Laser-Based Method and Sample Handling Protocol for Measuring

A robust method is demonstrated to measure acetone in human breath at sub parts-per-million by volume (ppmv) concentrations using diode laser cavity e...
0 downloads 5 Views 696KB Size
Article pubs.acs.org/ac

Laser-Based Method and Sample Handling Protocol for Measuring Breath Acetone Gus Hancock,‡ Cathryn E. Langley,‡ Robert Peverall,*,‡,† Grant A. D. Ritchie,‡ and David Taylor† ‡

Department of Chemistry, Physical and Theoretical Chemistry Laboratory, University of Oxford, South Parks Road, Oxford, OX1 3QZ, United Kingdom † Oxford Medical Diagnostics Ltd., Centre for Innovation and Enterprise, Begbroke Science Park, Begbroke Hill, Begbroke, OX5 1PF, United Kingdom ABSTRACT: A robust method is demonstrated to measure acetone in human breath at sub parts-per-million by volume (ppmv) concentrations using diode laser cavity enhanced absorption spectroscopy. The laser operates in the nearinfrared at about 1690 nm probing overtone transitions in acetone in a spectral region relatively free from interference from common breath species such as CO2, water, and methane. Using an optical cavity with a length of 45 cm, bound by mirrors of 99.997% reflectivity, a limit of detection of ∼180 parts-per-billion by volume (ppbv) (1σ) of breath acetone is achieved. The method is validated with measurements made with an ion−molecule reaction mass spectrometer. A technique to calibrate the optical cavity mirror reflectivity using a temperature dependent water vapor source is also described.

U

present with ketoacidosis17−19 and would therefore be expected to exhale higher concentrations of acetone than normally found in breath. There is some evidence that breath acetone is associated with concentrations of blood glucose among patients with type 1 diabetes. In 1969, Tassopoulos et al.1 noted corresponding raised breath acetone and blood glucose concentrations. More recently, Turner et al.8 measured the breath acetone concentrations from patients with type 1 diabetes using the selected ion flow tube (SIFT) method20,21 and monitored their blood glucose concentrations, which were controlled using a hypoglycemic clamp technique. From the small study (8 subjects), they demonstrated that there was a positive correlation between blood glucose and breath acetone (although the exact relationship was different for each patient). Another mass spectrometry-based study by Minh et al.22 has identified four molecular markers present in breath including acetone as indicators of absolute blood glucose concentrations. Wang et al.9 have used UV laser-based cavity ring-down spectroscopy to measure breath acetone and report a general trend among patients with type 1 diabetes, with a linear correlation between the average breath acetone concentrations and the average blood glucose concentrations when the patients are grouped by different blood glucose concentrations. In another study, Spanel et al.14 demonstrate that, although a

nder normal conditions, acetone is the most abundant volatile organic compound (VOC) found in human breath and, coupled with its strong link with diabetes and diet, represents a particularly interesting molecule to study with regard to breath analysis.1−15 It is essentially produced when the body turns to its fat deposits as a source of energy in the absence of glycogen stores (or release of glucose there from).16 In summary, the acetone generation process originates with the triglyceride molecules (predominantly from adipose tissue) which are hydrolyzed via lypolysis to produce glycerol and fatty acid chains. The fatty acids are then utilized in β-oxidation to generate energy and produce the acetyl coenzyme A (Acetyl CoA) required for the Krebs cycle. Excess acetyl CoA is converted into acetoacetate and D-β-hydroxybutyrate (ketone bodies) in the liver, which subsequently travel through the bloodstream to where they can be utilized as an energy source (oxidation in the Krebs cycle within cells). Acetoacetate can undergo decarboxylation which results in the production of acetone. When fat stores are used instead of carbohydrates, the rate of ketone body production increases and their concentration increases in the bloodstream. In the case of acetone, which is a small volatile compound that easily diffuses from the bloodstream and into the lungs, this leads to elevated concentrations of acetone in exhaled breath. These periods of stress which cause fat deposits to be used as energy sources could be the result of intense exercise, dieting, or a lack of insulin resulting in a reduced uptake of glucose by the liver. A serious and potentially fatal condition often found with diabetes is that of diabetic ketoacidosis (DKA), which usually results from a lack of insulin. Undiagnosed patients with diabetes often © 2014 American Chemical Society

