Spectroscopic and imaging methods in forensic investigations of modern paints and coatings

(1)

MSc in Chemistry

Analytical Science

Literature Thesis

Spectroscopic and imaging methods in forensic

investigations of modern paints and coatings

by

Vera Nordmann

UvA: 11127112

VU: 2587245

November 2016

12 ECTS

September 2016 - November 2016

Supervisor/Examiner:

Examiner:

Prof. Dr. Freek Ariese

Prof. Govert Somsen

Supervisor:

(2)

(3)

Preface

Paints can be found in every aspect of the daily life. The present thesis illustrates the different methods applicable for forensic investigations on paint traces of different sources. Forensic investigation does not only require rapid, non-destructive methods, which provide qualitative and quantitative data but also access to either reference material of the suspect (or victim) or to databases which contain information (such as manufacturer, brand and year of manufacture) on paint samples of the same type of origin (such as vehicles, wall paints, etc.). The description of three cases provides a deeper understanding on the applicability of the different analytical methods on the analysis of paint traces.

(4)

Abbreviations

BSE Backscatter Electrons

CIE Commission on Illumination

DFM Dark Field Microscopy

FTIR Fourier Transform Infrared Spectroscopy

GC/MS Gas Chromatography coupled with Mass Spectroscopy

ICP-MS Inductively Coupled Plasma Mass Spectroscopy

IR Infrared

m/z Mass-to-Charge ratio

NA Numerical Aperture

OPD Optical Path Difference

PDQ Paint Data Query

PIGE Particle Induced Gamma-ray Emission Spectroscopy

PIXE Particle Induced X-ray Emission Spectroscopy

RCMP Royal Canadian Mounted Police

RRS Resonance Raman Spectroscopy

S/N Signal to Noise ratio

SE Secondary Electrons

SEM-EDX Scanning Electron Microscopy with Energy Dispersive X-ray Spectroscopy

UV-Vis Ultraviolet to Visible light range

VOC Volatile Organic Compound

XRF X-ray Fluorescence Spectroscopy

(5)

1

Introduction

In forensic investigations, paint traces often play an important role. These samples can originate from different sources such as burglary, vandalism, vehicle hit-and run accidents as well as from buildings and tools. Paint evidences can vary from single or multi layer fragments and chips to smears and small droplets. In order to obtain optimal results, the method applied for the forensic investigation of these traces needs to fulfil six conditions. First, the data gained should provide a high degree of discrimination power between the samples. Furthermore, it should be non-destructive, rapid and applicable to very small areas. In addition, the method should require minimal or no sample preparation in order to avoid contamination as well as sample destruction. The ideal technique is expected to provide qualitative and quantitative data. Lastly, the analysis of paint traces should provide information on physical and chem-ical characteristics, including structural and elemental composition. Most, methods available cannot fulfil all these requirements, therefore combination of several complementary methods are used. The first step is usually to determine physical characteristics using optical microscopy as well as colourime-try. These methods provide information on the colour, the texture and the number of layers present in the sample. These information are then used to optimise the chemical characterisation, using FTIR or Raman for the structural analysis and XRF or SEM-EDX for the determination of the elemental compo-sition. The application of FTIR spectroscopy allows the identification of binders, extenders and to some extend inorganic or organic pigments present in a paint sample. Raman is a versatile tool in forensic investigation, also applicable for analysing drugs, blood, etc. It is complementary to IR and is especially sensitive to (organic) pigments. In the field of criminalistics, the determination of the composition of the pain traces has to be complemented by comparing the gained information of the trace with the infor-mation gained from analysing reference material from the suspect or victim. If no reference material is available or there is no match, the sample has to be compared to samples form a database which covers samples of different brands and years of manufacturing. This way the expert can at least narrow down the possible sources. If the analysis of the trace and the reference material matches, the expert’s opinion is required, to state if these two samples have the same origin. This decision is becomes more complex as mass production increases. Here, the influence of the environment, restoration etc. can play an important role.

(8)

This thesis aims to provide a greater understanding on the methods applicable for forensic analysis of paints and coatings by not only exploring the principle of the methods and how they can be used for paint analysis but also by providing examples on some of this method on real investigations.

(9)

2

Contents to be analysed

In this part of the thesis, the focus lies on the compounds of interest present in paints and coatings. Here, the chemical composition as well as the changes observed over the last recent decades are dis-cussed. In addition, health and environmental concerns are presented as well as the resulting changes in the legislation of paint composition as well as manufacture.

2.1. Composition of paints and coatings

Paints and coatings are part of every day live. They can be found in every aspect of our life, such as wall paints, painted surfaces and utensils. Depending on their area of application they fulfil different properties, such as decorative or protective functions. Paints or coatings are liquid or powder products, which are forming adherent films on the surface of substrates. The properties of paint are determined by their qualitative and quantitative composition, which influence the viscosity, electrical conductivity and drying behaviour. These factors as well as the substrates surface have an impact on the properties of film coatings, like elasticity, scratch resistance , adhesion and surface structure. Coating is generally defined as a surface covering material. Whereas, paint is expressed to be a pigmented material and varnish a clear lacquer (ISO 4618/1) [1].

In general, paints and coatings consist of several layers. Usually a base layer which provides the right surface adhesion as well as flattens the surface, a middle layer which introduces the colour of the desired paint and an outer layer (clear lacquer) which protects the colour from environmental influ-ences. The composition of each layer differs depending on the surface composition of the substrate, the desired properties of the paint as well as ecological and economic constrains. Therefore, the com-positional investigation of the different layers can contribute to the determination of the origin of the sample.

The raw materials used in paints are based on petrochemical primary products as well as vegetable and animal oils and resins. Over the last years the interest in renewable raw materials has dramatically increased as well as the effort in reducing or recycling organic solvents used during the manufacture of paints, in order to decrease environmental pollution and human exposure (see Chapter 2.2).

Paints can either contain volatile or nonvolatile compounds. The former includes organic solvents,

(10)

water and coalescing agents, whereas the latter includes binders, resins, plasticisers, paint additives, pigments and extenders. In general, there are four basic components in paints: binders, pigments, solvents and additives. Usually all of these components occur in different quantities according to the desired properties of the paint, but not all have to be present necessarily [2, 3].

2.1.1. Binders

Binder is the fluid in which the pigment is suspended. When the binder is a liquid it acts as a carrier for the colour, when it is dry it acts as a ’glue’, holding the colour in place by providing adhesion to the substrate. Most of the binders used are organic, consisting of natural resins or synthetic polymers. The molecular mass varies between 500 and 30000 g mol . An increase in the relative molecular mass of the binder improves the properties of the polymer film, e.g. elasticity, hardness and increase in viscosity. Some common binders are synthetic binders (e.g.alkyd resins, polyvinyl acetate and acrylic), plant oils, resins and natural binders, such as natural latex, casein and cellulose [2–5].

