July 2005 - Volume 7 - Number 3

Research and Technology

Forensic Analysis of Black Ballpoint Pen Inks Using Capillary Electrophoresis

James M. Egan
Research Chemist
Federal Bureau of Investigation
Laboratory
Counterterrorism and Forensic Science Research Unit
Quantico, Virginia

Kristin A. Hagan
Oak Ridge Institute for Science and Education Visiting Scientist
Federal Bureau of Investigation
Laboratory
Counterterrorism and Forensic Science Research Unit
Quantico, Virginia

Jason D. Brewer
Chemist
Federal Bureau of Investigation
Laboratory
Chemistry Unit
Quantico, Virginia

Abstract

Capillary electrophoresis with ultraviolet-visible photodiode array detection (190–600 nm) was studied as an alternative separation and identification tool for forensic ink examination. Two different buffer systems were designed to analyze dye compounds in various black ballpoint pen ink formulations. Results were compared to thin-layer chromatography experiments to evaluate the sensitivity and performance of capillary electrophesis. A database of ballpoint pen ink analyses and common-dye reference standards has been constructed for future forensic use. Capillary electrophoresis allows ease of sample preparation with the ability to separate and identify dye compounds based on a calculated electrophoretic mobility and a characteristic ultraviolet-visible spectrum. Protocols for capillary electrophoresis sample preparation were designed to closely mimic procedures already in place for common ink-evidence analysis. Because of the small volume necessary for analysis, the remaining solution could be further processed using current law enforcement procedures for confirmation.

Introduction

Determining ink sources used on a variety of documents is a key priority for forensic document examiners. The ability to distinguish different inks can be quite useful for several reasons. Document alteration (e.g., at a date later than indicated) by writing with a pen of similar color but different dye composition is one specific example when ink differentiation is crucial in criminal cases. Ink comparison also can determine the relationship between two samples in a forgery case that involves an original and a copy or two documents believed to have the same author. Sample authentication also can be tested based on ink analysis of raw colorant materials available during a specific historical period. To unambiguously identify specific inks, data interpretation is simplified if a chemical-separation step precedes detection of various components in ink formulations.

Inks are complex mixtures of colorants, vehicles, and additives, which are adjusted in composition to produce the desired writing characteristics (Industrial Dyes: Chemistry, Property, Applications 2003). Colorants are compounds that give ink the desired color and can include any or all of the following chemical classifications: pigments and/or acidic, basic, azoic, direct, disperse, reactive, and solvent dyes. Colorants are often the focus of ink analysis because of their light-absorption and emission properties that can be detected by various analytical methods. Vehicles or carriers are usually solvents that allow the ink to flow and carry the colorants to the material surface. Solvents are the typical ingredients analyzed in date-of-origin investigations because of their gradual evaporation from a document (Aginsky 1994; Brunelle 1992; Brunelle and Cantu 1987; Brunelle et al. 1987; Cantu 1996; Cantu 1991; Cantu 1988). Last, additives can serve as flow (viscosity) modifiers, surface activators, corrosion controllers, solubility enhancers, and preservatives (Brunelle and Reed 1984; Leach and Pierce 1993). Detection of these additive compounds can greatly aid forensic examiners because the compounds can be manufacturer-specific. Their identity is often a highly guarded secret in ink formulations, as are the colorants themselves.

Typical forensic techniques on questioned documents involve ink extractions followed by thin-layer chromatography analysis (Standard Guide for Test Methods for Forensic Writing Ink Comparison 1996; Tebbett 1991). Knowledge gained from this procedure is limited to colored spots that are correlated to a calculated retardation factor (R_f). Densitometry also can be used to probe the concentration of dye on a specific area of the thin-layer chromatography slide in reflectance mode (Aginsky 1994). Carefully prepared standards are required for quantitative analysis of an unknown sample. Problems with thin-layer chromatography reproducibility can be related to the difficulty in spotting uniform samples and maintaining constant environmental influences, which lead to changes in R_f values. Limited sensitivity, sample destruction, and extra processing required for additional information are also shortcomings of the method. Evidence handling requires that samples processed by thin-layer chromatography be placed under environmentally controlled storage conditions for future reference to prevent against color fading.

Other analytical techniques, such as infrared, mass spectrometry, gas chromatography, and high-performance liquid chromatography, have been used to study different ink components for identification purposes. All of the techniques differ in the sample processing (destructive versus nondestructive) and information obtained from ink analysis. Some of these procedures are nondestructive (Harris 1991; Trzcinska 1993; Varlaskin and Low 1986), whereas others are more destructive than thin-layer chromatography extractions because they require that more samples be physically removed from the document (Merrill and Bartick 1992; Trzcinska 1990). Often, only qualitative information is obtained from these techniques, relying on pattern-recognition methods to differentiate inks. Infrared spectroscopy has received attention as an analytical tool to elicit spectral information from written ballpoint pen ink. A limited sample set of available inks allows infrared patterns to be distinguished, but as more pens are added to a database, statistical analysis methods will become less useful. Laser-induced infrared luminescence has been investigated as another method to quickly distinguish two inks of the same color (Horton and Nelson 1991). However, the method can determine only if components luminesce, and the analysis was again limited to a small ink population. Microspectrophotometry provides information about the ink’s acting as a chemical mixture, but still does not provide the necessary detail that is important in component analysis for ink differentiation (Zeichner et al. 1988). Laser desorption/ionization mass spectrometry is extremely useful in detecting particular dyes in inks, but prior knowledge of the ink components is useful to aid interpretation of the complex mass spectrum that is acquired with ink mixtures (Grim and Allison 2003; Grim et al. 2002). One other useful mass spectrometric technique involves positive- and negative-ion mode electrospray-ionization to create ions through direct sample infusion to aid in compound identification (Ng et al. 2002). High-performance liquid chromatography methods also have been employed to separate different dye compounds with some success, although the technique has afforded little understanding of different inks other than the absence or presence of unassigned chromatographic peaks (Lyter 1982). More thorough analysis is required to provide unambiguous ink identification based on standard reference comparisons.

