Trade in Splitless for Split: Faster Speed & More Miles Analyzing 3-MCPD / Glycidyl esters

In my recent work with analysis of 3-MCPD and glycidyl esters, I reviewed the available methods by AOCS and they list splitless or pulsed splitless injection. Splitless injection theoretically transfers all of the sample onto the column, allowing the analyst to achieve trace levels of detection. However, there are limitations. Both splitless hold-time and focusing (or lack thereof) have an effect on peak shape (broadening and tailing) when set incorrectly. Another consideration is the presence of derivatization agents which can damage and decrease column lifetime. The split injection (shoot-and-dilute approach) overcomes these limitations, however, since the majority of the sample is vented there is concern about achieving adequate sensitivity. There is an excellent blog about using this approach for the analysis of used motor oil here!
Before I compare split versus splitless, I wanted to make sure the splitless parameters are set up correctly, namely the initial oven temperature. As I mentioned, focusing (solvent trapping) can make a large difference in chromatographic performance, especially for early eluting peaks, therefore it is recommended that the initial temperature to be ~20 °C below solvent’s boiling point. In this case, the solvent is isooctane (b.p. 99°C) and the initial oven temperature is set to 85°C. However, the peaks of interest are later eluting, therefore, the focusing might not affect those peaks at all. I have evaluated higher initial temperatures to find the best setting for both split and splitless (Figure 1).

Figure 1: 3-MCPD peak widths at different initial oven/inlet (PTV) temperatures

While using split injection, increased temperature improves peak width slightly, while splitless injection has a clear improvement in the peak width around 105-110°C. Therefore, I’ve used split with a temperature of 120°C and splitless with a temperature of 110°C and 120°C.

To further evaluate the split injection, I’ve compared instrumental limits of detection based on calibration curve, calculated as follows: LOD = 3sy/m, where sy is standard error of y-value and m is the slope of the calibration curve. But first I evaluated simply how the calculated values of 3-MCPD compare (Figure 2). The calculation was performed as described in AOCS method Cd 29c-13 using spiked 3-MCPD-d5.

Figure 2: Comparison of calculated values of 3-MCPD. Spiked amount: 0.04-12 mg/kg

When we look at the calibration curves (Figure 3), we see the same story – split and splitless give us very similar results.

Figure 3: Calibration curves for analysis of 3-MCPD

And lastly, when we examine calculated LODs using both free 3-MCPD and 3-MCPD ester (in the matrix), the results are comparable:

Injection Free 3-MCPD
LOD (mg/kg)
3-MCPD ester
LOD (mg/kg)
Splitless (110°C) 0.19 0.14
Splitless (120°C) 0.12 0.17
Split (120°C) 0.12 0.13


The limits of detection are higher than what is required for 3-MCPD and glycidyl esters detection in infant formula/food, however, these limits are more limited by the instrument itself (single quad GC-MS) and they would be vastly improved if GC-MS/MS was used.

In conclusion, the split injection is a viable alternative to splitless injection with all the benefits of the speedier run due to higher initial temperature and longer column lifetime due to dilution of the matrix.

Next time I’ll look into running the method with regular split injector (instead of PTV).

Leave a Reply