2.1.2. Pigments

Pigments are usually present in form of powder and provide colour, opacity, as well as corrosion resis-tance. They can be subdivided into two groups: organic and inorganic. Of the former a large number of different pigments exist. Some of these might be partially soluble in certain solvents and resins. They can be either natural or synthesised. Their advantages over inorganic pigments are their vibrant and rich colours, which are more durable but also more expensive. Inorganic pigments have the ability to protect from corrosion and are more ultraviolet light resistant as well as have a better heat stabil-ity compared to organic pigments. In addition, they offer reflective and pearlescent effects. Extender pigments are a subgroup of inorganic pigments, which neither provide colour nor corrosion resistance but strongly affect other coating properties such as density, flow, hardness and permeability. These properties are useful in order to reduce production costs by adding extender pigments instead of other additives [3–5].

2.1.3. Solvents

Solvents are mostly required in order to dissolve the binder and to modify the viscosity of the paint, enabling the application of coatings by conventional methods. After applying the coating the solvent evaporates. Most common solvents are either organic liquids or water. The solubility of binders and resins varies strongly, following the well know rule in organic chemistry: ’like dissolves like’, meaning that polar resins are better solved in a polar solvent. Not only the compatibility of the solvent is important but also its rate of evaporation. Solvents with a high vapour pressure, which are rapidly evaporating, are considered fast or hot solvents. The rate of solvent evaporation has a major influence on the coatings properties. Solvents are one of the critical aspects in the field of paints and coatings due to the fact that most of them are volatile organic compounds (VOC) which in high concentration are harmful for human health as well as the environment [4, 5]. This will be further discussed in Chapter 2.2 Changes concerning Paints and Coatings.

2.1.4. Additives

Additives constitute a wide range of chemicals as well as functions. They are typically added in small amounts in order to improve certain properties of the coating. These include thickeners and surfactants,

(11)

2.2. Changes concerning Paints and Coatings 5

which reduce the surface tension of a liquid, or driers, which act as catalyst to the natural process of oxidation improving the drying process. These additives are of health concern as they cause health problems by inhalation etc. For this reason alternatives are under investigation and the application of plasticisers and other additives is minimised as much as possible [2, 4, 5].

2.2. Changes concerning Paints and Coatings

The composition of paints and coatings on the market continuously changed over the last two decades. This process has been introduced by the amending directive 1999/13/EC. It postulates the prevention or reduction of direct and indirect effects of emissions of volatile organic compounds into the environment by certain activities and installations. The directive 1999/13/EC defines volatile organic compounds as any organic compound which has a vapour pressure of 0.01 kPa or more at 293,15 K or which has a corresponding volatility under the particular conditions used. This change in regulations is based on the realisation that the emission of volatile organic compounds contribute to the ozone levels, which has a negative health and environmental impact when present in high concentration at the ground level [6]. In 2004 this regulation has been further specified in the pain directive 2004/42/EC. The aim of this directive is to prevent the negative effect of emissions of volatile organic compounds from decorative paints and vehicle refinishing products. It includes new requirements, such as the labelling of products, stating the subcategory of the product, the legal limit of VOC and the maximal VOC content of the product. Furthermore, it states the monitoring programme for the purpose of verifying compliance, including the maximal VOC content limits for paints and varnishes, differentiating between water-borne and solvent-borne paints (Annex II, see Chapter A) [7].

VOCs are necessary compounds of all organic surface coatings irrespective of the type. Their presence and concentration is dictated by the functional requirements of the paint. Replacement of these paints require a great deal of research, as not only the solvents need to be replaced but also the invention of new additives is mandatory, in order to create new paints with similar functions and properties. The paint manufacturers spend a lot of research, time and costs into the development of new paints. Therefore, the replacement of solvent-borne paints by high solids and/or water-borne paint is relatively slow [7, 8].

Solvent-borne paints are based on hydrocarbons and might be expected to be flammable, have strong primary odours and exposure limits that can be exceeded in confined spaces. Furthermore, most components used during the manufacturing of these type of paints are considered toxic and harmful to the human health, including solvents, binder resins and pigments. Water-borne solvents will not be flammable or toxic but are susceptible to freezing. Most of the important types of modern solvent-borne coatings, like epoxies, alkydes and acrylics, are also available in water-borne formulation. However, these water-borne paints are completely new designed, due to the unique character of water. Paint chemist must reinvent every part from resins to additives, as the dielectric constant, the density, the surface tension as well as the thermal conductivity of water are greater than of solvents. Furthermore, many polymers used as resins in paints cannot be dissolve in water. In order to achieve this, the backbone of the polymers has been chemically altered. Still most water-borne latex coatings exist as solid polymer particles dispersed in water. To stabilise the particle in the dispersion suitable additives are required. Furthermore, additives are required to avoid clogging of pigments as well as pigments integration with binders. Other additives may be used to prevent flash rusting of the steel before the water has evaporated [9, 10].

(12)

As mentions previously, certain health concerns also originate from pigments, which includes es-pecially lead and chromium pigments. Since 1978, the lead-based paints are banned from housings, due to its extensive environmental and human exposure. Lead is taken up by humans mostly trough dust particles which can affect adversely the brain, the central nervous system, blood cells and kid-ney, already at small concentration. At high concentration it can cause convulsion or even death. The exposure of children to lead can result in development delays, as well as lower IQs and increasing behavioural problems.

(13)

3

Spectroscopic and Imaging methods

The following part describes the most commonly used spectroscopic and imaging methods in the field of forensic analysis of modern paints and coatings. The principle as well as the application of each method will be described and discussed. Furthermore, characteristic parameters will be discussed, such as spectral and spatial resolution, spot size, penetration depth e.g.

Spectroscopic analysis is the measurement of interaction between matter and electromagnetic ra-diation, focusing on only a small part of the sample. Here, the parameter spectral resolution, spot size and penetration depth are characteristic. Spectral resolution defines the ability of the measurement to resolve features in the electromagnetic spectrum. It is determined by the smallest difference in wave-length that can be distinguished at a wavewave-length of lambda (𝜆). The maximal spectral resolution is determined by the contrast in the image obtained from the interference pattern between two orders of diffracted light. As Abbe (1873) observed, a diffraction pattern is formed in the back focal plane of the objective lens, if the diffracted light originates from a periodic specimen. His theory states that light is diffracted and forms diffraction pattern, when the incident wave plane hits a grating-like object. Every point in the back focal plane can be considered a source of coherent secondary disturbance. An image is formed from these interferences of the secondary light wave source, in the plane of the objective. This relationship is represented in Equation 3.1:

𝑅 = 𝜆

2𝑁𝐴 ≈ 𝑠𝑝𝑜𝑡𝑠𝑖𝑧𝑒 (3.1)

where R is the spectral resolution,𝜆 the wavelength of the illumination source, and NA the numerical aperture of the objective lens used [11].

The spot size refers to the minimum size of the incident beam on the sample surface and is of great importance in Raman microscopy as well as X-ray fluorescence spectroscopy. The greater the incident wavelength the greater the spot size. As the spot size depends on the wavelength as well as the spectral resolution, these two parameters are correlated. With increasing wavelength the spot size also increases, whereas the spectral resolution decreases, as the resolution improves the lower the value is [12].

