Elsevier

Thin Solid Films

Volume 488, Issues 1–2, 22 September 2005, Pages 74-81
Thin Solid Films

Growth of Fe2O3 thin films by atomic layer deposition

https://doi.org/10.1016/j.tsf.2005.04.063Get rights and content

Abstract

Thin films of α-Fe2O3 (α-Al2O3-type crystal structure) and γ-Fe2O3 (defect-spinel-type crystal structure) have been grown by the atomic layer deposition (ALD) technique with Fe(thd)3 (iron derivative of Hthd = 2,2,6,6-tetramethylheptane-3,5-dione) and ozone as precursors. It has been shown that an ALD window exists between 160 and 210 °C. The films have been characterized by various techniques and are shown to comprise (001)-oriented columns of α-Fe2O3 with no in-plane orientation when grown on soda-lime-glass and Si(100) substrates. Good quality films have been made with thicknesses ranging from 10 to 130 nm. Films grown on α-Al2O3(001) and MgO(100) substrates have the α-Fe2O3 and γ-Fe2O3 crystal structure, respectively, and consist of highly oriented columns with in-plane orientations matching those of the substrates.

Introduction

ALD (atomic layer deposition) is a surface-controlled growth technique for thin films based on sequential pulsing of precursors, which carry the constituents of the desired material. The intrinsic composition-control mechanism is based on the saturation of individually performed surface reactions between the substrate (including the evolving film) and the precursors. The controlled growth mechanism provides highly conformal films without pinholes [1], in addition to the mentioned management of the composition. The present authors [2] have, e.g., recently utilized the sensitivity of the ALD technique to show that application of a magnetic field alters the growth rate of α-Fe2O3 films deposited on soda-lime-glass substrates. However, the preceding communication did not include documentation about the ALD window and other experimental features, which are of interest in relation to preparation of α-Fe2O3 films. This report aims at providing such information.

The iron–oxygen system contains several phases, the most important for the present study being Fe1−xO (NaCl like; cubic/rhombohedral, Fmm, a =  4.326 Å ( x = 0) [3]; wüstite), Fe3O4 (inverse spinel; cubic, Fdm, a = 8.396 Å [3]; magnetite), α-Fe2O3 (α-Al2O3 type; rhombohedral, Rc, a = 5.0353(3), c = 13.7495(5) Å in hexagonal setting [4]; hematite) and γ-Fe2O3 (defect inverse spinel; cubic, P4332, a = 8.3474 Å [5]; maghemite). The crystal structure of γ-Fe2O3 is closely related to the inverse spinel structure of Fe3O4, and oxidation of Fe3O4 to γ-Fe2O3 involves the introduction of vacancies on one sixth of the octahedral Fe sites. These vacancies are ordered and the ordering creates a threefold tetragonal super lattice with c = 3a and space group P41212 [5].

Numerous studies have been devoted to the growth and characterization of iron oxide thin films, including oxygen-plasma-assisted molecular beam epitaxy (MBE), radio frequency sputtering, sol–gel spin coating, oxidation of iron films, traditional chemical vapor deposition (CVD) and now ALD. Depending on the substrate and the deposition conditions, films of Fe3O4 and Fe2O3 have been grown with various thicknesses and microstructures [2], [6], [7], [8], [9]. Epitaxial α-Fe2O3(001) and γ-Fe2O3(001) films have been grown in an MBE reactor on α-Al2O3 and MgO substrates, respectively, under otherwise equal deposition conditions [10]. Physical-vapor-deposition-grown films of Fe1−xO are unstable at room temperature even under low oxygen pressures and will transform into Fe3O4 [11]. In fact, bulk samples of Fe1−xO are thermodynamically unstable below 560 °C [12].

α-Fe2O3, which is the main target for this study, is of great importance for many applications, e.g., hydrocarbon gas sensors, heterogeneous catalysis, and in relation to corrosion [13], [14]. The properties of α-Fe2O3 are crystal-plane dependent and the most studied surface is (001), which is found to be Fe terminated and remains unreconstructed even at high oxygen pressures [14]. γ-Fe2O3 has been studied for use as magnetic recording medium and also for possible use in gas sensors for organic vapors and hydrogen where it can be used without a noble metal catalyst [10], [13], [15], [16].

