On-sorbent derivatization of carbonyl compounds in exhaled breath

UNIVERSITY’ OF PISA

Department of Chemistry and Industrial Chemistry

Master degree in Chemistry

Curriculum: Analytical Chemistry

On-sorbent derivatization of carbonyl compounds in

exhaled breath

Supervisors

Dott. Fabio Di Francesco

Dott. Tommaso Lomonaco

Examiner

Dott. Alessandro Mandoli

Table of content

1.2 Volatile Organic Compounds in human breath ... 17

1.3 Clinical Application of VOCs in breath analysis ... 19

1.4 Breath Analysis ... 23

1.5 Breath Sampling ... 24

1.5.1 Mixed or alveolar breath fraction ... 24

1.5.2 Nose vs mouth sample collection ... 27

1.5.3 Single or multiple breathes ... 28

Chapter 2 ... 29

2.1 Thermal desorption technique to analyze VOCs ... 29

2.2 Derivatization methods for analyze carbonyl compounds in air ... 35

2.2.1. Comparison between PFBHA and DNPH ... 36

2.2.2. UNICHIM Method ... 36

Chapter 3 ... 39

3.1 Chemicals and Materials... 39

3.2 Preparation of standard mixtures ... 40

3.2.1 Liquid mixtures ... 40

3.2.2 Gaseous mixtures ... 41

3.3 Preparation of Tenax GR sorbent tubes ... 42

3.4 Instrumentation and procedures ... 43

3.5 Experimental design for the optimization of derivatization procedure into Tenax GR sorbent tubes ... 48

3.6 Addition of the internal standard ... 48

3.7 Effect of humidity level and sampling flow rate on the derivatization reaction ... 49

3.8 Blank samples ... 49

3.9 Breakthrough ... 50

3.11.1 Stability of PFBHA and Tol-D8 in MeOH solution ... 50

3.11.2 Stability of PFBHA loaded into Tenax GR sorbent tube ... 51

3.12 Calibration curve, limit of detection (LOD) and quantification (LOQ) ... 51

Chapter 4 ... 53

4.1 Optimization of TD-GC-MS method... 53

4.2 Experimental design for the optimization of derivatization procedure into Tenax GR sorbent tubes ... 58

4.3 Addition of the internal standard ... 63

4.4 Effect of humidity level and sampling flow rate on the derivatization reaction .... 65

4.5 Breakthrough ... 70

4.6 Carry over ... 72

4.7 Stability study ... 72

4.7.1 Stability of PFBHA and Tol-D8 in MeOH solution ... 72

4.7.2 Stability of PFBHA loaded into Tenax GR sorbent tube ... 73

4.8 Calibration curve, limit of detection (LOD) and quantification (LOQ) ... 73

4.9 Quality control of analytical data ... 74

4.10 Application to real samples ... 75

Conclusions ... 79

List of Figures

Figure 1. Scheme of the respiratory system. ... 13 Figure 2. Representation of the respiratory membrane of the alveoli. ... 14 Figure 3. Expirogram reprinted from []. Solid line represents a gas with high solubility in water exchanging in the upper airways, whereas poorly soluble gas exchanging in the alveoli are represented by parallel lines. ... 15 Figure 4. Figure reprinted from [4]. Relation between Exchange Ratio (ER) and blood air partition coefficient... 16 Figure 5. Figure Reprinted from [19]. VOCs percentages in bodily fluids based on data from [18]. ... 17 Figure 6. Figure reprinted from [27]. Metabolic pathway for the synthesis of Acetone, Butyrate and 1-Butanol from pyruvate. ... 19 Figure 7. Figure reprinted from [64]. Partial pressure of CO2 during a respiratory cycle. ... 25

Figure 8. Figure reprinted from [83]. Illustration of the concentration potential of multi-stage TD. ... 31 Figure 9. Reaction mechanism between DNPH and a generic aldehyde/ketone... 35 Figure 10. Reaction mechanism between PFBHA and a generic aldehyde/ketone... 35 Figure 11. Deflated Nalophan bag (47 cm × 47 cm, S/V = 0.3 cm-1_{) composed of (a) Nalophan}

