- Email: [email protected]
Comparison of pigment content of paint samples using spectrometric methods
Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 130 (2014) 534–538
Contents lists available at ScienceDirect
Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy journal homepage: www.elsevier.com/locate/saa
Comparison of pigment content of paint samples using spectrometric methods Beata Trzcin´ska ⇑, Rafał Kowalski, Janina Zie˛ba-Palus Institute of Forensic Research, Westerplatte 9, 31-033 Krakow, Poland
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Paint samples of different colour
pigments were studied by Raman and VIS spectroscopy. The pigment identification based on Raman spectrum decreased with its concentration. The homogeneity of paint samples is an important factor for colour identification.
a r t i c l e
i n f o
Article history: Received 10 December 2013 Received in revised form 21 March 2014 Accepted 23 March 2014 Available online 13 April 2014 Keywords: Pigments Paint Vis microspectroscopy Raman spectroscopy IR spectroscopy colour comparison
a b s t r a c t The aim of the paper was to evaluate the influence of pigment concentration and its distribution in polymer binder on the possibility of colour identification and paint sample comparison. Two sets of paint samples: one containing red and another one green pigment were prepared. Each set consisted of 13 samples differing gradually in the concentration of pigment. To obtain the sets of various colour shades white paint was mixed with the appropriate pigment in the form of a concentrated suspension. After solvents evaporation the samples were examined using spectrometric methods. The resin and main filler were identified by IR method. Colour and white pigments were identified on the base of Raman spectra. Colour of samples were compared based on Vis spectrometry according to colour theory. It was found that samples are homogenous (parameter measuring colour similarity DE < 3). The values of DE between the neighbouring samples in the set revealed decreasing linear function and between the first and following one – a logarithmic function. Ó 2014 Elsevier B.V. All rights reserved.
Introduction Fragments of paint coat are very often found at crime scene. Their examination is performed in order to establish their origin and to compare with paint fragment originating from the suspect vehicle or tool used [1–4]. Different methods of optical microscopy and spectroscopy using the whole range of electromagnetic ⇑ Corresponding author. Tel.: +48 12 6185785; fax: +48 12 4223850. E-mail address: [email protected] (B. Trzcin´ska). http://dx.doi.org/10.1016/j.saa.2014.03.099 1386-1425/Ó 2014 Elsevier B.V. All rights reserved.
radiation are applied in routine examinations of paint samples. Infrared spectrometry provides information about polymer binder and fillers [5], while Raman spectrometry [6–9] – about pigment and dyes. The initial criterion of paint samples comparison is an estimation of the similarity of their colour. For relatively big samples (ca. 1 cm2) colour atlas is used for colour establishing, but this method is subjective and depends on the observer. Microspectrometry in the visible range MSP-Vis (combination of optical microscopy and spectrometry) is an objective way for colour comparison. This method is often applied in the colour examination of
´ ska et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 130 (2014) 534–538 B. Trzcin
many types of criminalistic traces e.g. paint samples. It enables one to perform measurement of a very small sample, i.e. less than 100 lm2. A quantitative colour description is also possible. It is based on the colour theory accepted by CIE (Comission Internationale de l’Eclairage). One of colour spaces is CIEL⁄a⁄b⁄ space characterised by axis: L⁄ – for lightness, a⁄ – green–red, b⁄ – blue–yellow colour-opponent dimensions [10–12]. Colour coordinates are calculated from Vis spectra. In many papers [13–16] its superiority to the visual method of colour comparison has been discussed. Thin layer chromatography (TLC) method with Vis spectrometry detection could be also used not only for colour examination but also in the identification of organic pigments. By Vis spectroscopy are not analysed paint samples but their colour components (each chromatographic spot) which was previously separated on TLC plate [13]. This procedure can give more spectral information than is obtainable from the paint itself. Some problems concerning the repeatability of measurements can occur, when achromatic or pearlescent coatings are analysed [17,18]. Sometimes statistical methods are needed in the evaluation of the results obtained [19,20]. Limited number of commercially available pigments (dyes) caused that desirable colour could be obtained by mixing few pigments (dyes). Pigments are insoluble in solvents or binders and therefore in micro-scale they could not be evenly distributed in the binder. So, two small fragments coming from the same coat could have different colour parameters calculated on the base of their VIS spectra. The aim of the paper was to check, if the pigment concentration and its distribution in the polymer binder has significant influence on the identification of the colour and the comparison of paint samples.
Materials and methods Two sets of paint samples were prepared, red and green one, of gradually changing shade. Red and green pigment were used as concentrated suspension (Pigment Mix, Inchem Poland) and their chemical composition were not known. These pigments were mixed with white paint (JedynkaÒ, Tikkurilla Poland SA, De˛bica). The reference and at the same time the initial solution (sample no. 0) was mixture of white paint and water in volume ratio 4:1. Each set contained thirteen paint samples (nos. 1–13) gradually differing in the concentration of colour pigment. Both sets were created in the same way. The first sample was the mixture of defined quality of pigment solution and initial solution of white paint. Each next sample was mixture defined quality of previous sample and initial solution. The volume of each obtained samples
Table 1 Concentration of paint solutions. Sample no.
