ICL-based TDLAS sensor for real-time breath gas analysis of carbon monoxide isotopes

We present a compact sensor for carbon monoxide (CO) in air and exhaled breath based on a room temperature interband cascade laser (ICL) operating at 4.69 μm, a lowvolume circular multipass cell and wavelength modulation absorption spectroscopy. A fringelimited (1σ) sensitivity of 6.5 × 10 cmHz and a detection limit of 9 ± 5 ppbv at 0.07 s acquisition time are achieved, which constitutes a 25-fold improvement compared to direct absorption spectroscopy. Integration over 10 s increases the precision to 0.6 ppbv. The setup also allows measuring the stable isotope CO in breath. We demonstrate quantification of indoor air CO and real-time detection of CO expirograms from healthy non-smokers and a healthy smoker before and after smoking. Isotope ratio analysis indicates depletion of CO in breath compared to natural abundance. © 2017 Optical Society of America OCIS codes: (280.3420) Laser sensors; (140.5965) Semiconductor lasers, quantum cascade; (300.6340) Spectroscopy, infrared; (300.6360) Spectroscopy, laser; (280.1415) Biological sensing and sensors. References and links 1. A. Amann and D. Smith, Volatile Biomarkers: Non-invasive Diagnosis in Physiology and Medicine (Elsevier, 2013). 2. T. H. Risby and S. F. Solga, “Current status of clinical breath analysis,” Appl. Phys. B 85(2), 421–426 (2006). 3. A. S. Modak, “Stable isotope breath tests in clinical medicine: a review,” J. Breath Res. 1(1), 014003 (2007). 4. J. Li, U. Parchatka, R. Königstedt, and H. Fischer, “Real-time measurements of atmospheric CO using a continuous-wave room temperature quantum cascade laser based spectrometer,” Opt. Express 20(7), 7590–7601 (2012). 5. S. W. Ryter and A. M. K. Choi, “Carbon monoxide in exhaled breath testing and therapeutics,” J. Breath Res. 7(1), 017111 (2013). 6. E. O. Owens, “Endogenous carbon monoxide production in disease,” Clin. Biochem. 43(15), 1183–1188 (2010). 7. R. Gajdócsy and I. Horváth, “Exhaled carbon monoxide in airway diseases: from research findings to clinical relevance,” J. Breath Res. 4(4), 047102 (2010). 8. M. Sawano, “Demonstration and quantification of the redistribution and oxidation of carbon monoxide in the human body by tracer analysis,” Med. Gas Res. 6(2), 59–63 (2016). 9. F. Keppler, A. Schiller, R. Ehehalt, M. Greule, J. Hartmann, and D. Polag, “Stable isotope and high precision concentration measurements confirm that all humans produce and exhale methane,” J. Breath Res. 10(1), 016003 (2016). 10. T. Fritsch, P. Hering, and M. Mürtz, “Infrared laser spectroscopy for online recording of exhaled carbon monoxide-a progress report,” J. Breath Res. 1(1), 014002 (2007). 11. M. Metsälä, F. M. Schmidt, M. Skyttä, O. Vaittinen, and L. Halonen, “Acetylene in breath: background levels and real-time elimination kinetics after smoking,” J. Breath Res. 4(4), 046003 (2010). 12. J. H. Shorter, D. D. Nelson, J. B. McManus, M. S. Zahniser, S. R. Sama, and D. K. Milton, “Clinical study of multiple breath biomarkers of asthma and COPD (NO, CO2, CO and N2O) by infrared laser spectroscopy,” J. Breath Res. 5(3), 037108 (2011). 13. L. Ciaffoni, D. P. O’Neill, J. H. Couper, G. A. Ritchie, G. Hancock, and P. A. Robbins, “In-airway molecular flow sensing: A new technology for continuous, noninvasive monitoring of oxygen consumption in critical care,” Sci. Adv. 2(8), e1600560 (2016). 14. W. Miekisch and J. K. Schubert, “From highly sophisticated analytical techniques to life-saving diagnostics: Technical developments in breath analysis,” Trac-Trend. Anal. Chem. 25(7), 665–673 (2006). 15. C. Wang and P. Sahay, “Breath analysis using laser spectroscopic techniques: breath biomarkers, spectral fingerprints, and detection limits,” Sensors (Basel) 9(10), 8230–8262 (2009). 16. O. Vaittinen, F. Manfred Schmidt, M. Metsala, and L. Halonen, “Exhaled breath biomonitoring using laser spectroscopy,” Curr. Anal. Chem. 9(3), 463–475 (2013). #283155 https://doi.org/10.1364/OE.25.012743 Journal © 2017 Received 3 Apr 2017; revised 11 May 2017; accepted 16 May 2017; published 23 May 2017 Vol. 25, No. 11 | 29 May 2017 | OPTICS EXPRESS 12743