Capillary electrophoresis was chosen over high-performance liquid chromatography for the front-end separation technique because small sample volumes and on-capillary sample preconcentration are possible. Capillary electrophoresis is an excellent candidate to separate charged- and neutral-dye compounds for detection with ultraviolet-visible absorption spectroscopy. Two characteristic parameters can be obtained simultaneously: mobility factor and ultraviolet-visible spectrum (a photodiode array detector must be used). There have been numerous studies on inks performed on specific compound classifications or ink formulations (e.g., ballpoint, nonballpoint, fountain pens) (Burkinshaw et al. 1993; Croft and Lewis 1992; Fanali and Schudel 1991; Jandera et al. 1996; Rhode et al. 1998; Rhode et al. 1997; Vogt et al. 1999; Vogt et al. 1997; Zlotnick and Smith 1998). Most of the capillary electrophoresis investigations focus on a qualitative approach to distinguish among different brands of pens. Qualitative electropherogram analysis in a specific ink formulation does not allow identification of compounds resulting from an observed peak. Problems with this approach are the limited sampling taken from the general ink population and the high probability that many formulations are too similar to be differentiated unless specific information is known prior to analysis. A reason only qualitative information was investigated was the collection of single-wavelength data. A more technical approach to ink separation was performed with ultraviolet-visible spectral collection on reference standards, which resulted in a separation of many different dye compounds in a single experiment (Xu et al. 1997). However, implementation of this particular system into a casework scenario would be extremely difficult because of the demanding requirements for preparation of sample and buffer solutions.

To devise a capillary electrophoresis protocol that is compatible with casework, this investigation focused on the following procedural requirements:

• Minimal sample preparation comparable to thin-layer chromatography extractions.

• A faster method that would collect more useful data with respect to thin-layer chromatography.

• Ease of implementation without using excessive experimental measures.
• The ability to adapt the technique when extra knowledge is gained.

To this end, one buffer was designed to adequately separate cationic dyes (general class of basic dyes), and another buffer was designed to separate anionic dyes (general class of acidic dyes). These buffers were successful in investigating a number of black ballpoint pen inks and identified specific dye compounds by comparison of electrophoretic mobility (μ_ep) and absorbance spectra to standard references. Libraries are now being compiled that consist of dye reference standards and extracted ink samples with electropherograms and absorbance spectra. These libraries will be used to compare and characterize unknown samples in the future.

Experimental Samples

Various dye compounds and black ballpoint pen inks were investigated to survey possible samples encountered by forensic ink examiners. Barnstead (Dubuque, Iowa) NANOpure Infinity ultrapure water (ρ ≥ 18 MΩ-cm) was used for the preparation of all solutions. All dyes were initially prepared in methanol at 1.0 mg/mL concentrations. Dye injection solutions were prepared based on the capillary electrophoresis method performed and will be further explained in the capillary electrophoresis experimental section below. Table 1 contains all dye compounds investigated in this study, and Figure 1 provides some known dye molecular structures. This is not an exhaustive list of dyes used in ballpoint pen ink formulations, but it is a starting point for constructing a database of dye standards. Most dyes were chosen for their presence in multiple ink formulations and will be validated by the ink analysis performed on ballpoint pen samples. The first 11 dyes listed in Table 1 correspond to single-dye compounds, and the remaining compounds may correspond to single-dye molecules or a combination of dye molecules. These specific industrial samples were acquired from the U.S. Secret Service and were studied to understand the complexity of different dye manufacturers’ products.

Table 1: Dye Reference Compounds Investigated for Capillary Electrophoresis, Matrix-Assisted Laser Desorption/Ionization-Mass Spectrometry, and Thin-Layer Chromatography Analysis

Figure 1: Structure of Single-Dye Reference Compounds Purchased from Sigma-Aldrich

Ballpoint pen inks are not referenced to either a make or manufacturer because of the sensitive nature of ink formulations. A simple nomenclature is used instead to differentiate various investigated inks. Samples were obtained from a stockpile of pens located at the FBI Laboratory or acquired from the U.S. Secret Service ink library. Ink samples were prepared differently for each analytical method to study the components. Pen lines were drawn on Whatman (Ann Arbor, Michigan) filter paper for extraction with methanol after various drying times. Drawn pen lines were typically used for mock casework samples that are presented in the text. Inks were also directly removed from the ink cartridge with a capillary for laser desorption/ionization-mass spectrometry and thin-layer chromatography purposes to ensure components were not missed because of low dye concentrations. Inks were also soaked onto Whatman filter paper and placed in plastic bags for storage purposes to allow further studies. Ink-soaked filter papers provided the most concentrated sample extracts and were used to confirm that peaks were not missed or assigned improperly when the extraction process was performed.

Capillary Electrophoresis

Two different capillary electrophoresis instruments were used to perform ink and dye separations. A Beckman Coulter (Fullerton, California) P/ACE MDQ Series capillary electrophoresis system equipped with a photodiode array detector measured the basic dye components present in black ink formulations and dye standards. A Hewlett Packard/Agilent 3D (Palo Alto, California) capillary electrophesis system, also equipped with a photodiode array detector, was chosen to perform all of the acidic dye separations. Reasons for choosing two different systems to complete this study included comparing methods on different instruments and working on parallel solutions simultaneously. All buffer and separation methods could be done on either instrument under the employed conditions.

Two unique buffer systems were prepared to separate the distinct classes of basic and acidic dye compounds necessary for dye analysis of the inks. The main reason for the two buffers involves the variable interaction of two different dye classes with the silica capillary wall. Wall effects substantially dominate the separation efficiency and will be discussed further in the results section.

Cationic Dye Method

All cationic dye component analyses were performed at 30°C on a Beckman P/ACE MDQ capillary electrophoresis system. Glacial acetic acid (TraceMetal grade, 99.5 percent) and sodium acetate (high-performance liquid chromatography grade, 99.2 percent) were purchased from Fisher Scientific (Fairlawn, New Jersey). Hexadecyltrimethylammonium bromide (CTAB) (~ 99 percent) and methanol (A.C.S. high-performance liquid chromatography grade, 99.3 percent) were obtained from Sigma-Aldrich (St. Louis, Missouri). Direct-mode ultraviolet-visible absorbance spectra were obtained over the wavelength range of 200 to 600 nm at a data rate of 32 Hz. Separate data channels were employed to collect electrical current information and a representative electropherogram at 214 nm. Fused-silica-coated capillaries of 75 μm inner diameter (i.d.) were purchased from Phenomenex (Torrance, California). The separation length to the detector (L_d) was 30 cm, and the total length (L_t) of the capillary was 40 cm. Prior to each injection, the capillary was rinsed at 29.0 psi with 1 percent bleach (1.50 minutes), water (1.00 minute), 0.1 M HCl (2.00 minutes), water (1.00 minute), 0.1 M NaOH (1.50 minutes), water (1.00 minute), and run buffer (1.50 minutes) to regenerate the desired wall properties. Hydrodynamic injection was performed at 0.3 psi for five seconds unless denoted otherwise. A potential of 25 kV was applied in reverse-polarity mode, which produced an electric field of approximately –625 V/cm.

