Troubleshooting Injection Volume Variation in the Concurrent Solvent Recondensation - Large Volume Sample Injection Technique

I've started receiving a number of questions about the large volume injection technique called Concurrent Solvent Recondensation - Large Volume Sample Injection. Most recently, a question came through the Tech Service group about injection volume variation. I had encountered a similar problem when working on the 50 µL injection for the combined 1,4-Dioxane and Nitrosamines (EPA methods 522 and 521) application note titled "Combined Determination of 1,4-Dioxane and Nitrosamine Contaminants in Drinking Water"

I had written up an examination of the problem, but it didn't make it into the final draft. Fortunately, this blog is an ideal platform to share the work. 

The default speed for a fast autosampler injection using a 10 µL syringe is 6,000 µL/min. When you increase the syringe volume to 100 µL and leave the ALS configured for a fast injection, the injection speed increases to 60,000 µL/min. During method development, we found that a standard single taper liner with wool did not provide enough packing material to arrest the 60,000 µL/min, 50 µL injection of dichloromethane. The high inlet flow pushes the large volume of liquid sample through channels in the wool, causing some of sample to enter the head of the column as a liquid. This causes peak fronting and splitting; two defects which are not amenable to reproducible integration and quantification. We addressed this problem by slowing the injection speed to 4,000 µL/min and adding more wool to the single taper liner, bringing the total mass of in situ deactivated quartz wool to approximately 15 mg.

One of the major challenges of combining the highly volatile compounds from EPA Method 522 with the less volatile nitrosamines from EPA 521 was minimizing the retention time variability of the early eluting compounds (THF, 1,4-dioxane, and NDMA). Early runs collected under the final GC analytical conditions showed great retention time variation for these compounds. Figure 1 shows the retention time variation of 1,4-dioxane-d8 over the course of nine sequential injections to be approximately 0.5 minutes. This is unacceptable for any analysis, but especially so for the trace-level SIM analysis.

Figure 1 – 1,4-Dioxane-D8 retention time variation over the course of nine 50 µL injections

By watching the injection sequence, it was determined that the retention time variation was largely due to variations in the injected volume, indicated by bubbles of various sizes present in the syringe from injection to injection. Adding a viscosity delay to the ALS program slightly improved retention time reproducibility, but the root cause of the problem was determined to be linked to the large volume syringe plunger design. Figure 2 shows the problem syringe (bottom) and the solution (top).

Figure 2: Switching to the more rugged, hermetically sealed fixed-needle ALS syringe (top) greatly increased retention time reproducibility for the more volatile components by more precisely controlling the injected volume. Top 100 µL gas-tight syringe with fixed 26/23 gauge needle (cat.# 005668); bottom: 100 µL syringe with removable 23 gauge needle (cat.# 005665);

Keeping the short viscosity delay and switching to a fixed needle gas-tight syringe with a PTFE-tipped plunger significantly improved the retention time stability of the volatile components. In addition to the PTFE tip, the gas-tight syringe plunger maintains its diameter down the entire length of the syringe barrel. Still, internal standard and surrogate compounds eluting close to SIM window borders were occasionally lost due to retention time drift. Figure 3 illustrates the reduced retention time variation shown by tetrahydrofuran-d8 following the change in syringe design.

Figure 3: Tetrahydrofuran-d8 retention time variation following the change in autosampler syringe design used in the CSR-LVSI experiments.

Eluting more than a minute earlier than 1,4-dioxane-d8, tetrahydrofuran-d8 showed even more retention time variation than 1,4-dioxane. The peaks are still eluting in a 6 - 8 second window, but this is a large improvement over the 30+ second spread we were seeing in our initial results. This indicates there is still some volume variation in the injections, but the internal standard correction should be sufficient to account for this.

The less volatile analytes, such as N-nitroso-di-n-propylamine-d14 did not display the same retention time variation, indicating analyte focusing was occurring rather than solvent focusing (Figure 4).

Figure 4: N-nitroso-di-n-propylamine-D14 retention time variation over the course of nine 50 µL injections. Being much less volatile, the thick film of the analytical column traps this compound until the oven reaches a much higher temperature than 1,4-dioxane or THF. This makes its retention time dependent on the oven program, not the injected solvent volume. The peak area is still injection volume dependent, and its use as the internal standard for the nitrosamines normalized the calculated recoveries for the small variations in injection volume.