(14)

The penetration depth is a parameter which is especially important in XRF and SEM-EDX. It de-scribes the distance that the incident beam travels into the sample. The extinction coefficient (𝜀) and the concentration (c) of the absorbing molecules determine the penetration depth (L). The correlation of these parameters can be seen in Equation 3.2.

𝐿 = 1

𝜀𝑐 (3.2)

Therefore the light travels a shorter distance, if more absorbing molecules are present along the pathway, -higher concentration results in lower penetration depth. Furthermore, the penetration depth relies on the wavelength as the extinction coefficient depends on the incident wavelength. The scatter-ing of light represents the deviation of a ray from a straight pass, due to the texture and the dynamics of the material being examined. The penetration depth of the light also depends on the scattering, as scattering will reduce the energy of the ray and therefore its penetration [13]. Thus the spectroscopic methods focus only lies on a small part of the sample, through mapping: repeated measurements on several spots on the sample; an impression on the samples general composition can be drawn.

Imaging techniques consider a wider area of the sample. Here, the spatial resolution pays an important part, as it states the ability to differentiate two objects in space. It is determined by the smallest difference that can be distinguished at a wavelength of lambda (Abbe’s limit). These techniques usually consider a wider area of the sample in order to collect as many information as possible. Here, the inhomogeneity of the paint sample enables the imaging of the sample.

3.1. Optical Analysis

In the following sections of this paper the principle of two common optical methods for the analysis of paints and coatings are described and discussed. In addition, their possible application in forensic investigations of paint samples is described in general.

3.1.1. Dark Field Microscopy

Dark field microscopy (DFM) is a simple method, which enables an improved optical analysis com-pared to the naked eye. It requires a standard microscope equipped with a dark field objective mask. The latter prevents the direct light rays from entering the objective. As visualised in Figure 3.1, the illumination light passes through the outer hollow ring of the objective and is further directed onto the sample at a high angle by mirrors. The light reflected by the surface then passes through the interior of the objective lens and reaches the eye piece or camera of the DFM. This illumination principle visu-alises only textured surfaces, whereas flat surfaces will appear dark due to the fact that the reflected light under this high incident angle will not be captured by the interior of the objective lens. In cases of paint samples, the surface is usually flat as well, however the heterogeneity of the paint results in a diffraction of the incident light beam. This diffraction in collected by the objective [14–16].

In order to observe a specimen in DFM it has to either reflect, refract or diffract the incident light. The former can be observed on smooth reflective surfaces for example. In contrast, refraction of light is being observed in situations where the refractive index of the specimen is different from the refractive indexes of its surroundings, or where gradients of refractive indexes are present. Both cases result in small angular changes of the direction of the light. This illumination technique is highly sensitive due to the fact that the images obtained are based on small amounts of diffracted light from miniature sample

(15)

3.1. Optical Analysis 9

Figure 3.1: Schematic visualisation of the principle of dark field microscopy1

surfaces, which clearly visualise the surface structure of the specimen against the dark background [17]. The images obtained are broader and less distinct compared to bright-field microscopy, due to the removal of one order of light information (unscattered light) from the diffraction plane, which makes the definition of the edges less distinct. Two types of objects can be analysed. One type considers the truly self-luminous objects. The other includes objects which deflect the light reaching them from an outside source, but do not emit light themselves. Using dark field microscopy, sample sizes of 0.2 µm or more in diameter can be analysed [18], even surface structure below the resolution limit can be visualised. Therefore, sample preparation is an important chapter in this optical technique, as features which lie above or below the focus plane also scatter the incident light and contribute to image degradation, in the same way as any dust in the sample chamber. The advantages of this method are that it is inexpensive and simple to employ as well as that it allows the detection of weak diffracted light, resulting in fine structural details. The disadvantage is that it is a unidirectional illumination of highly refractive objects, introducing large amounts of flare [16, 17].

The resolution of dark field microscopy is determined by Abbe’s limit (see Chapter 3) which states that the spatial resolution depends on the wavelength as well as the numerical aperture, as it can be seen in Equation 3.1. From this principle it can be concluded that the spatial resolution does not directly depend on the magnification used but rather on the wavelength of the incident light as well as the NA. Therefore, a high NA is suggested in order to gain a good resolution. If one considers an NA close to one, the spatial resolution of the DFM is half the wavelength of the visible light (400-700 nm) used. In DFm the area under investigation should be as great as possible without sacrificing too much resolution in order to gain the most information possible [16, 19, 20].Using a stereo microscope the number, the sequence colour, the thickness and the texture of each paint layer of a sample can be determined (see Figure 3.2). In cases of paint chips for example, the cross-section being observed is prepared by cutting perpendicular to sample surface with a microtome, in order to determine the colour and the number of layers.

1

http://www.leica-microsystems.com/science-lab/metallography-an-introduction/ [Accessed on 28th November 2016]

(16)

Figure 3.2: Visualisation of the paint layers of a paint chip using a microscope2

3.1.2. Colourimetric Analysis

Colourimetric Analysis defines the colour of a physical sample in terms of numbers, according to the colour notation system (z.B. CIELAB, Chapter B). This method simulates human colour vision, using a light source to illuminate the sample, a receptor to collect the scattered light (equivalent to the human eye) and the software to calculate the colour (the human brain) [21].

Colourimetry is a simple technique to determine low concentrations in a sample. Colourimetry and UV-Vis absorption spectroscopy can be used interchangeably, each term refers to the principle: that changes in colour of a sample can be monitored using absorbance spectroscopy. These changes can either be induced by indicators, whose reaction with the sample results in a change of colour, or in case of analysing coloured compounds these changes are naturally present. By measuring the UV-Vis absorption of the sample information about the colour of the sample can be obtained, as the energy of the absorption colour is complementary to the colour visible and/or detected. The concentration of the absorbing substance is directly proportional to the absorbed fraction, as stated in Beer-Lambert’s law (Equation 3.3 to 3.5).

𝐼 = 𝐼 10 (3.3)

𝐴 = 𝛽 ⋅ 𝑑 ⋅ 𝑐 = − log 𝑇 (3.4)

𝑇 = 𝐼

𝐼 = 10 (3.5)

where I is the amount of light emerging form a solution after absorption,𝐼 is the intensity of the light source entering the sample,𝛽 is the molar absorptivity (or extinction coefficient) with units of L mol cm , c is the concentration of the substance in solution, expressed in mol L and d is the path length of the sample (diameter of the cuvette holding the sample) in centimetres. A is the absorbance and T the fraction of the transmitted light. Figure 3.3 provides a visual representation of Beer-Lambert’s law. [22].

If the parameters𝛽 and d are known during the measurement then the concentration can be deter-mined from the measured absorption. Next to holding these parameters constant, six further conditions

2

http://www.milczarek.eu/wp-content/uploads/2012/10/IX.B-2-Milczarek.pdf[Accessed on 28th Novem-ber 2016]

(17)

3.1. Optical Analysis 11

Figure 3.3: Representation of the principle of Beer-Lambert’s-Law

have to be fulfilled in order to determine the concentration of a substance by colourimetry. First, over a reasonable concentration range the analyte colour has to obey Beer’s law. Furthermore, the colour has to be stable for a reasonable length of time, sensitive to small concentration changes of the substance and reproducible. In addition, the loss in transmittance has to only result from the absorption by the substance of interest. The entire substance has to be available for the reaction with the agent for the colour development, if the sample itself is not coloured. Lastly, the absorbent of light needs to be mea-surable. According to Beer’s law the absorption-concentration curve should be linear but in practice that is not true. Mostly, the curve is linear at the low concentration end but deviates from it at high concentrations, this can be explained by coagulation of the sample and/or stray radiation interfering with the measurement.

