ANALYTICAL SCIENCES APRIL 1998, VOL. 14 269 1998 © The Society for Analytical Chemistry

Analysis of Writing Dyestuffs by TLC and FT-IR and Its Application to Forensic Science

Kazuhiro TSUTSUMI* and Kazuya OHGA**

*Criminal Investigation Laboratory, Oita Prefecture Police Headquarters, Otemachi, Oita 870, Japan **Department of Applied Chemistry, Faculty of Engineering, Oita University, Dannoharu, Oita 870Ð11, Japan

Black, blue and red writing were classified into various groups using the Rf values and color tones of dyestuff bands separated by normal-phase thin-layer chromatography (TLC) of their . The classification is effective for the prelimi- nary identification of pens used in crime scenes. A microsampling technique was proposed for the TLC analysis of minute quantities of inks on questioned documents. Furthermore, a combination of reflectance-mode microscope/Fourier transform infrared spectroscopy and the pin-point condensation technique was proved to be useful for the precise discrim- ination of trace amounts of analogous water-soluble dyestuffs on TLC plates.

Keywords Forensic science, ink, dyestuff, thin-layer chromatography, Fourier transform infrared spectroscopy

Various writing implements have often been used at of them were kindly supplied by their manufacturers. crime scenes in Japan. One of the purposes for the Reference dyestuffs were obtained from manufacturers examination of writing ink is to specify writing imple- and the Identification Reference Data Center, National ments. A number of papers have been published Police Agency, , Japan. Water was purified by regarding techniques for forensic examinations, includ- reverse osmosis (Millipore Milli RO15) and deioniza- ing visible spectrophotometry1,2, thin-layer chromatog- tion. Developing organic solvents for TLC were of raphy (TLC)1Ð4, X-ray microanalysis1, microspec- reagent grade, and were distilled before use. All other trophotometry3,5, high-performance liquid chromatogra- chemicals were also of reagent grade and were used phy (HPLC)6, gel electrophoresis7 and capillary zone electrophoresis.8,9 TLC is the simplest of those meth- ods, and is effective for separating dyestuff compo- Table 1 Number of pens used in this work and the number of nents. A systematic TLC study, however, has been pur- groups into which the pens were classified by the present sued only slightly concerning the analysis of Japanese TLC analysis writing inks. In the present paper we describe a prelim- Mnemonic Number inary preparation of a standard TLC library for inks, Number and also discuss the effectiveness and limitation of symbol of group TLC for the identification of writing implements. We Oil-based also report on a microsampling technique, that is often ball-point pen requested in the analysis of inks on questioned docu- Black BP-Bk 16 9 ments, and the application of a pin-point condensation Red BP-R 10 8 technique10 to the identification of dye TLC bands by Blue BP-Bu 9 5 microscope/Fourier transform infrared spectroscopy marking pen (FT-IR). Black OM-Bk 18 11 Red OM-R 16 10 Blue OM-Bu 16 11 Aqueous Experimental roller-ball pen Black WB-Bk 6 3 Pens and reagents Red WB-R 6 5 In this work were used 161 kinds of black, red and Blue WB-Bu 6 5 blue pens, which were classified for convenience into marking pen four groups, as shown in Table 1: oil-based ball-point Black WM-Bk 20 16 and marking pens as well as aqueous roller-ball and Red WM-R 20 14 marking pens. They are now on the market, and some Blue WM-Bu 18 11 270 ANALYTICAL SCIENCES APRIL 1998, VOL. 14 without further purification. was dried up on a 0.1-mm hydrophobic perfluorinated- polymer film, with which a stainless-steel mirror was TLC coated. This type of drying procedure has been named TLC was carried out using a Merck precoated silica- pin-point condensation by Ikeda and Uchihara.10 The gel 60 F254 backed with an aluminum sheet and a RP-18 residue was subjected to microscope/FT-IR in the F254s backed with a glass sheet. Chloroform solutions reflectance mode. The analysis was performed while were applied to the TLC plates for the analyses of oil- receiving the aid of a Spectra-Tech IR-PLAN based ball-point pen inks and oil-soluble dyestuffs. microspectrometer equipped with a mercury-cadmium- Aqueous solutions were used for water-soluble telluride detector. The resolution was 4 cmÐ1 and the dyestuffs and the other inks were directly spotted with scanning was performed 256 times. The spectra of ref- their refills. All of the spots