Mass (g)
Concentration (weight%)
Pigment
Solution
1 2 3 4 5 6 7 8 9 10 11 12 13
8.00 4.00 1.98 0.99 0.49 0.25 0.12 0.06 0.03 0.015 0.008 0.004 0.002
40.04 40.43 39.99 40.00 39.99 39.99 40.00 40.01 40.00 39.99 39.99 40.00 40.00
19.98 9.89 4.94 2.47 1.24 0.62 0.31 0.15 0.08 0.04 0.02 0.01 0.005
535
was the same. The range of the concentrations was 0.005–20 weight% (Table 1). The samples were deposited on microscopic glass slide and examined after the evaporation of solvents.
Infrared measurements Infrared measurements were performed by means of an FTS 40 Pro Fourier-transform infrared spectrometer (BioRad/Digilab, USA, MA) which was equipped with a water-cooled high temperature ceramic source (MIR) and coupled with a UMA 500 microscope equipped with 15 objective and an MCT detector. Each spectrum of paint was obtained in transmission mode by averaging 128 sample and 64 reference spectra at 4 cm1 resolution in 3800– 600 cm1 range. KBr disc with thin film of paint sample was placed on the stage of the microscope and IR spectra were measured in the transmission mode. The spectra were recorded in absorbance units (the transmittance and absorbance units are mutually related and could be used invariable).
Raman measurements Raman spectra were obtained using a Renishaw inVia spectrometer equipped with a confocal Leica microscope and three types of excitation source: Ar ion (514.5 nm), He–Ne (632.5 nm) and a near infrared (785 nm) semiconductor lasers. The laser beam was focused on the samples by a 100 (N.A = 0.9) objective lens, which give a theoretical spot size of approximately 2 lm in diameter. The samples were analysed in situ in reflection mode. The light was dispersed by a diffraction grating with 2400 grooves/mm for the 514.5 nm laser and 1200 grooves/mm for 628.5 and 785 nm. The signal was recorded using a Peltier-cooled charged coupled device (CCD). Spectral data were processed with Renishaw Wire 3.2 software. Spectra were recorded in the 2500–200 cm1 range with an acquisition time of 10 s and 5 accumulations.
MSP-Vis measurements J&M TIDAS microspectrometer combined with a C. Zeiss Axioplan 2 microscope was used to perform measurements in the reflection mode. The spectral range was 380–780 nm. For the measurements an objective of the magnification 20 was used. Two programs were used, i.e. one of them controlled the observation (TIDASCOPE) while the other controlled the measurement and mathematical evaluation of the obtained data (J&M Spectralys 1.81). The measurements were performed in 10 spots (4 lm 2 lm). Initial solution of white paint was used as a reference. On the base of spectra the colour parameters (chromaticity coordinates) and colour difference parameter DE were calculated from the following formula:
DE ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 ðL1 L2 Þ þ ða1 a2 Þ2 þ ðb1 b2 Þ
where L1 , a1 , b1 and L2 , a2 , b2 are the CIE Lab values obtained for two samples labelled as 1 and 2. Colour coordinates L , a , b are used for colour representation in space, while on plane x, y coordinates are used. These units systems are mutually related. Colour is a subjective response of an eye and brain to incident radiation illuminate an object. To eliminate subjectivity of human perception colour coordinates are calculated from VIS spectrum. Mathematical distance between two points in colour space (DE) is a numerical description of a difference in colour.
536
´ ska et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 130 (2014) 534–538 B. Trzcin
Results The infrared, Raman and VIS spectra obtained for samples no. 0 (initial solution) and no. 13 (paint samples with the least concentration of pigment) in both sets were almost identical. It means that influence of pigment of concentration equal 0.005% on the spectrum shape could be neglected. It is clearly visible also in the chromaticity diagram, where the points representing samples no. 0 and no. 13 were lying close to each other and close to the neutral (N) white point (Fig. 1). IR spectra obtained for both sets of samples were similar, irrespective of their colour. The very intensive bands originated from the main fillers (white pigments) - calcium carbonate (875, 1420, 1800, 2520 cm1) and titanium dioxide – rutile (450–700 cm1) were very well visible. Weak intensities bands (730 and 1734 cm1) suggested that isophtalate alkyd could also be present as the polymer binder. In 900–1380 cm1 region there were visible bands about very weak intensity originating most probably from the organic pigments (Fig. 2). Their intensity quickly decreased within the set. Beginning with the spectra of samples no. 5 they were practically invisible. The presence of bands assigned to one of white pigments – titanium dioxide (rutile – PW6) was observed in Raman spectra of the sample no. 0 as well as in spectra of samples with number greater than 9 (pigment concentration 0.08%). The higher the sample number, the more intensive bands originating from rutile were observed. It was noticed that the intensities of bands of red and green pigments depended on pigment type and wavelength of the excitation laser. The green pigment suspension (used in green set of samples) contained two pigments: pigment yellow number 74 (PY 74 – pigment from the group of triaminopyrimidinazobenzoic acid) and pigment green number 7 – Table 2 (PG 7 – pigment from group of chlorinated phthalocyanine). The bands attributed to PY74 are observed in the Raman spectra using 514 nm excitation laser whereas those assigned to PG7 are observed in the spectrum obtained using 633 and 785 nm excitation lasers (Figs. 3 and 4). No spectra of red paint samples, were obtained using 514 nm line (Fig. 5). Spectra obtained using 633 and 785 nm laser were similar (Fig. 6) and similar but not identical with spectrum of red pigment number 83 (PR83). The red pigment suspension (used in red set of samples) was also a mixture, similar to the green pigment suspension. The red pigment PR 83 is probably the main component of this mixture.