The separation buffer consisted of 70/30 (v/v) acetate buffer (25 mM sodium acetate, 25 mM glacial acetic acid, 10 mM CTAB)/methanol, which was vacuum-filtrated (0.45 μm cellulose acetate filter system), and then titrated to a pH of 4.45 with glacial acetic acid. Individual dye standard injection solutions were prepared by dissolving an appropriate amount of dye in an acetate solution consisting of 2.5 mM sodium acetate and 2.5 mM glacial acetic acid (pH = 4.5). Methanol was then added to the dye in acetate solution to prepare 70/30 (v/v) or 50/50 (v/v) acetate/methanol injection mixtures. A six-dye-mixture standard solution (Table 2) was prepared by mixing aliquots of the individual dye standards in acetate with methanol to form a 50/50 (v/v) injection solution. Inks were extracted from filter paper by removing 1.0 mm diameter punches (one to five punches as necessary) and then placing the punches in 30 μL of methanol. An ultrasonic cleaner was used to agitate the extraction solution for two minutes to assist ink removal. An aliquot of 30 μL of 2.5 mM acetate solution was then added to the extraction solution to form a 50/50 (v/v) injection solution.

Table 2: Capillary Electrophoresis Results of Individual and Mixtures of Basic, Cationic Dyes Performed on a Beckman P/ACE MDQ Capillary Electrophoresis System with the CTAB/Acetate Buffer

Anionic Dye Method

Anionic, acidic dyes were separated using a buffer composed of 25 mM CHES [2-(N-cyclohexylamino)ethanesulfonic acid] (Sigma-Aldrich) and 5 mM β-cyclodextrin (Sigma-Aldrich) and adjusted to a pH of 8.80 with 1.0 M NaOH. The effective length (L_d) of the 75 μm i.d. fused-silica capillary was 50 cm, and the total length was 58 cm (L_t). Methanol solutions of extracted and standard samples were injected in the capillary for ten seconds at a pressure of 25 mbar. Voltage was applied in the normal mode (with the outlet electrode negative with respect to the inlet electrode) at 25 kV and resulted in currents of approximately 19 μA (430 V/cm). Capillary temperature was maintained at 30°C with the Hewlett Packard Chemstation software package. All separations took place in less than ten minutes. Prior to injection, a series of rinse solutions was used to regenerate and clean the capillary for subsequent runs. These rinsing steps included a one-minute flush with 1.0 percent bleach, two-minute flush cycles with 0.1 M NaOH and water, and then a two-minute capillary fill with the separation buffer. Ultraviolet-visible spectra were collected from 190 to 600 nm, and data channels were used to monitor real-time absorbance at 214 nm and the electrical current trace. Recording the current during the experiment is a good diagnostic tool to evaluate system performance. The dye standards were analyzed individually at 100 μg/mL concentrations in methanol. Two different black pen ink analyses were performed for dye detection and identification purposes. A single 1 mm punch was removed from ink-soaked Whatman filter paper and extracted with 30 μL of methanol. The second experimental test involved a 1 cm drawn line and was completely extracted with 30 μL of methanol to simulate casework examples.

To demonstrate that this technique was amenable to a casework scenario when destruction of evidence has to be minimal, a casework protocol was proposed, and samples were analyzed with all of the procedures mentioned above. To begin, five 1.0 mm hole punches were removed from a piece of printer paper with written ink lines that were dried for one month. An additional five 1.0 mm hole punches, which were blank and spatially located close to ink lines, were removed simultaneously from the same piece of paper. Extraction was performed with 10 μL of methanol and ultrasonicated for three minutes.

Thin-Layer Chromatography

Thin-layer chromatography was performed on all black ballpoint pen inks and dye reference standards to provide information about the content and color of different components in extracted samples. All procedures followed for thin-layer chromatography can be found in the Standard Guide for Test Methods for Forensic Writing Ink Comparison (ASTM 1996). Minor modifications were made to some of the procedures to allow for compatible capillary electrophoresis sample preparation. Methanol, instead of pyridine, was chosen to perform the ink extraction from paper substrates because of high ultraviolet-absorbance for pyridine. The extracted ink was spotted on high-performance thin-layer chromatography slides (Whatman, 10 x 10 cm and 200 μm separation layer) with disposable micropipettes (VWR Scientific, West Chester, Pennsylvania). Solvent system I was chosen for all thin-layer chromatography experiments and consisted of a 70:35:30 mixture of ethyl acetate, ethanol, and water. Separation time was limited to a solvent front migration distance of 5.0 cm that required a 20-minute run. Color was noted for each dye compound, and the respective R_f was calculated.

Laser Desorption/Ionization Time-of-Flight Mass Spectrometry

All laser desorption/ionization-mass spectrometry data was acquired with an Applied Biosystems Voyager-DE Biospectrometry Workstation (Foster City, California) benchtop matrix-assisted laser desorption/ionization time-of-flight mass spectrometer. Both positive- and negative-ion modes were used to determine the presence of acidic, basic, or neutral dye compounds. Fifty laser shots were acquired for each spectrum with a pulsed nitrogen laser (337 nm) intensity varying at attenuator levels of 2000–2300 at a repetition rate of 3 Hz. Accelerating voltage for the positive mode was 20 kV and the negative mode was 15 kV. The extraction delay was set at 100 nsec and the mass/charge (m/z) range was 50 - 2000. An external calibrant of a saturated solution of CsI was used for mass assignment using the peaks for Cs (m/z 132.91), Cs₂I (m/z 392.71), and Cs₃I₂ (m/z 652.53). Different sample types were investigated to determine the best method of analysis. Ink lines on paper were prepared and mounted as 2 x 2 cm strips with tape to a 64-well sample plate. Pen ink straight from the cartridge and the extracted ink solution (methanol and ink) were placed in different sample wells on a matrix-assisted laser desorption/ionization sample plate, allowed to evaporate, and then probed. No additional matrix was added to the samples for laser desorption/ionization. The results from the laser desorption/ionization-mass spectrometry analyses were used as confirmation for dye assignments made based on capillary electrophoresis results.