piece, (b) PTFE tube and (c) stopcock. ... 40 Figure 12. Markes’ Calibration Solution Loading Rig (CSLR). (a) Injection port, (b) Tenax GR (60/80 mesh) sorbent tubes, (c) tube connected to a pocket pump for controlling the sampling flow rate (50 mL/min), (d) tube connected to Nalophan bag (47 × 47 cm, S/V = 0.3 cm-1) containing medical air or working standard gaseous mixtures of analytes and (e) needle valve to regulate the sampling flow rate trough the sorbent tube. ... 43 Figure 13. Extracted ion chromatogram (m/z 181) of Aldehydes-PFBHA derivatives. Peak 1) Formaldehyde; 2) PFBHA; 3) Acetaldehyde; 4) Propanal; 5) Acrolein; 6) Isobutyraldehyde; 7) Methacrolein; 8) Butanal; 9) 2-Methylbutyraldehyde; 10) 3-Methylbutyraldehyde; 11) Pentanal; 12) Hexanal; 13) Heptanal; 14) Octanal; 15) Benzaldehyde; 16) Nonanal; 17) Glyoxal; 18) Pyruvaldehyde. ... 54 Figure 14. Extracted ion chromatogram (m/z 181) of branched Ketones-PFBHA derivatives. Peak 1) Acetone; 2) 2-Butanone; 3) 2-Pentanone; 4) 3-Pentanone; 5) 2-Hexanone; 6) 4-Heptanone; 7) 3-4-Heptanone; 8) 2-4-Heptanone; 9) 3-Octanone; 10) 2-Methylcyclohexanone; 11) 2-Octanone; 12) 3-Methylcyclohexanone; 13) 5-Nonanone; 14)

4-Figure 15. Extracted ion chromatogram (m/z 181) of Ketones-PFBHA derivatives. Peak 1) 3-Methyl-2-butanone; 2) 3,3-Dimethyl-2-butanone; 3) 2-Methyl-3-pentanone; 4) 4-Methyl-2-pentanone; 5) 2-Methyl-3-hexanone; 6) 4-Methylpent-3-en-2-one; 7) 2,4-Dimethyl-3-pentanone; 8) 5-Methyl-2-hexanone; 9) Cyclo2,4-Dimethyl-3-pentanone; 10) 2-Methyl-3-heptanone; 11) 2,6-Dimethyl-4-heptanone; 12) Cyclohexanone; 13) Acetophenone. ... 55 Figure 16. Product ion mass spectrum for methacrolein-PFBHA oxime at three collision energy: 2 eV (A), 4 eV (B) and 6 eV (C). ... 56 Figure 17. Plot for Butanal (A), 2-Butanone (B) and 3-methyl-2-butanone (C): Replicates plot, Summary of fit - R2 (blank), Q2 (blue), Model validity (yellow) and Reproducibility (green) – Coefficient plot and Observed vs Predicted plot. ... 60 Figure 18. Average values of the areas of the aldehydes-PFBHA adducts (A), branched

ketones-PFBHA adducts (B) and ketones-ketones-PFBHA adducts (C) in the humid and dry gaseous standard mixture, normalized with respect to the area corresponding to the dry sample. Error bars correspond to the standard deviation of three replicates. ... 66 Figure 19. Effect of the relative humidity levels (90 and <15%) on the peak area for the following oximes: Butanal-PFBHA (A), 2-Eptanal-PFBHA (B), 2-Butanone-PFBHA (C) and 2-Eptanone-PFBHA (D). Error bars correspond to the standard deviation of three replicates. ... 67 Figure 20. Average values of the areas of the aldehydes-PFBHA adducts (A), branched

ketones-PFBHA adducts (B) and ketones-ketones-PFBHA adducts (C) collected at 30 and 130 mL/min, normalized with respect to the area corresponding to the sample collected at 30 mL/min. Error bars correspond to the standard deviation of three replicates. ... 69 Figure 21. Relationship between peak area and sampling volume for Propanal (A), 2-butanone (B) and Acetone (C) on PFBHA-coated Tenax GR sorbent tube. Error bars correspond to the standard deviation of three replicates. ... 71 Figure 22. Toluene-D8 control chart reporting the daily average values of the ratio between Toluene-D8 peak area and amount (red dots), the average of all the ratios obtained during the experimental period (dark line), the attention limits (average ± SD, blue line) and the control limits (average ± 2SD, red line). Error bars correspond to the standard deviation of three replicates. ... 75 Figure 23. Normalized data for the main ketones (A) and aldehydes (B) determined in exhaled breath collected from a patient suffering from heart failure. ... 76

