Using a sol-gel method, we have fabricated poly(dimethylsiloxane) (PDMS) microchips with SiO2 particles homogeneously distributed within the PDMS polymer matrix. These particles are approximately 10 nm in diameter. To fabricate such devices, PDMS (Sylgard 184) was cast against SU-8 molds. After curing, the chips were carefully removed from the mold and sealed against flat, cured pieces of PDMS to form enclosed channel manifolds. These chips were then solvated in tetraethyl orthosilicate (TEOS), causing them to expand. Subsequently, the chips were placed in an aqueous solution containing 2.8% ethylamine and heated to form nanometer-sized SiO2 particles within the cross-linked PDMS polymer. The water contact angle for the PDMS-SiO2 chips was approximately 90.2 degrees compared to a water contact angle for Sylgard 184 of approximately 108.5 degrees . More importantly, the SiO2 modified PDMS chips showed no rhodamine B absorption after 4 h, indicating a substantially more hydrophilic and nonabsorptive surface than native PDMS. Initial electroosmotic mobilities (EOM) of (8.3+/-0.2)x10(-4) cm2/(V.s) (RSD=2.6% (RSD is relative standard deviation); n=10) were measured. This value was approximately twice that of native Sylgard 184 PDMS chips (4.21+/-0.09)x10(-4) cm2/(V.s) (RSD=2.2%; n=10) and 55% greater than glass chips (5.3+/-0.4)x10(-4) cm2/(V.s) (RSD=7.7%; n=5). After 60 days of dry storage, the EOM was (7.6+/-0.3)x10(-4) cm2/(V.s) (RSD=3.9%; n=3), a decrease of only 8% below that of the initially measured value. Separations performed on these devices generated 80,000-100,000 theoretical plates in 6-14 s for both tetramethylrhodamine succidimidyl ester and fluorescein-5-isothiocyanate derivatized amino acids. The separation distance was 3.5 cm. Plots of peak variance vs analyte migration times gave diffusion coefficients which indicate that the separation efficiencies are within 15% of the diffusion limit.