Fig. 2. IR spectra of white paint (0) and samples no. 1.
Table 2 Pigments identified by Raman spectroscopy. Pigment
Chemical compound
Laser (nm)
Characteristic bands (cm1)
PW 6 PY74
Rutile Azo compounda
633 514
PG 7
Chlorinated copper phthalocyanine Alizarine
633
447, 608 801, 1262, 1325, 1352, 1402, 1510, 1592 684, 740, 776, 1214, 1282, 1338, 1446, 1538 526, 967, 1161, 1230, 1282, 1357, 1482, 1579
PR 83
a PY 74 = azo PYRIMIDINYL)AZO).
compound
633
(BENZOIC
ACID,
p-((2,4,6-TRIAMINO-5-
Fig. 3. Raman spectra of green paint nos. 0, 1, 4 and 8 (R – rutile, PY – pigment yellow).
Fig. 1. Chromaticity diagram for paint samples (1R – red sample no 1, 1G – green sample no. 1, 13R and 13G – red and green samples no. 13 respectively).
The second component was not identified. It not be excluded that it could be another form of this kind of pigment (PR83.1 or PR83.3). Raman pigment library does not contain spectra of all form of this pigment. It was observed that the shape of Vis spectra of samples within set were the same and only their intensity changed. No bands were present in the range above 600 nm in spectra of red paints. The main band for green paints was located at 645 nm. Colour coordinates for samples in both sets changed in different way (Table 3). In set of red samples (from sample no. 1 to sample no. 13) the value of ‘‘L⁄’’ parameter increased (lightness was greater), while parameter ‘‘a⁄’’ decreased (shade shifts towards more light but still in red area). In the set of green samples parameters ‘‘L⁄’’ and ‘‘a⁄’’ increased (shade still shifts towards more light but still in red
´ ska et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 130 (2014) 534–538 B. Trzcin
537
Table 3 Colour coordinates in space (L⁄ – lightness, a⁄ – green–red, b⁄ – blue–yellow axis) of the analysed samples. Sample no.
1 2 3 4 5 6 7 8 9 10 11 12 13
Fig. 4. Raman spectra of green paint nos. 0, 1, 4 and 8 (R – rutile, PG – pigment green).
L
a
b
Red
Green
Red
Green
Red
Green
56.69 64.18 73.34 77.54 83.03 90.05 93.08 95.04 97.06 98.34 99.05 99.54 100.62
54.93 65.76 75.18 81.00 88.30 91.48 93.43 96.16 97.65 98.66 99.65 99.63 99.59
60.82 55.31 48.05 38.25 29.10 21.56 14.35 11.01 7.68 4.37 1.82 1.41 0.75
50.84 48.50 44.26 35.54 29.05 21.47 15.5 11.17 7.43 4.82 2.78 1.73 0.96
23.13 14.85 9.41 4.54 2.17 0.01 0.10 0.55 0.18 0.59 0.96 0.03 0.63
14.95 12.92 11.20 8.44 5.77 4.21 2.56 1.90 0.88 0.46 0.16 0.38 0.08
Table 4 Values of DE parameters and errors of measurement. Sample pairs
Measurement replication Neighbouring The first sample and following one a
DE a
RSD
Red
Green
Red
Green
0.33–2.44
0.23–1.30
0.38–0.72
0.34–0.52
1.4–12.8 12.5–77.7
0.9–11.2 11.3–68.6
0.06–0.29 0.01–0.09
0.04–0.18 0.01–0.07
DE value <3 means that colour of samples are very similar or the same.
Table 5 DE (colour difference) value of the analysed pair of samples in relation to concentration (cn, cm concentrations of pigment for pair of samples).
Fig. 5. Raman spectra of red paint nos. 0, 1, 6 and 10 (R – rutile) at 785 nm.
Fig. 6. Raman spectra of red paint nos. 0, 1, 6 and 10 (R – rutile) at 514 nm.