Results and Discussion

Cationic Dyes

Cationic, basic dyes that are used as colorants in ballpoint pen inks were chosen as analytes for the cationic dye capillary electrophoresis method development. Initially, an acetate buffer without hexadecyltrimethylammonium bromide (CTAB) was used as the background electrolyte for separation in normal polarity (+25 kV). Significant tailing was observed due to the Coulombic interaction of the positively charged dyes with the deprotonated silanol groups of the capillary wall (Weinberger 2000). Poor resolution for sample mixtures was also observed for the simple acetate buffer. Therefore, additives were explored in an effort to reduce the analyte-wall interaction. CTAB was employed as a buffer additive to form a bilayer with the capillary wall, which reversed the effective surface wall charge. CTAB prevents the cationic species from adhering to the wall and has been shown to be effective for storing cationic dyes in glassware (Giles and McKay 1965). Changing the wall charge reversed the electroosmotic flow and forced experiments to be run in reverse-polarity mode (–25 kV). Tailing was eliminated upon addition of CTAB, and better resolution also was achieved. Complete baseline resolution of the six-dye cationic mixture was not observed until the buffer was adjusted to 30 percent methanol to modify specific compound mobilities (Weinberger 2000). Each dye standard was run, and the migration times (t_m) and ultraviolet-visible spectra were recorded. Table 2 lists the μ_ep of the individually run dye standards. Electrophoretic mobilities were calculated by reference to the mobility of a neutral marker peak and the apparent mobility of the analyte peak.

Capture of ultraviolet-visible spectra with the photodiode array detector provides full spectral absorbance for identification purposes. The reference dye ultraviolet-visible spectra were stored in a database along with t_m‘s to compare with spectra obtained for ink extractions. Figure 2 illustrates some of the ultraviolet-visible spectra obtained for standard dyes and represents a small sample of the current FBI database. Spectral analysis provides complementary data to peak μ_ep‘s for dye identification in various ink formulations. Both the P/ACE MDQ Series capillary electrophoresis and Agilent Hewlett Packard systems possess software that stores spectral references and uses search algorithms to quickly identify possible spectral matches. Many of the commercial dye compounds contain complex mixtures of simple dye compounds to produce the desired color and composition. Some of the commercial dyes were identified as containing dyes that were characterized from chemical standards, some that can be partially assigned, and still others that, because of the lack of a reference, cannot be assigned to a specific compound. Several commercial dyes that were studied will not be included because of proprietary formulations. However, knowledge of these specialty formulations is being acquired and can help interpret results obtained from ink extractions. Rhodamine Base B solution was used to estimate the concentration limit of detection and mass limit of detection of the optimized capillary electrophoresis method. A signal-to-noise ratio of 3:1 was chosen as the detection limit and corresponded to a Rhodamine Base B concentration limit of detection of approximately 3.5 μg/mL. Based on the 30.0 nL injection plug volume, the mass limit of detection was calculated to be ~ 100 pg.

Figure 2: Ultraviolet-Visible Spectra of Individual Reference Dye Compounds Collected with a Photodiode Array Detector During Capillary Electrophoresis Analysis

To test the resolving power of the capillary electrophoresis method, a six-basic-dye standard mixture was analyzed. Figure 3 displays an electropherogram of the six-dye mixture. Table 2 lists the concentrations, t_m‘s, and μ_ep‘s of the mixture. The μ_epep‘s of the individually run dyes and the peaks from the mixture allowed the assignment of each peak to its corresponding dye. The separation was completed in less than five minutes, and each of the dyes was baseline-resolved, including the Victoria Blue dyes (see Figure 1k for the structural similarity). In addition, the ultraviolet-visible spectrum for each peak was searched against the spectral reference library (contains all listed dyes in Table 1) and confirmed the migration order of the first three peaks with confidences of 99.32 percent, 99.4 percent, and 99.27 percent, respectively. The last three peaks could not be differentiated because nearly identical absorption spectra are observed because of the similarity of the Victoria Blue structures. values are all negative because the migration of the dye compounds is slower than neutral analytes and signifies that the dyes are traveling against the electroosmotic flow. Comparison of the μ

Figure 3: Six-dye standard mixture of cationic, basic dye compounds separated on the Beckman P/ACE MDQ system and labeled with peak migration times (minutes). (See Table 2.) The electropherogram was recorded at λ = 214 nm.

Anionic Dyes

Although the acidic buffer with the CTAB wall modifier works well for preventing the cationic species from sticking to the capillary wall, the anionic dyes are attracted toward the positive bilayer, resulting in extensive peak tailing for any anionic species that are analyzed with this method. Therefore, an alternative buffer had to be chosen to effectively separate acidic dye compounds in ink formulations. Success of organic buffers with the cationic organic molecules led to testing other organic salts with basic pK_a‘s to separate acidic dyes. Figure 4 depicts an electropherogram of a mixture of a single cationic dye (Crystal Violet) with six anionic dyes. Crystal Violet was added to the mixture because of the likelihood that the violet dye would be present in black ballpoint pen inks. Analyte migration toward the detector is typical for normal polarity with the cations migrating extremely fast (less than 2.4 minutes), neutrals migrating next, and the anions opposing the electroosmotic flow. The rapid electroosmotic flow caused by high pH conditions will overcome the anionic migration, resulting in anions reaching the detector based on the molecule’s shape/charge. Slower-moving anions will reach the detector sooner and have a smaller |μ_ep|, and the faster-moving anions will be detected later. Table 3 lists the t_m and μ_ep values for individual dye components.

Table 3: Results of the Seven-Dye Mixture in CHES/β-Cyclodextrin Buffer

Figure 4: Electropherogram at λ = 214 nm of a seven-dye mixture separated using the CHES/β-cyclodextrin buffer. Migration times indicate the main peaks obtained for all seven compounds. (See Table 3.)