List of tables

Table 1. Data from [90]. Sorbent materials for thermal desorption. ... 34 Table 2. Concentration of analytes in the glass flask calculated at 37 °C and ambient pressure. ... 42 Table 3. Retention times, quantifier ion, quantifier/qualifier transition and its collision energy

(CE) for all the investigated compounds. In bold the ion or MRM transition used as quantifier. ... 45 Table 4. Experimental range for the investigated factor levels. ... 48 Table 5. Optimized TD conditions for the desorption of oximes from Tenax GR sorbent tube. ... 54 Table 6. Qualifier to quantifier ratio (q/Q) for each investigated compound. ... 57 Table 7. Experimental table for 23 factorial design. ... 59 Table 8. Peak area of Butanal-PFBHA, 2-butanone-PFBHA and 3-methyl-2-butanone-PFBHA

observed for experiment 5 and 6. ... 62 Table 9. Amount and peak area of Acetone-D6-PFBHA oxime (m/z 259) obtained when 10, 25

and 50 μL of stock gaseous solution of Acetone-D6 (873 ppmv) were injected during the transfer (50 mL/min) of 250 mL of dried medical air into pre-coated sorbent tube. Data reported as mean ± standard deviation. ... 64 Table 10. Peak area of Butanal-, 2-Butanone-, 2-Eptanal- and 2-Eptanone-PFBHA under dry

Acronyms

CI = Chemical Ionization CE = Collision Energy CF = Cystic Fibrosis

DoE = Design of Experiments DVB = Divinylbenzene EI = Extract Ion

FSR = Full Scan Range

GC-MS = Gas Chromatography Mass Spectrometry HS = Head Space

IS = Internal Standard ISR = Inlet Split Ratio m/z = mass to charge ratio

MRM = Multiple Reaction Monitoring MW = Molecular Weight

NTD = Needle Trap Device OSR = Outlet Split Ratio PDMS = Polydimethylsiloxane

PFBHA = O-2,3,4,5,6-Pentafluorobenzylhydroxylammine hydrochloride PUFA = Polyunsaturated Fatty Acid

QI = Quantifier Ion qI = qualifier Ion

RH = Relative Humidity

ROS = Reactive Oxygen Species RT = Retention Time

RTemp = Room Temperature SIM = Selected Ion Monitoring SNR = Signal to Noise Ratio

SPME = Solid Phase Micro Extraction ST = Sorbent Tubes

TB = Tuberculosis

TD = Thermal Desorption TH = Trap Hold

Abstract

Over the last few years, breath analysis for monitoring metabolic disorders caused by specific diseases has become more and more attractive due to its non-invasiveness. Various classes of volatile organic compounds (VOCs) have been proposed for the early diagnosis of lung cancer and the monitoring of diabetes and asthma.

In this context, oxygen-containing substances (e.g. aldehydes and ketones) are interesting as related to oxidative stress, which is associated to serious illnesses such as arteriosclerosis, Parkinson syndrome and pulmonary embolism. Among these substances, aldehydes are particularly difficult to monitor in exhaled breath since they are reactive compounds and tend to decompose or react during sample preparation or storage. In addition, analytical problems come from their low concentrations (from low- to sub-ppbv levels) in exhaled breath. Chemical derivatization, ideally coupled with pre-concentration techniques, may offer a solution to overcome these problems. The most commonly used approach is to collect the analytes on solid sorbents coated with a suitable derivatization agent (i.e. 2,4-Dinitrophenylhydrazine), followed by solvent desorption using acetonitrile and liquid injection for analysis by high-pressure liquid chromatography.

Aim of this work was to develop an analytical procedure including an on-sorbent derivatization and a thermal desorption coupled to GC-MS/MS. Carbonyls in standard gaseous mixtures and/or breath samples were collected onto O-(2,3,4,5,6-pentafluorobenzyl) hydroxylamine (PFBHA)-coated Tenax GR sorbent packed in a stainless-steel tube. A mixture of carbonyl compounds, composed by linear aldehydes (C2-C9), linear (C3-C9) and branched ketones, were used to optimize the analytical performances of the on-sorbent technique. Pre-coated sorbent tubes were thermally desorbed by a TD-100 multi-tubes auto-sampler (Markes International) and oximes were analyzed by a 7890B GC System equipped with a DB5ms gas chromatographic column (60 m, 0.25 mm ID, 1.0 μm) and coupled to a 7010 MS Triple Quad mass spectrometer (Agilent Technologies). The effect of temperature and reaction time as well as the amount of PFBHA spiked into Tenax GR sorbent tubes was evaluated using a 23 full factorial design. In addition, the influence of the humidity level and sampling flow rate on the efficiency of carbonyl collection into pre-coated sorbent tubes was assessed.

