| Textile fibers found as trace evidence at crime scenes often provide confirmation or leads in criminal investigation. The development of forensic analytical methods for robust automated micro-extractions of dyes from trace fibers and analysis of extracts by capillary electrophoresis (CE) with UV/visible diode array (DAD) or mass spectrometric (MS) detection is described.;A liquid handling robot was employed to vary experimental conditions (solvent composition, temperature, and extraction time). Robust optimum micro-extraction conditions were identified for four subclasses of acid and basic dyes (anthraquinone, azo, methine, and metal complex) from nylon and acrylic, for direct, reactive, and vat dyes from cotton, and for disperse dyes from polyester. Acid dyes were extracted by water:pyridine:aqueous ammonia solvents; basic dyes were extracted with formic acid and water; disperse dyes (azo and anthraquinone) were extracted from polyester by chlorobenzene. Conditions for extraction of dyes from cotton depend on the dye type.;CE methods were developed for separation and identification six major textile dye classes. Separations of anionic dyes from cotton and nylon used a buffer containing 15 mM ammonium acetate in acetonitrile-water (40:60, v/v) at pH 9.3. Vat dyes from cotton were separated with sodium dithionite added to the same buffer. Cationic dyes from acrylic were separated using a 45 mM ammonium acetate buffer in acetonitrile-water (60:40, v/v) at pH 4.7. Disperse dyes from polyester were separated using a non-aqueous buffer of 80 mM ammonium acetate and 75% acetonitrile in methanol.;Although CE/DAD is viable for single fiber lengths to one cm, the more sensitive and selective CE/MS approach was needed for analysis of dye extracts from fibers down to 2 mm in length. The combination of MS (and UV/visible) information provides semi-quantitative estimates of dye amounts as well as qualitative information for dye identification.;The final study used UV/visible microspectrophotometry with linear discriminant analysis to confirm statistical significance of differences among spectra from different black and blue inks. A classification accuracy of 97.9% was achieved for a set of 10 black and 9 blue inks, demonstrating the potential usefulness of this approach for interpreting spectra from questioned documents. |