Other peaks not specifically labeled as dye peaks are present in the electropherogram and are associated with various impurities or decomposition products in the dye samples (Table 1). These extra peaks are important because ink manufacturers do not purify purchased dyes; thus these additional peaks provide a level of uniqueness to formulations. Slight tailing is observed for Crystal Violet (2.20 minutes) because of capillary wall interactions, but the rapid mobility limits the effect. Metanil Yellow (2.91 minutes) and Sulforhodamine B (2.96 minutes) are not baseline-resolved; however, this is not considered a problem because Sulforhodamine B is not commonly used as a dye in black pen ink formulations, but Sulforhodamine B has been observed for ink-jet printer inks analyzed with the same experimental setup. Alternative buffer recipes incorporating different micellar properties were able to resolve these two peaks; however, at this time the CHES/β-cyclodextrin buffer provides the best separation power. Fronting for some of the peaks was observed, particularly for Acid Blue 92 (4.62 minutes) and Tartrazine (5.52 minutes). Two factors are responsible for this phenomenon. First, both dyes are multiply charged (Figures 1a and 1h), which leads to electrodispersion (Weinberger 2000), whereas the second factor is the presence of unresolvable components, especially with Acid Blue 92 because it is reported to be only 40 percent pure (Table 1). Again, different buffer conditions (e.g., buffer concentration and pH) and additives (e.g., different micelle-forming detergents) allowed resolution of the Acid Blue 92 peaks. Negative effects caused by buffer alteration were resolution loss for the other dye peaks, detracting from any gain in chromatographic resolution of the Acid Blue 92 impurities.

Table 3 presents the individual dye concentrations in the standard mixture. Dye concentrations listed have been adjusted to the purity of dye material reported on the chemical container. Most dye concentration limits of detection were determined to be 2 μg/mL with the photodiode array under stated experimental conditions. Acid Blue 92 and Tartrazine have higher concentration limits of detection because of the broader peak shapes. Lower concentrations could be observed if a longer injection time was used, but extracted ink analysis did not warrant these procedures. Approximate mass limits of detection can be calculated from the total injection volume used for capillary electrophoresis analysis. For a ten-second injection at 25 mbar, approximately 55.5 nL of sample would be introduced to the capillary. For a 2 μg/mL concentration limit of detection, a 110 pg mass detection limit is estimated.

Reliability and Validation for the Anionic Dye Method

Mobility value reproducibility is important for forensic analysis because an intense ultraviolet-visible spectrum will not always be obtained for ink extractions in sample-limited scenarios. In limited sample cases, a reproducible μ_ep value will lend credibility to the assignment of a particular dye. Reproducibility was examined by day-to-day experiments with the prepared anionic buffer and with three standard mixture (Table 3 analytes) injections without changing the inlet and outlet buffer solutions. Day-to-day reproducibility reflects the capillary performance, as well as the stability of the buffer solution over time. Three repeated anionic dye mixture runs per day over a one-week time span were recorded, and the calculated μ_ep‘s were determined to be within ±2 percent for all dye compounds. Running three experiments on a single set of anionic buffer solutions would be advantageous for casework to ensure that a sample, a blank, and a reference mixture are run under identical conditions. Three consecutive runs on the same buffer with the reference mixture resulted in all μ_ep values falling within ±1.5 percent.

Ballpoint Pen Inks

Forensic analysis of ballpoint pen inks was the driving force in developing capillary electrophoresis buffers that adequately separate dye compounds for identification. Ink extractions were subjected to capillary electrophoresis buffer methods, as well as thin-layer chromatography and laser desorption/ionization-mass spectrometry analyses to characterize the dyes present in the pen inks studied. Thin-layer chromatography and laser desorption/ionization-mass spectrometry were used to complement the capillary electrophoresis analysis and to search for dye components that might not be detected with capillary electrophoresis. Figures 5 and 6 present two electropherograms of ink #1 extraction (Table 4). The U.S. Secret Service sample was prepared by saturating ink on Whatman filter paper, drying, and then storing the sample in a plastic bag. Because of the large ink concentration, a single 1.0 mm punch was removed and extracted with 200 μL of methanol for five minutes. Figure 5 shows the presence of two dye peaks that are close in μ_ep values and have spectral similarities greater than 90 percent for Crystal Violet and Rhodamine Base B. The first dye peak (t_m = 3.40 min, μ_ep = –1.08 x 10^–4 cm²V^–1sec^–1) agrees well with the μ_ep and ultraviolet-visible spectrum of Rhodamine Base B, whereas the second peak (t_m = 3.99 min, μ_ep = –1.41 x 10^–4 cm²V^–1sec^–1) agrees with the μ_ep and spectrum of Crystal Violet. This ink formulation was reported to possess Crystal Violet, Rhodamine Base B, and a yellow dye (personal communication with G. LaPorte, U.S. Secret Service, June 2004). The yellow dye was not detected using the cationic capillary electrophoresis method because it is an acidic dye that interacted strongly with the CTAB bilayer. Figure 6 shows the presence of one cationic (Crystal Violet, μ_ep = +0.84 x 10^–4 cm²V^–1sec^–1) and one anionic (Metanil Yellow, μ_ep = –1.45 x 10^–4 cm²V^–1sec^–1) dye. Also in Figure 6 is the disturbance of the neutral markers with absorbance in the visible region indicating the presence of the third dye compound, Rhodamine Base B, which was already detected with the cationic dye capillary electrophoresis method. Thin-layer chromatography results of ink #1 with the three reference dyes (commercial products of Crystal Violet, Metanil Yellow, and Rhodamine Base B) confirmed that these are the only dyes that are present in this ink.

Table 4: Results of complete analyses performed on commercial black ballpoint pen ink extractions. Two or three thin-layer chromatography spots were observed for Crystal Violet (Methyl Violet) because of the presence of demethylation products.

Figure 5: Ink #1 extraction was run on the P/ACE MDQ system at 214 nm with CTAB/acetate capillary electrophoresis conditions to identify possible cationic dye components.

Figure 6: Ink #1 extraction was also run on the Agilent 3D capillary electrophoresis system at 214 nm under CHES/β-cyclodextrin capillary electrophoresis conditions to identify possible anionic dye components.