Introduction
Breath gas analysis (BGA) has recently received wide attention and is increasingly used in medical research and diagnostics [1].Many of the volatile compounds exchanged at the alveolar interface in the lung are associated with health conditions and biological functions and thus can provide information on the physiological and metabolic state of the body [2].The main advantages compared to traditional clinical methods, such as blood sampling and tissue biopsy, are (i) completely non-invasive approach, (ii) unlimited sample volume and (iii) fast response time.Applications range from early disease detection, treatment monitoring and stable isotope tests to breathomics and validation of physiological models [1][2][3].
Atmospheric carbon monoxide (CO) arises primarily from incomplete organic combustion and appears in concentrations of 0.05-0.2parts per million by volume (ppmv) in clean outdoor air [4].The main endogenous source of exhaled breath CO (eCO) is systemic heme metabolism, catalyzed by heme oxygenase enzymes in response to oxidative stress, which typically results in end-tidal, mouth-exhaled eCO concentrations of 1-3 ppmv in the healthy population [5].Elevated eCO levels were found in smokers and diseased cohorts [6].In general, eCO is affected by exogenous sources, such as indoor air CO (up to several ppmv in poorly ventilated rooms), outdoor air pollution and cigarette smoke [5].A fraction of eCO may also originate from induced heme oxygenase in airway or lung tissue as a consequence of local oxidative stress and inflammation.If this contribution could be distinguished from systemic CO and exogenous sources, eCO may be useful as biomarker for air pollution health effects and related respiratory diseases [7].Possible strategies to discriminate between biomarker sources are real-time measurements, as well as isotope ratio and stable isotope tracer analysis [8,9].
Real-time BGA refers to online breath sampling with gas exchange and data acquisition times fast enough to resolve individual breath cycles (expirograms or exhalation profiles).This is in contrast to traditional offline sampling, where mixed or end-tidal breath is stored in sample containers for later analysis.Real-time BGA can be used to measure the response to exposure or interventions and extract physiological parameters [10][11][12][13], and to optimize breath sampling procedures.Another important, yet largely unexplored, feature is the possibility to obtain spatially resolved information about the respiratory tract.
Analytical techniques capable of real-time BGA with sufficient sensitivity are softionization mass spectrometry (normally not used for CO due to the low proton affinity of the molecule) and laser spectroscopy [14].In particular, laser absorption spectroscopy has had considerable impact on BGA owing to its inherent selectivity and accuracy [15].While cavity-enhanced techniques are often employed to reach the required sensitivities [16], many important biomarkers can today be accessed with mid-infrared tunable diode laser absorption spectroscopy (TDLAS) [15,17].The combination of quantum cascade lasers (QCLs) or interband cascade lasers (ICLs) with multipass cells (MPCs) and modulation techniques opens up for sensitive, yet robust and portable instrumentation useful for clinical applications [18].Distributed feedback ICLs are novel single-mode semiconductor lasers that cover the important spectral range 2.5-5 µm and offer narrow linewidth, wide current tuning, lowpower consumption and room-temperature operation with thermoelectric cooling [19].The rapid gas exchange times needed for real-time BGA can be achieved with low-volume MPCs [20].
Several cavity-enhanced absorption spectrometers [10,[21][22][23] and QCL-based TDLAS systems [12,23] have been utilized for detection of exhaled breath CO.Recently, we introduced a TDLAS sensor employing an external-cavity QCL for simultaneous detection of CO and carbon dioxide (CO 2 ) exhalation profiles [24].Isotopes of CO in breath have been measured with a lead salt diode laser system [25] and cavity ring down spectroscopy [10].Interband cascade lasers, however, despite their promising features, have so far rarely been applied to breath gas analysis [26] or CO detection [27].In general, using direct absorption spectroscopy (DAS), ICLs have shown excellent performance down to 10 −3 relative absorption in multipass cell-enhanced setups, but a sensitivity improvement due to wavelength modulation spectroscopy (WMS) could not be demonstrated [28].
In this work, we present a compact mid-infrared TDLAS spectrometer based on an interband cascade laser for real-time detection of CO in ambient air and breath.It is shown that wavelength modulation spectroscopy can significantly improve the sensitivity and precision of ICL spectrometers.The use of WMS in combination with a novel, circular multipass cell with low volume and fringe level enabled eCO detection with high precision, accuracy and time-resolution.The sensor is applied to real-time detection of 12 CO and 13 CO exhalation profiles from healthy subjects and to investigate breath CO isotope ratios.