The colourimetric measurement is usually done using a spectrophotometer, whose basic principle is visualised in Figure 3.4.

Grating

Figure 3.4: Schematic visualisation of the principle of a spectrophotometer

The spectrophotometer disperses the reflected light from the sample into the different wavelength of the visible range. Mostly, a CCD detector simultaneously measures the intensity of each wavelength or wavelength range in relation to the illumination intensity.

These intervals are specified in terms of bandwidth. For a part of the colour spectrum (e.g. red, green, blue) the bandwidth corresponds to about 100 nm . With the use of lenses, slits and filters it

(18)

is possible to measure the entire spectrum of colours in intervals of 10 nm or lower. The bandwidth achievable depends on the slit width used. By decreasing the slit width the bandwidth also decreases, resulting in more accurate data. The smaller the slit width the lower the amount of light reaching the detector and therefore the lower the S/N-ratio. How low this slit width can be set depends on the performance of the detector, in consideration of sensitivity and stability and the required spectral resolution. The resulting reflectance curve is characteristic for each colour.

Two general concepts should be considered. One considers the intensity at a certain wavelength. Thereby the change of transmittance of a colour at a specific wavelength can be observed. The other monitors the change in the maximal absorption wavelength, such as detecting the change of a colour to a different one. Every spectrophotometer measurement requires a blank sample in order to determine the zero optical density intensity. For the quantification of the absorbing components of the sample, a calibration curve is required, using standards of known concentration at the optimal wavelength . Fur-thermore, test considering the working range of the spectrophotometer should be done using standards of known concentration. The concentration of the unknown sample should lie inside this range in order to acquire a quantitative results. The detection limit strongly depends on the sample of interest, as well as the sensitivity of the detector [23–30].

In cases of paint analysis, the illumination light is rather reflected than transmitted by the sample. Therefore the experimental set-up needs to be adapted.

The measuring geometries for the determination of the samples’ are decisive. The impression of a colour and/or colour difference can vary by applying different illumination conditions and viewing angles. The measurement geometries are therefore specified in internal standards. However, for each application the geometry yielding the right results needs to be determined, as the usefulness of the colourimetric reading is critical affect by it. In the investigation of paint samples, the spherical geometry provides the right results to detect the samples’ reflected light. Figure 3.5 shows the principle of a reflection integrating sphere. Here, the illumination is directly aligned with the sample port, where the

Figure 3.5: Principle of a reflection integrating sphere3

3

http://www.avantes.com/products/accessories/item/269-integrating-spheres[Accessed on 28th Novem-ber 2016]

(19)

3.2. Vibrational Spectroscopic Analysis 13

light is reflected. The inside of the sphere is a highly reflective, Lambertian surface. The light coming from the sample is reflected multiple times and scattered uniformly around the interior of the sphere. A fiber-optic collects a homogenised light signal and carries it to the spectrometer. This optic is protect by an baffled port which prevents the entering of the first reflections. The baffled port is independent on the angular properties of the sample port. Using this set-up the Beer-Lambert’s law is no longer valid [21]. Colourimetry enables the determination of the colour of the sample as well as its concentration. In addition, differences in constituents composition can be distinguished between different samples.

3.2. Vibrational Spectroscopic Analysis

The following section describes the most commonly used spectroscopic and imaging methods in the field of forensic analysis of modern paints and coatings. The principle as well as the application of each methods will be described and discussed. In addition, the types of analytes which can be studied with each technique will be explained.

3.2.1. Fourier Transform Infrared Spectroscopy

In infrared (IR) spectroscopy the interaction of the sample with infrared light is measured, by determining the absorption of the sample. As already stated in Chapter 3, the amount of light absorbed by the sample depends on the concentration, the path length as well as the molar absorptivity (Equation 3.4). The infrared spectrum usually displays the measured light intensity (T or A) against the photon energy of the light. The x-axis is plotted with the low wavenumbers to the right and high wavenumber to the left, by convention. The height or peak area of the resulting peak, in the absorbance spectrum, is directly proportional to the concentration of the absorbing substance. In order to determine the concentration of the sample using IR spectroscopy, Beer-Lambert’s law (Equation 3.3 to 3.5) is applied. For this purpose the system is calibrated using standards of known concentrations. IR spectra can also be plotted in percent transmittance units (y-axis) but absorbance spectra are more common because they can be used for quantitative analysis as the absorbance is linearly proportional to the concentration. Hence, absorbance spectra can be used for spectral subtraction. Both settings are useful for library search, and qualitative analysis.

In an IR spectrum the peak position is correlated with the molecular structure of the substance of interest. By comparing the reference spectra with the unknown one, close attention is brought to how well the peak position, heights and width match. In addition, IR spectroscopy is considered a relatively fast and easy technique, which is relatively inexpensive. Furthermore, it is a sensitive technique, as only a minimal amount of sample is required (ideally 10 g to 10 g). The disadvantage of IR spectroscopy is that only substances with vibration modes can be detected, excluding e.g. monoatomic ions, homo nuclear diatomic molecules and Nobel gases. Furthermore, complex compositions of a sample result in complex spectra which are difficult to interpret. This can be avoided by purifying the mixture or doing library searches. Another disadvantage of IR spectroscopy is the fact that liquid water results in broad and intense peaks which mask the spectrum of the sample [31, 32].

The advantages of IR spectroscopy are that it is an almost universal technique with information rich spectra. From an IR spectrum the structure of the unknown substance can be determined by the peak positions. As previously mentioned, the concentration can be determined by the peak height, whereas the peak width gives information on the samples sensitivity towards the chemical matrix, such as pH and hydrogen bonding.

(20)

Fourier Transform Infrared Spectroscopy (FTIR) is a form of IR spectroscopy. In this technique, an interferometer is used to measure the interference pattern between two light beams. The most commonly applied interferometer is the Michelson interferometer, visualised in Figure 3.6. As can be seen in this graphic, this type of instrument consists of four parts. The top part contains the IR source as well as a collimating mirror. The latter is used to collect the light from the source and parallels the rays. The bottom part illustrates the fixed mirror, whereas the right part visualises the moving mirror. In the middle of the scheme the beamsplitter is shown. This optical device transmits and reflects some of the incident light. The former is directed towards the fixed mirror, whereas the reflected light is directed towards the moving mirror. These two light path are each reflected at the corresponding mirror and recombined in the beamsplitter to a single beam. This new light beam then leaves the interferometer and interacts with the sample before it is detected by the detector [31–33].

