Challenges Determining the Correct Deposition ... - ACS Publications

Subscriber access provided by AUBURN UNIV AUBURN

Article

Challenges determining the correct deposition order of different intersecting black inks by Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS) Robyn E. Goacher, Lauren G. DiFonzo, and Kathleen C. Lesko Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.6b03411 • Publication Date (Web): 23 Nov 2016 Downloaded from http://pubs.acs.org on November 30, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Challenges determining the correct deposition order of different intersecting black inks by Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS) Robyn E. Goacher*, Lauren G. DiFonzo, Kathleen C. Lesko Department of Biochemistry, Chemistry and Physics, Niagara University, 5795 Lewiston Road, Lewiston NY, 14109, USA ABSTRACT: The distinction of different inks and determination of their deposition order are important forensic tasks when evaluating questioned documents. Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS) has received attention for these tasks due to the technique’s non-destructive nature, rich mass spectral information, ability to provide chemical images, and excellent surface sensitivity. Prior literature results demonstrate the ability of ToF-SIMS to differentiate between many varieties of blue ballpoint pens, and to determine the correct deposition order for selected ink intersections. The current work therefore sought to further study the intersections of different black inks. Three black pens were initially tested and were successfully distinguished. However, interesting phenomena were observed where certain inks dominated over others, causing distinctly incorrect conclusions to be made regarding ink deposition order. To explore whether these incorrect results could be corrected, different primary ions (Bi3+, Bi3++, Ar1000+) were used and positive and negative secondary ions were evaluated. In data analysis, restriction of the mass range was considered and different multivariate analysis techniques, Principal Component Analysis (PCA) and Multivariate Curve Resolution (MCR), were performed on the secondary ion images and spectra from selected regions of interest. Issues regarding incorrect apparent order of deposition persisted through the variation of all these parameters. In addition to ink differentiation, and deposition order, evidence of ink ageing was also observed. These results raise several questions to be answered regarding the widespread use of ToF-SIMS for forensic purposes.

INTRODUCTION. Despite the popularity of electronic communication, there remain many important business and legal transactions that proceed via paper. Therefore the analysis of questioned documents continues to be relevant to many forensic cases. A review by Calcerrada and Garcia-Ruiz states that most recent research related to questioned documents has focused on determining the age of the inks present, and on differentiating between different inks.1 Fewer studies have focused on determining the deposition order (writing sequence) for questioned documents, and this remains an important task.1 The deposition sequence of overlapping lines, regardless of chemical similarity or dissimilarity of the inks, may be studied by topographical methods. The pliability of the paper surface results in its deformation from the pressure of subsequent pen strokes. SEM and AFM may be used to analyze microtopography but their scanning area is limited.2 A greater field of view is available via non-destructive 3D laser profilometry, which uses holography to determine the thickness profile of an intersection, providing either positive confirmation of writing sequence or inconclusive results when the ink layer is of commensurate thickness to the paper roughness.2 Furthermore, ballpoint pens tend to deposit more ink at the edges of the pen stroke, creating parallel ridges that appear like railway tracks.3 These ridges can be lifted via the Kromekote lifting technique, but this method is destructive and is sensitive to both the original writing pressure and the lifting pressure.3 For differentiation of inks that cannot be distinguished by a forensic examiner using optical microscopy, different instru-

mental approaches are required. Separation techniques (e.g. TLC, HPLC) have historically been used to determine whether there are differing ink compositions in a document, but the extraction step causes these approaches to be destructive. Therefore, the prevalence of separations-based reports has been decreasing.1 Non-destructive spectrometric chemical imaging approaches, such as infrared and Raman imaging, and mass spectrometry, are preferred for the preservation of unique forensic evidence, and reports of their use are increasing.1 A review by Braz et al indicates that imaging Raman spectroscopy can successfully distinguish different inks and their deposition order in some, but not all, cases.4 ATR-FTIR spectroscopy has also been demonstrated as able to differentiate ten unique pens using multivariate statistical methods, with correct prediction rates of greater than 90% for inks on differing papers.5 ATR-FTIR was also successfully used to distinguish intersections of ballpoint pen overlying laser toner and red sealing ink, although other intersections could not be successfully analyzed due to dominance of paper signals in the spectra.6 Overall, vibrational spectroscopies have promise but are limited in their chemical specificity due to the similar functional groups present within many inks. ToF-SIMS has been investigated as a technique for the chemical imaging of inks on paper for forensic purposes for many years.7-15 Like other mass spectrometry techniques, ToFSIMS provides more specific chemical information, including both inorganic and organic ions, than do Raman and IR spectroscopies. The excellent surface sensitivity of ToF-SIMS also allows for good exclusion of paper signals, and provides the possibility that the upper ink in an intersection would be dom-