area). Parameter ‘‘b⁄’’ decreased in both sets (shift towards yellow area). Colour difference parameter (DE) defines colour similarity between two samples. It was statistically calculated (for inexperienced observer) that DE = 3 is a limit (threshold) of this similarity for both sets of paint samples DE calculated for repeated measurement was less than limit value (DE < 3). So, it could be clearly seen (Table 4) that the analysed samples are homogenous. There is not simple relation between colour coordinates and concentration of colourants. Generally colour coordinates depend
Pair of samples (cn cm)
DE Red
Green
1 and 2 (10.09) 2 and 3 (4.95) 3 and 4 (2.47) 4 and 5 (1.23) 5 and 6 (0.62) 6 and 7 (0.31) 7 and 8 (0.16) 8 and 9 (0.07) 9 and 10 (0.04) 10 and 11 (0.02) 11 and 12 (0.01) 12 and 13 (0.005)
12.62 13.00 11.78 11.01 10.56 7.85 3.98 3.98 3.58 2.70 1.27 1.47
11.32 10.56 10.89 10.19 8.40 6.47 5.22 4.16 2.87 2.31 1.10 0.94
Pair of samples (cn cm)
DE Red
Green
1 1 1 1 1 1 1 1 1 1 1 1
12.62 25.15 35.96 46.29 56.50 63.45 66.82 70.71 74.07 76.54 76.83 77.76
11.32 21.66 30.93 40.91 48.11 53.69 58.70 62.51 65.12 67.30 67.99 68.62
and and and and and and and and and and and and
2 (10.09) 3 (15.04) 4 (17.51) 5 (18.74) 6 (19.36) 7 (19.67) 8 (19.83) 9 (19.90) 10 (19.94) 11 (19.96) 12 (19.97) 13 (19.98)
on wavelength of incident radiation and intensity of spectrum is function of compound concentration. Human experience shows that the grater colour difference is seen the grater difference of concentration of colourants ought to be. So, it was expected that colour difference between neighbouring samples (constant concentration ratio) will be smaller than this between the first samples and the following ones (decreasing concentration ratio). This supposition turn out to be right (Table 5). Colour difference parameter (DE) for the neighbouring samples could be described by linear decreasing function (Fig. 7). It means, the smaller is difference of pigment concentration for analysed samples the smaller is difference in colour. Also it turn out that colour difference for the first sample and the following ones could be described by logarithmic function (Fig. 8). It means, the grater is difference of pigment concentration for analysed samples the grater is difference in colour.
´ ska et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 130 (2014) 534–538 B. Trzcin
538
Fig. 7. DE value of compared neighbouring red paint samples.
colour comparison. For both sets of paint samples having in their composition different pigment (red or green), the obtained DE value (colour difference parameter) of repeated measurements pointed out on identity of colour on whole surface of sample (DE < 3). In that way the homogeneity of prepared samples was confirmed and comparison of samples colour was possible. Homogeneity of commercial paint is certainly better than paint samples prepared in laboratory. So, colour analysis of real samples coming from place of accident and suspected car is possible. If colour of these samples is similar (DE < 3) they could come from one source. The possibility of colour pigment identification basing on Raman spectra decreased significantly when pigment concentration is less than 0.08% of weight. Presence of white pigment (rutile) hinder identification of colour pigment. When colour pigment concentration is above 1% of weight only bands attributed to it are seen on Raman spectrum. References
Fig. 8. DE value of first and the following green paint samples.
Averaged ranges of DE and the relative standard deviation (RSD) for analysed types of samples pairs are given in Table 4. Obtained results will be useful for analysis of real paint samples which are often object of criminalistic cases. Conclusions It could be concluded that homogeneity (pigment distribution in binder) of paint samples is not a factor, which makes difficult
[1] A. Beveridge, T. Fung, D. MacDougall, in: B. Caddy (Ed.), Forensic Examination of Glass and Paints, Taylor & Francis, London and New York, 2001, pp. 183–241. [2] S.G. Roland, T.A. Jergovich, K.P. Kirkbride, Forensic Sci. Rev. 18 (2006) 98–117. [3] F. Govaert, M. Bernard, Forensic Sci. Int. 140 (2004) 61–70. [4] ASTM International, E1610-02 standard guide for forensic paint analysis and comparison, ASTM International, West Conhohoken, PA, 2005. [5] J. Zie˛ba-Palus, in: R.A. Meyers (Ed.), Encyclopedia of Analytical Chemistry: Instrumentation and Applications, J. Wiley, Chichester, 2000, pp. 4484–4497. [6] J. de Gelder, P. Vandenabeele, F. Govaert, L. Moen, J. Raman Spectroscopy 36 (2005) 1059–1067. [7] E.G. Bartick, P. Buzzini, in: R.A. Meyers (Ed.), Encyclopedia of Analytical Chemistry, John Wiley & Sons Ltd, U.K., 2009. [8] J. Zie˛ba-Palus, A. Michalska, A. Wesełucha-Birczyn´ska, J. Mol. Struct. 993 (2011) 134–141. [9] A. Colombini, D. Kaifas, e-PS 7 (2010) 14–21. [10] M.C. Taylor, D.R. Cousins, R.H. Holding, J. Locke, J.M. Wilkinson, Forensic Sci. Int. 40 (1989) 131–141. [11] B. Trzcin´ska, J. Zie˛ba-Palus, P. Kos´cielniak, J. Mol. Struct. 924–926 (2009) 393– 399. [12] P. Martin, in: J.A. Siegel, P.J. Saukko (Eds.), Encyclopedia of Forensic Sciences, second ed., Waltham, USA, 2013, pp. 148–154. [13] N.A. Fuller, Forensic Sci. Int. 27 (1985) 189–204. [14] W. Stoeklein, H. Fujiwara, Sci. Justice 39 (1999) 188–195. [15] G. Massonnet, W. Stoecklein, Sci. Justice 39 (1999) 128–134. [16] J. Zie˛ba-Palus, B. Trzcin´ska, P. Kos´cielniak, Anal. Lett. 43 (3) (2010) 436–445. [17] M.E. Nadal, E.A. Early, Colour Res. Appl. 29 (1) (2004) 38–42. [18] K.A. Kopchick, C.R. Bommarito, J. Forensic Sci. 51 (2006) 340–343. [19] A. Takagi, S. Sato, G. Baba, Colour Res. Appl. 32 (2007) 378–387. [20] A. Takagi, A. Watanabe, G. Baba, Colour Res. Appl. 30 (2005) 275–282.