Laser desorption/ionization-mass spectrometry was performed to provide complementary data for dye components in ink samples. Individual compounds were studied to determine molecules that would yield ions that could be correlated to a specific compound. Only a few dyes resulted in distinct singly charged ions in either positive- or negative-ion mode. Crystal Violet and Methyl Violet gave characteristic mass spectral patterns that have been studied previously for ink-dating purposes (Grim et al. 2002). A peak at m/z 372 due to the singly charged molecule minus the chlorine anion was present for inks containing Crystal Violet (data not shown). There were also peaks for demethylation products of the dye molecule. There is some variation in the detected m/z for compounds and is a result of small sample changes with respect to the calibration standard. All of the other dyes that were separated in the cationic mixture displayed characteristic peaks indicative of the single-charged cationic species minus the chlorine anion: Rhodamine 6G = m/z 444, Victoria Blue B = m/z 470, Victoria Pure Blue BO = m/z 478, and Victoria Blue R = m/zm/z 443 because it is a neutral molecule without any counterions. Metanil Yellow was the only anionic dye that had an easily interpretable laser desorption/ionization-mass spectrometry that could be assigned to the single-charged molecule at m/z 351. Metanil Yellow is also the only anionic dye that has a single charge associated with the parent molecule and should be the only anionic dye expected to have detectable singly charged molecular species (Karas et al. 2000). Further analysis of the other dyes did not uncover other spectral peaks that could be quickly identified with a particular species, and therefore, laser desorption/ionization-mass spectrometry interpretation was limited to the analysis of the above-mentioned dyes. Because of the complexity of the mass spectra resulting from numerous ink components, only specific spectral peaks were evaluated. The main goal in performing mass spectrometry of standard dyes and inks was to determine the potential advantages of linking mass spectrometry as a secondary detector after capillary electrophoresis separation. Observation of compound-specific masses has proven that compounds could be identified based on mass spectrometry and ultraviolet-visible spectra. Mass spectral information would increase the sample information obtained, although buffer modifications would have to be made for mass detector interfacing and would result in less efficient separations. 423. One peak appeared in each ionization mode for Rhodamine Base B at

Table 4 presents the complete analyses performed on all of the black ballpoint pen inks. Table 4 contains only those results obtained from the anionic capillary electrophoresis method because all of the black pens studied possess anionic dyes, with some neutral dyes and a single cationic dye (Crystal Violet/Methyl Violet). Ink samples were also investigated with the cationic capillary electrophoresis method but only yielded the presence of Crystal Violet/Methyl Violet. More interesting results will be obtained for blue ballpoint inks with the cationic method because Victoria Blue dyes are used to achieve blue ink colors. Capillary electrophoresis results were obtained by extracting a single 1.0 mm hole punch from soaked Whatman filter paper. The hole punch was extracted with 30 μL of methanol and directly injected for ten seconds at 25 mbar. Comparison between the hole-punch technique and one in which a 1.0 cm drawn line was extracted with the same volume of methanol led to comparable peak signal intensities. A 1.0 cm line was chosen to represent a typical casework scenario because normally five 1.0 mm punches are taken from a questioned document. To obtain the most concentrated sample, hole punches can be taken from pen lines that overlap because these spots will contain the largest amount of ink and result in minimal document destruction. Therefore, a 1.0 cm line is roughly equivalent to the above casework scenario. Table 4 contains the thin-layer chromatography results, along with dye assignments based on capillary electrophoresis results, and corroboration by laser desorption/ionization-mass spectrometry data when possible. For dyes that were assigned by capillary electrophoesis, a μ_ep value and an ultraviolet-visible confidence value were calculated. The confidence percentage, computed by the Agilent ChemStation software, determined how similar an ultraviolet-visible spectrum is with respect to a best match from a database of reference chemicals. In some extracted samples, dye identification of detected peaks could not be made at this time. The lack of a reference standard to compare with the capillary electrophoresis or thin-layer chromatography results is responsible for no chemical identification. As more dye references are obtained and analyzed, more complete details about ink formulations will be gained.

One advantage of performing capillary electrophoresis separation on dyes is the ability to effectively narrow the compound candidates responsible for a particular peak. One example of this intuitive approach was used to assign the chemical identity of an unknown yellow dye compound that was present in ink #7. The peak at 3.94 minutes in Figure 7 was not one of the original reference dyes analyzed. Based on the ultraviolet-visible spectrum obtained for that peak (shown in the inset of Figure 7) and the presence of a yellow spot on thin-layer chromatography plates, the search for a possible dye was narrowed to a yellow dye with the ultraviolet-visible spectrum depicted. The ultraviolet-visible spectrum obtained for Metanil Yellow (Figure 2c) is similar; however, the μ_epep value was very useful in predicting the possible molecular weight and charge of the compound by understanding that anionic dyes at shorter t_m‘s have large m/zm/z ratios. Based on the predictive nature of the technique, a m/z between 300 and 360 amu was estimated. A search of the Sigma-Aldrich Handbook of Stains, Dyes and Indicatorsm/z and the expected ultraviolet-visible spectrum: Acid Yellow 42. Acid Yellow 42 was purchased and analyzed with the capillary electrophoresis method and gave identical μ_ep and ultraviolet-visible spectral results as the dye found in ink #7. Spotting this dye solution on the same thin-layer chromatography plate as the ink verified that this dye had the same visual color and R_f value for the yellow dye present in ink #7 was very different. The μ ratios, whereas dyes appearing at longer times have smaller (Green 1990) gave only one candidate that had this value as the unknown. Based on thin-layer chromatography results, there is one additional purple dye that is present in this ink, and the capillary electrophoresis peak at 4.32 minutes has an ultraviolet-visible spectrum similar to other purple dyes. Identification of this dye has not yet been determined, but analysis of more standards will be crucial in this process.

Figure 7: Three dyes are observed in the CHES/β-cyclodextrin capillary electrophoresis separation method of an extracted sample of ink #7 at 214 nm. Inset is an ultraviolet-visible spectrum observed at 3.9 minutes.

Further data can be collected on black ballpoint inks with the possibility of extracting further information from the capillary electrophoresis results. As can be observed in Table 4, there are other peaks that correspond to dyes that have not yet been identified in the electropherograms. Further analyses of the electropherograms also indicate peaks that only exhibit absorbance in the ultraviolet region and could be representative of nondye components found in the specific ink, such as solvents or other additives. Only a limited sampling of inks has been investigated, but so far no ink has shown identical characteristics in the acquired electropherogram, and all could be differentiated based solely on the peak patterns observed. Although chemical identification of every peak detected is not possible, particular electropherogram patterns could be used to differentiate among inks until specific information is determined. Peak area analysis could also be used to calculate specific ratios among different detected ink components to increase confidence in ink assessment. The overall potential of this technique can be advanced when more standard dye samples, as well as other components, can be analyzed in tandem.