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 9

Profile “P”, BicTM Pro+ “B” and StaplesTM Motiva “S”). A sheet of Boise X-9 multi-use copy paper (92 brightness) was taken from the middle of the ream to minimize contamination due to handling. The first marks were made as vertical lines and either 1 minute, 1 hour or 1 day elapsed prior to making the second horizontal marks. Throughout the study the intersections were referred to as “Xvert1-Yhoriz2” to indicate the deposition order, where X and Y were replaced with the corresponding B, P, and S abbreviations for the pen brands. The 1-minute samples were produced three separate times, once for comparison of bismuth primary ions (1 day elapsed between writing and ToF-SIMS analysis for Pvert1-Bhoriz2, Pvert1-Shoriz2, Svert1-Bhoriz2, and Svert1-Phoriz2, while 5 weeks elapsed for Bvert1-Phoriz2 and Bvert1-Shoriz2), once for analysis with Ar1000+ primary ions (8 days elapsed between writing and analysis) and once for comparison with 1-hour and 1-day drying times (5 days elapsed between writing of first marks and analysis). Samples were stored in a covered polystyrene petri dish in dark conditions. ToF-SIMS Analysis. Each intersection was analyzed using a ToF-SIMS V instrument (IonTof GmbH, Muenster Germany) equipped with a reflectron type mass analyzer. Samples having 1-minute drying time were analyzed first, acquiring both positive and negative secondary ions with 25 keV Bi3+, 50 keV Bi3++ and 20 keV Ar1000+ primary ions. Next, the 1-hour and 1day drying times were examined with the 50 keV Bi3++ primary ion and positive secondary ions. For all samples, 20 eV electron flooding was used for charge neutralization. Images were acquired using macro scans covering a 1.5mm x 1.5mm area with 150x150 pixels. A single scan was obtained for each intersection with 1 shot per pixel and 500 frames per patch. The pressure in the analysis chamber was maintained below 2.5x10-7 mbar. SurfaceLab v.6.3 software (IonToF GmbH, Muenster Germany) was used to calibrate the spectra and to select regions of interest (ROIs) for the paper, vertical (first) ink, horizontal (second) ink, and center intersecting regions. Peak intensities were binned to nominal mass using a comprehensive peak list. Full-area image stacks were then exported as BIF6 files, and spectra from the selected ROIs were exported to Excel. Multivariate Statistical Analysis. Data was analyzed using PLS Toolbox v.7.0.3 and MIA Toolbox v.2.8.2 (Eigenvector, Inc) in MATLAB v.8.0 (The Mathworks Inc). Spectra from the ROIs were pre-processed using Poisson (square root mean) scaling, normalization and mean centering and were analyzed by PCA. Calibration models were built from the ROIs of the pure B, P, and S inks from all intersections within a data set (e.g. spectra of the same secondary ion polarity from the same primary ion). Then, all of the center intersection ROIs were placed into this model as test samples to see whether the intersections were correctly predicted to be the same as the horizontal ink written second, which should appear on top (see Fig 1). This was dubbed “all ROI” analysis. Next, the two pure ink ROIs from each individual intersection were used to build a two-point calibration model, into which the corresponding center ROI of the intersection was loaded (see Fig S-1). This was dubbed “single ROI” analysis. A percent error was calculated from the scores of these single ROI models as follows: |score of horizontal ROI – score of

inantly detected within the typical 1-2 monolayer sampling depth. ToF-SIMS has been called “reliable due to its high capability to discriminate among ink samples” but was also described as an expensive option “used only in some cases”.1 In 1994, Pachuta and Staral demonstrated that ToF-SIMS could be used to distinguish inks from different manufacturers (of the same or different colors from fountain and ballpoint pens), emphasizing that single paper fibers could be removed from a document and analyzed if the entire document could not be placed in the instrument.7 Another study found that by manual inspection of ToF-SIMS spectra from 13 blue ballpoint pen inks, characteristic organic and inorganic ions could be identified for each ink.8 Denman et al compared 24 blue ballpoint pens via ToF-SIMS, using Principal Component Analysis (PCA) to successfully distinguish 41 out of 45 ink pairs.9 Denman et al also observed that different batches of inks from the same manufacturer could be distinguished by ToF-SIMS and they provided a comprehensive list of characteristic secondary ions for different dyes.9 ToF-SIMS has furthermore been found to distinguish inks from inkjet-printed markings using PCA10 and has been used to study the diffusion of ink vehicles, or solvents, in inkjet printing11. While ink intersections were not evaluated in these reports, these studies indicated that the ToF-SIMS approach was promising for distinguishing different inks. Intersections of differing markings have been studied in a few variations. He et al used imaging ToF-SIMS to study intersections of different ink colors (e.g. red and black ink), concluding that ink deposition order could be correctly determined if sufficient drying time (1.5 hr in this study) was provided between the first and second inks.12 Preliminary results also reported the correct determination of the deposition order by ToF-SIMS between two overlapping black inks used on a signature line, which struck popular interest for its potential use in the court setting.13 Additional intersections have been reported to produce apparently correct deposition orders via ToF-SIMS analysis of black gel and black ballpoint pens14 and between red ballpoint and red fiber tip pens15. In the above reports, correct deposition order was determined when maps of the characteristic ion signal from the first line drawn appeared interrupted, and the signal from the second line drawn appeared continuous. The goal of the present research was to undertake a more extended study of the intersections of black ballpoint pen inks to determine whether the inks could be identified by ToFSIMS as different, and whether the correct deposition order would be determined. Black inks were chosen because ToFSIMS analysis of black inks appears to be relatively understudied in comparison to blue inks, and ballpoint pen inks are featured in ~80% of questioned document cases9. However, in the first group of three black inks selected at random for this study, we observed the as yet unreported result of clearly incorrect deposition orders, in which the bottom ink clearly appeared continuous, while the upper ink appeared interrupted. Therefore, the goals were modified in light of this unusual result to explore the influence of experimental parameters such as the primary ion used, the polarity of the secondary ions collected, and the mass range considered during analysis. The influence of drying time was also evaluated. EXPERIMENTAL. Sample Preparation. Three black retractable ballpoint pens were arbitrarily chosen (PapermateTM

