In Situ Synchrotron X-ray Diffraction Study of the Sol–Gel Synthesis of

Jun 16, 2015 - A fast position-sensitive detector (PSD) using Mythen-2 Si modules with 90° aperture was used to collect diffraction data (a few secon...
13 downloads 12 Views 4MB Size
Article pubs.acs.org/cm

In Situ Synchrotron X‑ray Diffraction Study of the Sol−Gel Synthesis of Fe3N and Fe3C Zoe Schnepp,*,† Ashleigh E. Danks,† Martin J. Hollamby,‡ Brian R. Pauw,§ Claire A. Murray,⊥ and Chiu C. Tang⊥ †

School of Chemistry, University of Birmingham, Birmingham, B152TT, United Kingdom Department of Chemistry, Keele University, Staffordshire, ST55BG, United Kingdom § National Institute for Materials Science, 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan ⊥ Diamond Light Source, Harwell Science and Innovation Campus, Didcot, OX11 0DE, United Kingdom ‡

ABSTRACT: Sol−gel chemistry is a powerful tool for the synthesis of porous or nanocrystalline structures of a wide range of materials. The morphology and composition of the final product can be strongly dependent on the mechanism that operates during the heating step, as has been demonstrated in many investigations of binary, ternary, and quaternary metal oxides. We demonstrate the complex phase transformations occurring in a transition-metal carbide (Fe3C, cementite) synthesis using in situ synchrotron X-ray diffraction. Our new study proves the existence of multiple intermediate phases and elucidates how oxide−nitride and nitride−carbide phase transformations can occur. This is particularly important as many transitionmetal nitrides and carbides are metastable, which renders their stability windows sensitive to small changes in sol−gel methodology.

1. INTRODUCTION

materials, mechanistic information about the structure and phase changes during synthesis must be identified. The challenge of understanding reaction mechanisms in sol− gel synthesis has been addressed in oxide systems through a range of in situ methods, such as synchrotron X-ray powder diffraction (SXPD), neutron diffraction, and small-angle scattering.12 In situ SXPD has also been combined with nonair systems, e.g., oxygen vacancies13 in Ce0.9Pr0.1O2−δ under O2, He, and H2 or the formation of a bismuth ferrite under argon or vacuum from solid oxide precursors.14 Given the wide expertise in these methods, it seems obvious to apply them to the sol− gel synthesis of nitrides and carbides. However, this is not a trivial experiment. Iron nitrides and carbides are easily oxidized and their formation during sol−gel synthesis is highly dependent on reaction conditions. Furthermore, the sol−gel reaction, which includes decomposition of organic precursors, releases many gaseous and reactive side-products such as NH3, CO, CO2, and H2O.8 In this paper, we describe an in situ SXPD study of the formation of Fe3N and Fe3C from sol−gel precursors. To the best of our knowledge, this is the first in situ (or “real-time”) observation of nitride and carbide formation from a sol−gel method. The methodological challenges we have overcome have given a detailed insight into the Fe3N/ Fe3C system.

Iron forms a class of interstitial compounds with nitrogen and carbon including θ-Fe3C (cementite), ε-Fe3(N,C)1+x, and Fe5C2 (Hägg carbide). These nitrides and carbides encompass a wide range of structures and stoichiometries that have been extensively characterized because of their importance in steels. More recently, iron nitrides and carbides with a porous or nanoparticulate structure have been pursued1 for key catalytic reactions such as the Fischer−Tropsch process,2 ammonia decomposition,3 and the oxygen reduction reaction.4 In another different field, films and nanoparticles of materials such as Fe3C and Fe16N2 have shown remarkable magnetic properties with potential applications as, e.g., magnetic resonance imaging contrast agents.5 For applications requiring particulate or porous carbides and nitrides, various synthetic methods have been developed. These include laser ablation,6 sol−gel synthesis,7 and ammonolysis of iron oxide.8 Sol−gel methods have been particularly widely investigated because of their relative simplicity. Generally, a sol−gel route involves heating a mixture of an iron salt (nitrate, acetate, chloride) with a molecular organic precursor (e.g., gelatin, urea)9,10 in an inert atmosphere (typically nitrogen). However, composition, and therefore properties, can vary significantly with different precursors or different conditions such as temperature and time.11 Iron nitrides and carbides are metastable, the thermodynamic product under inert atmosphere being metallic iron, and different conditions can dramatically alter the range of thermal stability of the phases. For better control of the properties and composition of these © 2015 American Chemical Society

Received: May 14, 2015 Revised: June 16, 2015 Published: June 16, 2015 5094

DOI: 10.1021/acs.chemmater.5b01811 Chem. Mater. 2015, 27, 5094−5099

Article

Chemistry of Materials

Figure 1. Schematic of the experimental setup viewed from (a) the top and (b) the side.