Figure 3.6: Schematic visualisation of the principle of a Michelson interferometer, see text [31]

The recombination of the two light beams in the beamsplitter is based on the property of waves that states: when waves are superimposed they will interfere with each other, resulting in a new wave with an amplitude equal to the sum of their amplitudes, represented in Equation 3.6

𝐴 = 𝐴 + 𝐴 (3.6)

where𝐴 is the amplitude of the superimposed wave and 𝐴 and 𝐴 are the amplitudes of each individual light beam.

If𝐴 is greater than the incident amplitudes a constructive interference is being observed. The two light beams are still in phase and the difference of cycle displacement is zero or an integer number times the wavelength. If𝐴 is smaller than 𝐴 or 𝐴 , the optical path difference (𝛿) of the two light beams is a fraction of the wavelength, resulting in destructive interferences [31, 32].

The fundamental measurement obtained by FTIR is an interferogram, which plots the light intensity (in voltage) versus the optical path difference (𝛿). The resulting graph is a cosine wave, representing the electrical signal coming from the detector. The interferogram is measured using the Michelson interferometer by moving the mirror back and forth, which corresponds to one scan. The change in distance of the moving mirror influences the light intensity exiting the interferometer. The light intensity is maximal, if the optical path difference is equal to zero or an integer number times the wavelength and minimal if it corresponds to a fraction of a wavelength, considering monochromatic light. In the case of polychromatic light, each wavelength results in a different interferogram due to their different

(21)

frequencies. The detector usually measures all these different interferograms simultaneously and sums them. Hence, the signal at zero path difference will be amplified due to constructive interference, whereas the rest will be minimised due to destructive interference (see Figure 3.7) Therefore an IR spectrum contains two aspects: the light intensity and the wavenumber. Both parameters are included in the interferogram, which is used to calculate the spectrum [31, 32].

Figure 3.7: Schematic of the signal reaching the detector using a polychromatic light source4

The advantage of FTIR spectroscopy over general (dispersive) IR spectroscopy is the better signal-to-noise (S/N) ratio . This can be explained by the higher throughput of the FTIR as more incident light reaches the detector. In FTIR the light does not pass through a prism, grating or slit, which would decrease the light intensity of the measurement and therefore provides a higher signal. Furthermore, the S/N ratio of a spectral region is directly proportional to the square root of the number of scans added together (Equation 3.7).

𝑆/𝑁 ∝ √𝑁 (3.7)

where N is the number of scans.

Concluding from this relation the S/N ratio improves when more scans are added together. In addition, FTIR usually uses an IR laser, which acts as an internal standard for the wavenumber, resulting in a determination precision of a wavenumber of± 0.01 cm .

The disadvantage of FTIR compared to IR spectroscopy is the artifacts. These are characteristics visible in the sample spectrum which do not originate from the sample, such as water vapour and carbon dioxide peaks. Usually, these background signals can be removed by subtracting the sample spectrum from the background spectrum (spectrum without a sample present). But in FTIR these spectra are not measured simultaneously but rather after each other, which requires the opening of the instrument. This might result in a change of the environment of the sample chamber causing a variation of the background spectra [31, 32].

In most of the cases studied here, a mercury cadmium telluride (MCT) detector was used. Further a spectral resolution of 4 cm was obtained in a spectral range of 4000 to 650 cm . The upper wavenumber limit results from the fact that above 4000 cm there are no fundamental absorbances present. For the lower limit, the reason is the fact that the MCT is not sensitive for wavenumbers below

4

(22)

650 cm . The spectral resolution in FTIR is influenced by the optical path difference (OPD) . If the OPD is 100 cm the resolution is 0.01 cm , implying that peaks which are as close as 0.01 cm can be distinguished using an OPD of 100cm. As the data points in an interferogram are obtained at evenly spaced intervals, more data points are required for a high-resolution scan and therefore a greater optical path difference than for a low-resolution measurements is essential [34]. This relationship can be written as follows

𝑅𝑒𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 ∝ 1

𝛿 (3.8)

In addition, the higher the resolution the noisier the spectrum, resulting in the fact that a trade-off between the resolution and the S/N ratio has to be obtained. The physical state of the sample has a strong influence on the right choice. For example, in solids and liquids molecules are packed closely together. These close molecules experience significantly different environmental conditions, resulting in varying energies and therefore wider IR bands (like 10 cm or more). Therefore the resolution of the measurement should be around 8 or 4 cm , because a higher resolution would not lead to more information, just more noise. For gaseous samples a resolution of 2 cm or higher is suggested as these vapour molecules are well separated and show much narrower bandwidths [31, 35–37].

FTIR requires sample preparation, which includes the flattening of the sample using a diamante or microtome. This step is necessary, as the absorbance measured in FTIR is defined by Beer-Lambert’s law. Hence, the path length has to be reduced as the extinction coefficient (𝛽) is very high of paint samples. FTIR enables the identification of the binders, resins, fillers and pigments (more likely inor-ganic than orinor-ganic) in a paint trace. Depending on the set-up of the method as well as the preparation of the sample information about the composition of the individual paint layers can be obtained. This information is easily stored and can be helpful in comparing samples of presumed same origin. Fur-thermore, with the help of data bases conjectures about the samples origin, such as brand, year and manufacturer can be withdrawn. More information on the necessity and the availability of databases concerning paint spectra can be found in Appendix III: Databases (see Chapter C) [31–33, 38].

3.2.2. Raman Spectroscopy

A complementary method to FTIR is Raman spectroscopy, which also detects vibrational transitions of molecules. This technique retains information on the chemical structure as well as physical form of the sample. Each substance has a characteristic spectral pattern by which the molecules present in the sample can be identified. Furthermore, this methods enables the quantitative determination of the amount of substances in a sample.

In Raman spectroscopy, a laser light source is focused on the sample by an objective. The Raman scattering resulting from the irradiation of the sample is collected by the same objective and directed further into the optical path by the beamsplitter. Before it reaches the CDD detector it passes through the notch filter, which is wavelength specific and blocks the unwanted Rayleigh scatter. Figure 3.8 visualises the principle set-up of a Raman spectrometer.

Coming back to the principle of Raman spectroscopy. The Raman effect, named after its discov-erer Chandrasekhara Venkat Raman (1880-1970), is based on the scattering of light by a polyatomic molecule, resulting in a change of frequency, if some energy is exchanged between the incoming pho-ton and the scattering molecule. Two different scattering processes can be observed: elastic and inelastic scattering. The former is the most common one and is called Rayleigh scattering. In this

(23)

Figure 3.8: Schematic visualisation of a Raman microscope5

case, the excitation frequency is the same as the scatter frequency of the molecule. The Rayleigh peaks therefore appear at the same frequency as the excitation. As the unit of a Raman spectrum is the shift in energy between the incident photon and the scattered molecule [𝜈 in cm ], the Rayleigh scattering can be found at the zero mark. The inelastic scattering, also called Raman scattering, can be further subdivided into two types: Stokes and anti-Stokes. The former corresponds to molecules which are initially in the ground state and are excited to the vibrational excited state. In this process the frequency of the scattered photon is lower than the incident photon frequency, resulting in a loss of incident energy. Stokes peaks can be found on the low-energy side of the Rayleigh peaks. Anti-Stokes scattering refers to molecules which are initially in the vibrational excited state and transferred back to the ground state by donating energy to the incident photon. The peaks of this type of scattering can be found on the high energy side of the Rayleigh scatter as they have a positive shift in energy. Figure 3.9 visualises the different scattering processes in form of a Jablonski diagram.