2


Page 3 of 9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry incorrect when S was on the bottom. Analysis of single ROIs from each individual ink intersection (see Fig S-1) corroborated the conclusions of the all-ROI analyses, clarifying some ambiguous results. For single ROIs, % errors 45% were called “incorrect”. The conclusions from the all-ROI and single-ROI analyses are combined with images of ink-descriptive MCR components in summary figures such as Figure 2, which is for full-mass analysis of 1-minute drying times analyzed with Bi3++ primary ions. Summary tables for other primary ions, mass ranges, and drying times can be found in the Supporting Information (Figs. S-6 to S-13).

center ROI| / |score of horizontal ROI – score of vertical ROI|. This is illustrated in Fig. S-1. Full-area images were pre-processed using Poisson (square root mean) scaling and normalization. For each image, PCA was applied as an unbiased method to determine the number of significant components to be used in Multivariate Curve Resolution (MCR). An example of a full MCR model appears in Fig S-2. Typically, two or more MCR components described paper, but these were not informative regarding ink distinction or deposition and so were not included in the summary tables. To examine the influence of more intact vs fragmented secondary ions, all PCA and MCR processes were performed on the full spectra (m/z 12-740 for negative ions, m/z 12-733 for positive ions), low mass range (m/z 12-50 for negative ions, m/z 12-200 for positive ions), and high mass range (m/z 51740 for negative ions and m/z 201-733 for positive ions). The negative ion cutoff was based on the greater prevalence of high-intensity peaks below m/z 50 that may experience detector saturation. The positive ion cutoff was based on observation of different sets of peaks in the spectra. RESULTS AND DISCUSSION. Ink differentiation and determination of deposition order. Figure 1 shows the all-ROI PCA model for full-mass positive spectra of intersections with a 1-minute drying time, analyzed with Bi3++ primary ions. Each group of pure ink replicates was distinct, with no overlap in their 95% confidence ellipses (Fig. 1a), supporting prior reports of successful differentiation of like-colored inks with ToF-SIMS. To evaluate the correctness of the apparent deposition order, the ROIs for the centers of the intersections were loaded into the model as test samples with the expectation that these central regions would have the same chemistry as the second ink applied (horizontal), which should be on top of the first ink applied (vertical). The center ROIs are therefore symbolized as circles of the color of the horizontal ink. Inspection of Fig. 1a indicates that this expectation was not realized for all intersections. Instead, it appears that the PapermateTM ink (P) dominated all intersections that it was involved in, with all centers from these intersections appearing in or close to the 95% confidence ellipse of the P ink. Therefore, the P ink correctly appeared on top when it was on top (Phoriz2) but incorrectly appeared on top when it was on bottom (Pvert1). When the P ink was not present, but the intersections of S and B inks were analyzed, both center regions scored closer to the S ink, with the S ink dominating over the B ink. The main peak in the loadings describing the B ink was sodium at m/z 23 (Fig. 1b). Exclusion of the sodium peak from the full-mass positive ion PCA (Fig S-3) did not alter any conclusions regarding the identity of the centers of the intersections and therefore removal of sodium will not be discussed further. Full-mass negative ion all-ROI results (Fig S-4a,b) were also consistent with full-mass positive ion results. In the negative ion mode, both of the centers from intersections between B and P inks scored with P inks. Both centers from intersections between P and S inks also scored closer to the P ink than to the S ink, supporting the dominance of the P ink. When S and B inks were combined, the results were ambiguous when S was on top (nearly half-way between S and B groups) and

Figure 1. PCA scores (a) and loadings (b) for model built on all ROIs of pure inks for intersections with 1-minute drying time, analyzed with Bi3++ primary ions, using full mass range of positive secondary ions.