Contents lists available at ScienceDirect
Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy journal homepage: www.elsevier.com/locate/saa
Comparison of pigment content of paint samples using spectrometric methods Beata Trzcin´ska ⇑, Rafał Kowalski, Janina Zie˛ba-Palus Institute of Forensic Research, Westerplatte 9, 31-033 Krakow, Poland
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Paint samples of different colour
pigments were studied by Raman and VIS spectroscopy. The pigment identification based on Raman spectrum decreased with its concentration. The homogeneity of paint samples is an important factor for colour identification.
a r t i c l e
i n f o
Article history: Received 10 December 2013 Received in revised form 21 March 2014 Accepted 23 March 2014 Available online 13 April 2014 Keywords: Pigments Paint Vis microspectroscopy Raman spectroscopy IR spectroscopy colour comparison
a b s t r a c t The aim of the paper was to evaluate the influence of pigment concentration and its distribution in polymer binder on the possibility of colour identification and paint sample comparison. Two sets of paint samples: one containing red and another one green pigment were prepared. Each set consisted of 13 samples differing gradually in the concentration of pigment. To obtain the sets of various colour shades white paint was mixed with the appropriate pigment in the form of a concentrated suspension. After solvents evaporation the samples were examined using spectrometric methods. The resin and main filler were identified by IR method. Colour and white pigments were identified on the base of Raman spectra. Colour of samples were compared based on Vis spectrometry according to colour theory. It was found that samples are homogenous (parameter measuring colour similarity DE < 3). The values of DE between the neighbouring samples in the set revealed decreasing linear function and between the first and following one – a logarithmic function. Ó 2014 Elsevier B.V. All rights reserved.
Introduction Fragments of paint coat are very often found at crime scene. Their examination is performed in order to establish their origin and to compare with paint fragment originating from the suspect vehicle or tool used [1–4]. Different methods of optical microscopy and spectroscopy using the whole range of electromagnetic ⇑ Corresponding author. Tel.: +48 12 6185785; fax: +48 12 4223850. E-mail address: [email protected] (B. Trzcin´ska). http://dx.doi.org/10.1016/j.saa.2014.03.099 1386-1425/Ó 2014 Elsevier B.V. All rights reserved.
radiation are applied in routine examinations of paint samples. Infrared spectrometry provides information about polymer binder and fillers [5], while Raman spectrometry [6–9] – about pigment and dyes. The initial criterion of paint samples comparison is an estimation of the similarity of their colour. For relatively big samples (ca. 1 cm2) colour atlas is used for colour establishing, but this method is subjective and depends on the observer. Microspectrometry in the visible range MSP-Vis (combination of optical microscopy and spectrometry) is an objective way for colour comparison. This method is often applied in the colour examination of
´ ska et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 130 (2014) 534–538 B. Trzcin
many types of criminalistic traces e.g. paint samples. It enables one to perform measurement of a very small sample, i.e. less than 100 lm2. A quantitative colour description is also possible. It is based on the colour theory accepted by CIE (Comission Internationale de l’Eclairage). One of colour spaces is CIEL⁄a⁄b⁄ space characterised by axis: L⁄ – for lightness, a⁄ – green–red, b⁄ – blue–yellow colour-opponent dimensions [10–12]. Colour coordinates are calculated from Vis spectra. In many papers [13–16] its superiority to the visual method of colour comparison has been discussed. Thin layer chromatography (TLC) method with Vis spectrometry detection could be also used not only for colour examination but also in the identification of organic pigments. By Vis spectroscopy are not analysed paint samples but their colour components (each chromatographic spot) which was previously separated on TLC plate [13]. This procedure can give more spectral information than is obtainable from the paint itself. Some problems concerning the repeatability of measurements can occur, when achromatic or pearlescent coatings are analysed [17,18]. Sometimes statistical methods are needed in the evaluation of the results obtained [19,20]. Limited number of commercially available pigments (dyes) caused that desirable colour could be obtained by mixing few pigments (dyes). Pigments are insoluble in solvents or binders and therefore in micro-scale they could not be evenly distributed in the binder. So, two small fragments coming from the same coat could have different colour parameters calculated on the base of their VIS spectra. The aim of the paper was to check, if the pigment concentration and its distribution in the polymer binder has significant influence on the identification of the colour and the comparison of paint samples.
Materials and methods Two sets of paint samples were prepared, red and green one, of gradually changing shade. Red and green pigment were used as concentrated suspension (Pigment Mix, Inchem Poland) and their chemical composition were not known. These pigments were mixed with white paint (JedynkaÒ, Tikkurilla Poland SA, De˛bica). The reference and at the same time the initial solution (sample no. 0) was mixture of white paint and water in volume ratio 4:1. Each set contained thirteen paint samples (nos. 1–13) gradually differing in the concentration of colour pigment. Both sets were created in the same way. The first sample was the mixture of defined quality of pigment solution and initial solution of white paint. Each next sample was mixture defined quality of previous sample and initial solution. The volume of each obtained samples
Table 1 Concentration of paint solutions. Sample no.
