Effect of cooling rate and chromium doping on the microstructure of Al-25 at.% Ni Raney type alloy

Highlights

• Microstructure and phase composition of Cr doped Raney catalyst is investigated.

• Microstructure appears more dendritic in particles with a higher cooling rate.

• 1.5 at.% Cr doping affects the microstructure with the appearance of more dendrites.

• A fourth phase, Al₁₃Cr₂, was found on the boundary between NiAl3 and Al eutectic.

• A model for the enhanced activity of Cr doped Raney catalyst is proposed.

Abstract

Al-25 at.% Ni and Al-23.5 at.% Ni-1.5 at.% Cr alloys were synthesised via gas atomisation to study the effect of rapid cooling on the microstructure and phase composition of Raney type catalyst precursor powders. In the undoped powders, the three phases, Al₃Ni₂, Al₃Ni, and Al-Al₃Ni eutectic were identified, while in the Cr-doped powders the additional phase Al₁₃Cr₂ was also identified. Extensive substitution of Ni onto the Cr lattice sites is observed, which generates the observed phase fraction of Al₁₃Cr₂. Elemental mapping and quantitative image analysis of backscattered electron micrographs indicates that the Al₁₃Cr₂ phase precipitates late in solidification, probably direct from the melt, during the final stages of Al₃Ni growth. As such, this explains previous observations that Cr is found on the surface of the activated catalyst without the need to invoke migration of Cr. Leaching of such an Al-rich compound offers a plausible explanation for the enhanced catalytic activity observed in Cr-doped Raney catalysts.

Introduction

Skeletal metal catalysts, such as Raney Ni, are a popular choice of catalyst for hydrogenation reactions. The traditional Raney Ni alloy, discovered by Murray Raney, comprises a 50-50 wt% mixture of Ni and Al, although due to the lower atomic mass of Al compared to Ni, this is approximately 68.5 at.% Al. The production of Raney Ni catalysts is traditionally performed by castings the alloy into ingots, which are then crushed into a coarse powder to allow activation by leaching.

It has been determined that the precursor Al-Ni alloy contains 3 phases, irrespective of the initial alloy composition; Al₃Ni₂ (Space group: P$\overline{3}$m1 [1]), Al₃Ni (Space Group: Pnma [1]) and an Al-Al₃Ni eutectic [2]. With reference to the Al-Ni phase diagram (Fig. 1), the first phase to appear from the Ni-68.5 at.% Al melt will be AlNi when cooling to 1623 K. At approximately 1406 K the AlNi will react with the remaining liquid via the peritectic reaction:

$$L+AlNi\leftrightarrow A{l}_{3}N{i}_2$$

to give Al₃Ni₂. It has been shown that during the slow cooling appropriate to producing large ingots, and also at higher cooling rates such as during gas atomisation, the final microstructure of the Ni-68.5 at.% Al alloy rarely contains AlNi. This has led to some debate as to whether the AlNi phase is bypassed as the primary solidification phase in favour of direct solidification of Ni₂Al₃ from the melt, or whether the peritectic reaction at 1406 K is sufficiently rapid so as to go to completion [2]. However, this question has been resolved using X-ray diffraction (XRD) experiments conducted by Lengsdorf et al. Using in-situ time-resolved XRD during solidification on levitated drops at the ESRF synchrotron radiation facility, the diffraction pattern of AlNi was detected, thereby establishing that for the Ni-68.5 at. % Al composition, Al₃Ni₂ must form via the peritectic, not direct from the melt [3].

At 1127 K the Al₃Ni₂ undergoes a second peritectic reaction to form Al₃Ni:

$$L+A{l}_{3}N{i}_2\leftrightarrow A{l}_{3}Ni$$

As with many peritectic reactions, this is observed to form a shell of the secondary phase (Al₃Ni) surrounding the core of the remaining unreacted primary phase (Al₃Ni₂). The contrast between the two peritectic reactions at 1406 K and 1127 K is that the former is a rapid reaction, therefore allowing the AlNi → Al₃Ni₂ reaction to go to completion. Conversely, the latter peritectic reaction at 1127 K is much slower, allowing some of the Al₃Ni₂ to be retained. The reason for this contrast may be due to the similarity between the structures of AlNi and Al₃Ni₂; Al₃Ni₂ is a trigonal extension of the cubic B2 AlNi phase in which every third plane of Ni atoms perpendicular to the trigonal axis is missing [4]. On the other hand, since Al₃Ni is a stoichiometric intermetallic, the rate of solid-state diffusion through the growing shell of Al₃Ni will be sluggish, as it is difficult for the line compound to support a concentration gradient to drive diffusion. This places the Al₃Ni₂ → Al₃Ni peritectic in Type C of the classification described by Kerr & Kurz [5]. This classification is characterised by slow transformation rates during the solid-solid peritectic transformation (SSPT) stage of the reaction.

The final stage of solidification occurs at 913 K, wherein the residual liquid solidifies to an Al-Al₃Ni eutectic [6]:

$$L\leftrightarrow A{l}_{3}Ni+Al$$

During the preparation of the catalyst, the precursor alloy is leached in concentrated sodium hydroxide solution, removing the Al from both the eutectic and the intermetallic phases, leaving nano-crystalline Ni, whilst also producing a suitable atmosphere to activate the Nickel catalyst. After the leaching process, aqueous washing is used to remove the dissolved aluminate and excess alkali [7]. The remaining catalyst, which is highly pyrophoric, can then be stored in water or alcohol [6].