The optimized method including derivatization, extraction, and analysis allows to successfully determine carbonyl analytes in breath. A quantitative linear response in the range 0.1-20 ppbv, detection limits close to 50 pptv and no carry-over effect were observed.

Chapter 1

Breath analysis: physiological basis and sampling techniques

1.1 Respiratory system

The respiratory system is the biological system that transfers O2 to the blood and removes

the CO2 produced by the cellular metabolism [1]. As shown in Figure 1, this system consists of

two sections: a) the upper respiratory tract and b) the lower respiratory tract.

Figure 1. Scheme of the respiratory system.

The primary function of the upper respiratory tract, composed by nose, pharynx, epiglottis and the larynx, is to filter, humidify and warm the air inhaled to body temperature (34-37 °C). On the contrary, in the lower respiratory tract, which is composed by trachea, bronchi, bronchioles and the lungs, the O2 inspired from the ambient air is exchanged with the CO2

produced by the cellular metabolism. During inspiration, air flows through the trachea and reaches the pulmonary alveoli, small sacs that passively expand and relax along with the movements of the rib cage. These sacs are rich in capillaries and the gas exchange happens across their thin walls. When the diaphragm contracts, a negative pressure is produced in the thorax, and the air rush to fill the alveoli, which cause the lungs to expand. During the

air filled with CO2 from the cells is pushed out. At rest, the minute ventilation (VE), which is

the volume of air moved in and out of the lungs per minute, it’s usually around 6 L/min. During respiration, a healthy subject normally exchanges 500 mL of air in 5 s. This volume is called tidal volume (VT) [2] and it is made up of two fractions: the former (about 150 mL)

represents the dead space section contained the upper airways, whereas the latter (about 350 mL) represents the alveolar fraction directly involved in the gas exchange in the alveolar sacs. During inspiration, 500 mL of ambient air are inhaled, 350 mL of which reach the lung together with 150 mL of alveolar air left in the dead space from the previous exhalation. This air is mixed and diluted with the alveolar air contained in the lungs, while the remaining 150 mL of ambient air fills the dead space. During expiration, the ambient air contained in the dead space is emitted through the mouth and/or nose together with 350 mL of air coming from the lungs. During respiration, the dead space is alternatively filled with ambient or alveolar air derived from the previous expiration [1]. Normally ambient air contains about 21% of O2, 78% of N2,

and other gases at trace levels such as CO2 (0.04%). Due to gas exchange in the lung, the expired

air contains much more CO2 (up to 5%) and other gases at trace levels that reflect the metabolic

and biological activity as well as the present and past exposure to ambient air [3]. Such exchange is caused by the diffusion across the respiratory membrane (Figure 2), and includes both respiratory and non-respiratory gases.

Figure 2. Representation of the respiratory membrane of the alveoli.

These volatile organic compounds (VOCs) can be classified according to their origin as

ambient air. Anderson et al.[4] stated that “a comprehensive understanding of gas exchange is necessary to interpret the breath concentrations, to compare them to blood concentrations and to identify possible breath biomarkers”. In fact, the gas exchange also occurs in the conducting airways if some conditions are met. The location of this process depends on the water solubility of the gas, described by the water-air partition coefficient (λw:a). Gas with a λw:a < 10 are almost

exclusively exchanged in the alveoli, whereas gases with λw:a > 100 remarkably exchange also

in the airways. Gases with intermediate values of water-air partition coefficient exchange in both locations. A difference in the value of λw:a entails a different behavior of these gases during

the exhalation. In the following graph (namely expirogram) proposed by Anderson (Figure 3) this difference is clearly visible. This graph shows the relationship between the partial pressure of a gas (pE) and the exhaled volume. Phase I represents the initial volume that does not contain soluble gases, i.e. the anatomic dead space air. The rapidly rising phase labeled as phase II represents the transition between dead space and alveolar air, whereas phase III represents the air coming from alveoli.

Figure 3. Expirogram reprinted from [3]. Solid line represents a gas with high solubility in water exchanging in

the upper airways, whereas poorly soluble gas exchanging in the alveoli are represented by parallel lines.