The ratio between the two inelastic scattering principles is temperature dependent. At ambient temperature most molecules are present in the ground state, resulting in a more dominant Stokes scattering. This process is considered a weak process because only 10 to 10 of the incident photons scatter inelastically.

Raman scattering provides information on the vibrational levels of a given electronic state, usually the ground state. These vibrational levels are characteristic for a specific molecule as they depend on the chemical structure and conformation of the molecule [39–43].

Raman and IR are complementary methods. There are two advantages of the former over IR: (i) there is no overlapping of bands like in IR in case of RRS and (ii) the Raman spectrum extends well below 600 cm , where inorganic pigments and extender show characteristic Raman bands [44].

Raman spectroscopy is commonly used in forensic applications, such as in the analysis of modern paints and coatings. It enables the identification of the main pigments, resins and extenders. Raman spectroscopy is a non-destructive, non-invasive technique which obtains a good spatial resolution with minor or even no sample preparation [35, 44]. The spectral resolution in Raman microscopy is limited by the optical system used, such as objectives and filters. As already explained in Chapter 3 Spectroscopic

5

(24)

Figure 3.9: Schematic drawing of the Jablonski diagram showing the different types of Raman scatter-ing, as well as the phenomenon of fluorescence, resonance Raman and IR

and Imaging methods the spectral resolution is defined by Abbe’s limit, stating that the resolution is depending on the wavelength as well as NA. If a high NA is given then the spectral resolution is about half the excitation wavelength, which would be in the sub-micron range [45].

By the use of different excitation sources it is possible to identify several or even mixtures of pig-ments in a sample. This is very useful in the field of paint and coating analysis, due to the fact that paints often do not consist of only one pigment but rather a mixture of pigments [46]. The identification of these mixture is critical, in order to determine the manufacturer, brand and even year of manufacturing of the given sample. The quality of the gained Raman spectra depends on the excitation wavelength [47]. In some aspects, it might be useful to use resonance Raman spectroscopy (RRS) in order to improve the information gained from the spectra. RRS is a special type of Raman spectroscopy, where the choice of excitation wavelength can provide an improvement of the information gained by a spectra of a mixture. In classical Raman spectroscopy, only the frequency (𝜈 ) of the inducing light influences the signal. Here, the intensity of the scattered signal will vary according to𝜈 . However, in RRS the incident frequency (𝜈 ) corresponds to the irradiated molecules’ frequency of an electronic transition, resulting in an enhancement of an subset of Raman active modes, of up to six orders of magnitude (see Figure 3.9. Hereby, not the entire spectrum is improved but rather a single peak, which corresponds to the molecule whose electronic transition frequency matches the incident photon frequency. RRS en-ables a selective observation of a molecule in a complex matrix. In addition, information on the nature of the transition, the nuclei involved as well as the coupling between the transition and the different vibrational modes of the molecule are obtained. In classical Raman spectroscopy, fluorescence is a great disadvantage as the fluorescence signal masks possible Raman signals. Here, the origin of the fluorescence is the electronic excited state of the matrix or impurities which have the same frequency as the incident photon. As visualised on the right side in Figure 3.9, fluorescence only originates from the lowest vibrational state of the electronically excited state and results in a variety of possible emission energies, usually visualised in a broad peak. This phenomenon can be limited by several

(25)

possibili-3.2. Vibrational Spectroscopic Analysis 19

ties, either by changing the excitation wavelength (towards the IR-range) or by using a quencher or by gating the detector so that only the incident Raman scatter are detected or by measuring at two different wavelength so that the fluorescence background stays constant whereas the Raman peaks shift. [43, 48].

Coming back to the type of samples in this research, pigments are identified by their characteristic maximal absorption wavelength. This parameter determines the position as well as the intensity of the Raman signal and can be further improved using RRS, as most organic pigments are able to induce the resonance effect allowing an improved discrimination of the pigments signal from the complex matrix. Raman spectra show a high degree of discrimination between the samples. In forensic applications, the gained spectra of the sample are not only compared to reference spectra but also to possible samples of the suspect. Raman spectra are easily stored in order to form personal databases (possible databases are discussed in more detail in Section A) [49]. Most forensic applications make use of of several excitation laser as well as several objective lenses in order to gain the most information possible. Usually, the spectral range is between 400 and 2000 cm with varying resolutions [34–37, 45]. In most forensic application, a second method such as IR or XRF is used to support and extend the gained information.

In the field of forensic analysis of modern paints and coatings, Raman spectroscopy finds a broad range of applications, such as the analysis of paint flakes or smears of hit-and-run accidents or the analysis of documents and paintings. Hereby, the identification of (inorganic) pigments, binder and extenders can be achieved whereas the information on the former are of great interest as they contribute to the data obtained using FTIR.

3.2.3. Pyrolysis GC-MS

The third spectroscopic method is pyrolysis gas chromatography coupled to mass spectrometry (GC/MS) . In the analysis of paints and coatings, pyrolysis GC/MS plays an important role in the examination and comparison of organic portions, such as binder and additives. As the name already indicates it is a combination of three different principles: pyrolysis, chromatographic separation and mass specific detection.

During analytical pyrolysis the sample is subjected, under controlled conditions, to a sufficiently high temperature in the pyrolyzer (see Figure 3.10). During this process the constituent molecules break down into smaller fragments, called pyrolysates, which can be separated by GC. The pyrolysis reaction involves thermal cleavage of C-C bonds. Furthermore, thermally induced chemical reactions within the pyrolysates can occur. This is not desired and can be prevented by reaching the pyrolysis temperature as rapidly as possible. The reason why pyrolysis is necessary for the GC/MS analysis of paint binders is their high-molecular weight. Even though they are soluble in volatile organic solvents they can still not be volatilised in a common GC injection port, whereas the pyrolysates of the polymer can be volatilised and separated by the GC [50, 51].

The amount of sample used must corresponds to the capacity of the commercially available fused silica capillary column. This usually corresponds to a microgram range or less. This small sample range of the GC/MS improves the pyrolysis as the heat transport in small samples is enhanced, decreasing the probability of side reactions [51].

The pyrolysates are then introduced into the GC where there are mixed with the carrier gas and separated. The carrier gas, helium if combined to MS, constitutes the mobile phase and transports

(26)

Figure 3.10: Schematic of the principle of a pyrolyzer [51]

the analytes into an analytical column of which the inner surface is covered with a chemical film, also called stationary phase. The pyrolysates migrate through the column at various speeds, resulting in a separation of the molecules in time. The time each molecule requires to go through the column depends on its volatility as well as its interaction with the stationary phase. The separation speed of each molecule is related to its boiling point. This thermodynamic measurement depends on the molecular weight and the polarity. In order to gain an optimal separation, the choice of stationary film is strongly dependent on the nature of the substance of interest. In order for GC to be applicable, the substance of interest should be volatile and thermally stable with a low polarity. The separated molecules are then introduced into the MS [52].