Received: February 13, 2014 Accepted: May 15, 2014 Published: May 15, 2014 5838

dx.doi.org/10.1021/ac500614n | Anal. Chem. 2014, 86, 5838−5843

Analytical Chemistry

Article

predicted increase in the concentration of acetone is observed in the breath of those on a ketogenic diet, there is a wide variation observed in the concentrations as a result of natural intraindividual biological and daily variability. Therefore, it is clear from these studies and recent reviews23,24 that more work is needed to firmly establish the exact link between blood glucose and breath acetone. We note, however, that the relationship between blood ketones and breath acetone is far more secure (see, for example, refs 12 and 25). Including the work of Wang et al., there have been several optical methodologies described to measure acetone. Wang et al. use a frequency quadrupled YAG laser to generate radiation at 266 nm in their cavity ring-down system,26,27 which was used in the 2010 study, examining breath acetone and blood glucose correlations.9 However, breath acetone samples were not verified against a secondary device in order to assess accurately the sensitivity of the instrument to acetone within a breath matrix. Lewicki et al.28 have demonstrated the use of a widely tunable quantum cascade laser (QCL) at 8.4 μm with claimed capability29 to detect 4%, and methane can vary enormously between individuals, from background concentrations (1.8 ppmv) up to 100 ppmv in archaea positive humans.33 One VOC with relatively high basal concentration in human breath is isoprene, which is generally exhaled by adults at concentrations of between 50 and 300 ppbv34 and to a lesser extent in juveniles,35 and its first overtone C−H stretch spectrum will occur in this region.30 All these species can cause some interference at the small levels of absorption required to measure changes in signal from acetone at sub-ppmv concentrations. Clearly, for the species with a high level of spectral structure (CO2, CH4, and H2O, see below), we can select regions where spectral absorption lines are weaker and potentially sparsely spaced. For acetone in a breath matrix, the region around 1690 nm is suitable from this respect, except where isoprene is concerned. However, detailed analysis reveals two unassailable issues, even around 1690 nm, especially since measuring with an atmospheric (or near atmospheric) pressure sample is desirable for as large an acetone signal as possible. One issue is the range of methane concentrations experienced, possibly up to 50 ppmv or thereabouts, which could still cause observable absorption between line centers. The other is potentially more problematic and is associated with the water vapor continuum. The first experimental evidence that water vapor absorbs infrared radiation was presented by Tyndall,36 but it was not until 1918 that the first detailed water vapor spectrum over a broad spectral range was determined by Hettner.37 In 1938, Elsässer noted that there was absorption at wavelengths at which water did not possess rotation or ro-vibrational bands. 38,39 This broad, slowly varying component of absorption, found in addition to the individual spectral lines, is known as the water continuum and it is split into two components: the self-continuum and the foreign-continuum. 5839

dx.doi.org/10.1021/ac500614n | Anal. Chem. 2014, 86, 5838−5843

Analytical Chemistry

Article

molecular sieve (Hydranal, 3 Å, ∼10 g). The molecular sieve removes most VOCs and water from the breath including acetone but allows methane and CO2 through. Another absorption measurement is then made on this dry, scrubbed sample and constitutes a breath background (I0). I and I0 are then used in the following equation to determine the quantity of acetone43,44

The former is a result of the (binary) interactions between water molecules while the latter describes the interaction of water molecules with “foreign” molecules, commonly those found in the atmosphere, such as N2 and O2. The water continuum has been the subject of countless studies thanks to its importance in atmospheric science: its physical origin has been a point of debate over the years, and a discussion of which is beyond the scope of this paper but can be found in a number of reviews.40−42 For the partial pressures of water found in breath (∼25 mbar at 21 °C) and for a total pressure of 1 atm, the apparent absorption cross-section for the water vapor continuum can be estimated to be in the region of 2.5 × 10−25 cm2 at 1690 nm.41,42 This would yield an absorption equivalent to 14 ppmv of acetone at the same wavelength.

I0 − I αL = I 1−R

where L is the physical length of the optical cavity, R is the geometric-mean mirror reflectivity, and α is the absorption coefficient. The internal volume of the cavity is 115 cm3, and so, there is ample breath available for both measurements and the mass spectrometric analysis. The mass spectrometer is calibrated using standard gas samples of 1 and 5 ppmv acetone in air or nitrogen which are commercially available (BOC speciality gases Ltd., UK) and certified to within 5%. The mass spectrometer was calibrated with these mixtures directly from the cylinder which showed good agreement with the same gas mixtures introduced to the mass spectrometer via a breath bag. The calibration of the mass spectrometer was checked at regular intervals and adjusted as required, although it was generally found to be stable over the course of a day, after an initial period (∼10 min) following the application of the full filament current to its ionization source (used to generate the reactant ions). Figure 2 shows a schematic of the experimental arrangement.