Line selection
The fundamental rotation-vibrational P(3) transition of 12 CO at 2131.6316 cm −1 (4.6912 µm) with an actual (abundance-unweighted) line strength of 2.503 × 10 −19 cm −1 /(molecule•cm −2 ) was selected to maximize sensitivity to CO, while minimizing spectral interference due to other breath constituents.Figure 1(a) shows a HITRAN [29] simulation of the expected absorption spectrum around the P(3) transition for a CO concentration of 2 ppmv at a total pressure of 100 Torr and a temperature of 296 K.As potentially interfering species in this region, 5% CO 2 and 5% water vapor (H 2 O) were included in the simulation.Evidently, overlap with CO 2 is negligible, but there could be some minor spectral interference due to H 2 O.If necessary, exhaled H 2 O can be removed prior to analysis using a cold trap or Nafion tube.On the other hand, if the breath sampling system is unheated, as in this work, the H 2 O concentration in the sample cell might be considerably lower than 5% due to condensation [24].

Experimental TDLAS setup
A schematic drawing of the experimental setup is shown in Fig. 2(a).The thermoelectrically cooled, continuous-wave distributed-feedback ICL (Nanoplus GmbH) emitting around 2131 cm −1 was driven by a commercial low-noise laser current and temperature controller (ILX Lightwave, LDC-3724C).At a diode temperature of 4 °C and a current scan interval of 38-80 mA, a mode-hop-free tuning range of ~50 GHz (2130.09-2131.87cm −1 ) at an average output power of 1.4 mW was achieved.The laser was scanned across the absorption lines using a 140 Hz triangular waveform supplied by a function generator (Agilent, 33522A).In order to perform WMS, a sinusoidal waveform with a frequency of 44.13 kHz generated by an external lock-in amplifier (Stanford Research Systems, SR830 DSP) was superimposed on the scan waveform using a power combiner (Mini-Circuits, ZFRSC-2050 + ).The modulation depth was optimized to ~2.2 times the half-width half-maximum of the absorption line.The relative frequency scale of the laser scan was obtained prior to each set of measurements employing a solid Germanium etalon (LightMachinery, OP-5483-50.8) with a free-spectral-range of 734.2 MHz.
A plano-convex lens with a focal length of 100 mm was used to couple the laser beam to a circular multipass cell (MPC, IR Sweep, IRcell-4M), which provided an effective absorption path length of 399 cm in a configuration with 51-reflections, and had a volume of 38 ml.The fringe level was specified to below 0.39 ‰ rms of the dc intensity by the manufacturer.The output beam from the MPC was focused on a mid-infrared photodetector (VIGO System, PVI-2TE-5-1x1) using an off-axis parabolic mirror.An optical attenuator prior to the MPC was used to adjust the laser power to match the detector specifications (<1 mW).The detector signal was recorded at an acquisition rate of 16 MHz by a PC equipped with a 16-bit data acquisition card (Spectrum, M2i.4963-exp).In WMS mode, the detector signal was first demodulated by the lock-in amplifier, and the resulting 2f-WMS signal was acquired.
Online breath sampling was achieved using an unheated, home-built buffer tube made of Teflon with a volume of 30 ml, shown in Fig. 2(b), on which a disposable antibacterial filter (GVS, Eco Maxi Electrostatic Filter, 4222/701) was mounted as mouthpiece and to prevent contamination.An inline capnograph (Phillips Respironics, Capnostat 5) and a flow meter (Phillips Respironics, FloTrak Elite) were installed between filter and buffer tube to monitor exhaled carbon dioxide (eCO 2 ) as well as exhalation flow rate and volume.Subjects performed free tidal, mouth-exhaled breathing (both inhalation and exhalation) through the buffer tube, while a part of the sample was continuously drawn to the MPC.During inhalation, the tube was quickly flushed with ambient air, which marked the end of exhalation and improved the measurement of indoor air CO and the subsequent expirogram.The pressure in the MPC was monitored by a gauge (Leybold, Ceravac CTR100) and kept at 100 Torr with the help of a shut-off valve (Swagelok, EL3233) installed at the MPC inlet.The MPC outlet was connected to a vacuum pump (Leybold, Divac 1.4HV3C) with maximum pumping speed of 360 ml/s at 100 Torr, which resulted in MPC gas exchange times of the order of 0.1 s, and a sample flow rate of 50 ml/s from the atmospheric buffer tube to the cell.