Figure 3.11: Principle set-up of a GC/MS6

The MS consists of at least five elements: the analyte introduction system, here the GC outlet, the source, which produces gaseous ions, the analyser which allows the separation of ions, the detector which detects and quantifies the ions, as well as the acquisition and data treatment system [52]. After the injection from the GC into the source of the MS, without causing any disturbances in the high vac-uum of the latter, the separated compounds are ionised prior to the analysis in the mass analyser (see Figure 3.11). There is a variety of ionisation sources available, which each have different advantages

6

(27)

3.3. Elemental Analysis 21

and limitations and should be chosen depending on the physical and chemical properties of the ionised sample as well as on the internal energy transfer during the ionisation process. Ions are mainly pro-duced through the ejection or capture of electrons, protonation or deprotonation and adduct formation of neutral molecules in the gas phase. The mass analyser measures the physical properties of the ions in form of their mass-to-charge ratio (m/z) . Therefore it is important to consider that ions might also contain multiple charges resulting in an small apparent m/z value, which is a fraction of the actual mass.

Five main characteristics are important for the mass analyser. First, the m/z range over which a mass analyser can measure ions. Furthermore, the rate at which the analyser measures over a particular mass range, also called analysis speed, is important. Another main characteristic is the transmission ratio of the number of ions reaching the detector and the number of ions entering the mass analyser. This parameter includes the ions lost throughout other sections in the mass analyser, such as electric lenses. Mass accuracy represents the difference between the theoretical m/z value and the measured one. The last characteristic to consider is the resolution. It defines the ability of the mass analyser to yield distinct signals for two ions of close m/z values. Therefore, ions with smaller mass difference require a greater resolving power in order to be distinguished.

The next element of the MS is the detector. Here, an electric current is generated from the incident ions, which is proportional to their abundance. Due to the fact that the number of ions leaving the mass analyser at a particular instant is generally quite small, acquisition of an usable signal requires a significant amplification. High resolution MS instruments are able to measure the mass of ions and their associated isotopes with sufficient accuracy, in order to distinguish between isotopes of different molecules. As each element has characteristic mass defects elemental compositions are identifiable using MS [53]. In general, MS produces gas phase ions which undergo characteristic fragmentation. These products are then separated according to their m/z ratio and are then detected according to their abundance. The gained mass spectrum visualises the relative intensity or number of ions versus the m/z ratio. This type of instrument is able to achieve detection limits at pico- and femtomole level. This can be further optimised by coupling the MS to a chromatographic system [53]. Combining this three techniques resolves in a reliable technique to identify and compare different analytes. This method has a good discrimination power and requires only minimal sample destruction. From the pyrolysis product itself no conclusion about the structure can be drawn. The identification is empirical and is based on fingerprint patterns which are characteristic for a particular parent sample [54–56].

In the field of paint and coating analysis, pyrolysis GC/MS can be a useful tool for the characterisa-tion, classification and comparison of binders, additives and solvents. The degradation of the polymers present in these samples allows a comparative analysis of organic components of paint. This can be done for example by a visual comparison of the pyrolysis pattern, looking at the retention time, ab-sence or preab-sence of peaks as well as their relative intensity. In combination with the MS it is possible to not only compare the degradation products obtained but also to identify them [57]. In addition, this technique is suitable for the determination of organic compounds and residual monomers in paints.

3.3. Elemental Analysis

The information gained by spectroscopic analysis of paint samples, such as FTIR and/or Raman, are not allways sufficient to determine the origin of a sample. Therefore analytical methods which enable the determination of the elemental composition of a sample are used to complement or advance the

(28)

re-sults obtained by the methods described in Chapter 3 Vibrational Spectroscopic Analysis. The following section will describe their principle as well as their ability with the focus on analysing paint traces.

3.3.1. X-ray Fluorescence

X-ray fluorescence spectroscopy (XRF) is a non-destructive method to quantitatively and qualitatively analyse chemical elements based on the measurement of wavelength and intensity of their charac-teristic X-ray spectral lines. The primary beam of the light source irradiates the specimen, exciting each chemical element to emit second spectral lines, which have a characteristic wavelength for the element of interest, useful for the qualitative analysis of the sample. The intensity of the spectral line of an element is related to its concentration (quantification) [58, 59]. XRF is based on the principle that electromagnetic radiation occurs whenever an electron is removed from an electron shell of the target atom, by an electron beam. The resulting hole is filled by an electron from a higher energy level by releasing the dispensable energy in the form of characteristic X-rays. This can be explained in detail considering Bohr’s atomic model (Figure 3.12) [60].

Figure 3.12: Schematic of the Bohr’s atomic model

The positively charged nucleus is surrounded by electrons which move within defined areas. De-pending on the level that the electrons occupy on an atom, the binding strength with the nucleus varies. In order for an electron to be released, the primary beams energy has to be at least the binding energy of the electron. This binding energy increases with decreasing distance from the nucleus. The individ-ual shells are labelled with letters K, L, M. The innermost shell is K which contains two electrons, the second shell from the nucleus is L with eight electrons, and then M with 18 electrons. Using XRF every element can be clearly defined by its atomic number Z or by the number of electrons in the neutral state. Each element has a characteristic binding energy resulting from varying numbers of negative and positive charges in the nucleus. If an electron from an outer shell fills the hole in the inner shell the energy difference between these two levels is emitted in the form of X-rays, also called fluorescence yield (𝜔). This parameter depends on the atomic number of the element and the shell from which the electron has been removed. Hence, 𝜔 is low for light elements. The nomenclature of the released radiation considers the shell which loses the electron as well as the outer shell from which the electron

(29)

is transferred to stabilise the atom. Therefore, K-radiation considers the radiation emitted by filling up the K-shell, whereas the Greek letters (𝛼, 𝛽) as well as the numbers specify from which shell the filling electron originates (Figure 3.13) [59–61].

K L M

Figure 3.13: Schematic representation of the excitation and emission of XFR

While passing through the sample the X-ray intensity is weakened. The degree of weakening de-pends on the energy of the radiation as well as the chemical composition of the absorbing material. Heavier elements absorb better than light ones. The lines obtained are characteristic for the element and are independent on the chemical environment. The spectrum is a superposition of a continuous part and the characteristic lines. The continuous part is a broad wavelength band of radiation which is also called continuum or Bremsstrahlung. It arises due to deceleration of the accelerated electrons within the target material by interacting with the elemental electrons. This accelerated electrons origi-nate from the X-ray source [59]. The penetration depth in XRF strongly depends on the energy of the incident X-rays. The higher the energy the further the X-ray beam can travel inside the sample. In XRF there are two main aspects to consider when discussing the penetration depth: the penetration of the primary X-ray beam into the sample and the escape depth from which the fluorescent X-rays can be detected. Usually the penetration depth varies between 𝜇m and mm. In addition to the distance travelled by the X-ray beam inside the sample, the energy of the X-ray beam also influences which lines will be excited. The higher the energy of the beam the closer is the ejected electron to the nucleus. Therefore, high energy irradiation causes K-lines, whereas low energy irradiation causes more likely, L- or M-lines. Furthermore, the previously mentioned Bremsstrahlung is also influenced by the energy of the X-ray beam. The continuum can be reduced by reducing the energy of the incident beam, which causes a reduction in penetration depth [62–64].