Figure 2. (a) Powder diffraction patterns taken at 10 or 20 °C intervals during heating, with reference peaks for Fe3C and Fe3N. (b) Patterns for the sample during the early stages of heating with reference peaks for Fe3O4 and FeOx. (c) Data for the range 600 and 670 °C with the ε-Fe3N and θFe3C peaks, marked by a cross symbol (+) and an asterisk (*), respectively. viscous orange gel, followed by drying in air to produce a brittle orange-brown foam. Calcination of iron nitrate/gelatin mixtures is associated with a large mass loss on dehydration. Thus, to prevent water buildup, sample expansion and iron oxidation during the capillary experiments, the foam samples were heated to 250 °C under nitrogen in a muffle furnace to produce black solids. Powdered samples of this black solid were loaded into quartz capillaries (0.7 mm diameter, 0.02 mm wall thickness) and packed either side with quartz wool to prevent movement of the powder during heating under gas flow. Each prepared capillary was loaded between Swagelok T-pieces into a custom-built holder and sealed with ferrules. This holder was connected to a nitrogen cylinder and to an outgas flowmeter to enable

2. EXPERIMENTAL SECTION The in situ SXPD experiments were conducted at beamline I11 (Diamond Light Source, U.K.) a dedicated instrument for highresolution and time-resolved powder diffraction. The monochromatic incident beam of λ = 0.825721(1) Å was calibrated using a high-quality Si powder standard (SRM640c). A fast position-sensitive detector (PSD) using Mythen-2 Si modules with 90° aperture was used to collect diffraction data (a few seconds per pattern). The technical details and detector systems of the beamline are described elsewhere.15,16 Samples were prepared by combining hot aqueous gelatin solution (10% w/w, 10 g; Sigma−Aldrich, Lot No. G2500) with aqueous iron nitrate (10% w/v, 20.2 mL, Fe(NO3)3·9H2O) to form a 5095

DOI: 10.1021/acs.chemmater.5b01811 Chem. Mater. 2015, 27, 5094−5099

Article

Chemistry of Materials

Figure 3. Diffraction patterns showing disappearance of (a) the (110) and (002) ε-Fe3N peaks and (b) the (111) ε-Fe3N peaks with increasing temperature, as well as (c) data for the (110) peak at three different temperatures showing the peak split. Panel d shows the trend in lattice dimensions for the two phases, and panel e shows the structure of ε-Fe3N with N occupying octahedral interstices and θ-Fe3C with C in trigonal prismatic sites. Panel f shows reference patterns for ε-Fe3N1+x, calculated from ref 21. a slow nitrogen flow over the sample, as shown schematically in Figure 1. During the experiment, the sample was heated using a hot-air blower (Cyberstar). The use of this heating device enables fast heating (required for the sol−gel experiments), an unobstructed incident and diffracted beam path, and fast cooling for sample changeover. A disadvantage is that the region of heating was relatively narrow (∼6 mm), resulting in a temperature gradient along the capillary. For many systems, this would not affect the experiment, provided the hot air blower and beam are aligned to the same position. However, for sol− gel reactions, this can result in delayed reactions and, therefore, the release of reactive gases from regions “upstream” from the region of interest in the sample. To overcome this problem and ensure a consistent reaction, at each 10 °C interval, the capillary was moved to record a diffraction pattern 3 mm on either side of the central point. During each 10 °C heating period, diffraction patterns were recorded automatically every ∼1.2 s. To reduce measurement time, the sample was initially heated at 10 °C min−1 to 400 °C. This is based on quenching studies showing an absence of crystalline content during these early stages of the reaction.

this would allow a very long time for the Fe3O4/FeOx particles to sinter or crystallize. This may result in samples with considerably different structure and reflects the limitations of quenching for probing intermediate states in high-temperature reactions. Analysis of the three main peaks at 540 °C using the Scherrer equation gives a crystallite size of ∼3 nm, which is consistent with the theory that polymer “sol−gel” precursors constrain the nucleation of crystalline precipitates.17 It is possible that the broad peaks actually derive from disorder in the FeOx phasei.e., the FeOx phase actually consists of larger particles that are poorly crystalline or are composed of multiple crystalline domains. However, if this was the case, one might expect substantial peak sharpening during the heating as the particles crystallize. Furthermore, the very striking difference in peak width between the FeOx phase and the Fe3N phase should eliminate any concerns that peak broadening is substantially affected by instrumental effects. From 560 °C, peaks for iron nitride (ε-Fe3N, ICDD File Card No. 01-083-0877) begin to form, concurrent with a gradual reduction in intensity of the FeOx peaks. The Fe3N peaks persist until 650−660 °C and are gradually replaced by cementite (θ-Fe3C, ICDD File Card No. 01-074-3843) from ∼620 °C. At 680 °C, the crystalline part of the sample is composed entirely of cementite. Scherrer analysis of the Fe3N peaks at 580, 590, and 600 °C indicates a particle size of ∼30 nm, and analysis of the Fe3C peaks at 660 and 680 °C indicates a particle size of ∼60 nm. These values are consistent with particle sizes estimated from transmission electron microscopy (TEM) studies of samples in previous studies of this system.9 Further information about the transitions in this system can be inferred from analysis of the full dataset. Figure 2c shows diffraction patterns taken at 1.2 °C intervals offset along both the x- and y-axes for clarity. From this graph, the disappearance of the ε-Fe3N peaks and emergence of sharp peaks for θ-Fe3C

3. RESULTS In Figure 2a, the SXPD patterns show an overview of the main crystallographic transformations that occur during this synthesis, i.e., FeOx → Fe3N → Fe3C. Data from 250 °C to 550 °C (Figure 2b) shows that the very early stages of the reaction involve the nucleation of magnetite (Fe3O4, ICDD File Card No. 01-074-0748), followed by reduction to the iron(II) oxide phase wüstite (FeOx ICDD File Card No. 01-074-1886). The broad peaks indicate a very small particle size (