Autocorrelation infrared analysis of mineralogical samples: The influence of user controllable experimental parameters

https://doi.org/10.1016/j.aca.2007.03.030Get rights and content

Abstract

Autocorrelation infrared (ACIR) analysis is based upon the application of the autocorrelation function corr(α,ω)=α(ω+ω)α(ω)dω to standard Fourier transform infrared (FTIR) transmission spectra. We present a rigorous examination of the effect of experimental parameters such as dilution ratio, spectral resolution, grinding time and pressing conditions upon the ACIR analysis of haematite. Results were found to vary by less than 4.5% irrespective of sample preparation, instrumental and data collection parameters. For a series of perovskite samples, the relationship between the measured effective linewidth and material composition appears to be reproducible, even though the absolute magnitudes of Δcorr values do not. Our results further indicate that the ACIR technique is indeed valid for comparative analysis of synthetic sample sequences that vary slightly in composition or structural state, provided that primary spectra are all recorded by the same instrument.

Introduction

An infrared transmission spectrum may be used to determine the constituents of a sample, both qualitatively and quantitatively. However, this becomes challenging for the analysis of complex spectra, such as those obtained from minerals or mixtures of pure substances, as broad phonon bands in their vibrational spectra tend to overlap [1]. Consequently scientists at Cambridge University developed the autocorrelation infrared (ACIR) technique [2], [3] for the quantification of linewidth variations in powder absorption spectra from mineralogical samples. ACIR analysis relies on the application of the autocorrelation function corr(α,ω)=α(ω+ω)α(ω)dω to primary infrared (IR) absorption spectra, where α(ω) is the primary IR spectrum itself and α(ω + ω′) is the primary spectrum offset in frequency by ω′[3], [4]. This function serves to parameterise the effective linewidths (Δcorr) of individual absorption bands or groups of bands within a spectrum, without the need for any peak fitting to the primary spectrum itself [3], [4], [5], [6]. The width of the autocorrelation spectrum offers a statistical measurement that is proportional to the average linewidth of absorbance bands in the region of the spectrum under investigation [5]. Accordingly, ACIR is applicable to samples with varying composition, degree of cation order or structural state, and to complex IR spectra containing broad overlapping peaks [3], [4], [7]. Applications include the analysis of garnets [7], [8], feldspars [3], [5], [9], perovskites [4], [10] and a broad range of other solid solutions such as alumino-silicates and Mg–Fe bearing silicate solutions [11], [12], [13], [14], [15], [16]. An advantage of the autocorrelation analysis is that it directly measures peak linewidth without requiring any direct peak fitting to the primary spectrum (which is susceptible to ambiguity when the background near an absorption peak changes sharply). Hence, the autocorrelation method provides a more dependable measure of slight changes in linewidth between samples (or across a series of complex spectra) than that which can be achieved using conventional methods [5], [6]. A comprehensive description of the autocorrelation method, derivation of values, and theory behind its application toward the measurement of effective linewidths of primary IR absorption spectra was reported by Salje et al. [3].

At present, doubts remain about the sensitivity of line width parameters obtained by the ACIR method to sample preparation and instrumental parameters. In this article, we report a thorough examination of the sensitivity of the technique to experimental parameters, and an investigation of reproducibility through comparison of new results with previously published data from independent research.

Section snippets

Sample preparation

Reagents were all of analytical grade unless otherwise specified. Synthetic haematite (Fe2O3, calcined, 97%, BDH Chemicals Ltd., England) was ground using an agate mortar and pestle for 5–15 min. A known quantity of this sample was subsequently diluted in a known mass of either potassium bromide (KBr, Uvasol®, Merck, Germany) or caesium iodide (CsI, Suprapur®, Merck, Germany) and ground for a further 3–5 min. A known amount (150–300 mg) of the mixed powder is then transferred to a hydraulic die

Preliminary experiments

Preliminary experiments were focussed on obtaining primary infrared spectra for synthetic haematite and determining the appropriate regions for application of autocorrelation analysis. As can be seen in Fig. 1, the infrared spectrum of an haematite pellet prepared at a sample to KBr ratio of 1:250 (w/w) exhibits two prominent peaks centred at ∼470 and ∼550 cm−1 and a broad peak at ∼1000 cm−1. There were no significant spectral features observed above 1200 cm−1. Consequently the region of 422–675 cm

Conclusions

A thorough investigation of the influence of several experimental parameters upon the autocorrelation analysis of synthetic samples of Fe2O3 has been performed. In particular, their influence on the value of the effective linewidth of peaks in IR absorption spectra has been determined. It is evident that Δcorr values exhibit only small variations for haematite pellets that have been prepared at different sample to matrix dilutions or with different pressing conditions. In the results shown here

Acknowledgements

The authors would like to thank Dr. Milena Ginic-Markovic for her assistance with the Nexus IR spectrometer. This work was funded in part by a Flinders UCIRG grant and by ARC project DP0342934.

References (19)

  • C. Tenailleau et al.

    J. Solid State Chem.

    (2005)
  • T. Boffa Ballaran et al.

    Chem. Geol.

    (2006)
  • Y. Wang et al.

    Colloids Surf. A: Physicochem. Eng. Aspects

    (1998)
  • X. Wang et al.

    J. Cryst. Growth

    (2004)
  • S.C. Tarantino et al.

    Phys. Chem. Miner.

    (2005)
  • T. Boffa Ballaran et al.

    Am. Miner.

    (1998)
  • E.K.H. Salje et al.

    Eur. J. Miner.

    (2000)
  • A. Atkinson et al.

    Eur. J. Miner.

    (1999)
  • S. Tarantino et al.

    Eur. J. Miner.

    (2002)
There are more references available in the full text version of this article.

Cited by (8)

View all citing articles on Scopus
View full text