Multivariate analysis (e.g. MCR) of a ToF-SIMS image can be advantageous when there is only one questioned intersection in a document because the many pixels provide replicates from which patterns can be extracted. These images would also likely be persuasive for jurors, helping them visualize the data. Momentarily disregarding the apparent deposition order, and focusing on ink differentiation, it is notable that all fullmass MCR images supported the ability of ToF-SIMS to distinguish between the different black inks. Sometimes two

3



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 9

Figure 2. Summary tables for intersections with 1-minute drying time analyzed with Bi3++ primary ions, using full mass range of positive secondary ions (a) and negative secondary ions (b). Each intersection is represented by the abbreviations B=Bic, S=Staples, P=Papermate brand, with the first line labeled “vert1” and the second line labeled “horiz2”. For each ink intersection, the images of MCR scores for inkrelated components are shown (where red indicates the highest scores and blue indicates the lowest scores). MCR annotations indicate whether 2 inks were distinguishable, and their apparent correct, ambiguous or incorrect deposition order. Conclusions are also included from “All ROI” analyses (e.g. Fig.1) and single ROI “sROI” analyses (with % error, e.g. Fig. S-1).

mutually exclusive components were identified for the two inks, with one component describing the vertical ink and another describing the horizontal ink (e.g. intersections of P and

S inks in Fig. 2). Other intersections were also distinguishable as two inks but one or both of the ink components would have some intensity in both the vertical and horizontal lines (e.g.

4


Page 5 of 9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry due to matrix effects, it was decided to investigate which ions were characteristic of which inks, and whether restriction of the mass range in either positive or negative mode might uncover a “sweet spot” wherein the correct deposition order might be obtained. Furthermore, the identity of the primary ion can alter the information depth of the analysis, and produce different levels of fragmentation of the secondary ions. Thus, different primary ions were compared for the analysis. Characteristic ions for the three inks were identified through inspection of the raw ToF-SIMS spectra (Fig. S-14, S-15) and of all available PCA and MCR loadings. The positive ink spectra all had peaks at m/z 340-344 (pararosaniline8), 358-359 (Basic Violet 18) and 372-373, associated with the Basic Violet 3 dye8-9 (Fig. S-14), which are common dyes in black and blue inks9. PCA at full and high mass (Fig.1, S-5c,d) differentiated the P ink as having less signal for Basic Violet 3 and more signal for p-rosaniline and Basic Violet 1. Furthermore, the P ink had unique peaks at m/z 140, 158, 174, 230 and 274, which contributed to its separation from the B and S inks. The S ink furthermore had greater intensity of aromatic peaks at m/z 77, 91, 105 and 117. The B ink had no unique peaks and tended to be described by higher relative intensity of sodium at m/z 23 or of many other peaks common to several of the inks. Manual observation of the ToF-SIMS spectra (Fig. S-14) showed other peaks that differed between the inks but none of these loaded strongly in the multivariate analyses. Additionally, no peaks higher than m/z 400 distinguished the inks in the multivariate analyses. Given the similarities among the inks for positive ions greater than m/z 200, it is understandable that when the positive ion mass range was limited to high mass, the inks could not be distinguished by MCR using Bi3++ primary ions (e.g. Fig. S-6c). Only one component with a + shape was observed, describing both inks. High-mass positive ion all-ROI analysis also had poorer separation between B and S inks than at full mass (Fig. S-5c vs Fig 1). Positive MCR images for Bi3++ at low mass (≤ m/z 200) continued to distinguish the two inks, producing the same conclusions as at full mass, except that the MCR image for Pvert1-Bhoriz2 was clearly incorrect at low mass (Fig S-6b) when it had been ambiguous at full mass (Fig 2). Low-mass all-ROI and single-ROI conclusions also agreed completely with full-mass conclusions. When cluster primary ions with lower energy per incident atom were used (Bi3+ and Ar1000+), a similar trend in useful mass range was observed for positive ions, where the low mass range positive ions were more effective at distinguishing the different inks than were the high mass range (Fig. S-7, S8). However, for these “softer” primary ions, MCR images from high mass positive ions could distinguish the P and B inks, and in some cases could distinguish the P and S inks. Conclusions regarding the deposition order were nearly identical for Bi3++ (Fig S-6) and Bi3+ (Fig S-7) at low and full masses. Conclusions regarding the deposition order using positive full-mass spectra from Ar1000+ primary ion bombardment (Fig S-8a) were more ambiguous for intersections of B and P inks, and were more incorrect for intersections of S and P inks. Ar+ 1000 conclusions at low mass (Fig S-8b), however, agreed with + Bi3 and Bi3++ conclusions with the exception that some ambiguous MCR images were more clearly incorrect.