were 0.5 Ð 0.8 mm in erence samples were taken without the attachment in diameter, and the amounts of the inks and reference the transmission mode using the KBr tablet. dyestuffs applied were 1.0 Ð 1.5 mg. The origins were at 1.0 cm from the bases of the plates. The developing HPLC solvents used were ethyl acetate/ethanol/water (14:7:6), HPLC was performed with a Hitachi L-6000 ethyl acetate/methanol/28% aqueous ammonia (5:2:1) equipped with an Ohtsuka Denshi MCPD-3600 photo- and trichloroethylene/1,1,1-trichloroethane/ethyl diode array detector. The column was a Kagakuhin- acetate (10:1:1) for the normal-phase mode chromatog- Kensa Kyokai L-column ODS (4.6mmf´250 mm) and raphy, and acetonitrile/3% aqueous KBr (7:1) for the the eluent was acetonitrile/aqueous 3% KBr (7:1). reversed-phase mode; they were allowed to creep up to the plates a distance of 6.0 cm. The uniform origin position on the plate, proper loading at the origin and Results and Discussion the solvent saturation in the developing tank, which had been recommended by Lewis4, were kept during all of TLC the TLC analyses in order to obtain a high reproducibil- There are many reference books13 and reports1Ð4 on ity. the TLC of inks and dyes, in which proper developing solvents have been noted, principally for normal-phase Microsampling silica-gel plates. As a result of a thorough examination A procedure devised by Golding and Kokot11,12 for made for most of those solvents with the Merck normal dyestuff extraction from a filament was modified in phase plate, the one reported by Brunelle and Pro2, order to accommodate to an imitative microsampling ethyl acetate/ethanol/water (14:7:6, abbreviated as the experiment in this work. A fraction of a line (ca. 3 mm B.P. solvent hereafter), was found to provide high long) drawn on filter paper (Toyo Roshi #2) was put in degrees of dye separation to most of the present inks. a haematocrit glass capillary (1 mmf´75 mm) together The results obtained for oil-based and aqueous red with 5 mm3 of a given solvent. The capillary was marking pens with the B.P. solvent are typically sum- sealed, and then sonicated in a Yamato 1210 ultrasonic marized in Tables 2 and 3, respectively. The relative bath at 60ûC for 5 to 10 min. The use of filter paper standard deviations for the Rf values were below 3.1% was grounded on the fact that suspicious letters are usu- for 5-times repeated analyses of the same samples. ally written on paper made from wood pulp. An effec- Frequently observed overlapping of dyestuff bands tive extraction solvent was searched by paper chro- was confirmed through their separation with ethyl matography according to the criteria that it should leave acetate/methanol/28% aqueous ammonia (5:2:1) or a no colors due to the dyestuffs at the starting point, give less polar solvent, trichloroethylene/1,1,1-trichloro- large Rf values to the dyestuffs and have a high volatili- ethane/ethyl acetate (10:1:1). Less-resolved bands hav- ty. The solvents examined were methanol, ethanol, 2- ing high Rfs were separated into more bands with the propanol, aqueous ammonia, acetic acid, pyridine, less-polar solvent; for example, the R1 dye band having DMF, THF, acetone, chloroform, acetonitrile, toluene, Rf of 0.96 in Table 2 yielded four bands having Rfs of ethyl acetate and a mixture of two of these. On the nor- 0.53, 0.47, 0.39 and 0.31. Also, the use of aqueous 3% mal-phase TLC plate was applied a dot of an extract in KBr instead of the water appreciably improved the tail- the above-mentioned capillary tube, whose tip had been ing bands, such as the Y24 and Bw4 bands in Table 3, previously stretched with a small flame and cut off. which were probably due to ionic dyes. This modified solvent system only slightly affected the Rfs and color FT-IR tones of the other bands. FT-IR spectra were obtained with a Nicolet System 710 spectrometer. Among the dyestuff bands obtained Classification from aqueous pen inks through the above-mentioned The Rf values and color tones of the bands separated microsampling/TLC procedure, the desired one was by the normal-phase TLC analysis using the B.P. sol- scraped up from the normal-phase TLC plate onto a fil- vent permitted us to classify the writing pens into vari- ter of 0.45 mm pore size (Millipore Sumplep HV4), and ous groups, as listed in Table 1. Thus, standard TLC then extracted with ca. 40 mm3 of water. The extract tables, such as Tables 2 and 3, appear to be available ANALYTICAL SCIENCES APRIL 1998, VOL. 14 271