EXPERIMENTAL CONSIDERATIONS Any method that purports to measure acetone vapor in a breath matrix must clearly circumvent any of these confounding issues. Here, we present a cavity enhanced laser absorption technique coupled to a breath handling algorithm that gives reasonable accuracy for breath acetone at sub-ppmv concentrations. The absorption coefficient for 1 ppmv of acetone at the selected wavelength (1688.45 nm in vacuo) is 2.65 × 10−8 cm−1, and so, we have chosen a cavity with mirrors of nominal reflectivity of 99.995% (Layertec GmbH radii of curvature, −1 m; diameter, 25 mm) and a cavity length of 45 cm, capable of detecting absorptions with minimum detectable absorption coefficients of ∼10−9 cm−1. The laser used in this study is a distributed feedback diode laser (NEL NLKU5E1AAA) delivering ∼10 mW of power through polarization maintaining single mode fiber. The cavity consists of a solid aluminum tube of internal diameter of 18 mm with the mirrors at either end held in recesses by clamping rings, and the cavity has no moving parts. The only fine optical alignment possible is by an external turning mirror. The decision to make the cavity as mechanically robust as possible stems from the nature of the measurement, which is a dc level measurement (rather than a measurement of a spectral line shape). We have found, during the development of this method, that mirrors held in vacuum compatible adjustable mounts have a tendency to move when the pressure of the cavity/sample cell is cycled, which in this case can lead to erroneous results. There are entrance and exit ports for sample introduction and extraction. In order to obtain data on acetone absorption, the diode laser is repetitively scanned over a range of about 0.1 cm−1 around 5922.6 cm−1 (1688.45 nm) at a frequency of 500 Hz with a triangular waveform. An average over the middle 70% of the up-ramp of the waveform is recorded over 1000 scans using a digital oscilloscope to produce a number representative of the light intensity transmitted through the cavity. How the breath samples are handled and introduced into the cavity has as much importance as the optical measurement. In this study, breath is collected in aluminized breath bags (FAN GmbH) and within 5 min is analyzed on an ion−molecule reaction mass spectrometer (AirSense, V&F Analyse- und Messtechnik GmbH) and with the optical apparatus. Initially, the cavity is evacuated and the breath from the bag drawn into the cavity via a thermo-electrically cooled chiller device which is maintained at −20 °C. This removes a substantial fraction of the water but has little or no effect on the acetone in the sample. An absorption measurement (I) is then made on this dry sample. The cavity is then evacuated, and more breath is drawn into the device but this time through a small quantity of



RESULTS AND DISCUSSION The performance of the chiller arrangement was initially tested by measuring absorption on a nearby water transition at 1689.45 nm (formally denoted as the (12, 3, 9) ← (11, 1, 0)

Figure 2. Schematic of the experimental arrangement. Sample is passed into the cavity via the chiller (−20 °C) to measure the signal and then via the molecular sieve to measure a background. The laser is temperature and current controlled (∼25 °C and ∼90 mA) to obtain a central wavelength of 1688.45 nm (in vacuo) or 5922.6 cm−1 and is scanned over a range of 0.2 cm−1 at 100 Hz. The central 70% of the scan is used to establish the average signal (over ∼1000 scans). The preamplifier amplification is 108 V/A. 5840

dx.doi.org/10.1021/ac500614n | Anal. Chem. 2014, 86, 5838−5843

Analytical Chemistry

Article

transition within the (0, 1, 1) ← (0, 0, 0) vibrational combination band), for various chiller temperatures, using laboratory air as the sample, and filling the cavity to 50 mbar. The dependence of the measured partial pressure of water on chiller temperature is shown in Figure 3. Also plotted in this

Figure 4. Measurements of the cavity enhanced signal as a function of acetone concentrations. For these data, mixtures of 5 and 25 ppmv acetone in dry air (BOC) have been used to generate the different acetone concentrations. Data were recorded at a total pressure of 1 atm; the straight line fit returns a mirror reflectivity of 99.9965 ± 0.0005%. Figure 3. Partial pressure of water vapor in the cell as a function of chiller temperature as measured using cavity enhanced absorption spectroscopy of a water transition at 1689.45 nm. Total pressure is 50 mbar. The first four and the final measurements are labeled for clarity. The solid line is the predicted water partial pressure from the Antoine relations: see text for more information. The inset shows examples of water absorption spectra at high and low chiller temperature from which this plot was constructed.