intersections of S and B inks in Fig. 2). Returning to the question of writing sequence, examination of the MCR images (Figs. 2, S-6 to S-13) makes it clear that some of the intersections between dissimilar inks appeared correct, with continuous signal for the horizontal ink that was written second, and interrupted signal for the underlying vertical ink that was written first (e.g. Svert1-Phoriz2 in Fig. 2a). Other intersections appeared ambiguous, with interruptions in both lines (e.g. Pvert1-Shoriz2 in Fig. 2a) or in neither ink line (e.g. Bvert1-Shoriz2 in Fig. 2a). Such correct results are as described in prior literature12-15, and similar ambiguous results have been reported due to mixing from inadequate drying times12. However, this is the first time that clearly incorrect results have been reported for the ToF-SIMS determination of deposition order, where the first line (in this case, the vertical line) that is below, appears continuous, and the second line (in this case, the horizontal line) that is above, appears interrupted (e.g. Pvert1-Bhoriz2 and Svert1-Bhoriz2 intersections in Fig. 2b). Some of the intersections exhibited the “railroad tracks” characteristic of ballpoint inks3 but these were not clearly underlying or overlying. Curvature of the ink signals at the intersections was also sometimes observed, but occurred both for the underlying vertical and overlying horizontal inks in different intersections. Given the dominance of correct writing sequence results in previously reported ToF-SIMS literature12-15, it is alarming to find so many incorrect determinations of deposition order via all three methods of analysis (all-ROI, single-ROI and MCR of images). The assumption that the lower ink in an intersection will have interrupted signal and that the upper ink will have continuous signal depends on the presence of distinct ink layers (a double layer). Ozbek et al16 used visual light microscopy to examine cross-sections from layered inks from different ink categories: oil-based, liquid-based and gel-based. Their observations showed that in many cases, double layers were not formed, but rather that there was either a partial or completemixing of the ink layers.16 When oil-based inks were deposited first, double layers tended to not form, but there was no firm predictive trend in their data.16 They did not report any inversions of the applied ink layers. Thus, it is surprising to obtain anything other than a correct (double-layer) or ambiguous (mixed) result. All three pens used here were categorized as “retractable ballpoint pens” on the Staples website. However, the Papermate ink was described as having “the smooth feeling of a gel pen, with a quick drying, super bold ink”17. The Staples pen was described as an “advanced ink” pen whose “hybrid ink combines the functionality of a ballpoint pen with the smoothness of gel ink”.18 Therefore, these pens may not have the same oil-based ink that is typical of ballpoint pens.1 The dominance of a given pen, such as the P ink, could be the result of matrix effects, where compounds in the underlying P ink suppress the ionization of compounds in the overlying S and B inks, or may be the result of hydrophobic/hydrophilic interactions where an underlying P ink layer repels the deposition of overlying S or B ink layers. Characteristic ions and role of mass range and primary ion. Given the potential that the incorrect deposition order may be

5



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

In the negative polarity, PCA loadings (Fig. S-4) indicated that a main difference between the inks was that the S ink contained greater intensity of peaks at m/z 63 (PO2-) and 79 (PO3-) while the B and P inks had higher intensity at m/z 26 (CN-) and 80 (HPO3- and/or SO3-). These differences were also apparent in manual inspection of the spectra (Fig. S-15). The B and S inks produced significant negative ion peaks at m/z 156, 171 and 352 (unidentified), while the P ink tended to not have these peaks. As with the positive inks, no peaks higher than m/z 400 described the inks in the multivariate analyses. Restriction of the mass range for negative ions was made at m/z 50 and was done primarily for the reason of excluding highly intense low-mass peaks whose intensities may vary due to detector saturation. For Bi3++ primary ions, restriction of the negative mass range to high mass (> m/z 50, Fig S-9b) provided equivalent results to the full mass analysis (Fig 2b), while restriction to low mass (≤ m/z 50) made differentiation of the inks slightly more ambiguous (Fig S-9c). Comparing Bi3+ primary ions to Bi3++ primary ions for negative spectra, full, low and high mass conclusions were less consistent and more ambiguous with Bi3+, including some intersections that could not be distinguished in the MCR images at full and low mass (Fig S-10). The negative ions at full and high mass for Ar1000+ were more incorrect for intersections of S and P inks and were more correct for the Bvert1-Shoriz2 intersection, than were the Bi3++ conclusions. Low Mass Ar1000+ analyses frequently led to no distinction of the inks and mixed ROIs results, likely due to the relatively low signal from low-mass peaks with the massive argon clusters. In summary, ToF-SIMS was effective at distinguishing the different black inks at intersections for positive and negative secondary ions. In the case of the inks studied, this distinction was mainly due to lower-mass characteristic positive ions and higher-mass characteristic negative ions. No primary ion or mass range tested produced completely correct ink deposition orders. However, the rate of correct deposition orders for positive and negative ions was highest with the Bi3++ primary ion, and this was used to further examine the influence of drying time on the conclusions. Effects of Drying Time. The conclusions reached through positive ion analyses of samples with 1-hour (Fig S-12) and 1day (Fig S-13) drying times were nearly identical to those discussed so far for 1-minute drying times (Fig S-6). The only differences observed were the interchange between ambiguous and incorrect conclusions for MCR images for full-mass and low-mass ranges. The full-mass and low-mass conclusions from all-ROI and single-ROI analyses remained consistent across the three time scales, indicating that the dominance of the P ink over the S and B inks, and of the S ink over the B ink is maintained over a range of drying times between ink deposition. The high-mass range for all three time scales was unable to distinguish the different inks by MCR images, and conclusions reached through ROI analysis of the high-mass range were inconsistent across the time scale. This is consistent with the poorer distinction for positive ions at high mass. Halo around the PapermateTM ink. Although it was less effective for determining the ink deposition order, the high-mass positive ion analysis of the inks revealed an interesting phenomenon. Particularly for longer drying times, positive highmass MCR images revealed a separate component that ap-