Table 2 TLC of oil-based red marking pens

Dyestuff banda Marking pene

b c Number Color Rf 1 2 3 4 5 6 7 8 9 10111213141516

Y1 yellow 0.96 R1 yellowish red 0.96 R2 red 0.91 R3 red 0.91 Y3 yellow 0.89 Y4 yellowF 0.89 Y9 orangeT (0.79)d R9 yellowish red 0.68 R12 yellowish red 0.64 Y15 yellow 0.62 R15 yellowish redF 0.61 R21 yellowish redF 0.46 R22 pinkF 0.46 Y24 yellowT (0.11)d a. Obtained on a Merck silica-gel 60 F254 plate using ethyl acetate/ethanol/water (14:7:6) as the developing solvent. b. Serial numbers assigned to dyestuff bands in each color series in order of magnitude in Rf. Yellow and red series are abbreviated as Y and R. For typi- cal color series, see Tables 4 and 5. c. Colors observed on the TLC plates. Superscripts F and T represent fluorescent and tailing bands, respectively. d. Rf of the tailing band obtained with ethyl acetate/ethanol/aqueous 3% KBr (14:7:6). e. Pens are, in numerical order, N50, Mitsubishi A-50, PIN-10 and PA-121T, Sakura PK, PGK and YK, M-10EF, OS-FF1, Konishi 400- 0009 and 400-0709, Lion 241-30, 150-MC-R, Kokuyo PM-41, and Shachihata POM-2A and K-36T. Marks and represent dark and light colors, respectively, and blank spaces express the lack of bands.

Table 3 TLC of aqueous red marking pens

Dyestuff banda Marking pene

b c Number Color Rf 1234567891011121314151617181920

Y7 reddish yellow 0.85 R5 yellowish redF 0.77 R7 purplish red 0.71 R8 yellowish redF 0.69 Y13 yellow 0.69 R15 yellowish redF 0.61 Y17 orangeF 0.58 R16 purplish redF 0.58 Y19 yellow 0.52 R20 yellowish red 0.50 Y22 orange 0.47 Y23 yellow 0.46 R21 yellowish redF 0.46 R25 pink 0.26 Bw4 brownT (0.19)d Y24 yellowT (0.11)d a, b, c and d. See the footnotes a, b, c and d in Table 2, respectively. Bw is the abbreviation of the brown series. e. Pens are, in numeri- cal order, Pentel S520, Mitsubishi OFD-20BOXY and MyT-7, Sakura AK and WK, Pilot S-10PP, Tombow WS-100FA, WP-FN and WM-480P, Konishi 400-0339, Lion P-55, Zebra MWS-101-R, MWS-102-R and MWS-100F-R, Kokuyo PM-3, Shachihata PM-20B, K-803 and K-210, and Platinum SPM-150N and FE-200-3. Marks and represent dark and light colors, respectively, and blank spaces express the lack of bands. for a preliminary examination of pens used in crime marking pens listed in Table 2. It should be noted here scenes. The incomplete pen identification was due to that indistinguishable pens were not differentiated, even the presence of pens whose chromatograms were the if they were analyzed with different developing sol- same as one another in the separation pattern: for vents and/or in reversed phase mode. A complete iden- example, #1, #5, #10 and #12 of the oil-based red tification thus requires an additional analysis, such as 272 ANALYTICAL SCIENCES APRIL 1998, VOL. 14