Sensor performance
The performance of the TDLAS sensor was evaluated in terms of sensitivity, precision and linearity.Figure 3 presents experimental DAS and 2f-WMS raw signals (red markers, not all data points shown for clarity) of the P(3) transition at 2.76 ppmv 12 CO, recorded with 0.07 s integration time (10 averages).The sample was derived from a 28.24 ppmv CO gas standard.Each spectrum also shows a least-squares fit (solid line) of a simulated Voigt line shape to the raw data, as well as the corresponding fit residual.Direct absorption spectra were calculated using Beer-Lambert's law, with the nitrogen-filled MPC background signal as incident intensity and including a first-order polynomial function to account for drifts in the wavelength-dependent background.The analyte concentration was directly obtained from the fit.2f-WMS Voigt line shapes were simulated following the methodology proposed by Westberg et al. [30], and the 2f-WMS peak value extracted from the curve fit was used to obtain the CO concentration via calibration with DAS.
As exemplified in Fig. 3(a), the DAS scheme was limited by 1/f-noise, and, in practice, the detection limit was 230 parts per billion by volume (ppbv) at 0.07 s acquisition time, which implied a (3σ) sensitivity of 1.8 × 10 −6 cm −1 Hz -1/2 .In WMS mode, the 1/f-noise was efficiently removed, yielding a detection limit of 9 ppbv at 0.07 s integration time, and a corresponding (1σ) sensitivity of 6.5 × 10 −8 cm −1 Hz -1/2 .Thus, a sensitivity improvement of about 25 was achieved with WMS compared to DAS.The WMS sensitivity was ultimately limited by drifts and fluctuations in the fringe-dominated background signal.The overall fringe-to-signal level of 0.45‰ rms was close to the minimum value specified by the MPC manufacturer.
Precision and stability of the sensor were investigated by means of the Allan-Werle deviation recorded at CO concentrations of 2.2 ppmv and 510 ppbv for DAS and WMS, respectively, and plotted as a function of integration time in Fig. 4(a).The plots follow the general behavior; first, white noise dominates and the precision improves with integration time (slope~1/τ), but above 10 s integration the precision is limited by thermal drifts and fluctuations in the background signal (slope~τ 2 ).At 0.07 s integration time, the plots indicate a precision of 18 ppbv for DAS and 5 ppbv for WMS.The plots further suggest that a DAS precision of 2 ppbv and a WMS precision of 0.6 ppbv can be achieved at 20 s and 10 s integration time, respectively.The linearity of the instrument response to changes in 12 CO concentrations was tested for both detection schemes using 11 different CO/air mixtures at concentrations between 0.8 and 28.24 ppmv derived from the CO gas standard.The gas mixtures were generated using digital mass flow controllers (MKS Instruments, GM50A) at a total flow rate of 10 ml/s.While the DAS calibration curve in Fig. 4(b) exhibits a linear behavior in this concentration range, the 2f-WMS peak value is linear with concentration only in the optical thin limit, up to an absorbance of about 0.1 (10 ppmv).Considering typical eCO concentrations expected from healthy and diseased subjects and smokers, this upper limit for WMS detection is sufficient.Throughout the paper, 5 ppmv CO signals were used for 2f-WMS calibration.In case a breath test would require a larger dynamic range, rapid time-division multiplexed methods could be used for quasi-simultaneous DAS and WMS detection [31].