Depending upon the preparation of the precursor alloy, it is observed that there is a variation between the amounts of each phase present, even if the composition remains the same. Cooling rate in particular has been shown to affect the phase composition of the precursor alloy, with higher cooling rates tending to supress the Al₃Ni₂ → Al₃Ni peritectic, leading to greater retained Ni₂Al₃ with consequently less Al₃Ni. The phase composition of the precursor has an important effect on the properties of the activated catalyst. It has been found that Al₃Ni is the least resistant phase to leach, resulting a highly active catalyst but with poor mechanical strength [2]. Consequently, catalysts based on precursor alloys with high Al₃Ni concentrations are not suitable for applications such as slurry and tubular bed reactors [8]. Al₃Ni₂ is more resistant to leaching, but when leached successfully can maintain greater structural integrity than an Al₃Ni based catalyst, due mainly to incomplete leaching. In a commercial catalyst produced by cast-crush processing of slowly cooled ingots, it is common that there is more Al₃Ni₂ in the microstructure than Al₃Ni. Around 58 wt% of Al₃Ni₂ is expected from the commercial alloy using regular cooling rate [9], with the remaining 42 wt% made up of Al₃Ni. Very small amounts of Al eutectic may be present (<1 wt%) [9].

Rapid solidification has been investigated as a route to producing improved Raney type catalysts, both because the higher cooling rates alter the distribution of phases, and because it provides a route to the utilisation of more Al-rich compositions [9] [10]. The performance of gas atomised Raney powders has been investigated extensively as part of the IMPRESS project, with cooling rates between 200 and 5000 K s⁻¹ for typical gas atomised powders in the size range 38–212 μm [11]. It was found that for the hydrogenation of nitrobenzene, relative to a bench mark activity for cast-crush 50-50 wt% catalyst of 3.5 mol kg⁻¹ min⁻¹, the gas atomised powder (106–150 μm size fraction) showed an activity of only 1.4 mol kg⁻¹ min⁻¹ for the 50-50 wt% composition, though this could be increased to 4.8 mol kg⁻¹ min⁻¹ for a 75 at.% Al catalyst (106–150 μm size fraction) [12].

Similar conclusions have been drawn from theoretical studies, such as that conducted by Tourret et al. Simulations were performed using the public binary database PBIN within the ThermoCalc software¹ [13]. The alloy thermo-physical properties used were the same as a previous solidification model for the solidification of electromagnetically levitated Al-25 at.% Ni droplets [14]. Particle sizes of 10 μm, 60 μm and 120 μm diameter were modelled, wherein solidification times ranged from 1.3 × 10⁻³ s (10 μm) to 0.2 s (120 μm). It was found that in an Ni-80 at.% Al alloy, the increase in particle size resulted in an increase in the Al₃Ni phase fraction and a decrease in the Al₃Ni₂ phase fraction at all particle sizes. When more Al was introduced to the initial alloy composition, this was only observed in larger particles.

The activity of Raney Ni catalysts can be improved by introducing a 3rd metal, known as a promoter, or dopant. The most common promoters used with Raney Ni are Cr, Fe and Mo, but research has been conducted on other promoters such as Cu, La, Co and Ti. During the preparation of the alloy, the promoter can be added to the melt, usually in small quantities (1–3%) depending on the reaction to be catalysed [6]. One study found that all of the aforementioned promoters increase the activity of the catalyst, but that Molybdenum was found to be the most effective for the hydrogenation of butyronitrile and acetone [15].

Research has been carried out on the influence of Chromium as a promoter in Raney Ni catalysts. It has been found that the optimum level of Chromium required in the Raney Ni alloy is 1.5 wt% for the hydrogenation of butyronitrile, acetone and sodium p-nitrophenolate [6]. Bonnier et al. [16] studied the activity of Cr-doped Raney Ni, prepared by mechanically mixing Al-Ni and Cr-Al alloys, on the hydrogenation of acetophenone. It was found that there was indeed an increase in activity, with the Chromium being strongly segregated to the surface in an oxidised state. It was further found that due to the presence of chromium oxide, the residual aluminium, in a metallic state, is retained more than that of an undoped AlNi alloy. Metallic aluminium seems to inhibit reactions such as the hydrogenation of carbonyl groups [17] which are promoted by the chromium oxide [6]. Work by Pisarek et al. also found that surface segregated Cr promoted catalytic activity for the hydrogenation of isophorone to dihydroisophorone, but also that Cr deactivates the Ni–Al catalyst for the reaction of dihydroisophorone hydrogenation [18]. In previous work it has been assumed, due to the chemical similarity between Ni and Cr, that Cr substitutes randomly for Ni [16].

The IMPRESS project, as well as looking at the effect of high cooling rate on the activity of Raney Ni catalysts, also investigated the effect of dopants upon activity. It was found that, by the addition of 1.5 at.% Cr to a 75 at.% Al alloy, the catalytic activity for the hydrogenation of nitrobenzene was increased from 4.8 mol kg⁻¹ min⁻¹ to 11.6 mol kg⁻¹ min⁻¹ (both 106–150 μm size fraction). This is comparable to the effect of adding 1.5 at.% Cr to a standard 50-50 wt% cast-crush catalyst, wherein the catalytic activity increases from 3.5 mol kg⁻¹ min⁻¹ to 7.35 mol kg⁻¹ min⁻¹. However, the underlying mechanism for this astonishing 140% increase in catalytic activity of the doped, gas atomised powders, was never established. In this paper, we reanalyse the original gas atomised precursor alloy powders in order to elucidate the mechanism for this increased catalytic activity.