The overall goal of this technique is facile introduction into everyday casework protocols. To accomplish this, the sample preparation had to mimic currently performed extraction procedures (see Experimental Samples Section). The ink extraction was injected into the capillary electrophoresis instrument and separated in fresh buffer, and the result is presented in Figure 8. The presence of two dyes for ink #6 was observed, and chemical identity was determined by ultraviolet-visible spectral database searching and μ_ep value comparison with reference dye compounds. Crystal Violet was the ultraviolet-visible spectral match (93.4 percent) for the peak with a μ_ep of +0.80 x 10^–4 cm²V^–1sec^–1. Metanil Yellow was assigned to the second peak,

μ_ep = –0.45 x 10^–4 cm²V^–1sec^–1,

with an 83.9 percent ultraviolet-visible spectral match. The low ultraviolet-visible spectral similarity of the Metanil Yellow peak is due to the reduced absorbance detected because of the lower concentration of the mock casework sample. The remaining solution was spotted on a thin-layer chromatography plate and matrix-assisted laser desorption/ionization sample plate in an effort to confirm the capillary electrophoresis results. Thin-layer chromatography analysis of the remaining ink injection solution resulted in faint spots that were hard to discern. The second run performed on the same capillary electrophoresis buffer was a blank paper extraction to indicate possible sample carryover and to observe peaks attributable to the paper material. No interference peaks were observed for the blank material. The last injection was a standard mixture containing the two suspected dyes contained in the unknown ink formulation and mimicked the sample electropherogram.

Figure 8: Capillary Electrophoresis Results at 214 nm of a Casework Protocol Used to Extract Five 1 mm Hole Punches of Ink #6

The above demonstration indicated that capillary electrophoresis analysis could be performed in addition to techniques already used in evidence analysis. No additional material is necessary for increased sensitivity. In fact, if thin-layer chromatography spots can be observed for the extracted solution, capillary electrophoresis will be successful. The lone disadvantage of capillary electrophoresis is the necessity that the analyte be soluble in the separation buffer. Many dyes are at least partially soluble, but some dye classes are fairly insoluble. One investigated ink in this report resulted in particle formation in the injection solution. Although this occurred, the dyes that remained in solution could still be investigated.

The third injection could be modified based on casework requirements. The one above was chosen to demonstrate that reproducible results could be obtained from successive injections, and simple data overlay could produce confirmatory results. However, the absence of alternative dye possibilities in the standard mixture could lead to arguments made for biasing of the comparison standard. In this case, a full standard library could be developed to analyze for comparison, but a much larger database needs to be developed. Another possibility for a confirmation run could be the suspected pen obtained from a crime scene. Matching peaks from the evidence and the suspected source would provide confirmation that the same dyes are present, although this would not necessarily prove that the particular pen was used for the evidence being processed. Exclusion of a pen could also provide strong evidence in the case of inconsistent dye ingredients.

Other Applications

Ballpoint pen ink analysis is not the only area that could be investigated by capillary electrophoresis (Liu et al. 1995; Masar et al. 1996). Yellow Food Dye #5 is a purified form of Tartrazine, one of the yellow dyes successfully separated in the CHES/β-cyclodextrin capillary electrophoresis buffer for anionic, acidic dyes. This food coloring is found in Mountain Dew soda (PepsiCo, Chicago, Illinois) and was examined with the capillary electrophoresis method. A sample of Mountain Dew was boiled down to a small volume for increased concentration of the food-coloring additive. The presence of a Tartrazine peak was observed in the Mountain Dew sample (data not shown). Other peaks also are detected by the photodiode array, but examination of the ultraviolet-visible spectra indicates that only the peak responsible for Tartrazine has significant absorption in the visible region. Printer inks and dyes used in the textile and currency industries are also possible analytes for the methods developed in this report.

Conclusion

Capillary electrophoresis has been shown to be a powerful analytical tool for forensic analysis of dyes contained in ballpoint pen inks. Two unique pieces of information (ultraviolet-visible spectrum and μ_ep value) can be gained from the analysis. These chemical characteristics can be used in ink identification or differentiation on a document. Protocols have been developed for casework scenarios while preserving extracted solutions to be further analyzed with other complementary techniques. The anionic capillary electrophoresis buffer was found to effectively separate black ballpoint ink dye components with higher sensitivity, faster analysis time, and more definitive chemical identification than thin-layer chromatography procedures. Capillary electrophoresis also can detect solvents and other ink additives that are not dyes in the same experiment. Electronic data storage also provides the benefit of database and library search algorithms. Although ballpoint pen ink analysis was the primary focus of this report, these formulated buffers could be used in a multitude of other forensic analyses, such as food and textile dyes.

Acknowledgments

The authors thank Dr. Gerry LaPorte for generously donating ink and dye samples from the U.S. Secret Service ink library. Funding, in part, was obtained from the Oak Ridge Institute for Science and Education, Oak Ridge, Tennessee, for postdoctoral research projects.

References

Aginsky, V. N. Determination of the age of ballpoint pen ink by gas and densitometric thin-layer chromatography, Journal of Chromatography A (1994) 678:119-125.

ASTM. Standard Guide for Test Methods for Forensic Writing Ink Comparison, ASTM-E1422-96, 1996.

Brunelle, R. L. Ink dating: The state of the art, Journal of Forensic Sciences (1992) 37:113-124.

Brunelle, R. L. and Cantu, A. A. Critical evaluation of current ink dating techniques, Journal of Forensic Sciences (1987) 32:1522-1536.

Brunelle, R. L. and Reed, R. W. Forensic Examination of Ink and Paper. Charles C. Thomas, Springfield, Illinois, 1984.

Brunelle, R. L., Breedlove, C. H., and Midkiff, C. R. Determining the relative age of ballpoint inks using a single-solvent extraction technique, Journal of Forensic Sciences (1987) 32:1511-1521.