Mass (g)
Concentration (weight%)
Pigment
Solution
1 2 3 4 5 6 7 8 9 10 11 12 13
8.00 4.00 1.98 0.99 0.49 0.25 0.12 0.06 0.03 0.015 0.008 0.004 0.002
40.04 40.43 39.99 40.00 39.99 39.99 40.00 40.01 40.00 39.99 39.99 40.00 40.00
19.98 9.89 4.94 2.47 1.24 0.62 0.31 0.15 0.08 0.04 0.02 0.01 0.005
535
was the same. The range of the concentrations was 0.005–20 weight% (Table 1). The samples were deposited on microscopic glass slide and examined after the evaporation of solvents.
Infrared measurements Infrared measurements were performed by means of an FTS 40 Pro Fourier-transform infrared spectrometer (BioRad/Digilab, USA, MA) which was equipped with a water-cooled high temperature ceramic source (MIR) and coupled with a UMA 500 microscope equipped with 15 objective and an MCT detector. Each spectrum of paint was obtained in transmission mode by averaging 128 sample and 64 reference spectra at 4 cm1 resolution in 3800– 600 cm1 range. KBr disc with thin film of paint sample was placed on the stage of the microscope and IR spectra were measured in the transmission mode. The spectra were recorded in absorbance units (the transmittance and absorbance units are mutually related and could be used invariable).
Raman measurements Raman spectra were obtained using a Renishaw inVia spectrometer equipped with a confocal Leica microscope and three types of excitation source: Ar ion (514.5 nm), He–Ne (632.5 nm) and a near infrared (785 nm) semiconductor lasers. The laser beam was focused on the samples by a 100 (N.A = 0.9) objective lens, which give a theoretical spot size of approximately 2 lm in diameter. The samples were analysed in situ in reflection mode. The light was dispersed by a diffraction grating with 2400 grooves/mm for the 514.5 nm laser and 1200 grooves/mm for 628.5 and 785 nm. The signal was recorded using a Peltier-cooled charged coupled device (CCD). Spectral data were processed with Renishaw Wire 3.2 software. Spectra were recorded in the 2500–200 cm1 range with an acquisition time of 10 s and 5 accumulations.
MSP-Vis measurements J&M TIDAS microspectrometer combined with a C. Zeiss Axioplan 2 microscope was used to perform measurements in the reflection mode. The spectral range was 380–780 nm. For the measurements an objective of the magnification 20 was used. Two programs were used, i.e. one of them controlled the observation (TIDASCOPE) while the other controlled the measurement and mathematical evaluation of the obtained data (J&M Spectralys 1.81). The measurements were performed in 10 spots (4 lm 2 lm). Initial solution of white paint was used as a reference. On the base of spectra the colour parameters (chromaticity coordinates) and colour difference parameter DE were calculated from the following formula:
DE ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 ðL1 L2 Þ þ ða1 a2 Þ2 þ ðb1 b2 Þ
where L1 , a1 , b1 and L2 , a2 , b2 are the CIE Lab values obtained for two samples labelled as 1 and 2. Colour coordinates L , a , b are used for colour representation in space, while on plane x, y coordinates are used. These units systems are mutually related. Colour is a subjective response of an eye and brain to incident radiation illuminate an object. To eliminate subjectivity of human perception colour coordinates are calculated from VIS spectrum. Mathematical distance between two points in colour space (DE) is a numerical description of a difference in colour.
536
´ ska et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 130 (2014) 534–538 B. Trzcin
Results The infrared, Raman and VIS spectra obtained for samples no. 0 (initial solution) and no. 13 (paint samples with the least concentration of pigment) in both sets were almost identical. It means that influence of pigment of concentration equal 0.005% on the spectrum shape could be neglected. It is clearly visible also in the chromaticity diagram, where the points representing samples no. 0 and no. 13 were lying close to each other and close to the neutral (N) white point (Fig. 1). IR spectra obtained for both sets of samples were similar, irrespective of their colour. The very intensive bands originated from the main fillers (white pigments) - calcium carbonate (875, 1420, 1800, 2520 cm1) and titanium dioxide – rutile (450–700 cm1) were very well visible. Weak intensities bands (730 and 1734 cm1) suggested that isophtalate alkyd could also be present as the polymer binder. In 900–1380 cm1 region there were visible bands about very weak intensity originating most probably from the organic pigments (Fig. 2). Their intensity quickly decreased within the set. Beginning with the spectra of samples no. 5 they were practically invisible. The presence of bands assigned to one of white pigments – titanium dioxide (rutile – PW6) was observed in Raman spectra of the sample no. 0 as well as in spectra of samples with number greater than 9 (pigment concentration 0.08%). The higher the sample number, the more intensive bands originating from rutile were observed. It was noticed that the intensities of bands of red and green pigments depended on pigment type and wavelength of the excitation laser. The green pigment suspension (used in green set of samples) contained two pigments: pigment yellow number 74 (PY 74 – pigment from the group of triaminopyrimidinazobenzoic acid) and pigment green number 7 – Table 2 (PG 7 – pigment from group of chlorinated phthalocyanine). The bands attributed to PY74 are observed in the Raman spectra using 514 nm excitation laser whereas those assigned to PG7 are observed in the spectrum obtained using 633 and 785 nm excitation lasers (Figs. 3 and 4). No spectra of red paint samples, were obtained using 514 nm line (Fig. 5). Spectra obtained using 633 and 785 nm laser were similar (Fig. 6) and similar but not identical with spectrum of red pigment number 83 (PR83). The red pigment suspension (used in red set of samples) was also a mixture, similar to the green pigment suspension. The red pigment PR 83 is probably the main component of this mixture.
