UMaine CCI Stable Isotope Laboratory Procedures

Measuring the isotopic composition of high latitude and high altitude snow and ice is one of the routine analyses used in paleoclimate reconstructions. Measuring water isotope signals in modern precipitation and other natural waters allows us to better understand the meteoric processes by which the isotopic signature is imparted to those historical records. Measuring both signals allows us to calibrate and interpret better those historical climate records contained in various ice cores and snow pits.  The Climate Change Institute at The University of Maine, Orono, ME. uses a Picarro L-2130-I Isotopic H2O Laser Ring Down Spectrometer to measure the isotopes of current and ancient waters by laser adsorption spectrophotometry. Snow, ice and water is recovered by Climate Change Institute (CCI) researchers from all over the globe and is stored at the CCI, as either frozen snow and/or ice in a -20C freezer facility, or it is melted and then stored at room temperature in glass vials with special polyvinyl conical inserts in the caps.  These caps prevent evaporation and fractionation of the samples during storage, analysis and sample archiving. Our tests have shown that there is no measureable fractionation in samples that have been archived in these vials for at least 20 years. The Picarro instrument is fitted with an autosampler and vaporizer peripheral. 1.6 ml of sample water is aliquoted from the storage vials, to a 2 ml vial with a septum and placed in the autosampler. A high tolerance syringe then delivers 5 ul of water to the vaporizer and sequentially measures both the delO18 and delH/D as water vapor in a heated ring down chamber until there is no significant sample memory from the previously injected sample. Samples are reported to the universally accepted international standard VSMOW (Vienna Standard Mean Ocean Water) and expressed in delta notation normalized to the VSMOW-SLAP scale. 1 Three internal lab standards with widely varying isotope signatures have been calibrated to VSMOW, NIST-SLAP (Standard Light Antarctic Precipitation) and NIST-GISP (Greenland Ice Sheet Precipitation). 2 We have also calibrated these internal standards and the instrument to the consensus values of IAEA-OH-1, IAEA-OH-2, IAEA-OH-3 and IAEA-OH-4. 3 (see Table 1)  The lab standards are used on a daily basis for daily calibration purposes and instrument monitoring. The standard deviation of measurement is reported as +/- 1 per mille for delta D/H and +/- 0.1 per mille for delta 18O. Although individual analytical runs are often better than reported we report the more conservative error. We base this determination on instrument factory specifications but also because several of the listed secondary standards are only known to .1 per mille for 18O and 1 per mille for D/H. (see Table 1)  UMaine’s participation in 2 IAEA interlaboratory comparisons using our calibration, have shown that the error associated with the calibration is on the order of what we report. 2, 3 The natural variations in both modern and historic water isotope signals exceed these reported errors and obviate underestimating the standard uncertainty.

Update 2024

In 2023-24 we completed an interlaboratory comparison with two other isotope laboratories to improve our measurements of highly depleted Antarctic samples. Given the different analytical procedures and instrumentation used in each lab to calibrate standards and measure samples, we consider the observed offset among the lab results to be acceptable. However, we found a systematic more depleted offset in UMaine reported values for dD.

Discussion.

The dD offset may be caused by 1) memory effects inherent in laser ring down spectrophotometry measurements, 2) larger than estimated standard error, and/or 3) linear instabilities inherent in continuous flow mass spectrometry (particularly for dD) when originally calibrating our internal standards with earlier instrumentation.

1)    Memory effect in laser ring down spectrophotometry is never completely removed. However, our experiments have shown that with multiple sample injections for both d¹⁸O and dD that the measured value goes asymptotic within several sample injections. The number of repeat injections depends on how isotopically far apart adjoining samples are and how much water vapor is injected into the ring-down chamber. We adjust the number of injections to allow for high throughput analysis by reducing the volume of water vapor carried into the ring-down chamber to approximately 14,000 ppm. Memory was determined not to be a cause for the offset.

2)    The UMaine dD averaged standard deviation is +/- 1.00 ‰. This is comparable to the generally accepted and reported standard deviation for dD measurements in other laboratories. Since this is an average value, the standard deviation can range from +/- 0.5 ‰ for enriched measurements, to +/- 1.5 ‰ or greater, for more depleted values. Greater error for more depleted measurements may be introduced by calculating a calibration regression line that is not constrained by a more highly depleted end member. Therefore, we developed and calibrated a new internal standard SPI (South Pole Ice) that is approximately 60 ‰ more depleted in dD and 8 ‰ more depleted in d¹⁸O than our previously used most depleted internal standard.

3)    The calibration for our Picarro Laser Ring Down Spectrophotometer was heldover from a previous calibration conducted on an Isoprime continuous flow mass spectrometer linked to a Eurovector chromium reduction peripheral. We used the same calibration between the older and the newer instrument to allow for continuity, comparability and agreement with all samples analyzed by our lab, within our lab through both time and space. This was verified by repeated analysis of 1000’s of samples that were run both on the previous instrument and the new instrument. All repeat analyses were comparable and spot on. If the original calibration was influenced by (unknown to us at the time) instrumental non-linearities we decided to recalibrate all our internal standards on the current instrument to update our calibration curve.

Recalibration results.

The recalibration showed negligible change for d¹⁸O and therefore no update is applied.

For any UMaine dD analyses made prior to April 1, 2024, sample values can be updated to the new dD calibration following a y = mx + b linear regression where:

y = newly updated dD value

m = 0.98797495

x = older delta dD value

b = 1.15357797

An example follows:

If the previous dD value is -205.79 ‰, then the updated value is -205.79 * 0.98797495 + 1.15357797 = -202.16 ‰

Because the updated values are within the range of standard deviation for dD measurements, this adjustment is optional.