III. CYCLIC MOLECULES
2. METHODS
3.1. I SOTOPE EXCHANGE EXPERIMENTS
3.1.2. Experimental measurements of equilibrium fractionation for H α
During the second stage of experiments, incubations were carried out with all substrates at pH 12 for 6 days at 70°C, 10 days at 50°C, and 75 days at 25°C, more than 40 times longer than the estimated t1/2 at each temperature to ensure the attainment of 2H/1H exchange equilibrium. The values of the isotopic enrichment factor εeq for Hα are derived from the regression slope between δ2H values of the ketone and water at equilibrium (Chapter 1, Section 3.1.5.). This yields the εeq value that averages all exchanging Hα
positions in each molecule.
3.1.2.1. Effect of isotopic normalization on the value of equilibrium fractionation. In the isotopic analysis of the exchanged ketones, we co-injected a series of multiple organic standards with each sample and used the normalization curve based on these standards to correct for memory effect. Detailed procedures have been described in Chapter 1, Section 3.1.3. The six cyclic ketones used in this study have very different original δ2H values (before exchange with water), ranging from −256‰ to −148‰ for cyclohexanone, 2- methyl-cyclohexnaone, and 3-methyl-cyclohexanone, and from −99‰ to −26‰ for 4- methyl-cyclohexnaone, 2,6-dimethyl-cyclohexnaone, and 2,2,6-trimethyl-cyclohexnaone.
Therefore we designed two series of organic standards to better fit the δ2H value of each ketone. The first series (Standard 1) consists of three n-alkane standards, nC17 (−142‰), nC22 (−62.2‰), and nC25 (−256.4‰), and was co-injected with the first three cyclic ketones. The second series (Standard 2) consists of two n-alkane standards, nC17 (−142‰),
114 nC22 (−62.2‰), and two ester standards, palmitic acid methyl ester (C16M, 88.0‰) and eicosenoic n-butyl ester (C20B, 1.5‰). It was co-injected with the last three cyclic ketones.
Theoretically, both standard series should give statistically consistent normalization curves. However, the analyses using Standard 1 produced normalization curves with slope of 0.97±0.02 and intercept of 6±3, whereas those using Standard 2 produced normalization slopes of 0.88±0.02 and intercept of -9±2. For the same ketone samples, application of the second normalization gives equilibrium fractionation (εeq values) that are consistently higher by 40–60‰ than the εeq values using the first one. This offset in normalization curves could not be due to random error or system drifting, because it is observed in successive measurements during the same day. In fact, this offset is primarily caused by the large decrease in measured δ2H values relative to the “known” values for the ester standards in Standard 2 (16–20‰ for C16M and 7–9‰ for C20B). By deleting them, the normalization of Standard 2 becomes consistent with Standard 1. Since the ester standards are synthesized via transesterification (Chapter 4, Section 2.1.) while the n-alkane standards are purchased pure products, it is possible that impurities during the synthesis and purification processes had biased the offline measurements for the ester standards (i.e., the
“known” δ2H values). Another possibility is the variation in memory effects. Since the measured δ2H values are sensitive to peak size, especially for 2H-enriched analytes (Chapter 4, Section 3.3.), we intended to maintain the same peak size for each standard and the ketone so that all of them were subject to equivalent memory effects from the preceding methane peaks. Nevertheless, small variations are still present. The peak size of C16M is smaller than nC17 by ~8%, which probably biased the measured δ2H of C16M towards lower values.
In general, the normalization based on Standard 1 appears to be more reliable and is consistent with previous normalization records. We therefore applied this normalization to all exchanged cyclic ketone samples. R2 values for the regressions between the water δ2H values and the normalized ketone δ2H values are all greater than 0.99. Complete δ2H data are presented in Appendix Table 3-A1.
115 3.1.2.2. Equilibrium fractionation factors for Hα. Derived values of the equilibrium fractionation (εeq values) for Hα positions in the cyclic ketone substrates are summarized in Table 3-2 and Fig. 3-4. The uncertainties in εeq values, generally between 5–15‰, are estimated from the standard deviation of the ketone-water regressions. They are slightly smaller than the uncertainty for acyclic ketones (Table 1-2), because we used 8 waters of varying δ2H values to exchange with each cyclic ketone, while 7 waters were used for each acyclic ketone. The relatively large uncertainty for 2,2,6-trimethylcyclohexanone is due to its low Hα content (fex = 0.063), which produces a much smaller δ2H range after the incubations, compared to other cyclic ketones. The values of non-exchangeable H derived at different temperatures are statistically identical for each molecule. Similar to acyclic ketones, this confirms the assumption that only Hα is undergoing appreciable exchange.
The values of εeq for cyclohexanone (around −150‰) are consistent with the results measured in our previous study (Table 1-2). The two ketones that contain non-α methyl group, 3-methylcyclohexanone and 4-methylcyclohexanone, yield statistically consistent εeq values ranging between −133‰ and −154‰ with little temperature variation. They are similar to cyclohexanone, indicating that alkyl groups on non-α C atoms do not effectively change the equilibrium fractionation of Hα, consistent with the “cutoff” effect (Stern and Wolfsberg 1966a,b; Hartshorn and Shiner, 1972) and our previous results for acyclic ketones (Chapter 1). Molecules containing only tertiary Hα atoms, cis-2,6- dimethylcyclohexanone and 2,2,6-trimethylcyclohexanone, have similar εeq values between 7‰ and −42‰ with negative temperature dependence. The results also suggest that the measured δ2H values for cis-2,6-dimethylcyclohexanone are not affected by the slight coelution with the trans-isomer in GC/P/IRMS analysis (see Section 2.1.1.). The εeq values for 2-methylcyclohexanone which contains one tertiary and two secondary Hα atoms agree well with the weighted average of tertiary Hα and secondary Hα. This trend, wherein εeq for Hα increases substantially with more alkyl substituents on Cα, is again consistent with the observation for acyclic ketones. It is caused by the electron-donating effect of alkyl groups that acts to enhance the electron density of the Cα−Hα bond and thus increases its bond stiffness.
116 Table 3-2. Experimental isotopic enrichment factors (εeq) for Hα and δH values for non- exchangeable H (δN) in the six cyclic ketones
Uncertainties (1σ) are given in brackets. They are propagated from the standard error of the ketone- water regression slope and intercept.
Figure 3-4. Equilibrium isotopic enrichment factors (εeq) for Hα in cyclic ketones. Symbols represent experimental results measured at 25, 50, and 70°C with error bars of ±1σ. Curves represent theoretical results from 0–100°C for secondary (2°) and tertiary (3°) Hα at the equatorial or axial position in the same molecules.
ketone substrate 25°C 50°C 70°C
εeq (‰) δN (‰) εeq (‰) δN (‰) εeq (‰) δN (‰) Cyclohexanone -148 (11) -196 (8) -158(7) -191 (5) -152 (6) -195 (4)
2-methylcyclohexanone -102(7) -203 (2) -105 (11) -203 (4) -103 (6) -205 (2)
3-methylcyclohexanone -148 (5) -234 (3) -154 (13) -234 (7) -139 (9) -243 (4)
4-methylcyclohexanone -133 (10) -97 (5) -142 (10) -92 (5) -137 (10) -91 (5)
2,6-dimethylcyclohexanone -8 (12) -13 (2) -30 (10) -18 (2) -42 (18) -16 (3) 2,2,6-trimethylcyclohexanone 7 (26) -104 (2) -31 (19) -102 (1) -34 (19) -102 (1)
117