Correction algorithm for online continuous flow δ 13 C and δ 18 O carbonate and cellulose stable isotope analyses

For the LSVEC reference material, the precision achieved is better than reported in Stichler [1995], which was 0.15 permil for d13C and 0.25 permil for d18O.

“For the LSVEC reference material, the precision achieved is better than reported in Stichler [1995], which was 0.15 and 0.25& for d C and d O , respectively.”

• cites Stichler, 1995Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• AssumptionRests on an assumption: The known d13C and d18O values for the working standard a-celluloses SAC and AKC with respect to V-PDB and V-SMOW are accurate.

For carbonate d18O data, median precision of validation estimates for two reference materials and two calibrated working standards is 0.07 permil.

“For carbonate d 13 C (d 18 O) data, median precision of validation estimates for two reference materials and two calibrated working standards is 0.05& (0.07&);”

• Also stated in: § 5. Discussion · plain

  “For the carbonate materials we analyzed, we found that the median validation precisions across the carbonates analyzed are 0.05 and 0.07& for d C and d O , respectively (Table 1).”

• AssumptionRests on an assumption: The known d13C and d18O values for the working standard a-celluloses SAC and AKC with respect to V-PDB and V-SMOW are accurate.

These biases are small relative to the calibrated ranges in δ13C and δ18O across all materials analyzed (16.1 and 19.1‰, respectively).

“but these biases are again small relative to the calibrated ranges in d C and d O across all materials analyzed (16.1 and 19.1&, respectively; Table 2 and Figures 1c and 1d)”

These validation results are within the 5th-95th percentile range of subsequent routine runtime validation exercises in which one working standard is used to calibrate the other.

“These results are within the 5th-95th percentile range of subsequent routine runtime validation exercises in which one working standard is used to calibrate the other.”

For carbonate d13C data, median bias is 0.04 permil over a range of 49.2 permil.

“median bias is 0.04& (0.02&) over a range of 49.2& (24.3&).”

• Also stated in: § 5. Discussion · plain

  “However, the median biases (0.04 and 0.02& for d C and d O , respectively; Table 1) are small relative to the known range across in d C and d O across working standards (49.2 and 24.3&, respectively; Table 1 and Figures 1a and 1b).”

• AssumptionRests on an assumption: The known d13C and d18O values for the working standard a-celluloses SAC and AKC with respect to V-PDB and V-SMOW are accurate.

For carbonate d18O data, median bias is 0.02 permil over a range of 24.3 permil.

“median bias is 0.04& (0.02&) over a range of 49.2& (24.3&).”

• Also stated in: § 5. Discussion · plain

  “However, the median biases (0.04 and 0.02& for d C and d O , respectively; Table 1) are small relative to the known range across in d C and d O across working standards (49.2 and 24.3&, respectively; Table 1 and Figures 1a and 1b).”

• AssumptionRests on an assumption: The known d13C and d18O values for the working standard a-celluloses SAC and AKC with respect to V-PDB and V-SMOW are accurate.

For carbonate d13C data, median precision of validation estimates for two reference materials and two calibrated working standards is 0.05 permil.

“For carbonate d 13 C (d 18 O) data, median precision of validation estimates for two reference materials and two calibrated working standards is 0.05& (0.07&);”

• Also stated in: § 5. Discussion · plain

  “For the carbonate materials we analyzed, we found that the median validation precisions across the carbonates analyzed are 0.05 and 0.07& for d C and d O , respectively (Table 1).”

• AssumptionRests on an assumption: The known d13C and d18O values for the working standard a-celluloses SAC and AKC with respect to V-PDB and V-SMOW are accurate.

Of 12 comparisons, 5 are statistically distinguishable from zero bias.

“include 5 of 12 comparisons that are statistically distinguishable from zero bias”

For a-cellulose d18O data, median bias is 20.10 permil over a range of 19.1 permil.

“median bias is 0.13& (20.10&) over a range of 16.1& (19.1&).”

• AssumptionRests on an assumption: The known d13C and d18O values for the working standard a-celluloses SAC and AKC with respect to V-PDB and V-SMOW are accurate.