Page 6 of 9

peared as a diffuse “halo” around the P ink. The progression of this “halo” from a faint diffuse region at 1 minute drying time to a more distinct zone along the P ink for 1 hour and 1 day drying times is illustrated for the Bvert1-Phoriz2 intersection in Figure 3 and can be further observed for other intersections involving the P ink in Figs S-6, S-12, and S-13. Loadings for these components show that the feature is due to the peaks at m/z 230, 231, 274 and 275 (Fig. 3d).

Figure 3. MCR scores images (yellow pixels have higher scores than blue) for high-mass positive ions from Bi3++ analysis of the Bvert1-Phoriz2 intersection after drying times of 1-minute (a), 1hour (b), and 1 day (c). Representative MCR loadings for these components (d).

Filenkova et al previously observed separation between the ink vehicle (solvent) and crystal violet dye in their study of inkjet-printed formulations, observing greater spread of the solvent on multipurpose paper than on glossy photo paper.11 Thus, the “halo” observed around the P ink may be the result of a unique solvent used in this formulation. It is known that various guanidines are used to raise the pH of inks containing acidic dyes, and these guanidines often have characteristic masses in the mid-200s with even-numbered masses due to the nitrogen in their structures.9 Although the peaks at m/z 230 and 274 do not coincide with previously reported guanidines in ballpoint inks, these peaks may arise from a similar type of additive, with tentative peak assignments of C12H28N3+ (-22.8 ppm) at m/z 230.22 and C14H32N3+ (-6.2 ppm) at m/z 274.24. If this halo is the result of diffusion or absorption from the P ink into the paper, it is somewhat peculiar that the spread of the m/z 230 and 274 peaks did not occur to the same extent for 1-minute drying times as it did for 1-hour and 1-day drying times. All samples were all analyzed together, 5 days after the deposition of the original lines and so these inks should have had the similar amounts of time to diffuse. If this drying time dependence is replicated in future studies, it may be that other solvent components in the B and S inks inhibit the diffusion of these components from the P ink. For the 1-hour and 1-day drying times, diffusion from the P ink would be temporally

