Medical Diagnostics Based on Exhaled-Air Analysis with Laser Absorption Spectroscopy

Michele Gianella and Grant A.D. Ritchie

European Flag

This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no 327441.

Objective & Overview

The goal of project "MEDIXHALE" is to develop a laser absorption spectroscopy sensor to assist physicians in the diagnosis and/or monitoring of patients with cystic fibrosis (CF). To do so, the sensor measures a specific chemical compound in exhaled breath.

A cavity-enhanced laser absorption spectrometer with a sample preconcentration stage has been built to quantitatively measure acetonitrile in exhaled breath. The system has been fully characterized and tested with breath sampled from two healthy adults.

Our results have been published in Analytical Chemistry:

Michele Gianella and Grant A.D. Ritchie, "Cavity-Enhanced Near-Infrared Laser Absorption Spectrometer for the Measurement of Acetonitrile in Breath", Analytical Chemistry (2015).


For further information please feel free to contact us:

Michele Gianella |
Grant Ritchie |

University of Oxford | Department of Chemistry
Physical and Theoretical Chemistry Laboratory
South Parks Road
Oxford, OX1 3QZ, United Kingdom
P: +44 (0) 1865 275489 P: +44 (0) 1865 285723

History of Breath Analysis and Diagnostics

The potential of breath analysis for medical diagnostics has been known and recognized for millennia [1].

But while traditionally physicians had to rely on their sense of smell, in more recent times a number of highly sensitive and selective analytical techniques have become available. In 1971 about 250 substances were discovered in breath, and many more were identified in the following years [2]. The concentrations of this large number of compounds varies greatly from person to person, but also for the same person throughout the day, depending on activity and metabolism.

Diseases can cause significant changes in the concentrations of one or more substances. The variations in these so-called biomarkers therefore have diagnostic potential. Biomarker concentrations are typically very small, below 1 ppm (part per million).

Lavoisier's experiments with human respiration Antoine Lavoisier carries out an experiment to study the oxygen content of air exhaled from a man's lungs.

Cystic Fibrosis and Acetonitrile

Cystic fibrosis is a hereditary disease caused by a fault in the CFTR gene (Cystic Fibrosis Transmembrane Conductance): it causes an excessive production of mucus that clogs the airways and the digestive system.

The number of newborns with CF varies greatly across different ethnic groups: in the UK 1 in 2500 babies is born with the disease. There is yet no cure for CF, but a number of treatments to manage the symptoms are available.

In a small-scale study conducted in our group [3], acetonitrile (methyl cyanide, CH3CN, abbreviated here as ACN) was identified as a good discriminator between a group of CF patients and a control group. It is likely that acetonitrile is produced, alongside with hydrogen cyanide (HCN), by bacteria that often infect the lungs and airways of CF patients. The concentrations of acetonitrile in breath are of the order of 10 ppb (parts per billion).

Healthy and CF lung Healthy airway (above) and airway with cystic fibrosis (below). Source: National Heart, Lung, and Blood Institute; National Institutes of Health; U.S. Department of Health and Human Services.
The acetonitrile molecule The acetonitrile (methyl cyanide) molecule.
The acetonitrile permeation device An acetonitrile permeation device, used to produce a known ACN concentration.

Absorption Spectroscopy

Every molecule has a set of discrete and unique allowed energies. Given certain selection rules are fulfilled, transitions from a lower to a higher energetic state can be induced by the absorption of a photon, but only if the photon has an adequate energy. The absorbed photons are missing from the light transmitted through the sample: this creates a signature - called an absorption spectrum - that is unique to the molecule.

One of the most important laws in absorption spectroscopy is Beer-Lambert's, which states that the optical power P1 transmitted through a sample cell depends exponentially on the product of length L of the cell and absorption coefficient a of the sample.

The absorption coefficient depends on the wavelength and is a property of the sample. It depends on the structure of the molecule, the amount (concentration), and on environmental factors such as pressure and temperature. The absorption coefficient is measured in cm - 1 (inverse centimetres). A value of α = 1 cm - 1 means that for every centimetre of cell length, the optical power decreases by a factor of e (approximately 2.72).

In the gas-phase spectroscopy of trace compounds we generally deal with absorption coefficients of α = 10-9 cm-1 or less, which means that for every centimetre of cell length the optical power decreases by one billionth (0.000 000 1 %) of its initial value. Such weak absorptions can only be measured if the path length L is sufficiently long. Indeed, after a path length of L = 10 km the optical power decreases by 0.1 %, which is a small but measurable change. Because building kilometre-long sample cells is not practical, a different approach based on an optical cavity is often chosen.

Beer-Lambert Absorption of light during a single pass through a cell filled with an absorber.
Carbon dioxide absorption spectrum The absorption spectrum of carbon dioxide at 67 mbar.

Optical Cavities

An optical cavity consists of two slightly concave highly reflective mirrors facing each other and separated by some distance L.