In practice over additional routine tests, the amplitude correction is rarely significant at P < 0.10 for both correction standards.

“In practice over additional routine tests (last two columns of Tables 3 and 4), the amplitude correction is rarely significant at P < 0.10 for both correction standards.”

If validation precisions are not already small, routine cross-validated precision estimates are improved by up to 50% for d13C and 80% for d18O.

“If validation precisions are not already small, routine cross-validated precision estimates are improved by up to 50% (80%).”

Differential reaction and equilibration times for batch acid versus sequential acid injection cause a systematic drift in d18O measurements.

“analyses of batch acid versus sequential acid injection and analyses (results not shown) suggest that differential reaction and equilibration times for the former procedure cause a systematic drift in the d 18 O measurements.”

In the single validation studies presented, some spline and amplitude corrected validation precisions are higher than raw precisions.

“In the results of the single validation studies presented (Tables 1 and 2), some spline and amplitude corrected validation precisions are higher than raw precisions”

Because precisions are small, the biases in some of the validation data sets (JTB1, EACARB) are statistically different from zero.

“Because precisions are small, the biases in some of the validation data sets (JTB1, EACARB) are statistically different from zero.”

• AssumptionRests on an assumption: The known d13C and d18O values for the working standard a-celluloses SAC and AKC with respect to V-PDB and V-SMOW are accurate.

For a-cellulose d18O data, median precision of validation estimates for one reference material and five working standards is 0.27 permil.

“For a-cellulose d 13 C (d 18 O) data, median precision of validation estimates for one reference material and five working standards is 0.11& (0.27&);”

• AssumptionRests on an assumption: The known d13C and d18O values for the working standard a-celluloses SAC and AKC with respect to V-PDB and V-SMOW are accurate.

For a-cellulose d13C data, median precision of validation estimates for one reference material and five working standards is 0.11 permil.

“For a-cellulose d 13 C (d 18 O) data, median precision of validation estimates for one reference material and five working standards is 0.11& (0.27&);”

• AssumptionRests on an assumption: The known d13C and d18O values for the working standard a-celluloses SAC and AKC with respect to V-PDB and V-SMOW are accurate.

For a-cellulose d13C data, median bias is 0.13 permil over a range of 16.1 permil.

“median bias is 0.13& (20.10&) over a range of 16.1& (19.1&).”

• AssumptionRests on an assumption: The known d13C and d18O values for the working standard a-celluloses SAC and AKC with respect to V-PDB and V-SMOW are accurate.

Validation study precisions and biases for both carbonate and cellulose experiments fall within the 5th-95th percentile ranges of the routine runtime precision and bias estimates.

“Validation study precisions and biases for both carbonate and cellulose experiments (Tables 1 and 2) fall within the 5th-95th percentile ranges of the routine runtime precision and bias estimates (Figures 2 and 3).”

Two-point precision is approximately that achieved after the drift and amplitude correction in the validation study and in routine cross-validation of many batch runs.

“Two-point precision is approximately that achieved after the drift and amplitude correction in the validation study and in routine cross-validation of many batch runs.”

Reversing the order of drift and amplitude corrections for both the carbonate and cellulose validation experiments gives similar validation precision and bias estimates reported in the abstract and as shown in Tables 1 and 2.

“Reversing the order of drift and amplitude corrections for both the carbonate and cellulose validation experiments gives similar validation precision and bias estimates reported in the abstract and as shown in Tables 1 and 2.”

The algorithm described here detects and corrects for three potential sources of systematic uncertainty in CF-IRMS data streams.

“Although the algorithm described here detects and corrects for three potential sources of systematic uncertainty in CF-IRMS data streams”

Measurements in cellulose pyrolysis are commonly subject to runtime isotopic drift, arising from changes in the state of the thermal conversion reactor as consumables are used and elemental sample carbon is added to the system.

“Measurements are also commonly subject to runtime isotopic drift, arising from changes in the state of the thermal conversion reactor as consumables are used and elemental sample carbon is added to the system [e.g., Young et al., 2011].”