After inspired air reaches the alveoli, poorly blood soluble gases diffuse into the airstream. During expiration, gases present in the mouth and in the upper airways are expelled first, causing the phase I of the expirogram. The air contained in the alveoli, as well as the low blood soluble gases, is the last to exit during the phase III. As it can be seen in the expirogram, there

exchange. Therefore, an anatomical dead space cannot be defined for these gases. The high blood soluble gases are absorbed by the inspired air from the mucus layer of the airways and deposited back to the airway wall during the expiration. The release is caused by the gradient generated between the air, saturated with gases, and the conducting airways, where the partial pressure of gases is low due to the previous absorption step. This wash-in/wash-out cycle causes the partial pressure of high blood soluble gases, in the end-exhaled breath (circle), to always be less than the partial pressure of low blood soluble gases present in the alveoli at the same time. This leads to a problematic quantification of the relation between the amount of these gases present in the end-exhaled breath, and the same gases found in the blood. To overcome this issue, Anderson et al. defined an Exchange Ratio (ER) between the airway gas exchange and total pulmonary gas exchange (alveolar + airways). When ER = 0 or ER = 1, 100% of gas exchange occurs in the alveoli or in the airways respectively [4].The same research group has determined the relation between the Exchange Ratio and the value of λb:a for a bunch of gases,

characterized by different blood solubility values (Figure 4).

1.2 Volatile Organic Compounds in human breath

Nowadays, with the Directive 2004/40/EC of the European Community (approved in Italy with the D.Lgs 161/06) indicates as Organic Compound “any compound containing at least the element carbon and one or more of hydrogen, oxygen, sulphur, phosphorus, silicon, nitrogen, or a halogen, with the exception of carbon oxides and inorganic carbonates and bicarbonates”, and with the name of Volatile Organic Compounds “any organic compound having an initial boiling point less than or quale to 250 °C measured at a standard pressure of 101.3 kPa”[5]. Normative ISO 16000-6 and EN 16516 define the Total VOCs as “all VOCs or as the sum of all detected volatile organic compounds, sampled on Tenax TA, which elute from a nopolar or slightly polar gas chromatographic separation column between and including n-hexane and n-hexadecane (n-C6 – n-C16), measured by mass selective detector (MSD), and

quantified as toluene equivalent” [6, 7]. Several compounds, some of which can be considered as VOCs, where found in the human body at some concentration level. For example, ammonia [8, 9, 10], dimethylamine and trimethylamine [11, 12], isoprene [13, 14, 15, 16], as well as methanol, ethanol, propanol and acetaldehyde [17, 18, 19] can be found in human specimens (such as breath and saliva). The first comprehensive review of breath composition was presented by Manolis in 1983 [17]. This review discussed the presence of many compounds in human breath and proposed the use of breath analysis to quantify exposure to various exogenous toxic compounds. Recent studies have reported up to 1764 different VOCs out of a total 2577 found inside the human body. Some of these can be found only in certain body fluids (e.g. 1-hexene found only in breath), whereas others can be found in more than one body fluid (e.g. acetaldehyde can be found in saliva, blood, breath) [18, 20].

As described by several papers, those compounds have a great variability in function of numerous factor, such as age, diet, smoking habits, metabolism, physical condition, diseases, medication use and contaminants exposure [21, 22, 23, 24].

In 2013, Mochalski et al tried to distinguish between endogenous and exogenous VOCs, comparing the concentration found in body fluid, such as blood and exhaled breath, with the values found in ambient air [25]. For this purpose, the breath of 28 healthy volunteers was analyzed using HS-SPME (blood samples) and NTD (breath samples) techniques coupled with a GC-MS, to identify and quantify the VOCs. Within the group, 11 VOCs (isoprene, acetone, limonene, dimethyl selenide, p-cymene, 2-pentanone, methyl propyl sulfide, dimethyl sulfide, n-octane, 4-heptanone and methyl acetate) showed a very high concentration in the blood, suggesting an endogenous origin. Other 10 compounds (propanal, decane, 2-butanone, benzaldehyde, hexanal, 3-methyl thiophene, methyl butane, ethylacetate, acetonitrile and 2-methylhexane), instead exhibited a higher concentration in the ambient air, suggesting an exogenous origin. Exogenous compounds can be found in human breath due to: i) environmental pollution for both indoor and outdoor air (e.g. benzene, toluene, BTEX and so on…), ii) emissions from combustion process (e.g. gasoline and diesel engines), iii) incinerators (e.g. benzaldehyde and propanal), iv) plasticizers (acrolein and methacrolein), v) disinfectants and solvents (e.g. isopropanol), vi) ingestion of food and beverages (e.g. alcohols, ketones, sulfur containing compounds) and vii) active or passive smoking (e.g. acetonitrile, furans, BTEX, unsaturated hydrocarbons) [10]. Some bacteria and fungi may be also produce some VOCs: Pseudomonas aeruginosa (ethyl 2-methylbutirate) [26], Staphylococcus aureus ((Z)-2-methyl-2-butenal) [25] and Streptococcus pneumoniae (3-phenylfuran) [27].