the data in Figure 4 verifies the calibration mirror reflectivity yielding R = 99.9965 ± 0.0005%. The scatter in the plot reflects errors in sample handling rather than the sensitivity of the technique; however, from repeated measurements at below 2 ppmv acetone concentration, we estimate the limit of detection to be no worse than 250 ppbv, corresponding to an absorption coefficient αmin of ∼6 × 10−9 cm−1 (1σ). The results from some 37 different samples of breath taken over about 2 weeks from eight healthy individuals measured in our CEAS instrument following our handling protocol and with the mass spectrometer are shown in the correlation plot, Figure 5A. Individuals were asked to take a deep breath and hold that breath for about five seconds and then expel perhaps half their breath before filling a breath bag with the remainder. This ensured end-tidal breath conditions but strictly speaking was not necessary for testing this device and protocol (since we measure the concentration within the bag anyway). Included in the plot are four premixed acetone in dry air/nitrogen samples that have undergone exactly the same handling protocol, which are the highest four readings. Two of the individuals underwent an overnight fast before giving some of the samples, and these are the next six highest data points. Isoprene has been identified as the predominant interfering species in these measurements and can be taken into account in two different ways. The first of these is to remove the contribution as ascertained from the measured isoprene concentration (measured with the mass spectrometer), and these data are shown in Figure 5B. Obviously, this is impractical for this optical only instrument, but it does serve as a useful indicator of the scatter in the data solely due to the measurement protocol as applied to breath. The second is to remove a fixed percentage of the absorption as ascertained from the mean value of isoprene determined from these data sets (180 ± 60 ppbv). (Note: The effective contribution to the recorded acetone concentration is 1.5 times higher than this because of the larger isoprene absorption cross section.) On average, these concentrations of isoprene are higher than those reported by Turner et al.34 (118 ± 68 ppbv) but within the range of values reported using the SIFT technique. Applying

figure is the predicted partial pressure from the Antoine relations for a saturated atmosphere.45 The experiment was started at room temperature, and the chiller temperature decreased to a minimum of −30 °C and then increased back to room temperature, with data recorded at different temperatures. The absorption measurement returns the laboratory partial pressure of water (adjusted for the reduction of total pressure) until the chiller reaches the dew point of the laboratory air (in this case about 5 °C). At temperatures below this (with reducing temperature), the amount of water entering the cavity is consistent with the saturation partial pressure of water for the chiller temperature. As the chiller temperature is raised, the measured partial pressure follows the saturation partial pressure all the way up to room temperature, as clearly some water is trapped and stored within the device. This property of the chiller, as a temperature dependent water vapor source, is useful and provides a known concentration of water with which to calibrate the optical cavity without having to measure the mirror reflectivity some other way, e.g., by taking a cavity ringdown measurement. Calibrating the optical cavity in this way revealed that the mirror reflectivity (the geometric mean thereof) was 99.9968 ± 0.0008%. Initially, a series of measurements was performed of various concentrations of acetone in a dry air/N2 matrix. These were prepared by diluting samples from calibrated mixes of 5 ppmv acetone in dry air and 25 ppmv acetone in dry air (both BOC Special Gases) with N2, and the resulting plot of cavity enhanced absorption signal against acetone concentration is given in Figure 4. The samples were prepared directly in the optical cavity at 1 atm total pressure, rather than going through a chiller device and the breath handling protocol, to assess the baseline response of the optical cavity to acetone. The slope of 5841

dx.doi.org/10.1021/ac500614n | Anal. Chem. 2014, 86, 5838−5843

Analytical Chemistry

Article

uncertainties in quadrature). Nevertheless, isoprene has been identified as having slightly different exhalation characteristics compared to acetone,47 and while it has not caused a significant problem for the eight or so subjects who have provided breath for this study and nor would it cause problems for the majority of those subjects studied in Turner et al.,34 the presence of isoprene still limits the universality of the method. One possible solution to this problem would be to find a zeolite that traps acetone but not isoprene, so that isoprene would be present in the background measurements. Another is to measure the presence of isoprene as well, which would require an additional light source. For many applications, however, such as dietary monitoring or ketosis control, where perhaps the concentrations of breath acetone are higher than ∼0.5 ppmv, the presence of isoprene is not an issue.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare the following competing financial interest(s): G.H. and G.A.D.R. act as consultants for Oxford Medical Diagnostics Ltd.; R.P. is the science director of Oxford Medical Diagnostics Ltd., and G.H. is a director of Oxford Medical Diagnostics Ltd.

