal clusters into crystallites.

Since H₂ is often in excess, the kinetics of the overall process can be interpreted as pseudo-first-order

$$\text{−r}{}_{\mathrm{<mtext>NiO</mtext>}}\text{=(kP}{}_{\mathrm{<mtext>H</mtext>}}\text{)C}{}_{\mathrm{<mtext>NiO</mtext>}}$$

where k is the intrinsic chemical rate constant, P_H the partial pressure of H₂, and C_NiO is the concentration of NiO. The common practice has been to find (kP_H) from curves such as Fig. 1 by taking the slope at the point where it is most linear, usually at about 50% conversion.

Koga and Harrison, reviewed general reactions between solids and H₂ and represented the induction process as generation of nickel atoms on the outer surface of NiO grains [15]. Following nucleation, Ni clusters grow two-dimensionally across the surface until they overlap, at which point H₂ rapidly dissociates on Ni and the interface proceeds quickly into the grain. This growth of clusters is a special case of a class of solid phase mechanisms discussed in 1940 by Avrami, who developed the relationship given in Eq. (3) [16], [17], [18]:

$$\text{x}{}_{\mathrm{<mtext>NiO</mtext>}}\text{=1−}\text{exp}\text{(−kt}{}^{\mathrm{m}}\text{)}$$

where x_NiO is the fractional conversion at time t, k an overall rate constant, and m is an exponent whose value (0.5–4) depends on grain geometry and the limiting step (i.e. chemical nucleation or diffusion). Eq. (3) is sigmoidal in shape for low values of k and can account for many observed reduction curves. It is interesting that the curve through the points in Fig. 1 is Eq. (3) fitted to the data with m=1.34, suggests three-dimensional growth and instantaneous nucleation with the rate controlled by diffusion across the interface, but there is no other evidence to support this.

Bandrowski et al. [19] measured water produced during the reduction to generate sigmoidal conversion curves they explained with a two-step kinetic model. The first step, reaction between NiO and hydrogen atoms adsorbed on NiO, predominates in the early part of the reduction and is proportional to the square root of the H₂ pressure. The second step is reaction at the metal-oxide interface between NiO and hydrogen atoms adsorbed on previously reduced NiO. This step is independent of H₂ pressure and dominates in the final stages of reduction. Although this model fits the data above 300 °C, the authors concluded that H₂ diffusional effects occurred below 300 °C. They verified reduction retardation by added water and introduced a Langmiur–Hinshelwood adsorptive term to the rate equation:

$$\text{−r}{}_{\mathrm{<mtext>NiO</mtext>}}\text{=}\text{(kP}{}_{\mathrm{<mtext>H</mtext>}}\text{)C}{}_{\mathrm{<mtext>NiO</mtext>}}\text{1+K}{}_{\mathrm{<mtext>w</mtext>}}\text{P}{}_{\mathrm{<mtext>w</mtext>}}$$

where P_w is the partial pressure of water vapor and K_w an adsorption coefficient. However, the Avrami model also gives just as good a fit (m=2.07) for their published data at 295 °C.

These early models emphasized the chemical mechanism and kinetics of the reduction process, but there was little uniformity in the physical properties of NiO grains and pellets used. Nevertheless, it was suspected that morphological factors were just as important as topological properties in determining the course of the reduction. This was demonstrated by Moriyama and Yamaguchi [20], who found that reduction rate constants were inversely proportional to grain size above a diameter of about 10 μm [21]. Such dependence is predicted by the classical shrinking core model [22], in which the Ni–NiO interface moves towards the center of the grain, leaving behind a porous metallic product layer through which H₂ diffuses in and H₂O out. Various rate equations can be derived, depending on which process is slowest. With chemical reaction at the interface controlling, the model leads to Eq. (5):

$$\text{x}{}_{\mathrm{<mtext>NiO</mtext>}}\text{=1−(1−k}{}_{\mathrm{<mtext>c</mtext>}}\text{t)}{}^3$$

where k_c=(kP_H)/R_g (R_g is the grain radius).

Delmon [23] argued that grain size dependence is consistent with nucleation at the interface. As R_g decreases, the rate increases, first according to the shrinking core model and then progressively slower, since nucleation becomes increasingly rate-limiting and k_c decreases at lower sizes. According to Delmon and Pouchot [24], the limiting value for uniform behavior (constant k_c) is R_g=10 μm.

The simple shrinking core model assumes the NiO grain or particle is non-porous, and Eq. (5) applies only under these circumstances. For example, Sridhar et al. [25] studied the reduction of NiO particles calcined at 1000 °C and presumably non-porous. These authors found satisfactory fits to Eq. (5) up to NiO conversions of 0.8, above which reduction became slower. They speculated that the increasingly thick (and perhaps dense) product layer made diffusion of water out of the particle more difficult.

Reduction of porous particles was addressed by Szekely and coworkers [26], [27], [28], [29] and applied to experimental measurements on NiO [31]. These authors viewed the pellet or particle as a porous agglomeration of NiO grains each undergoing a microscopic shrinking core reduction. Depending on temperature, pellet size and porosity, asymptotic solutions reflect either chemical reaction control (low temperatures, small pellets, high porosities) or internal diffusion control (high temperatures, large spheres, low porosities, towards the end of the reaction). For the first case, the reaction proceeds uniformly throughout the pellet, with each grain undergoing the same shrinking core kinetics, unless there is a distribution of grain sizes. For the second, there exists a sharp reaction front or interface in the pellet that moves progressively towards the center. Predictions for mixed control show a diffuse front moving inward. The most serious restriction of the model is the requirement that the pellet maintain its structure, i.e. the individual grains retain approximately the same size and location as NiO converts to Ni. Since the model is empirical without analytical solutions, it has not been a convenient tool for kinetic interpretation.

In studies of NiO reduction Richardson et al. conducted at series of experiments using isothermal H₂ consumption and magnetization measurements to determine the Ni–O bond rupture (NiO conversion) and the growth of nucleated Ni⁰ atoms (Ni growth), respectively [32]. They found the growth process lagged NiO conversion by a time interval that increased with decreasing temperature, lower gas flow rates and the presence of H₂O added to the reducing gas. They suggested that adsorbed H₂O molecules, produced by the reaction or added to the reducing gas, retard the growth process by limiting diffusion of Ni⁰ atoms to appropriate nucleation centers. However, since the H₂ consumption and magnetization measurements were separate experiments, both using small packed beds of NiO grains, there was some concern that the observed “lag” could be an experimental artifact.

In this paper, we report the results of a NiO reduction study using a technique that avoids this objection: in situ hot-stage X-ray diffraction (XRD) that measures NiO loss and Ni appearance simultaneously. Together with the ability to provide NiO and Ni crystallite sizes during reduction, this method uses a thin slab sample geometry that avoids problems inherent to packed beds.