A laser beam is aimed at the first mirror. Because of the high reflectivity of the mirror, R, most of the incident laser light is reflected back and lost. The remainder, however, makes it into the cavity and circulates, on average, for several kilometres before leaving through the mirror on the right (and of course also through the mirror on the left, but that signal is not measured). The relevant path length is no longer the mirror separation distance, but the effective distance travelled by the light while bouncing between the two mirrors. The decrease in transmitted power due to the presence of an absorber in the cavity takes on the form [4]

(Pna - P)/Pna = (GαL)/(1 + GαL),

where G = R / (1 - R) is called the gain of the cavity, and Pna is the transmitted power if there were no absorption. The product GL is the effective absorption path length of the cavity. For example, for R = 99.99 %, G = 104, so that the effective path length is 10 000 times the physical length of the cavity!

Cavity-enhanced absorption spectroscopy A light beam injected into a linear cavity consisting of two slightly concave mirrors with an absorber held in between.
Cavity-enhanced absorption spectroscopy When the laser wavelength is tuned across an absorption line, the power transmitted through the cavity dips by a small amount.

Cavity-Enhanced Absorption Spectroscopy

When we employ optical cavities for absorption spectroscopy we speak about cavity-enhanced absorption spectroscopy.

The cavity-enhanced absorption spectrometer employs a near-infrared distributed feedback (DFB) diode laser (lambda = 1654 nm) and an optical cavity with a physical length of 80 cm. The reflectivity of the mirrors is approximately 99.99 %, so the cavity gain is G = 104 and the effective path length measured with a ring-down experiment is 7.3 km.

Optical setup Laser-based cavity enhanced absorption spectrometer.
DFB laser mount Distributed feedback laser in 14-pin butterfly package.
Optical cavities Optical cavities.


To further increase the sensitivity of the system we preconcentrate the gaseous samples by employing an adsorption/thermal desorption cycle with a carbon molecular sieve (CMS).

During adsorption, the sample is flushed through the CMS, which acts as a retaining filter for ACN (and other substances), but not for nitrogen and oxygen, which make up about 90 % of the breath volume.

Later, the CMS is heated very quickly to 200 degrees Celsius or more and a small volume of nitrogen is flushed through it and into the evacuated cavity. ACN desorbs easily from the CMS at these temperatures, so only a very small volume of nitrogen is required to extract most of the ACN from the CMS. The whole amount of ACN that was present in the initial sample (with a volume of, say, 1 L), is now available in a much smaller volume (say, 20 mL), which result in an increase of the ACN concentration by, theoretically, a factor of 50. In practice, starting with 1 L we achieved preconcentration factors (PCF) of 30.

Conditioning simply involves heating the CMS to 360 degrees Celsius and flush helium through it for 15 min to clean and prepare the CMS for the next run.

Thermal desorption setup The preconcentration stage used to increase acetonitrile concentrations.
Carbon molecular sieve A carbon molecular sieve (Carbosieve S-III).


The limit of detection is the smallest ACN concentration that can be reliably detected.

If the concentration is below a certain value, the limit of detection (LOD), the sample is undistinguishable from a blank sample.

By measuring a series of blanks (with a known concentration of zero) we obtain an error distribution. From it we can gather the average measurement error and, consequently, the LOD.

For instance, if the error in the concentration were ±0.1 ppm, then we couldn't distinguish between a non-blank sample with a concentration of 0.1 ppm and a blank sample (with a concentration of 0 ppm), since the two measurements would not be significantly different (0.1±0.1 ppm for the non-blank and 0±0.1 ppm for the blank).

The LOD of acetonitrile in nitrogen is 72 ppb. When large amounts of carbon dioxide are present, as is the case in breath, the LOD worsens to 114 ppb. Unfortunately this is not enough to detect ACN in breath (we need 10 ppb). However, the preconcentration stage allows us to increase the original concentrations by at least a factor of 30. Starting with 10 ppb we would then have, after preconcentration, 300 ppb, a concentration that is easily measured.

Effective path length 7.3 km
Sample pressure 66.7 mbar
Cavity volume 350 cm3
Acquisition rate 138 spectra / s
Measurement time 4 min 51 s
Limit of detection 72 ppb ACN in N2 / 114 ppb ACN in breath
Sensitivity Error distribution.

Breath Measurements

To test the system under real conditions, we measured the absorption spectra of preconcentrated breath samples taken from the two authors.

Absorption features due to water vapour and carbon dioxide can easily be identified in the upper panels. In the lower panels of the second figure the vertical axis is scaled so that the ACN absorption feature becomes visible. The measured concentrations were 690 ppb and 870 ppb. However, as the samples had been concentrated, the concentrations originally present in breath were only 23 ppb and 29 ppb, respectively, which are close to values we expect in the breath of healthy individuals.

Breath sample bags Special breath sample bags.
Absorption spectra of breath Absorption spectra of preconcentrated breath samples.

References / Further Reading