Figure 5. Acetone concentration data from 37 breath samples (from eight healthy individuals) and four premixed calibration samples (the four highest points). (A) Data uncorrected for breath isoprene (+) and a straight line fit to the data including the four calibration points. The breath data have a standard deviation of 180 ppbv around the straight line fit. (B) Data corrected for the contribution from the measured breath isoprene (standard deviation = 150 ppbv).



ACKNOWLEDGMENTS C.E.L. would like to acknowledge the South East England Development Agency (SEEDA) for a CASE DPhil studentship.



this operation to the data in Figure 5A removes much of the offset and has no effect on the statistics of the fit but adds a further slight uncertainty to the total measurement as a consequence of the distribution of exhaled isoprene concentrations. Further care would be needed when measuring juvenile subjects who invariably exhibit lower isoprene exhalation concentrations. Overall, the standard deviation for the data in Figure 5A is 180 ppbv; for the data below 1 ppmv (70% of the data), this increases slightly to 190 ppbv. For individuals with ketogenic dietary requirements, who have fasted, or who have undergone heavy exercise or patients with diabetes approaching ketosis, we expect that breath concentrations of acetone can far exceed 1 ppmv.2,4,12,46

REFERENCES

(1) Tassopoulos, C. N.; Barnett, D.; Russel Fraser, T. Lancet 1969, 293, 1282−1286. (2) Goschke, H.; Lauffenburger, T. Res. Exp. Med. 1975, 165, 233− 244. (3) Crofford, O. B.; Mallard, R. E.; Winton, R. E.; Rogers, N. L.; Jackson, J. C.; Keller, U. Trans. Am. Clin. Climatol. Assoc. 1977, 88, 128−139. (4) Jones, A. W. J. Anal. Toxicol. 1987, 11, 67−69. (5) Kinoyama, M.; Nitta, H.; Watanabe, A.; et al. J. Health Sci. 2008, 54, 471−477. (6) Teshima, N.; Li, J.; Toda, K.; Dasgupta, P. K. Anal. Chim. Acta 2005, 535, 189−199. (7) Minh, T. D. C.; Blake, D. R.; Galassetti, P. R. Diabetes Res. Clin. Pract. 2012, 97, 195−205. (8) Turner, C.; Walton, C.; Hoashi, S.; Evans, M. J. Breath Res. 2009, 3, 046004. (9) Wang, C.; Mbi, A.; Shepherd, M. IEEE Sens. J. 2010, 10, 54−63. (10) Righettoni, M.; Tricoli, A. J. Breath Res. 2011, 5, 037109. (11) Deng, C. H.; Zhang, J.; Yu, X. F.; et al. J. Chromatogr., B 2004, 810, 269−275. (12) Musa-Veloso, K.; Likhodii, S.; Cunnane, S. C. Am. J. Clin. Nutr. 2002, 76, 65−70. (13) Nelson, N.; Lagesson, V.; Nosratabadi, A. R.; Lugvisson, J.; Tagesson, C. Pediatr. Res. 1998, 44, 363−367. (14) Spanel, P.; Dryahina, K.; Rejskova, A.; Chippendale, T. W. E.; Smith, D. Physiol. Meas. 2011, 32, N23−N31. (15) Smith, D.; Spanel, P.; Davies, S. J. Appl. Physiol. 1999, 87, 1584− 1588. (16) Laffel, L. Diabetes/Metab. Res. Rev. 1999, 15, 412−426.



CONCLUSIONS We have presented here an optical-based method for determining breath acetone concentrations at sub-ppmv having developed a sample handling protocol that uses the sample to be measured to generate an absorption baseline by scrubbing the sample of VOCs and water. This has proven to be important because of the variability of different compounds in breath samples such as water, methane, and CO2 that are present to an extent that can yield comparable absorptions to breath acetone in the near and mid-infrared spectral regions. Once the problems with water, methane, and CO2 are mitigated, the predominant interfering species is identified as isoprene. In the incarnation of the optical apparatus presented here, the presence of isoprene decreases the precision of the acetone measurement only by about 20 ppbv (taking the 5842