Burkinshaw, S. M., Hinks, D., and Lewis, D. M. Capillary zone electrophoresis in the analysis of dyes and other compounds employed in the dye-manufacturing and dye-using industries, Journal of Chromatography (1993) 640:413-417.

Cantu, A. A. Analytical methods for detecting fraudulent documents, Analytical Chemistry (1991) 63:A847-A854.

Cantu, A. A. Comments on the accelerated aging of ink, Journal of Forensic Sciences (1988) 33:744-750.

Cantu, A. A. Sketch of analytical methods for document dating: The dynamic approach: Determining age dependent analytical profiles, International Journal of Forensic Document Examiners (1996) 2:192-208.

Croft, S. N. and Lewis, D. M. Analysis of reactive dyes and related derivatives using high-performance capillary electrophoresis, Dyes and Pigments (1992) 18:309-317.

Fanali, S. and Schudel, M. Some separations of black and red water-soluble fiber-tip pen inks by capillary zone electrophoresis and thin-layer chromatography, Journal of Forensic Sciences (1991) 36:1192-1197.

Giles, C. H. and McKay, R. B. Adsorption of cationic (basic) dyes by fixed yeast cells, Journal of Bacteriology (1965) 89:390-397.

Green, F. J. Sigma-Aldrich Handbook of Stains, Dyes, and Indicators. Aldrich Chemical, Milwaukee, Minnesota, 1990.

Grim, D. M. and Allison, J. Identification of colorants as used in watercolor and oil paintings by UV laser desorption mass spectrometry, International Journal of Mass Spectrometry (2003) 222:85-99.

Grim, D. M., Siegel, J., and Allison, J. Evaluation of laser desorption mass spectrometry and UV accelerated aging of dyes on paper as tools for the evaluation of a questioned document, Journal of Forensic Sciences (2002) 47:1265-1273.

Harris, J. A preliminary report on the nondestructive examination of ballpoint pen ink on questioned documents by FT-IR spectroscopy, Journal of the Canadian Society of Forensic Science (1991) 24:5-21.

Horton, R. A. and Nelson, L. K. Evaluation of the use of laser-induced infrared luminescence to differentiate writing inks, Journal of Forensic Sciences (1991) 36:838-843.

Industrial Dyes: Chemistry, Properties, Applications. Wiley-VcH, Weinheim, Germany, 2003.

Jandera, P., Fischer, J., Stanek, V., Kucerova, M., and Zvonicek, P. Separation of aromatic sulphonic acid dye intermediates by high-performance liquid chromatography and capillary zone electrophoresis, Journal of Chromatography A (1996) 738:201-213.

Karas, M., Gluckman, M., and Schafer, J. Ionization in matrix-assisted laser desorption/ionization: Singly charged molecular ions are the lucky survivors, Journal of Mass Spectroscopy (2000) 35:1-12.

LaPorte, G. U.S. Secret Service, personal communication, June 2004.

Leach, R. H. and Pierce, R. J. Printing Ink Manual. Chapman and Hall, London, 1993.

Liu, H., Zhu, T., Zhang, Y., Qi, S., Huang, A., and Sun, Y. Determination of synthetic colourant food additives by capillary zone electrophoresis, Journal of Chromatography A (1995) 718:448-453.

Lyter, A. H. Examination of ball pen ink by high pressure liquid chromatography, Journal of Forensic Sciences (1982) 27:154-160.

Masar, M., Kaniansky, D., and Madajova, V. Separation of synthetic food colourants by capillary zone electrophoresis in a hydrodynamically closed separation compartment, Journal of Chromatography A (1996) 724:327-336.

Merrill, R. A. and Bartick, E. G. Analysis of ballpoint pen inks by diffuse reflectance infrared spectrometry, Journal of Forensic Sciences (1992) 37:528-541.

Ng, L. K., Lafontaine, P., and Brazeau, L. Ballpoint pen inks: Characterization by positive and negative ion-electrospray ionization mass spectrometry for the forensic examination of writing inks, Journal of Forensic Sciences (2002) 47:1-10.

Rhode, E., McManus, A. C., Vogt, C., and Heineman, W. R. Separation and comparison of fountain pen inks by capillary zone electrophoresis, Journal of Forensic Sciences (1997) 42:1004-1011.

Rhode, E., Vogt, C., and Heineman, W. R. Analysis of fountain pen inks by capillary electrophoresis with ultraviolet/visible absorbance and laser-induced fluorescence detection, Electrophoresis (1998) 19:31-41.

Tebbett, I. Chromatographic analysis of inks for forensic science applications, Forensic Science Review (1991) 3:71-82.

Trzcinska, B. M. Infrared spectroscopy of ballpen paste, Forensic Science International (1990) 46:105-109.

Trzcinska, B. M. Writing materials examination in criminalistic research by FTIR spectroscopy, Journal of Molecular Structure (1993) 294:259-262.

Varlaskin, P. G. and Low, M. J. D. FT-IR photothermal beam deflection spectroscopy of black inks on paper, Applied Spectroscopy (1986) 40:507-513.

Vogt, C., Becker, A., and Vogt, J. Investigation of ball point pen inks by capillary electrophoresis (CE) with UV/Vis absorbance and laser induced fluorescence detection and particle induced X-ray emission (PIXE), Journal of Forensic Sciences (1999) 44:819-831.

Vogt, C., Vogt, J., Becker, A., and Rhode, E. Separation, comparison and identification of fountain pen inks by capillary electrophoresis with UV-visible and fluorescence detection and by proton-induced X-ray emission, Journal of Chromatography A (1997) 781:391-405.

Weinberger, R. Practical Capillary Electrophoresis. 2nd ed., Academic Press, San Diego, California, 2000.

Xu, X., Koeijer, J. A., Moel, J. J. M., and Logtenberg, H. Ink analysis for forensic science applications by micellar electrokinetic capillary chromatography with photodiode array detection, International Journal of Forensic Document Examiners (1997) 3:240-260.

Zeichner, A., Levin, N., Klein, A., and Novoselsky, Y. Transmission and reflectance microspectrophotometry of inks, Journal of Forensic Sciences (1988) 33:1171-1184.

Zlotnick, J. A. and Smith, F. P. Separation of some black rollerball pen inks by capillary electrophoresis: Preliminary data, Forensic Science International (1998) 92:269-280.