Effects of oxygen plasma ashing treatment on carbonate clumped isotopes

Rationale: For clumped isotope analysis ( Δ 47 ), hydrocarbon and organic molecules present an important contaminant that cannot always be removed by CO 2 purification through a Porapak-Q trap. Low-temperature oxygen plasma ashing (OPA) is a quick and easy approach for treatment; however, the impact of this treatment on the original carbonate clumped isotope values has never been fully studied. Methods: We tested the isotopic impact of OPA using three natural samples with a large range of initial Δ 47 values. Crushed and sieved (125 μ m mesh) samples were placed into a Henniker Plasma HPT-100 plasma system and treated at a flow rate of 46 mL/min and a power of 100 W at a vacuum of 0.2 mbar for 10, 20, 30 and 60 min before clumped isotope analysis using two MAT 253 isotope ratio mass spectrometers modified to measure masses 44 – 49. Results: OPA treatment for 30 min or more on calcite powder samples has the potential to alter the clumped isotopic composition of the samples beyond analytical error. A systematic positive offset is observed in all samples. The magnitude of this alteration translates to a temperature offset from known values ranging from 4 (cid:1) C to 13 (cid:1) C. We postulate that the observed positive offset in Δ 47 occurs because the bonds within lighter isotopologues are preferentially broken by plasma treatment, leading to an artificial increase in the ‘ clumping ’ value of the sample. Conclusions: We recommend that any laboratory performing OPA treatments should reduce the runs to 10 – 20 min or carry out successive runs of 10 min followed by sample stirring, as this procedure showed no alteration in the initial Δ 47 values. Our


| INTRODUCTION
Since the introduction of carbonate clumped isotope thermometry more than a decade ago, the number of laboratories, as well as the number of geoscience applications using this technique, has increased rapidly. The naturally occurring, multiply substituted isotopologues (clumped isotopes) of 13 C and 18 O are ideal for paleotemperature reconstruction because no information on the oxygen isotope composition of the water is required. 1 Carbonate clumped isotopes have wide applications such as reconstructing burial history, [2][3][4][5][6][7][8][9][10] carbonate diagenesis, [11][12][13][14] constraining the conditions for dolomitization, [15][16][17][18] reconstructing subsurface fluid flow, [19][20][21] estimating the burial pressure regime of carbonates, 22 reconstructing past glacial/interglacial temperature variations, 23,24 constraining the uplift rates on continents, 25 and estimating fault-related temperature and fluid composition. 26 As carbonates are widespread minerals found in sedimentary basins and crystalline rock complexes, further applications of this technique can provide important information on conditions for economic mineral deposits 27,28 and characterization of hydrocarbon reservoirs. 29,30 A wide range of applications implies a wide range of sample types used for clumped isotopes, and therefore it has become increasingly more important to evaluate possible sources of uncertainties related to sample contamination. The most common problem in stable isotopic measurements, other than the presence of water, is the presence of hydrocarbon (eg, bitumen) and adhering organic material, which can affect the measurements of both carbonate clumped isotopes 30,31 and conventional bulk isotopes. [32][33][34] These organic components react with phosphoric acid (H 3 PO 4 ) during acid digestion to generate molecular gaseous species that have similar molecular weights to the carbon dioxide molecules. These radicals and ions in the 44-46 mass range (eg, NO 2 ) produce isobaric effects that interfere with the mass spectrometric determination of δ 13 C and δ 18  In the field of conventional bulk isotope analysis, it has become common practice to perform a treatment to remove organic matter. 32,[37][38][39][40][41][42][43][44] Low-temperature oxygen plasma ashing (OPA) is widely used to remove organic matter from samples using ionized oxygen 38,45 with a working temperature of 60 C-70 C. 46 The effect of low-temperature OPA treatment on the initial bulk isotopic composition has been demonstrated as being neglectable or within the limit of analytical error. 44 A recent study, however, indicated that low-temperature OPA treatment produced a maximum change of the initial bulk isotopic ratio of +0.30‰ for δ 13

| Stable isotope measurement
Measurements of clumped isotopes, δ 18 O and δ 13 C, were carried out in the Qatar Stable Isotope Laboratory at Imperial College London using our fully automated prototype IBEX (Imperial Batch EXtraction) system. Samples of calcite powder (4 mg) were preloaded on a 40-position carousel, and each sample was individually dropped into a vacuum-sealed common acid bath for reaction with 105% orthophosphoric acid at 90 C for 10 min. 51

| Δ 47 alteration
The Multiple comparison LSD tests on IOL and JMF-6A (Table 2) provide a numerical analysis that quantifies what we visually observe in the box plot, even if the overall ANOVA results are not statistically The total duration of OPA pre-treatment was 30 min; the treatment was ceased every 10 min to stir the sample powder. c The total duration was 60 min; the treatment was ceased every 10 min with no stirring procedure introduced. d The total duration was 60 min; the treatment was ceased and stirred every 10 min.