Fig. 2. IR spectra of white paint (0) and samples no. 1.
Table 2 Pigments identified by Raman spectroscopy. Pigment
Chemical compound
Laser (nm)
Characteristic bands (cm1)
PW 6 PY74
Rutile Azo compounda
633 514
PG 7
Chlorinated copper phthalocyanine Alizarine
633
447, 608 801, 1262, 1325, 1352, 1402, 1510, 1592 684, 740, 776, 1214, 1282, 1338, 1446, 1538 526, 967, 1161, 1230, 1282, 1357, 1482, 1579
PR 83
a PY 74 = azo PYRIMIDINYL)AZO).
compound
633
(BENZOIC
ACID,
p-((2,4,6-TRIAMINO-5-
Fig. 3. Raman spectra of green paint nos. 0, 1, 4 and 8 (R – rutile, PY – pigment yellow).
Fig. 1. Chromaticity diagram for paint samples (1R – red sample no 1, 1G – green sample no. 1, 13R and 13G – red and green samples no. 13 respectively).
The second component was not identified. It not be excluded that it could be another form of this kind of pigment (PR83.1 or PR83.3). Raman pigment library does not contain spectra of all form of this pigment. It was observed that the shape of Vis spectra of samples within set were the same and only their intensity changed. No bands were present in the range above 600 nm in spectra of red paints. The main band for green paints was located at 645 nm. Colour coordinates for samples in both sets changed in different way (Table 3). In set of red samples (from sample no. 1 to sample no. 13) the value of ‘‘L⁄’’ parameter increased (lightness was greater), while parameter ‘‘a⁄’’ decreased (shade shifts towards more light but still in red area). In the set of green samples parameters ‘‘L⁄’’ and ‘‘a⁄’’ increased (shade still shifts towards more light but still in red
´ ska et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 130 (2014) 534–538 B. Trzcin
537
Table 3 Colour coordinates in space (L⁄ – lightness, a⁄ – green–red, b⁄ – blue–yellow axis) of the analysed samples. Sample no.
1 2 3 4 5 6 7 8 9 10 11 12 13
Fig. 4. Raman spectra of green paint nos. 0, 1, 4 and 8 (R – rutile, PG – pigment green).
L
a
b
Red
Green
Red
Green
Red
Green
56.69 64.18 73.34 77.54 83.03 90.05 93.08 95.04 97.06 98.34 99.05 99.54 100.62
54.93 65.76 75.18 81.00 88.30 91.48 93.43 96.16 97.65 98.66 99.65 99.63 99.59
60.82 55.31 48.05 38.25 29.10 21.56 14.35 11.01 7.68 4.37 1.82 1.41 0.75
50.84 48.50 44.26 35.54 29.05 21.47 15.5 11.17 7.43 4.82 2.78 1.73 0.96
23.13 14.85 9.41 4.54 2.17 0.01 0.10 0.55 0.18 0.59 0.96 0.03 0.63
14.95 12.92 11.20 8.44 5.77 4.21 2.56 1.90 0.88 0.46 0.16 0.38 0.08
Table 4 Values of DE parameters and errors of measurement. Sample pairs
Measurement replication Neighbouring The first sample and following one a
DE a
RSD
Red
Green
Red
Green
0.33–2.44
0.23–1.30
0.38–0.72
0.34–0.52
1.4–12.8 12.5–77.7
0.9–11.2 11.3–68.6
0.06–0.29 0.01–0.09
0.04–0.18 0.01–0.07
DE value <3 means that colour of samples are very similar or the same.
Table 5 DE (colour difference) value of the analysed pair of samples in relation to concentration (cn, cm concentrations of pigment for pair of samples).
Fig. 5. Raman spectra of red paint nos. 0, 1, 6 and 10 (R – rutile) at 785 nm.
Fig. 6. Raman spectra of red paint nos. 0, 1, 6 and 10 (R – rutile) at 514 nm.