Table 4 Red TLC band series of ink dyestuffs quantitative HPLC, which is mentioned later. The Red band Pen occurrence of the same separation patterns is possibly attributed to the situation that plural products from one a b c d Number Color Rf Type Manufacturer manufacturer contain common dyestuff components, and ink compositions depend on the limited raw R1 yellowish red 0.96 OM-R Pe, Sa, Uc, Li dyestuffs provided by dyestuff makers, which are R2 red 0.91 OM-R Pe, Sa, Uc, Li, Ko smaller in number compared to writing-pen makers. R3 red 0.91 OM-R Mi R4 red 0.90 OM-Bk Pe, Mi, Sa, Pi, To, All of the dyestuff bands were divided into seven Uc, Li, Ze, Sh color band categories: 26 yellow, 25 red, 24 blue, 10 R5 yellowish redF 0.77 WB-R Pe, Mi, Sa, Pi, Ze, Ko purple, 2 green, 4 black and 4 brown bands. The red WM-R Pe, Mi, Sa, Pi, To, and purple series are typically shown in Tables 4 and 5, Uc, Li, Ze, Sh, Pl respectively, where the dyestuff bands are numbered R6 pink 0.71 OM-Bu Uc serially and the types of pens are affixed together with R7 purplish red 0.71 WM-R Mi the manufacturers. These color series tables are useful R8 yellowish redF 0.69 WB-Bk Ko for narrowing questioned inks down to limited candi- WB-R Pe, Mi, Sa, Pi, Ze, Ko dates. In addition, all of the chromatograms were WM-Bk Pl stored in the form of digital color images in library WM-R Pe, Mi, Sa, Pi, To, Uc, Li, Ze, Sh, Pl disks for a future visual collation with those from ques- R9 yellowish red 0.68 OM-R Sa tioned documents. R10 red 0.65 WM-Bk Mi R11 pinkF 0.64 BP-R Pi, Sh, Pl Identification of dyestuffs R12 yellowish red 0.64 OM-R Sa Some dyestuff components were identified by TLC R13 yellowish redF 0.63 BP-R To analyses using the four developing solvents in the two R14 purplish red 0.61 WM-Bk To, Sh modes described in the Experimental section. The R2 R15 yellowish redF 0.61 BP-R Pe, Mi, Pi, Sh, Oh, Pl, band in Table 2 was due to Color Index (C.I.) Solvent To, Ze, Ko Red 18. Nigrosine was a major component of most oil- OM-R Mi, Sa, Pi, To, Uc, based black felt-tip marking pens and a Tombow BC- Ze, Ko, Sh JN black ball-point pen. Most oil-based red ball-point WM-R Ko R16 purplish redF 0.58 BP-R Mi, Oh pens contained R15 and R22 (Table 4); the R22 WB-R Pe, Mi, Sa, Pi, Ko dyestuff was Rhodamine B Base. A yellow band of Rf WM-R Pe, Mi, Sa, Pi, To, Sh, 0.41 (Rfs mentioned in this section are the ones ob- Pl tained in the normal-phase mode with the B.P. solvent) WM-Bu Sh for a Kokuyo PR-P2 red ball-point pen was due to R17 red 0.57 OM-Bk Sa, Li Tartrazine. A blue band dyestuff of Rf 0.60 was C.I. R18 yellowish red 0.56 WB-R Ze Solvent Blue 5, which was a major dyestuff of many R19 red 0.52 WB-Bk Pe, Mi, Sa, Pi, Ze oil-based blue pens. The P2, P5 and P6 bands given in WM-Bk Uc, Ze Table 5 were due to Tetramethyl-p-rhosaniline, Methyl R20 yellowish red 0.50 WM-R Ze, Sh Violet and Crystal Violet, respectively, major dyestuff R21 yellowish redF 0.46 BP-R Ze OM-R Mi, Sa, Pi, To, Ze, Sh components of the black ball-point pens, except for the WM-R Ko above-mentioned Tombow pen. Some of the black R22 pinkF 0.46 BP-R Pe, Mi, Pi, To, Ko, pens contained C.I. Acid Yellow 42 or Metanil Yellow. Sh, Oh, Pl It should be emphasized here that a determination of BP-Bu Pe the content ratios of the above three purple-series dyes OM-R Ko (P2, P5 and P6) by HPLC (not shown) permitted us to R23 purplish redF 0.44 WB-R Pi identify all of the black ball-point and aqueous marking WB-Bu Mi pens. WM-Bk To, Mi Some of the aqueous black pens contained deeply WM-Bu Pi, To green C.I. Direct Black 154 having a Rf of 0.54. The R24 red 0.40 WM-R Li, Pl fluorescent R8 and R16 bands given in Table 4 were R25 pink 0.26 WM-R Ze, Sh due to Eosin and Phloxine B, respectively. These were a. Red bands abbreviated as R are numbered in order of magni- major dyestuffs of most aqueous red pens, which tude in the Rf. enabled us to specify the red pens through a determina- b. See the footnote c in Table 2. tion of their content ratios by HPLC. A major dye of c. See Table 1. many aqueous blue pens was Brilliant Blue-FCF of Rf d. Pentel, Sakura, Mitsubishi, Pilot, Tombow, Lion, Zebra, 0.37. Kokuyo, Shachihata, Ohta, Platinum, Saler and Konishi are abbre- viated as Pe, Sa, Mi, Pi, To, Li, Ze, Ko, Sh, Oh, Pl, Sl and Uc, respectively. Application of microsampling The microsampling depends primarily on the selec- tion of effective extraction solvents. Among the extrac- ANALYTICAL SCIENCES APRIL 1998, VOL. 14 273