On the contrary, endogenous compounds are produced by normal or abnormal (pathologic) physiology according to several metabolic pathways. For example, acetone is produced via the spontaneous decarboxylation of the excess of Pyruvate [28]; isoprene is generated along the mevalonic pathway of cholesterol synthesis; saturated hydrocarons (e.g. ethane and pentane) are formed via the lipid peroxidation initiated by ROS (e.g. superoxide, hydrogen peroxide, singlet oxygen), which transform PUFAs into alkanes, dienes and aldehydes [29]. Figure 6 shows the metabolic pathway for the synthesis of Acetone, Butyrate and 1-Butanol from pyruvate.

Figure 6. Figure reprinted from [27]. Metabolic pathway for the synthesis of Acetone, Butyrate and 1-Butanol

from pyruvate.

1.3 Clinical Application of VOCs in breath analysis

The possible relationship between breath odor and disease was known since the age of Hippocrates, who described fetor oris and fetor hepaticus in his treatise on breath aroma. In mid 1800s, Nebelthau showed that diabetic patients emit acetone through their breath [30], whereas Anstie in 1874 isolated ethanol from breath [31]. All these discoveries may seem a

analysis is generally attributed to L. Pauling, who reported in 1971 the presence of 250 substances in the exhaled breath [32].

In 2003, Corradi et al. presented a work on the monitoring aldehydes (i.e. malondialdehyde, hexanal, heptanal and nonanal) in exhaled breath condensate (EBC) collected from patients with chronic obstructive pulmonary disease. Aim of their work was to evaluate whether individual aldehydes resulting from lipid peroxidation could be measured with the EBC and to compare aldehyde levels of patients with different grades of COPD with those of smoking and nonsmoking control subjects. These compounds were measured by liquid chromatography-tandem mass spectrometry. The results showed that malondialdehyde (57.2  2.4 nmol/L), hexanal (63.5  4.4 nmol/L) and heptanal (26.6  3.9 nmol/L) were increased in patients as compared with nonsmoking control subjects (17.7  5.5 nmol/L, 14.2  3.5 nmol/L, 18.7  0.9 nmol/L, respectively). Malondialdehyde showed increased levels in COPD patients when compared with smoking control subjects (35.6  4.0 nmol/L) [33].

In 2007, the First Breath Analysis Summit was held on the Cleveland Clinic Campus in Cleveland, Ohio, USA. The Summit brought together industry executives and entrepreneurs with scientists and clinicians to discuss key trends, future directions, and upcoming technologies in breath analysis and medicine. The major focus of the Summit was on medical applications. Topics included exhaled nitric oxide, exhaled breath condensate, electronic nose and sensor arrays, mass spectrometry and bench-top instrumentation, and innovative sensor technologies. Medical applications that were covered included asthma, COPD, pulmonary hypertension, other respiratory diseases, gastrointestinal diseases, occupational diseases, critical illness, and cancer. That was the first step made toward the development of breath analysis techniques for the investigation of VOCs.

Volatile organic compounds collected from the headspace of stool samples, collected from 35 patients suffering from infectious diarrhea and 6 healthy controls, were examined by using the SPME-GC-MS technique [34]. Characteristic volatile profiles were related to the presence of different bacteria, e.g. the absence of hydrocarbons and terpenes indicated an infection with

Campylobacter, and the absence of furan species without indoles was indicative of infection

with Clostridium difficile. In another study, SPME-GC-MS was used to analyze volatiles emitted by fecal samples from patients with ulcerative colitis, a disease characterized by inflammation of the colonic mucosa (n = 18), infection with Campylobacter jejuni (n = 31); and Clostridium difficile infection (n = 22) and from 30 asymptomatic donors [35]. Around 300 volatiles could be identified by using a mass spectral NIST 05 library in cohort and longitudinal studies. On the contrary, at least 40 VOCs could be detected in all samples irrespective of the

underlying clinical disease, distinct VOC marker patterns were linked to specific clinical disease compared to the VOC marker patterns emanating from the feces of healthy donors [35].