dx.doi.org/10.1021/ac500614n | Anal. Chem. 2014, 86, 5838−5843

Analytical Chemistry

Article

(17) Klocker, A. A.; Phelan, H.; Twigg, S. M.; Craig, M. E. Diabetes Med. 2013, 10, 818−824. (18) Vanelli, M.; Chiari, G.; Capuano, C.; et al. Diabetes Nutr. Metab. 2003, 16, 312−316. (19) Ali, Z.; Levine, B.; Ripple, M.; et al. Am. J. Forensic Med. Pathol. 2012, 33, 189−193. (20) Spanel, P.; Smith, D. Med. Biol. Eng. Comput. 1996, 34, 409− 419. (21) Smith, D.; Spanel, P. Rapid Commun. Mass Spectrom. 1996, 10, 1183−1198. (22) Minh, T. D. C.; Oliver, S. R.; Ngo, J.; et al. Am. J. Physiol. Endcrinol. Metab. 2011, 300, E1166−E1175. (23) Spanel, P.; Smith, D. Curr. Opin. Clin. Nutr. Metab. Care 2011, 14, 455−460. (24) Smith, D.; Spanel, P.; Fryer, A. A.; Hanna, F.; Ferns, G. A. A. J. Breath Res. 2011, 5, 022001. (25) Musa-Veloso, K.; Likhodii, S. S.; Rarama, E.; Benoit, S.; et al. Nutrition 2006, 22, 1−8. (26) Wang, C.; Mbi, A. Meas. Sci. Technol. 2007, 18, 2731−2741. (27) Wang, C.; Scherrer, S. T.; Hossain, D. Appl. Spectrosc. 2004, 58, 784−791. (28) Lewicki, R.; Wysocki, G.; Kosterev, A. A.; Tittel, F. K. Opt. Express 2007, 15, 7357−7366. (29) McCurdy, M. R.; Bakhirkin, Y.; Wysocki, G.; Lewicki, R.; Tittel, F. K. J. Breath Res. 2007, 1, No. 014001. (30) Denzer, W.; Hancock, G.; Islam, M.; et al. Analyst 2011, 136, 801−806. (31) Ciaffoni, L.; Hancock, G.; Harrison, J. J.; et al. Anal. Chem. 2013, 85, 846−850. (32) Kjaergaard, H.; Henry, B.; Tarr, A. J. Chem. Phys. 1991, 94, 5844−5854. (33) Fernandes, J.; Wang, A.; Su, W.; et al. Nutrition 2013, 143, 1269−1275. (34) Turner, C.; Spanel, P.; Smith, D. Physiol. Meas. 2006, 27, 13−22. (35) Smith, D.; Spanel, P.; Enderby, B. J. Breath Res. 2010, 4, No. 017101. (36) Tyndall, J. Philos. Trans. R. Soc. London 1861, 151, 1−36. (37) Hettner, G. Ann. Phys. 1913, 360, 476−496. (38) Elsasser, W. M. Mon. Weather Rev. 1938, 66, 175−178. (39) Elsasser, W. M. Astrophys. J. 1938, 87, 497−507. (40) Ptashnik, I. V.; Shine, K. P.; Vigasin, A. A. J. Quant. Spectrosc. Radiat. Transfer 2011, 112, 1286−1303. (41) Paynter, D. J.; Ramaswamy, V. J. Geophys. Res.: Atmos. 2011, 116, No. D20302. (42) Shine, K. P.; Ptashnik, I. V.; Raedel, G. Surv. Geophys. 2012, 33, 535−555. (43) Berden, G.; Peeters, R.; Meijer, G. Int. Rev. Phys. Chem. 2000, 19, 565−607. (44) Mazurenka, M.; Orr-Ewing, A. J.; Peverall, R.; Ritchie, G. A. D. Ann. Rep. Prog. Chem. C: Phys. Chem. 2005, 101, 100−142. (45) Linstrom, P. J., Mallard, W. G., Eds. NIST Chemistry WebBook, NIST Standard Reference Database Number 69; National Institute of Standards and Technology: Gaithersburg, MD; http://webbook.nist. gov (accessed December 13, 2013). (46) Sulway, M. J.; Malins, J. M. Lancet 1970, 296, 736−740. (47) King, J.; Kupferthaler, A.; Unterkofler, K. J. Breath Res. 2009, 3, No. 027006.

5843

dx.doi.org/10.1021/ac500614n | Anal. Chem. 2014, 86, 5838−5843