Detection of CO isotopes
Measured DAS spectra from a scan across the entire mode-hop-free current tuning range of the ICL at a diode temperature of 4 °C are displayed in Fig. 5(a), demonstrating detection of the three most important CO isotopes.The upper curve (solid black line) represents mixed breath collected from a smoker during smoking, whereas the lower spectrum (inverted; solid red line) is from the undiluted 28.24 ppmv CO gas standard, both acquired with an integration time of 10 s.Here, mixed breath was exhaled into an aluminum sample bag directly after inhaling cigarette smoke.The sample contained 117.80 ppmv 12 CO, 1.38 ppmv 13 CO and 274 ppbv 12 C 18 O, which exemplifies the high CO levels in cigarette smoke.The gas standard sample contained 300 ppbv of 13 CO and 63 ppbv 12 C 18 O.The noisy baseline in the breath sample suggests residual absorption from unidentified breath or smoke constituents.  1CO in the gas standard diluted to 2 ppmv 12 CO, together with Voigt fit and residual.For both spectra, the sample pressure was 100 Torr.  1CO in a diluted gas standard sample (2 ppmv 12 CO) flown through the MPC.In accordance with the spectrometer sensitivity, the detection limit for 13 CO in case of a stable background signal was 9 ppbv.This means that, for healthy subjects, e 13 CO concentrations will be close to the detection limit, and e 12 C 18 O will not be measurable.However, exposure [11] and stable isotope-labeled tracer studies [8] could readily be performed with this setup.
Isotopic abundances are usually expressed in terms of isotope ratios, and deviations from a standard ratio are referred to using the delta notation (δ, in units of per-mil) [32].In the case of 13 C/ 12 C, the internationally recognized standard ratio is set to 0.0112372 based on CO 2 in Pee Dee Belemnite (PDB) limestone.Since the PDB standard contains exceptionally high 13 C levels, almost all natural materials show negative δ 13 C values.Here, the CO-based δ 13 C values found in the standard gas and breath during smoking were −34 ± 25 ‰ and −8 ± 16 ‰, respectively, where the precision stems from the standard deviation of 10 samples.The obtained δ 13 C values are well within the range expected for clean air and biomass burning [33].