Helicobacter pylori, which is the cause of a common bacterial infection of the stomach,

plays an important role in gastric cancer [36]. The SPME-GC-MS analysis of exhaled breath samples from six patients infected with H. pylori allowed the detection of isobutane, 2-butanone, and ethyl acetate that were absent in breath samples from healthy controls and uninfected healthy controls (n = 23) and present in the headspace of cultured H. pylori strains [37].In a different study, PTR-MS allowed to measure elevated levels of HCN and hydrogen nitrate in exhaled breath of patients with H. pylori gastritis compared to healthy controls [38].

Urinary tract infections (UTIs) are frequently caused by bacterial pathogens such as

Escherichia coli, Proteus species, enterococci, Klebsiella species, and Staphylococcus saprophyticus. Different urine sampling procedures and analytical methodologies have been

used to identify potential volatile markers of specific bacterial species responsible for UTI [39]. Urine is known to contain a complex mixture of VOCs [29], which show significant alterations after infection. Several studies have investigated the use of secondary metabolites produced by bacterial strains to detect UTI. Analytical techniques such as GC-MS have demonstrated the feasibility of detecting UTI via the identification of bacterial volatile metabolites in urine samples following short incubation periods with appropriate precursor compounds (e.g. Ethanol and Dimethyl disulfide) [40, 41, 42, 43]. This approach allowed the identification of several characteristic high-abundance volatiles produced by specific pathogens, although this procedure lacked sufficient accuracy and suitability for routine clinical use [36].

Tuberculosis (TB), an important infectious disease caused by the bacterium Mycobacterium

tuberculosis, is responsible for considerable morbidity and mortality worldwide [44]. Currently

available tests for the rapid detection of M. tuberculosis are inadequate, and there is an urgent need for improvements in TB diagnosis and for new methods to determine the efficacy of treatment, particularly in developing countries [45, 46, 47]. TB diagnostic tests based on VOC biomarkers offer substantial advantages over other approaches because of simplicity, low invasiveness and possibility of developing POC devices (e.g., e-noses). As early as 1923, the article entitled “Aroma-Producing Microorganisms” on the Journal of Bacteriology reviewed odor association with microorganisms and listed, among others, M. tuberculosis for its foul smell [48]. Headspace GC-MS analysis of VOCs from in vitro cultured Mycobacterium species revealed several metabolites of nicotinic acid, such as methyl phenyl acetate, methyl p-anisate, methyl nicotinate, and o-phenylanisole, which were considered specific for M. tuberculosis complex strains. These compounds represent derivatives of nicotinic acid, which can also

nonsmoking patients with TB and those without TB showed a striking difference in levels of nicotinic acid between the two groups [49]. In a subsequent study, this same group reported the detection of methylnicotinate, a metabolite of nicotinic acid, in the exhaled breath of TB patients, achieving a sensitivity of 84% but a low specificity of 64% among patients with smear-positive TB [50]. In a separate study, 10 most abundant VOCs were selected among 130 VOCs identified by GC-MS from the headspace of cultured M. tuberculosis [51]. GC-MS analyses were performed on exhaled-breath samples of 42 patients with clinical suspicion of pulmonary TB (sputum culture positive, n = 23; negative, n = 19) and 59 healthy controls. Breath samples of all patients with TB contained the M. tuberculosis-associated markers naphthalene, 1-methyl- and cyclohexane, 1,4-di1-methyl-, which were identical, as well as VOCs which were structurally similar to those observed in in vitro cultures. In addition, they also revealed increased markers of oxidative stress (C4 to C20 alkanes and monomethylated alkanes). Pattern recognition analysis of the data identified patients with a positive sputum culture with sensitivities and specificities of 95.5% and 78.9% and 82.6% and 100%, respectively [51].The same group extended the investigations to a multicenter international study in which breath samples collected in portable breath collection devices from a total of 226 symptomatic high-risk patients (United States, Philippines, and United Kingdom) were investigated by using an automated thermal desorption system coupled to GC-MS [52]. Breath volatiles contained components related to oxidative stress, such as alkanes (e.g. tridecane) and methylated alkane derivatives (4-methy-dodecane), as well as in vitro-defined VOCs of M. tuberculosis origin (e.g., cyclohexane, benzene, decane, and heptane derivatives). This VOC pattern in the breath of patients with active TB identified M. tuberculosis infection in this randomly selected population with an overall accuracy of 85.5%, compared to diagnosis based on sputum culture, microscopic smear, chest radiograph, and clinical symptoms.