| Testing variations of the OPA protocol on ICM
A variation of the OPA protocol (Table 1) Table 3). Even in the 5 min OPA treatment, both these contamination indicators were reduced significantly to within acceptable values (Δ 48 offset, 0.270‰-0.407‰; 49 parameters, 0.087-0.174; Table 3). Due to the low-temperature OPA treatment, the proportion of acquisitions showing an acceptable level of contamination increased to 77.8% 30 (Figure 3).

| Effect of the OPA treatment on Δ 47
Our results reveal that for the RF of our instruments, a 30 min OPA treatment has the potential to cause an observable systematic positive offset in clumped isotopes (Δ 47 values).
However, if the total duration of the 30 min of OPA treatment is applied in successive 10 min treatments followed by stirring, no significant increase in the Δ 47 value is observed. This suggests that for an offset to exist, the same mineral surface needs to be exposed to the plasma for 30 min, as stirring between runs effectively exposes new crystal surfaces between each run.
Although not statistically significant in the ANOVA tests (Δ 47 offset of 0.014‰ for IOL, 0.024‰ for JMF-6A and 0.011‰ for T A B L E 2 ANOVA and post hoc LSD analysis of IOL and JMF-6A samples  A difference in mineral reactivity (hypothesis 2) could, for instance, impact the parameters of solid-state reordering. 3 However, the low temperature recorded in the plasma chamber (<100 C) effectively precludes this process from happening. Other potential mineral reactivity mechanisms to explain how OPA treatment impacts clumped isotopes include, for instance, thermal dilatation. It is known that thermal dilatation or expansion of calcite minerals may occur in the powdered samples during the plasma treatment: a heating/cooling cycle of +20 C, −60 C, −20 C on Italian marble resulted in thermal expansion leading to a reduction in the cohesion strength of the grains. 66 The volumetric thermal expansion of limestones is also 14.5% larger 67 than that of marbles for the temperature range of 20 C-100 C; as a consequence, the thermal  Figure 4).

| Implications for paleotemperature reconstructions
As Δ 47 has a quantitative relationship with temperature, the implications of the observed offset caused by OPA treatment can be quantified using the following temperature calibration 49 : The difference between the initial temperature of the sample and the temperature after 30 min of continuous oxygen plasma treatment induces a bias towards a colder temperature of 3.9 C for IOL, 9.1 C for JMF-6A and 13.3 C for ICM (Table 4). Whether this bias is significant will depend on the particular application. In paleoclimate studies, the order of magnitude of change in Δ 47 for 30 min corresponds to a large enough temperature difference to potentially mask the paleoclimatic signal investigated. However, we also note again that shorter OPA treatments of 10 min do not significantly alter the Δ 47 value and can thus be considered safe for clumped isotope applications.

| CONCLUSIONS
An effective treatment to remove organic material before the carbonate clumped isotope analysis must maintain the original isotopic composition of the mineral while removing contaminants.
Based on our data, 30 min or more of OPA on calcite powders has the potential to alter the initial isotopic composition beyond the analytical error, thus biasing paleotemperature studies. The small data populations used for the test prevented us from proving the statistical significance (α = 0.05) of the OPA alteration.
F I G U R E 4 Hypothetical model for Δ 47 enrichment by preferential breaking of the weaker bonds (e.g., 13 C- 16  We postulate that this positive offset occurs because the bonds between the lighter isotopes are preferentially broken by OPA treatment, leading to an increase in the 'clumping' of the sample. Because the magnitude of the observed mean value can range from 4 C to 13 C in the 30 min treatment, we recommend that any laboratory performing OPA should reduce the runs to 10-20 min, as the plasma treatment of this duration on our instrument shows no alteration of the initial value of Δ 47 . In addition, it is possible to safely remove additional organic matter by performing successive runs of 10 min followed by a break and stirring of the sample, as this procedure with a total OPA time of 30 min, and 60 min on Carrara marble, shows no offset in the Δ 47 value. Our results thus validate the use of OPA for clumped isotope applications and will allow future research using clumped isotopes in challenging samples, such as oil-stained carbonates, bituminous shales and very high organic carbon content host rocks. Future research should consider the potential OPA effects on different minerals (eg, aragonite).