• cites Young et al., 2011 sourceNot checked — no full text of the source

Instrument linearity for the carbonate system was confirmed to be less than 0.03 permil/nA over a range of approximately 2-14 nA.

“Prior to batch analysis, instrument linearity (dependence of d 18 O and d 13 C on signal at the collectors) was confirmed to be <0.03&/nAover a range of approximately 2-14 nA.”

If systematic errors in isotopic analyses can be detected, they can be modeled and corrected for, resulting in more precise and accurate results.

“In principle, if systematic errors in isotopic analyses can be detected, we can model and correct for them, resulting in more precise and accurate results.”

Dual C/O measurements on cellulose samples in a low-temperature pyrolysis system at 1080 degrees C may be subject to scale compression and bias in both isotopes.

“Similarly, dual C/O measurements on cellulose samples in a low-temperature pyrolysis system (10808C) [Werner et al., 1996] may be subject to scale compression and bias in both isotopes.”

• cites Werner et al., 1996 sourceNot checked — no full text of the source

The spline and two-point corrections are most valuable when raw precisions are poor.

“Results as a function of correction step (Tables 3 and 4) suggest that the spline and two-point corrections are most valuable when raw precisions are poor (Tables 3 and 4, last two columns).”

Uncertainty in the correction scheme presented here arises from both empirical and structural uncertainties.

“Uncertainty in the correction scheme presented here arises from both empirical and structural uncertainties.”

Separation of CO and N2 peaks was verified by analysis of the chromatogram.

“Separation of CO and N 2 peaks (a small amount of N, with characteristically low 30/28 ratio, may be present in industrial or extracted a-cellulose [Anchukaitis et al., 2008]) was verified by analysis of the chromatogram.”

There might be artificial skill in these corrections if raw precision is already low.

“suggesting that there might be artificial skill in these corrections if raw precision is already low (first four columns of Tables 3 and 4).”

Drift and scale-compression corrections are the most reliable and valuable correction steps.

“Analysis of the relative importance of correction steps suggests that drift and scale-compression corrections are most reliable and valuable.”

Correction for systematic error may enable these particular CF-IRMS systems to produce d13C and d18O carbonate and cellulose isotopic analyses with higher validated precision, accuracy, and throughput than is typically reported for these systems.

“may enable these particular CF-IRMS systems to produce d 13 C and d 18 O carbonate and cellulose isotopic analyses with higher validated precision, accuracy, and throughput than is typically reported for these systems.”

The algorithm corrects for scale compression, runtime drift, and amplitude effects in carbonate and cellulose oxygen and carbon isotopic analyses made on two online CF-IRMS systems using gas chromatographic separation.

“We describe an algorithm to correct for scale compression, runtime drift, and amplitude effects in carbonate and cellulose oxygen and carbon isotopic analyses made on two online continuous flow isotope ratio mass spectrometry (CF-IRMS) systems using gas chromatographic (GC) separation.”

• AssumptionRests on an assumption: The known d13C and d18O values for the working standard a-celluloses SAC and AKC with respect to V-PDB and V-SMOW are accurate.

The long-term systematic bias in corrected data is not significantly different from zero.

“and that the long-term systematic bias in corrected data is not significantly different from zero.”

No measurement blank correction was applied to carbonate analyses because routine analyses on empty, He-flushed and acidified sample vials give signals of 0.05-0.1 nA.

“No measurement blank correction has been applied, as routine analyses on empty, He-flushed and acidified sample vials give signals of 0.05-0.1 nA.”

Using working standards of known isotopic composition and similar chemical composition as the basis for corrections, and analyzing them in batches together with unknowns, automatically removes differential working-standard-versus-unknown-sample biases (the IT principle).

“Because we use working standards of known isotopic composition and similar chemical composition as the basis for these corrections, and because these working standards are analyzed in batches together with the unknowns, differential (working standard versus unknown sample) biases are automatically removed, an approach known as the principle of Identical Treatment (IT principle) [Werner and Brand,”

• cites Werner and Brand, 2001 sourceNot checked — no full text of the source
• cites Brand et al., 2014 sourceNot checked — no full text of the source

• AssumptionRests on an assumption: The working standards used for correction have chemical composition similar to the unknown samples.

