Raman spectroscopy and chemometrics for materials identification from fire debris and its application of forensic fire modelling.

Fire investigation is often deemed the most difficult aspect of forensic science as evidence of events which occurred prior to and during the fire is destroyed. Fire investigation is multidisciplinary and must take into consideration the chemistry and physics of fires. Often times the fire scene is very complex by nature of the building design, the materials present and the breaking of windows that the investigator at the end of his assessment, may have more than one possible hypothesis as to origin and cause.

In recent years fire modeling has become a part of the investigator’s tool kit for selecting an appropriate hypothesis. Fire models offer the ability to simulate flames, plumes, hot layers and smoke movement during different stages of fire development, from ignition to flame spread, through flashover and extinction. The ability to model these events allows for fire scene reconstruction [Rein] and as such has become a very useful tool.

Computer model of fire at The Station nightclub showing temperature variation after 90 seconds at 1.5 meters (5 feet) above the floor. Credit: NIST

Shen et al. (2006) used the fire dynamics simulator (FDS) to successfully simulate a hotel arson fire in Taiwan. Its results demonstrated good prediction of fire development and smoke movement when compared to the combustion evidence of the scene [1]. Posteriori modelling of compartment fires however have a number of limitations which could call into question the validity of conclusions drawn from a model’s output. The posteriori modelling done in Dalmarnock fires showed that models had many inaccuracies due to uncertainties in the input parameters and the model had to be adjusted to fit the experimental result [2].

Cox (1994) pointed out that the challenge to the modeler will not only be attempting to model the physical and chemical processes occurring in the fire but if these are capable of being modeled there will always be the uncertainties in the input data for the model like the location of the fire within the enclosure, the very nature of the fuel involved, the configuration of the ventilation and the external wind conditions [3]. These challenges are further supported by Jahn, Rein and Torero (2008), they studied the effects of model parameters on the simulation of fire dynamics. They showed that the model output would be affected by the input of different ignition sources, fire location, fire area and heat of combustion, thermal and ignition material properties and flame radiative fraction. The most significant effect on the output of the model was found to be the evolution of heat release rate (HRR) with time [4]. HRR is highly dependent on the thermal and ignition properties of the materials existing in the compartment. Delemont et al. (2007) also shows that of the problems with using models in forensic investigations is that it assumes an evaluation of combustibles involved was done and that the nature of the heat source which started the fire was determined [5]

As pointed out the accuracy of the models are highly dependent on knowledge of the combustibles as it is there material properties that determine fire growth rate, spread, toxicity and smoke development. This forms the foundation for a study to be carried out to develop a method for determining the combustibles present in the compartment prior to the fire. To the best of our knowledge, only one research work was reported on the use of Raman spectroscopy for the identification of materials from fire debris. This work was done by Gonzalez-Rodriquez et al. which was geared towards providing a methodology for identifying ignitable liquids and determining the original material that had been ignited. They acquired Raman spectra for materials burnt with different ignitable liquids using a confocal Raman microscope instrument. It could be seen from their work that it was possible to have some degree of identification of the ignitable liquids with the materials they were burnt with by applying principal component analysis (PCA) to the respective Raman spectra [6]. The relative success of their experiments was significantly affected by material fluorescence after the materials were burnt.

Example Raman spectrum and PCA score plot

The idea of utilizing Raman spectroscopy coupled with PCA for identification and differentiation between burnt materials is supported by studies done by Brody et al. (2001), they used a combination of Raman spectroscopy and PCA as a means of discriminating and classify ivory from different species [7]. Later Yao et al. (2009) used this combination to successfully identify and differentiate apoptotic cells and control cells in the analysis of apoptosis of single human gastric cancer cells [8]. Degardin (2011) used the method for chemical profiling of medicine counterfeits [9]. This technique has been adopted by coffee researchers, medical researchers and other areas of science, thus its application to identifying and discriminating between polymers from burnt debris should be possible.

PCA is a widely used chemometric tool used to reduce the dimensionality of data. Its application to spectral data allows for the computation of principal components that present the maximum of the variance between the data given that the variables are numerous and often correlated [9].

The paper presented by Gonzalez-Rodriquez et al discussed the analysis of three materials, namely polypropylene (PP), polystyrene (PS) and Nylon (Ny). Their work has provided the foundations for examining several other common polymers which exist in households leading to the creation of a burnt materials Raman spectral library which could prove very useful in improving the accuracy of posteriori fire modeling. Imagine being able to determine the materials which existed in a compartment prior to a fire by applying a simple forensic analysis of the fire debris taken from a scene. A technique which requires no sample preparation which is most ideal for forensic samples.

Conclusion

Posteriori fire modeling has much to offer the forensic community of fire investigators and the justice system. It is not only a means of validating the investigators hypothesis or a learning tool but also gives an investigator the ability for to clarify and present clear information to the officers of the courts and the jurors of the events which occurred. Gonzalez-Rodriquez et al created an opportunity for others to explore the reality of post fire materials identification and the subsequent improvement of forensic fire models.

Raman spectroscopy may be a very good tool for this as it capable of providing detailed information about the materials be analysed. It is a simple, quick and non-destructive tool. Its greatest problem being the possibility of spectra being overwhelmed by fluorescence from the samples under study. A number of strategies have been worked on for reduction and possible elimination from spectra; these can be explored to determine which reduction method is ideal for burnt debris to get the best that Raman spectroscopy has to offer.

References:

1. Shen, T., Huang, Y. & Chien, S. Using Fire Dynamic Simulation (FDS) to Reconstruct an arson Fire Scene Building and Environment, 2006, Vol. 43, pp. 1036 – 1045

2. Jahn, W., Rein, G. & Torero, J.L. A posteriori modelling of the growth phase of Dalmarnock Fire Test One Building and Environment, 2011, Vol. 46(5), pp. 1065 – 1073

3. Cox, G. The Challenge of Fire Modelling Fire Safety Journal, 1994, Vol. 23, pp. 123 – 132

4. Wolfram Jahn, Guillermo Reina, J.L.T.The Effect of Model Parameters on the Simulation of Fire Dynamics Fire Safety Science, 2008, Vol. 9, pp. 1341-1352

5. Delémont, O. & Martin, J.-C. Application of Computational Fluid Dynamics modelling in the process of forensic fire investigation: Problems and solutions Forensic Science International, 2007, Vol. 167(2-3), pp. 127 – 135

6. González-Rodríguez, J., Sissons, N. & Robinson, S. Fire debris analysis by Raman spectroscopy and chemometrics Journal of Analytical and Applied Pyrolysis, 2011, Vol. 91(1), pp. 210 – 218

7. Brody R., Edwards, H., Pollard, A. Chemometric methods applied to the differentiation of FT Raman Spectra of Ivories Analytica Chimica Acta, 2001, 427, pp. 223 – 232

8. Yao, H., Tao, Z., Ai, M., Peng, L., Wang, G., He, B., Li, Y. Raman Spectroscopic Analysis of Apoptosis of single Human Gastric Cancer Cells Vibrational Spectroscopy, 2009, Vol. 50, pp. 193 – 197

9. Degardin, K., Roggo, Y., Been, F., Margot, P. Detection and Chemical Profiling of medicine counterfeits by Raman Spectroscopy and chemometrics Analytica Chimica Acta, 2011, 705, pp. 334 – 341