area). Parameter ‘‘b⁄’’ decreased in both sets (shift towards yellow area). Colour difference parameter (DE) defines colour similarity between two samples. It was statistically calculated (for inexperienced observer) that DE = 3 is a limit (threshold) of this similarity for both sets of paint samples DE calculated for repeated measurement was less than limit value (DE < 3). So, it could be clearly seen (Table 4) that the analysed samples are homogenous. There is not simple relation between colour coordinates and concentration of colourants. Generally colour coordinates depend
Pair of samples (cn cm)
DE Red
Green
1 and 2 (10.09) 2 and 3 (4.95) 3 and 4 (2.47) 4 and 5 (1.23) 5 and 6 (0.62) 6 and 7 (0.31) 7 and 8 (0.16) 8 and 9 (0.07) 9 and 10 (0.04) 10 and 11 (0.02) 11 and 12 (0.01) 12 and 13 (0.005)
12.62 13.00 11.78 11.01 10.56 7.85 3.98 3.98 3.58 2.70 1.27 1.47
11.32 10.56 10.89 10.19 8.40 6.47 5.22 4.16 2.87 2.31 1.10 0.94
Pair of samples (cn cm)
DE Red
Green
1 1 1 1 1 1 1 1 1 1 1 1
12.62 25.15 35.96 46.29 56.50 63.45 66.82 70.71 74.07 76.54 76.83 77.76
11.32 21.66 30.93 40.91 48.11 53.69 58.70 62.51 65.12 67.30 67.99 68.62
and and and and and and and and and and and and
2 (10.09) 3 (15.04) 4 (17.51) 5 (18.74) 6 (19.36) 7 (19.67) 8 (19.83) 9 (19.90) 10 (19.94) 11 (19.96) 12 (19.97) 13 (19.98)
on wavelength of incident radiation and intensity of spectrum is function of compound concentration. Human experience shows that the grater colour difference is seen the grater difference of concentration of colourants ought to be. So, it was expected that colour difference between neighbouring samples (constant concentration ratio) will be smaller than this between the first samples and the following ones (decreasing concentration ratio). This supposition turn out to be right (Table 5). Colour difference parameter (DE) for the neighbouring samples could be described by linear decreasing function (Fig. 7). It means, the smaller is difference of pigment concentration for analysed samples the smaller is difference in colour. Also it turn out that colour difference for the first sample and the following ones could be described by logarithmic function (Fig. 8). It means, the grater is difference of pigment concentration for analysed samples the grater is difference in colour.
´ ska et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 130 (2014) 534–538 B. Trzcin
538
Fig. 7. DE value of compared neighbouring red paint samples.
colour comparison. For both sets of paint samples having in their composition different pigment (red or green), the obtained DE value (colour difference parameter) of repeated measurements pointed out on identity of colour on whole surface of sample (DE < 3). In that way the homogeneity of prepared samples was confirmed and comparison of samples colour was possible. Homogeneity of commercial paint is certainly better than paint samples prepared in laboratory. So, colour analysis of real samples coming from place of accident and suspected car is possible. If colour of these samples is similar (DE < 3) they could come from one source. The possibility of colour pigment identification basing on Raman spectra decreased significantly when pigment concentration is less than 0.08% of weight. Presence of white pigment (rutile) hinder identification of colour pigment. When colour pigment concentration is above 1% of weight only bands attributed to it are seen on Raman spectrum. References
Fig. 8. DE value of first and the following green paint samples.
Averaged ranges of DE and the relative standard deviation (RSD) for analysed types of samples pairs are given in Table 4. Obtained results will be useful for analysis of real paint samples which are often object of criminalistic cases. Conclusions It could be concluded that homogeneity (pigment distribution in binder) of paint samples is not a factor, which makes difficult
[1] A. Beveridge, T. Fung, D. MacDougall, in: B. Caddy (Ed.), Forensic Examination of Glass and Paints, Taylor & Francis, London and New York, 2001, pp. 183–241. [2] S.G. Roland, T.A. Jergovich, K.P. Kirkbride, Forensic Sci. Rev. 18 (2006) 98–117. [3] F. Govaert, M. Bernard, Forensic Sci. Int. 140 (2004) 61–70. [4] ASTM International, E1610-02 standard guide for forensic paint analysis and comparison, ASTM International, West Conhohoken, PA, 2005. [5] J. Zie˛ba-Palus, in: R.A. Meyers (Ed.), Encyclopedia of Analytical Chemistry: Instrumentation and Applications, J. Wiley, Chichester, 2000, pp. 4484–4497. [6] J. de Gelder, P. Vandenabeele, F. Govaert, L. Moen, J. Raman Spectroscopy 36 (2005) 1059–1067. [7] E.G. Bartick, P. Buzzini, in: R.A. Meyers (Ed.), Encyclopedia of Analytical Chemistry, John Wiley & Sons Ltd, U.K., 2009. [8] J. Zie˛ba-Palus, A. Michalska, A. Wesełucha-Birczyn´ska, J. Mol. Struct. 993 (2011) 134–141. [9] A. Colombini, D. Kaifas, e-PS 7 (2010) 14–21. [10] M.C. Taylor, D.R. Cousins, R.H. Holding, J. Locke, J.M. Wilkinson, Forensic Sci. Int. 40 (1989) 131–141. [11] B. Trzcin´ska, J. Zie˛ba-Palus, P. Kos´cielniak, J. Mol. Struct. 924–926 (2009) 393– 399. [12] P. Martin, in: J.A. Siegel, P.J. Saukko (Eds.), Encyclopedia of Forensic Sciences, second ed., Waltham, USA, 2013, pp. 148–154. [13] N.A. Fuller, Forensic Sci. Int. 27 (1985) 189–204. [14] W. Stoeklein, H. Fujiwara, Sci. Justice 39 (1999) 188–195. [15] G. Massonnet, W. Stoecklein, Sci. Justice 39 (1999) 128–134. [16] J. Zie˛ba-Palus, B. Trzcin´ska, P. Kos´cielniak, Anal. Lett. 43 (3) (2010) 436–445. [17] M.E. Nadal, E.A. Early, Colour Res. Appl. 29 (1) (2004) 38–42. [18] K.A. Kopchick, C.R. Bommarito, J. Forensic Sci. 51 (2006) 340–343. [19] A. Takagi, S. Sato, G. Baba, Colour Res. Appl. 32 (2007) 378–387. [20] A. Takagi, A. Watanabe, G. Baba, Colour Res. Appl. 30 (2005) 275–282.