The α-cellulose validation precisions are comparable to those achieved routinely in high-temperature pyrolysis systems with near-quantitative conversion of α-cellulose into analyte CO above 1400°C.

“These precisions are comparable to those achieved routinely in high-temperature pyrolysis systems with near-quantitative conversion of a-cellulose into analyte CO above 14008C [Leuenberger and Filot, 2007].”

• cites Leuenberger and Filot, 2007 sourceNot checked — no full text of the source

These biases may be negligible in practical measurement and application.

“and may be negligible in practical measurement and application”

Runtime isotopic drift in oxygen isotopic composition may arise in continuous flow preparation systems from systematic differences in the time between acidification and analysis of evolved CO2 analyte gas.

“Depending on analytical method, runtime isotopic drift in oxygen isotopic composition may arise in continuous flow preparation systems from systematic differences in the time between acidification and analysis of evolved CO 2 analyte gas [Spotl, 2011];”

• cites Spotl, 2011 sourceNot checked — no full text of the source

Amplitude or integrated peak area biases in measured isotopic composition may arise in the analysis of carbonate isotopic composition by continuous flow mass spectrometry.

“For example, it has been observed that amplitude or integrated peak area biases in measured isotopic composition may arise in the analysis of carbonate isotopic composition by continuous flow mass spectrometry [Spotl and Vennemann, 2003].”

• cites Spotl and Vennemann, 2003 sourceNot checked — no full text of the source

Instrument linearity for the cellulose system was tested daily and confirmed to be less than 0.03 permil/nA over the 1-10 nA range.

“Instrument linearity (dependence of d 18 O and d 13 C on signal amplitude at the collectors) was tested daily and confirmed to be <0.03&/nAover the 1-10 nA range.”

The algorithm was validated by correcting measurements of samples of known isotopic composition that were not used to estimate the corrections.

“We validate the algorithm by correcting measurements of samples of known isotopic composition which are not used to estimate the corrections.”

• Also stated in: § PUBLICATIONS · plain

  “We validate the algorithm in two online CF-IRMS contexts-measurement on carbonate-derived analyte CO 2 and measurement on a-cellulose-derived analyte CO-by applying corrections to samples of known isotopic composition but which were not used to estimate the corrections themselves.”

Results are similar to those achieved by Young et al. [2011].

“results are similar to those achieved by Young et al. [2011].”

• cites Young et al. [2011] sourceNot checked — no full text of the source

Empirical sources of uncertainty include the statistical estimation of the spline, amplitude, and two-point corrections.

“Empirical sources of uncertainty include the statistical estimation of the spline, amplitude, and two-point corrections.”

CF-IRMS analyses using GC separation may be subject to runtime drift, signal amplitude and scale compression biases arising from changes in preparation system and mass analyzer conditions.

“CF-IRMS analyses using GC separation may be subject to runtime drift, signal amplitude and scale compression biases arising from changes in preparation system and mass analyzer conditions.”

Detection and correction of these biases may be used to produce corrected data sets with improved and validated precision and accuracy.

“Detection and correction of these biases may be used to produce corrected data sets with improved and validated precision and accuracy.”

There is no structural reason why the amplitude correction might not be performed prior to the drift correction.

“However, there is no structural reason why the amplitude correction might not be performed prior to the drift correction.”

The data suggest that results are not sensitive to order of operations.

“the data suggest that results are not sensitive to order of operations.”

Together with objective and automated data correction algorithms such as those presented here, high-throughput CF-IRMS systems enable rapid development of validated, precise, and accurate data streams in support of a wide range of data-intensive geoscientific applications.

“Together with objective and automated data correction algorithms such as those presented here, high-throughput CF-IRMS systems enable rapid development of validated, precise, and accurate data streams in support of a wide range of data-intensive geoscientific applications.”

The small effect of amplitude scaling reflects scaling factors routinely determined to be within 10% of unity in both systems studied here.

“This reflects the small effect of amplitude scaling by scaling factors routinely determined to be within 10% of unity in both systems studied here.”