Carbon monoxide in indoor air and exhaled breath
Measurement sensitivity and time-resolution of the presented sensor are clearly suitable for real-time CO monitoring in ambient air and detection of eCO exhalation profiles.Knowledge of the inhaled biomarker concentration is a prerequisite for accurate and reliable BGA studies.
A typical 2f-WMS signal from 140 ppbv CO in indoor air sampled online in a wellventilated laboratory is shown in Fig. 6(a) together with a Voigt fit. Figure 6(b) presents a 2f-WMS spectrum from 1.01 ppmv 12 CO in alveolar breath of a healthy non-smoker.Spectral interference due to H 2 O was not observed.The structure in the fit residual is the result of a slight difference between the nitrogen-filled MPC and analytical background signals.A 2f-WMS signal from 38 ppbv e 13 CO in alveolar breath of a healthy smoker after smoking is shown in Fig. 6(c).Since e 13 CO detection was performed close to the detection limit, the difference between the MPC and analytical background signals, which can probably be attributed to changes in fringe structure and interference from other volatile breath compounds, was more pronounced.To improve e 13 CO quantification, the backgroundsubtracted background signal was again subtracted from the background-subtracted analytical 2f-WMS spectrum.Typical e 12 CO and e 13 CO exhalation profiles from a healthy occasional smoker before smoking (19 h after the last cigarette) and 15 s after smoking are displayed in Fig. 8.The e 13 CO breath cycles were recorded about 30 s after e 12 CO.While the eCO levels are in the healthy population range before smoking, a two-fold increase was found after smoking.The shapes of the e 12 CO and e 13 CO expirograms are similar.Isotope ratio analysis reveals δ 13 C values of −268 ± 84 ‰ and −168 ± 39 ‰ before and after smoking, respectively.The precision was determined based on 10 pairs of spectra.After smoking, δ 13 C is less negative probably due to the high levels of 13 C in cigarette smoke.Typical δ 13 C values in the breath of healthy non-smokers (data not shown) were around −200 ‰.
Although, with the present setup, the isotopic species could not be analyzed simultaneously and the 13 CO concentrations had a relatively high uncertainty (5 ppbv), the results indicate depletion of 13 CO in the breath of healthy subjects.The deviation from the standard 13 C/ 12 C ratio seems larger for eCO than for eCO 2 , for which a δ 13 C value of −20‰ was reported [32].A probable 13 CO sink is oxidation to CO 2 [8].The results presented here are in contrast to the findings by Lee et al., who reported positive δ 13 C values in non-smokers and smokers [25].Overall, the successful combination of a low-volume multipass cell with ICL-based WMS enabled detection of eCO exhalation profiles of a quality previously only reached with cavityenhanced absorption techniques [10,22].Similar TDLAS systems are feasible for detection of other interesting biomarkers with exhaled breath concentrations in the high ppbv to ppmv range, such as ammonia (NH 3 ), methane (CH 4 ) and nitrous oxide (N 2 O).Precise real-time BGA with TDLAS will enable detailed clinical studies of biomarker physiology and toxicology, as well as coupling of experimental data to models of gas exchange in the respiratory tract.

Conclusions
A compact mid-infrared TDLAS sensor for carbon monoxide in air and breath was realized based on an interband cascade laser operating at 4.69 µm.Wavelength modulation spectroscopy and a novel, low-volume multipass cell were implemented to achieve a detection limit and precision in the low ppbv range, while maintaining sub-second gas exchange times.Compared to direct absorption spectroscopy a sensitivity improvement of 25 was reached with WMS.The sensor was applied to real-time detection of 12 CO and 13 CO exhalation profiles from non-smokers and an occasional smoker before and after smoking.Elevated eCO levels were found after smoking.Isotope ratio analysis revealed depletion of

Fig. 1 .
Fig. 1.HITRAN2012 simulation of absorption spectra expected from a breath sample containing 2 ppmv CO, 5% CO 2 and 5% H 2 O at 100 Torr and 296 K. (a) Around the P(3) 12 CO transition, (b) for a 50 GHz ICL current tuning range, also covering 13 C 16 O and 12 C 18 O.The wide mode-hop-free current tuning range (typically 50 GHz) of ICLs also allows detection of the stable isotopes 13 C 16 O and 12 C 18 O, with actual line strengths of 3.914 × 10 −19 cm −1 /(molecule•cm −2 ) and 3.596 × 10 −19 cm −1 /(molecule•cm −2 ), respectively, in the spectral region around the P(3) transition.Figure 1(b) depicts a corresponding HITRAN simulation, again including 5% CO 2 and 5% H 2 O.The close-up on the isotope species in the inset suggests that 13 C 16 O can be detected with minimal spectral interference, whereas some overlap with H 2 O may occur for a 12 C 18 O measurement.

Fig. 3 .
Fig. 3. Measured DAS (a) and background-corrected 2f-WMS (b) signals (markers) of the P(3)12 CO transition recorded at 0.07 s integration time (10 averages), with least-squares Voigt fits (blue, solid line) and fit residuals.The sample was a 2.76 ppmv gas standard at 100 Torr.For clarity, only every 10th and 30th data point is shown in the DAS and WMS spectra, respectively.