6


Page 7 of 9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


removed from the presence of these S and B components as they would evaporate prior to deposition of the P ink as a second ink, and would not have been present for the first hour or day when the P ink was deposited first. Unique features such as the P halo could provide additional clues to the identification of different inks, where some ink formulations might have similar dyes but different solvents. However, although the diffuse halo clearly ran alongside the P ink, it could confound interpretation of ink deposition order, as the halo appeared interrupted regardless of whether the P ink was written first or second. When the P ink is written first, this may be reasoned to be the result of masking of the halo molecules by the second ink. When the P ink is written second, this may be reasoned as resistance to the spread of these molecules due to the other ink changing the absorption properties of the paper. While these are reasonable conjectures, the presence of interrupted halos could confound jurors in reaching a conclusion, since in this work, it always appeared that the halo, associated with the P ink, was “below” since it was interrupted. Aging effects in intersections of like inks. The intersections of like inks (B over B, P over P, S over S) were analyzed for samples with 1-day dry time using Bi3++ primary ions and positive secondary ions. This was done to determine whether the two inks could be differentiated, pointing to either description of the age of the inks, or to a “false positive” situation where inks of the same type appear different. When the B and S inks were intersected with themselves, PCA and MCR of the images revealed differences between the vertical and horizontal lines. No distinction was evident for the P ink intersected with itself. The distinction of the first and second B inks was illustrated most efficiently by full-mass PCA after mean centering. PC1 (not shown) contrasted paper from the overall ink crossing (+ shape) describing 77.44% of the variance. This was a common occurrence for PCA of dissimilar intersections as well. The difference between vertical B and horizontal B inks appeared on PC5, which described 0.73% of the variance (Fig. 4). For comparison, about 9.5% of the variance was described for intersections of P and B inks and S and B inks in PCA models run under the same conditions. Therefore, the differences observed for the self-crossings were relatively small. Due to the 1 day elapsed drying time, the vertical B ink was 5 days old and the horizontal B ink was 4 days old when the intersection was analyzed. The older vertical ink was characterized by positive scores (yellow-red colors) and the newer horizontal ink was characterized by negative scores (blue colors) in Fig. 4a. The loadings in Fig. 4b indicated that the difference was mainly due to differences in the dye peaks between m/z 316 and 372, with the older ink (positive loadings) having greater abundance of peaks at m/z 316, 330, 344 and 358, and the newer ink having greater abundance of peaks at m/z 340, 356 and 372. Additional low-mass peaks also contributed to the distinction. This included higher signal in the older ink of peaks at m/z 73 and 147, indicative of the common surface contaminant PDMS. Siegel et al analyzed blue and black ballpoint inks using laser desorption-ionization mass spectrometry and found that the Crystal Violet (Basic Violet 3) dye common to these inks degrades via oxidative demethylation, with the m/z 372 peak decreasing over time, and peaks at m/z 358, 344, 330, 316, and

302 increasing over time due to the successive loss of methyl groups.19 This is consistent with the present ToF-SIMS results where greater intensity of the higher-mass peaks was observed in the newer B ink and greater intensity of the lower-mass peaks was observed in the older B ink (Fig. 4).

Figure 4. PCA scores (a) and loadings (b) illustrating differences between the first and second B inks for the Bvert1-Bhoriz2 selfcrossing with 1-day drying time, using full-mass positive ions from Bi3++ analysis. Mean centering was used in addition to Poisson scaling and normalization.

PCA for the 1-day S ink self-crossing was more complicated, having three PCs distinguishing the vertical and horizontal inks (Fig S-16). These PCs cumulatively described about 1.2% of the total variance after mean centering. The scores images were not as sharp, possibly due to what appeared to be several different trends. As with the B ink, all 3 distinctive PCs for the S ink had higher intensity of higher mass m/z 356-358 and 372 peaks for the newer S ink and had higher intensity of lower mass peaks at m/z 316, 330, and 344 for the older S ink. However, the PCs differed in the many peaks < m/z 200 that contributed to the loadings. Among these, differences in PDMS were again observed, but in this case more PDMS was observed along the newer S ink line. Additionally, within the low-mass peaks it appeared that the older S ink had greater intensity of the S-characteristic m/z 91 and 117 peaks. CONCLUSIONS. ToF-SIMS analysis of the three random-

7



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 9

The Supporting Information is available free of charge on the ACS Publications website. Goacher-Supporting Information (PDF)

ly-chosen black ballpoint inks studied here confirmed that ToF-SIMS is able to distinguish between two different inks used in each ink intersection. While further experimentation would be required to determine if it is a general rule, this differentiation was mainly based on positive ion peaks < m/z 200 and negative ion peaks > m/z 50. While different inks were clearly distinguished, the deposition order, or writing sequence, could not be correctly determined for all ink pairings. The PapermateTM brand ink dominated all of its intersections, resulting in clear, but incorrect conclusions when the P ink was below the S and B inks. Dominance of the StaplesTM brand ink over the BicTM ink was also observed, producing the incorrect order of deposition when the S ink was under the B ink. These alarming incorrect results indicate that further study is needed to more fully understand the limitations in determining the correct ink deposition order with ToF-SIMS. The incorrect deposition orders were obtained using positive and negative secondary ions, and were reproduced with several different primary ions. Clearer results were not obtained with the popular cluster ions despite their lower energy per incident atom, and presumably greater surface sensitivity. Analysis of the ink intersections with imaging MCR and PCA provided a quick, unbiased method for identifying patterns in the samples. However, in the determination of deposition order, MCR images sometimes appeared ambiguous, and the analysis of selected regions of interest provided a more definitive answer regarding correct or incorrect deposition order. Multivariate examination of high mass positive ion images was often unable to distinguish different inks within the ink lines, but did reveal the diffusion of selected components from the P ink into the surrounding paper. This spread appeared to depend on the drying time. The formation of a “halo” around certain inks may help in identifying different ink formulations. However, in this case, interruption of the diffuse halos regardless of the ink deposition order could cause confusion regarding writing sequence. Although further research is needed, it is observed in this data that the presence of an interrupted halo does not indicate for or against the correct deposition order. Finally, intersections of identical inks with just one day difference in drying time revealed the sensitivity of the ToFSIMS method to the aging of the ink dyes. This observation could help in cases where documents are altered using the same ink but at a later date. Further research into this phenomenon would be required to understand the kinetic profile of this ageing and whether the observed differences remain after the samples have aged for longer than the 5 days in this study.

