Rationale
The accuracy determined in the routine analysis of water isotopes (δ17O, δ18O, δ2H) using cavity ring-down spectroscopy is greatly affected by the memory effect (ME), a sample-to-sample carryover that biases measurements. This study aims to develop a simple method that rapidly removes the ME.

Methods
We developed a method, designed for the Picarro L2140-i, that removes the ME by injecting small amounts of water with an extreme isotopic value (“kick”) in the opposite direction of the ME. We conducted 11 experiments to identify the optimal kick for pairs of isotopically enriched and depleted samples. Once quantified, the optimal kick was used to create an ME-free, unbiased calibration curve, which was verified using international and internal lab standards.

Results
Our kick method removes the ME very efficiently in half the time it takes for experiments without a kick. The optimal number of kick injections required to minimize stabilization time between standards of different compositions is three injections of δ2H ≈ −1000‰ water per a 100‰ difference between standards. Three runs of routine measurements using the kick method resulted in uncertainties of 0.03‰, 0.2‰, and 5 permeg for δ18O, δ2H, and 17O-excess, respectively.

Conclusions
This study demonstrates a new method for rapidly removing the ME. Our kick protocol is a readily available, cheap, and efficient approach to reduce instrumental bias and improve measurement accuracy.