Fig. 4 .
Fig. 4. (a) Allan-Werle deviation plots for DAS at 2.2 ppmv and 2f-WMS at 510 ppbv, (b) calibration curve for DAS (blue, triangular markers) and 2f-WMS (red, circular markers) obtained by dilution of a 28.24 ppmv CO gas standard.Solid line -linear fit to DAS response.

Fig. 5 .
Fig. 5. (a) Measured DAS spectra over the full ICL current tuning range from a breath sample collected during smoking (black line) and from a 28.24 ppmv CO gas standard (inverted, red line), (b) Measured 2f-WMS signal (markers) from13 CO in the gas standard diluted to 2 ppmv12 CO, together with Voigt fit and residual.For both spectra, the sample pressure was 100 Torr.

Figure 5 (
Figure 5(b) shows a background-corrected 2f-WMS signal from 19 ppbv13 CO in a diluted gas standard sample (2 ppmv 12 CO) flown through the MPC.In accordance with the spectrometer sensitivity, the detection limit for13 CO in case of a stable background signal was 9 ppbv.This means that, for healthy subjects, e13 CO concentrations will be close to the detection limit, and e 12 C 18 O will not be measurable.However, exposure[11] and stable isotope-labeled tracer studies[8] could readily be performed with this setup.Isotopic abundances are usually expressed in terms of isotope ratios, and deviations from a standard ratio are referred to using the delta notation (δ, in units of per-mil)[32].In the case of13 C/ 12 C, the internationally recognized standard ratio is set to 0.0112372 based on CO 2 in Pee Dee Belemnite (PDB) limestone.Since the PDB standard contains exceptionally high13 C levels, almost all natural materials show negative δ 13 C values.Here, the CO-based δ 13 C values found in the standard gas and breath during smoking were −34 ± 25 ‰ and −8 ± 16 ‰, respectively, where the precision stems from the standard deviation of 10 samples.The obtained δ 13 C values are well within the range expected for clean air and biomass burning[33].

Fig. 6 .
Fig. 6.Typical 2f-WMS signals (markers) recorded during online sampling at 100 Torr, with least-squares Voigt fits (blue solid lines) and residuals (a) 12 CO in indoor air, (b) e 12 CO in alveolar breath of a healthy non-smoker, (c) e 13 CO in alveolar breath of a healthy smoker after smoking.For clarity, only every 40th (a, b) and 20th (c) data point is shown in the raw spectra.

Figure 7
Figure7presents typical sequences of e12 CO expirograms from two healthy non-smokers exhaling at flow rates of around 250 ml/s (a) and 150 ml/s (b) while performing tidal plus expiratory reserve volume breathing and tidal breathing, respectively.The observed end-tidal eCO concentrations are within the range expected for healthy subjects (1-3 ppmv).During the inhalation period, indoor air was sampled.Each marker in Fig.7(a) corresponds to the CO concentration derived from one recorded and evaluated raw spectrum.No data smoothing was applied in the e12 CO expirograms shown in this work.The noise level on the breath cycles lies within the precision determined from the Allan-Werle plots for 0.07 s integration time.In Fig.7(b), eCO 2 expirograms recorded simultaneously with eCO by capnography close to the mouth are shown for comparison.The identical exhalation time for the two gases confirms true real-time detection without sample line delay, as previously verified for this sampling system[24].

Fig. 7 .
Fig. 7. Sequences of e 12 CO expirograms obtained from two healthy non-smokers during tidal plus expiratory reserve volume breathing (a) and tidal breathing (b).No data smoothing was applied.(a) Each data point corresponds to a raw spectrum.(b) CO 2 exhalation profiles simultaneously obtained by capnography are shown for comparison.

Fig. 8 .
Fig. 8. Sequences of eCO expirograms from a healthy occasional smoker before smoking (19 h after the last cigarette) and 15 s after smoking; (a) e 12 CO profiles, (b) e 13 CO expirograms recorded 30 s after e 12 CO.Gray -raw data, colored lines -smoothed raw data.