Minimizing ultraviolet noise due to mis-matches between detector flow cell and post column mobile phase temperatures in supercritical fluid chromatography: Effect of flow cell design

• Suppression of UV noise caused by mis-match between the mobile phase temperature and the flow cell temperature.
• Noise with most older flow cells and detectors used in SFC can be significantly improved.
• Conical flow path not as important as in HPLC since refractive index in SFC is much lower.
• Thermal isolation of the flow cell from the optical bench is important.
• Quantify a 0.1% trace component with S/N > 10 at 80 Hz (200 bar).

A mis-match between the post-column mobile phase temperature and the UV detector flow cell temperature can cause significant UV noise in supercritical fluid chromatography (SFC). Deviations as little as 5 °C can increase noise as much as 5 times, making the detector unsuited for trace analysis. Two approaches were used to minimize this noise. When a flow cell was in direct thermal contact (metal on metal) with the detector optical bench, the mobile phase temperature was actively controlled to the measured flow cell temperature, by using one of the heat exchangers (HX) in the column compartment. However, with some older, but still widely used flow cell designs, this required repeated, hourly monitoring of the flow cell temperature and repeated manual adjustment of the heat exchanger temperature, due to thermal drift. Flow cell design had a strong influence on susceptibility to this thermally induced noise. Thermally insulating the flow cell from the optical bench made some cells much less susceptible to such thermally induced noise. Five different flow cells, some insulated, some un-insulated, were evaluated. Most had a truncated conical flow path, but one had a cylindrical flow path. Using either approach, the ASTM noise, with a 10 mm, 13 μL conical flow cell, could be optimized to ≈0.007 mAU at 2.5 Hz, in SFC, which is very near the 0.006 mAU manufacturer's specification for HPLC. The insulated version of this flow cell required far less optimization, compared to the un-insulated version. At 150 bar, an experimental 3 mm, 2 μL flow cell, with only one side insulated, yielded noise slightly too high (≈0.16–0.18 mAU) for trace analysis, at 80 Hz. However, at 200 bar, noise at 80 Hz was <0.06 mAU, which should allow quantification of a 1 mAU tall trace component with a signal to noise ratio (S/N) >10. Even partially un-insulated, this flow cell design was much less susceptible to thermally induced noise. Further insulating this flow cell design failed to improve performance.