AUTHOR INFORMATION Corresponding Author * [email protected], Fax (716) 286-8254.

Author Contributions All authors contributed to the design of the experiment. LGD performed preliminary analysis on positive secondary ions and KCL performed preliminary analysis on negative secondary ions for 1-minute samples comparing different primary ions and different mass ranges. REG completed these analyses and performed analyses of the 1-minute, 1-hour and 1-day samples. REG drafted the manuscript and all authors have reviewed and approved the final version of the manuscript.

ACKNOWLEDGMENT We gratefully acknowledge funding from the Niagara University Research Council and Academic Center for Integrated Sciences. We also appreciate access to the argon cluster ion source provided by Tong Leung at the University of Waterloo. REG further acknowledges an adjunct professor appointment at the University at Buffalo, which permits affordable access to the ToF-SIMS there. Lastly, the preliminary work on this project by Samantha Livingston is recognized.

REFERENCES (1) Calcerrada Calcerrada, M.; Garcia-Ruiz, C. Anal. Chim. Acta 2015, 853, 143-166. (2) Spagnolo, G. S. Forensic Sci. Int. 2009, 164, 102–109. (3) Leung, S. C.; Leung, Y. M. Sci. Justice 1997, 37, 197–206. (4) Braz, A.; López-López, M.; García-Ruiz, C. Forensic Sci. Int. 2013, 232, 206–212. (5) Silva, C. S.; de Souza L. B., F.; Pimentel, M. F.; Pontes, M. J. C.; Honorato, R. S.; Pasquini, C. Microchem. J. 2013, 109, 122-127. (6) Lee, J.; Kim, S. H.; Cho, Y.-J.; Nam, Y. S.; Lee, K.-B.; Lee, Y. Surf. Interface Anal. 2014, 46, 317-321. (7) Pachuta, S. J.; Staral, J. S. Anal. Chem. 1994, 66, 276-284. (8) Coumbaros, J.; Kirkbride, K. P.; Klass, G; Skinner, W. Forensic Sci. Int. 2009, 193, 42-46. (9) Denman, J. A.; Skinner, W. M.; Kirkbride, K. P.; Kempson, I. M. Appl. Surf. Sci. 2010, 256, 2155–2163. (10) Sodhi, R.N.S.; Sun, L. Sain, M.; Farnood, R. J. Adhes. 2008, 84, 277-292. (11) Filenkova, A.; Acosta, E.; Brodersen, P. M.; Sodhi, R. N. S.; Farnood, R. Surf. Interface Anal. 2010, 43, 576-581. (12) He, A.; Karpuzov, D.; Xu, S. Surf. Interface Anal. 2006, 38, 854–858. (13) Arnaud, C. Mass Spec Imaging Goes to Court. Chem. Eng. News, Jun 6, 2011, p. 42. (14) Lee, J.; Kim, S. H.; Cho, Y.-J.; Nam, Y. S.; Lee, K.-B.; Lee, Y. Surf. Interface Anal. 2014, 46, 317-321. (15) Lee, J.; Nam, Y. S.; Min, J.; Lee, K.-B.; Lee, Y. J. Forensic Sci. 2016, 61, 815-822. (16) Ozbek, N.; Braz, A.; López-López, M.; García-Ruiz, C. Forensic Sci. Int. 2014, 234, 39–44. (17)http://www.staples.com/Paper-Mate-Profile-RetractableBallpoint-Pens/product_SS1028017 (accessed Jul, 2016). (18)http://www.staples.com/Staples-Motiva-Advanced-InkRetractable-Ballpoint-Pens-Fine-Black-12-Pack/product_326482 (accessed Jul, 2016) (19) Siegel, J.; Allison, J.; Mohr, D.; Dunn, J., Talanta 2005, 67, 425–429.

ASSOCIATED CONTENT Supporting Information The following may be found in the Supporting Information: Example single-ROI PCA and % error calculation; Example MCR model; All-ROI PCA models for positive and negative ions with the Bi3++ primary ion; Summary figures for all drying times, primary ions, secondary ion polarities, and mass ranges studied; Raw ToF-SIMS spectra of the pure inks and paper for all primary ions and secondary ion polarities studied; PCA model for the selfcrossing of the S ink.

8


Page 9 of 9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Insert Table of Contents artwork here

9


Challenges Determining the Correct Deposition ... - ACS Publications

Recommend Documents