Section snippets

Experimental details

Thin films were deposited in a commercial flow-type F-120 Sat reactor from ASM Microchemistry. The precursors were alternately pulsed into the reactor, separated by purging with N2 gas (claimed purity > 99.999%) produced with an N2 generator. N2 also functioned as the carrier gas. The iron source was Fe(thd)3 (iron derivative of Hthd = 2,2,6,6-tetramethylheptane-3,5-dione) synthesized as described in Ref. [17] from FeCl3 (99.0%, Fluka) and Hthd (98%, Fluka). Ozone was supplied from an OT-20 ozone

Results

The ALD window for the deposition of the iron-oxide films was explored by varying one parameter at the time. The other parameters were then kept at values believed to be within the ALD-growth regime. These assumptions were confirmed by the experiments. Fig. 1 shows that a 1 s Fe(thd)3 pulse is sufficient using a precursor sublimation temperature of 114 °C. The subsequent N2 purging may conveniently be performed for 1 s (which is certainly more than sufficient to prevent CVD growth). Similarly,

Discussion

As shown in Fig. 1, there exists an ALD window for growth of α-Fe2O3 with Fe(thd)3 and ozone as precursors. For this window, the gas-flow parameters should be kept larger than 1 s for the iron-precursor pulse, 1 s for the iron-precursor purge, 2 s for the ozone pulse and 1 s for the ozone purge for self-limiting growth. Good quality films are obtained for reactor temperatures between 160 and 330 °C on Si(100) substrates and between 160 and 310 °C on soda-lime-glass substrates. At lower

Acknowledgements

The authors are grateful to Ola Nilsen for his contribution to the X-ray measurements and helpful discussions. The authors also wish to thank Oddvar Dyrlie for his help with the AFM images, Turid Winje for XRF measurements, and Steinar Foss for TEM analyses. This work has received financial support from the Research Council of Norway.

References (22)

  • M. Ritala et al.
  • O. Nilsen et al.

    Appl. Surf. Sci.

    (2004)
  • H. Sawada

    Mater. Res. Bull.

    (1996)
  • F. Yubero et al.

    Surf. Sci.

    (2000)
  • J. Sarradin et al.

    Solid State Ionics

    (1998)
  • Y. Gao et al.

    J. Cryst. Growth

    (1997)
  • C. Ruby et al.

    Thin Solid Films

    (1999)
  • K. Siroky et al.

    Thin Solid Films

    (1994)
  • S.A. Chambers et al.

    Surf. Sci.

    (1999)
  • S.N. Malchenko et al.

    Thin Solid Films

    (1993)
  • B.J. Kim et al.

    Thin Solid Films

    (1999)
  • Cited by (106)

    • Advanced deposition tools for the development of oxide thin films

      2023, Defect-Induced Magnetism in Oxide Semiconductors
    • Atomic layer deposition of ferrite thin films

      2023, Ferrite Nanostructured Magnetic Materials: Technologies and Applications
    • Atomic layer deposition of iron oxide on a porous carbon substrate via ethylferrocene and an oxygen plasma

      2021, Surface and Coatings Technology
      Citation Excerpt :

      Other iron precursors explored in the literature include bis(N,N′-di-tert-butylacetamidinato)iron [8,15], iron(III) tert-butoxide [16,17], and iron(III) tris(2,2,6,6-tetramethyl-3,5-heptanedionate) [18]. These precursors are typically lab-synthesized directly for ALD use and exist in the solid state [19–23]. Iron(III) chloride has also seen use in ALD and is readily reactive with a water co-reactant [24,25].

    • Group 7 and 8 compounds for chemical vapor deposition

      2021, Comprehensive Coordination Chemistry III
    • A highly transparent thin film hematite with multi-element dopability for an efficient unassisted water splitting system

      2020, Nano Energy
      Citation Excerpt :

      In order to achieve high STH efficiency of the tandem PEC system by improving solar cell performance, a highly transparent hematite photoanode has been an inevitable requirement for the solar cell to effectively use light. Many hematite thin films have been reported and shown excellent structural stability and high reproducibility, prepared by pulsed laser deposition (PLD) [30–32], atomic layer deposition (ALD) [33–36], spray pyrolysis [37,38] and ultrasonic spray pyrolysis (USP) [39–41]. However, the performance of the thin film hematite in these studies has been limited to low photocurrent density under 1mAcm−2 at 1.23V vs. RHE (1.23VRHE), because insufficient active area and low light absorption in the thin film hematite photo-absorber resulted in low performance of the photoanode.

    View all citing articles on Scopus
    View full text