Table 5 Purple TLC band series of ink dyestuffs

Purple band Pen

a b c d Number Color Rf Type Manufacturer

P1 reddish purple 0.81 BP-Bu Pe P2 reddish purple 0.54 BP-Bk Pe, Mi, Sa, Pi, Li, Ze, Ps, Ko, Sh, Oh, Sl, Pl BP-Bu Mi, Pi, To, Ko, Oh WM-Bk Sa, Ko, Sh P3 violet 0.53 BP-Bu Pe, Ze WB-Bk Ko P4 purple 0.51 WM-Bk Pe, Li, Ma, Pl P5 purple 0.50 BP-Bk Pe, Mi, Sa, Pi, Li, Ze, Pu, Ko, Sh, Oh, Sl, Pl, Bc BP-Bu Mi, Pi, To, Ko, Oh OM-Bu Ko WM-Bk Sh P6 violet 0.47 BP-Bk Pe, Mi, Sa, Pi, Li, Ze, Pu, Ko, Sh, Oh, Sl, Pl, Bc BP-Bu Mi, Pi, To, Ko, Oh OM-Bu Ko WM-Bk Sa, Ko, Sh WM-Bu Pe P7 violet 0.47 WM-Bu Ze P8 violet 0.43 WB-Bu Pe, Ko WM-Bk Pe, Pi, To, Li, Ma, Pl WM-Bu Mi, Sa, Pi, Li, Ze, Sh, Pl P9 violet 0.39 WB-Bk Ko P10 violetT 0 Ð 0.2 WM-Bk Sh

a. Purple bands abbreviated as P are numbered in order of magnitude in the Rf. b. See the footnote c in Table 2. c. See Table 1. d. See the footnote d in Table 4. Ma, Pu and Bc in the Manufacturer is the abbreviations of Marby, Plus and Bic. tion solvents examined, DMF gave the largest paper- guished from each other by a comparison of their color chromatographic Rf values to aqueous ink dyestuffs, but tones. The P8 micro-reflectance spectrum was in was hard to evaporate. Pyridine was judged to be the agreement with the reference spectrum, being distin- most favorable for the extraction of all oil-based inks, guishable from the P6 reference spectrum. This finding while it was unsuitable for some aqueous inks. In such unfavorable instances, a mixture of ethanol and 28% aqueous ammonia (1:1) was adopted as an alternate sol- vent, because the mixture gave comparatively good results to them. Two pens were selected at random from each of the 12 types (Table 1), and their inks were extracted with the corresponding suitable solvents. The extracts were subjected to TLC analysis according to the procedure described in the Experimental section. The chro- matograms obtained with the B.P. solvent were almost the same as those in the standard library, proving valid- ity of the present microsampling procedure.

Discrimination of dyestuffs by FT-IR Figure 1C typically shows the reflectance spectrum of a trace of a P8 dyestuff (Table 5) isolated from a 3-mm line of a Pilot aqueous black marking pen. Included in Fig. 1 for a comparison are the transmission spectra of references P6 (A) and P8 (B) dyestuffs, which were isolated through TLC of aqueous black marking pens with the ethyl acetate/methanol/28% aqueous ammonia developing solvent. The P6 and P8 present in many Fig. 1 Transmission FT-IR spectra of purple-series dyestuffs black and blue pens had Rf values close to each other, P6 (A) and P8 (B), and the micro-reflectance spectrum of the as can be seen in Table 5, and also were not distin- P8 (C). 274 ANALYTICAL SCIENCES APRIL 1998, VOL. 14 suggests that the present combination of reflectance- 6. I. R. Tebbett, C. Chen, M. Fitzgerald and L. Olson, J. mode microscope/FT-IR and pin-point technique is Forensic Sci., 37, 1149 (1992). effective for discriminating extremely small amounts of 7. H. W. Moon, J. Forensic Sci., 25, 146 (1980). dyestuffs. 8. K. Tsutsumi and K. Ohga, Anal. Sci., 12, 997 (1996). 9. S. Fanali and M. Schudel, J. Forensic Sci., 36, 1192 (1991). 10. M. Ikeda and H. Uchihara, Polyfile, 29, 32 (1992). 11. G. M. Golding and S. Kokot, J. Forensic Sci., 35, 1310 References (1990). 12. S. Suzuki, Y. Higashikawa, T. Kishi and Y. Marumo, 1. H. Harada, J. Forensic Sci. Soc., 28, 167 (1988). Reports of NIPS, 44, 50 (1990). 2. R. L. Brunelle and M. J. Pro, J. Assoc. Off. Anal. Chem., 13. For example, “Eisei Shikenhou/Chukai (Standard Methods 55, 823 (1972). of Analysis for Hygienic Chemists/Commentary, in 3. R. N. Totty, M. R. Ordidge and L. J. Onion, Forensic Sci. Japanese)”, ed. The Pharmaceutical Society of Japan, p. Int., 28, 137 (1985). 530, Kaneharashuppan, Tokyo, 1990. 4. J. A. Lewis, J. Forensic Sci., 41, 874 (1996). 5. A. Zeichner and B. Glattstein, J. Forensic Sci., 37, 738 (Received September 1, 1997) (1992). (Accepted October 29, 1997)