When considering the escape depth of the fluorescent X-rays, the distance travelled also depends on their energy, which is directly related to the molecular weight of the element being detected. The energy of the escaping fluorescent X-rays increases with increasing molecular weight. Therefore, low absorption elements, such as sodium, magnesium and aluminium, are more difficult to detect, even close to the surface of the sample. X-rays of heavier elements (e.g. Cu, Ag, Au) pass through longer distances in the sample as there are more energetic.

The advantages of XRF are that it is relatively simple, cheap and allows a quick analysis. Further, it requires minimal sample preparation and is therefore suitable for trace analysis. One disadvantage is that very light elements (H to Ne) are not detectable using this technique. Furthermore, the analysis of liquid samples requires a time consuming sample preparation.[59, 65]

(30)

Figure 3.14: Schematic representation of the SEM-EDX set-up7

content of a paint trace. Further it can be used for the identification of pigments, filler and metals.

3.3.2. SEM-EDX

Another very common method for the determination of the elemental composition of a trace sample is scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDX). This technique is a non-destructive analytical method applicable for imaging and analysing very small bulk samples. The principle set-up of SEM-EDX is represented in Figure 3.14.

In the electron gun, electrons are emitted when an electric field concentrated at the sharp metal (usually tungsten) tip of the cathode is formed due to the potential difference between the tip and the anode, facilitating an emission current. By the use of electron lenses the beam diameter is controlled as well as focused on the specimen. These lenses have three kinds of defects, spherical aberration, chromatic aberration and astigmatism, which influence the properties of the electron beam.

The former is a limiting factor with respect to the resolving power of the SEM due to formation of a series of focal points caused by the off-set of the optical axis. These effects can be limited by the use of apertures, which are metal disks with circular micron sized holes. They reduce the diameter of the disk of minimum confusion. But they also reduce the beam current as some parts of the electron beam are blocked by the apertures (Figure 3.15A). The chromatic aberration considers energy distribution of the generated electron beam, as electrons of different energies experience different forces at the same location in the column. This phenomenon can not be avoided, but it only gets problematic when low accelerating voltages are used for imaging (Figure 3.15B). Lastly, the astigmatism describes the lack of symmetry in the electromagnetic lenses, which also results in a disk of minimum confusion (Figure 3.15C).

The electron beam interacts with the Coulomb field of the specimens nucleus and electrons, result-ing in different signal types: secondary electrons, backscattered electrons, X-rays and Auger electrons [66]. Secondary electrons (SE) are formed when the electron beam interacts with the electric field of the specimen, transferring its energy and causing a potential expulsion of an electron from the atom of

7

(31)

A B _C

Figure 3.15: Schematic of the different lens aberrations: (A) represents the spherical aberration, (B) the chromatic aberration and (C) the astigmatism8

the specimen. This is an inelastic effect. By definition, SEs have less than 50 eV and are mainly gen-erated from the electrons at the surface. The resulting image is a property of the surface structure of the specimen of interest rather than any underlying structure. Backscatter electrons (BSE) are formed during the interaction of the incident beam with the specimen causing a change in direction without a loss in energy (elastic effect). Therefore BSE have an energy range from 50 eV up to about incident energy. These electrons originate from a deeper sample depth. The formed X-rays have an even greater energy range and therefore a greater penetration depth. Combining the inelastic and elastic interactions results in a distribution of the electron beam over a three dimensional interaction volume (Figure 3.16)

Figure 3.16: Schematic representation of the interaction volume in SEM-EDX9

As it can be seen in Figure 3.16 SE and BSE have different penetration as well as escape depths,

8

http://www.charfac.umn.edu/[Accessed on 28th November 2016]

9

(32)

resulting from their different energies. The escape depth of SE is about 5 to 50 nm, while it is 100 times greater for BSE and even greater for X-rays. With increasing escape depth the lateral dimension from which the signal can be generated gets wider, resulting in a decrease of potential resolution. These values depend on the accelerating voltage of the electron beam as well as on the atomic number. With increasing the former, the interaction volume of the specimen increases due to the fact that the energy of the beam increases, resulting in a decrease in energy loss in the specimen and therefore greater penetrate depth. The atomic number has the opposite effect. Increasing the atomic number results in a decrease of interaction volume and an increase in energy loss rate [66, 67].

In general, the spot size is below 10 nm per diameter and an interaction of the electron beam with the specimen is observable up to a depth of approximately 1 µm.

In SEM the signals are generated by the electron-sample interaction introduced by rastering an electron beam over the sample from left to right and top to bottom, generating an image of the sam-ple in the vacuum chamber. The obtained signals provide great information on external morphology, crystalline structure and chemical structure of the sample. SEM in combination with EDX can perform elemental mapping and point analysis with a high accuracy and sensitivity. EDX is based on the same principle as explained in Chapter 3.3.1 X-ray Fluorescence. The incident beam eliminates an electron form the stronger bounded shells of the target atom. The formed hole is filled with an electron from a higher electronic level to stabilise the atom. This transition results in a loss of an X-ray with an energy equal to the difference of the two shells involved. This X-ray energy is characteristic for a specific atom, allowing an elemental identification of the atom present in the target specimen [66, 67].

The intensity is measured by counting the photons reaching the detector. The precision is limited by the statistical error, resulting in an accuracy of about±2%. The limit of detection of SEM-EDX is about 1000 ppm (by weight), while the spectral resolution is governed by the size of the incident electron beam, the stability of the microscope and the sample as well as the inherent properties of the signal-generation process. In SEM the resolution and the probe size are interrelated. At low magnification, the pixel size is usually determining the image resolution, as features smaller than the pixels cannot be observed. At high magnification, resolvable features can be displayed by several pixels as the probe size is bigger than the pixel size. In order to provide a better depth resolution, the usage of low energy electron beam is suggested, as the penetration depth is lower. The latter depends on the electron beam irradiance as well as the physical characteristics of the sample [68–70]. In order to analyse specimen under the SEM-EDX it has to be conductive. Most samples do not have this property, therefore their surface has to be coated to provide a path for the incident electrons to flow to the ground. This is usually done by vacuum evaporation of carbon, which forms a thin layer of about 10 nm around the specimen. This coating minimises interferences with the intensity of the X-ray detection.[66, 67, 70] SEM-EDX is suitable to provide an elemental analysis of each layer of a given paint sample. This technique also provides information on the morphology of the trace and can identify binders and pigments.

3.3.3. Inductively Coupled Plasma Mass Spectroscopy

Over the last years, the following technique has emerged in different fields of forensic investigations. This technique shows also a great potential in the analysis of paints and coatings. Inductively Cou-pled Plasma Mass Spectroscopy (ICP-MS) is a fast growing multi-element technique for trace level detection. The technique can be subdivided into two parts, the ICP and the mass spectrometer. The former is used for the sample introduction and ionisation, whereas the latter is used as an identification