Cystic fibrosis (CF) is a genetic disease caused by mutations of the gene encoding the CF transmembrane regulator [53]. Lungs of CF patients are either colonized or infected from opportunistic bacteria such as P. aeruginosa, which lead to an unfavorable prognosis [50, 54]. Early detection of this bacterium is crucial for initiation of an appropriate therapy and is therefore of high clinical relevance. Several studies have reported the detection of high concentrations of hydrogen cyanide (HCN) in the headspace samples of in vitro cultures of P.

aeruginosa strains [55, 56, 57, 58]. Besides HCN, the production of methyl thiocyanate has

been reported in the headspace of P. aeruginosa cultures (28 out of 36 strains), as monitored by SPME-GC-MS and real-time SIFT-MS quantification methods. Methyl thiocyanate was also identified with the aid of SIFT-MS in the breath samples of 28 children with CF. The authors of this study noted that to determine the clinical relevance of this VOC marker for P. aeruginosa

infection, a large clinical investigation is necessary [59]. There is accumulating evidence that

P. aeruginosa infection may be better detected by using a VOC biomarker profile rather than

by using a single-biomarker. An ion mobility spectrometer coupled with a multicapillary column was used to examine the exhaled breath of 53 individuals, including 24 individuals infected or colonized with P. aeruginosa and 29 healthy controls. Twenty-one out of 224 signals enabled the discrimination of healthy and P. aeruginosa-infected groups with sensitivity and specificity values of 89% and 77%, respectively, and positive and negative predictive values of 83% and 86%, respectively [60]. In a different methodological approach, breath samples from 105 children, comprising 48 children with CF and the respective controls, were examined for VOC profiles using GC-time of flight-mass spectrometry (GC-TOF-MS). Based on 14 VOCs, it was possible to correctly identify CF patients with (n = 23) or without (n = 17) positive P. aeruginosa cultures with sensitivity and specificity ranges of 34 to 100% and 29 to 100%, respectively. Next, GC-TOF-MS was used to analyze exhaled breath from 105 children, comprising 48 children with CF and 57 controls. A selection of 22 distinctive VOCs enabled 100% correct discrimination of children with CF from healthy controls. Within the CF group, 100% correct identification of patients with or without a P. aeruginosa-positive culture was possible with only 14 VOCs. The discriminatory compounds were mostly hydrocarbons with 5 to 16 carbon atoms [61].

1.4 Breath Analysis

For centuries, doctor have used the odor of patient breath to diagnose liver and renal diseases. Nowadays, with the modern technologies on which we can rely, the breath analysis would be one of the most useful method of clinical diagnosis, thanks to its being non-invasive, low cost and high efficiency. The difficulties of reaching a standardized sampling and analyzing method prevents the comparison of work form different authors. Although the method may seem simple, there are many critical aspects to be considered during the development of a breath analysis. For example, one of the most influent factor is which fraction of the expired breath to collect, since different analytes may be found in different fractions of the exhaled air. Other factor may be the activity of patient during the sample collection, or what is the optimal flow for achieve the best result, it is better to collect through the mouth or the nose, with one long breath or more than one small breath? By now there’s no standardized method for the analysis of VOCs in breath samples.

In 1999, the American Thoracic Society (ATS) released the first guidelines concerning sample collection for breath analysis of nitric oxide [62]. In 2005, recommendations were published for both the measurement of NO in mouth or nose exhaled breath [63] and for sampling and analysis of exhaled breath condensate [64].

During the Breath Analysis Summit of 2013 in Saarbrücken, Germany, Professor Risby suggested the creation of a Task Force aiming to develop a standardized approach for the individual disciplines of breath analysis. This task force should provide guidance toward consistency so that studies from different laboratories could be compared. This includes also the guidance for choosing sample collection strategies to optimize data quality (number of subject, how many samples per subject, time of sampling) and other statistical issues, such as Limit of Detection and Quantification, variability and so on [65].

As first venture, this task force has conducted a survey of the current method employed in breath sampling and analysis, by asking to the leading researchers of this field to complete a questionnaire. In 2015, a perspective article [66], which summarized the responses and feedback received was published.

This poll evidenced that VOCs are the most investigated breath markers and that they are usually sampled under controlled conditions either from mixed-expiratory (64%) and alveolar exhaled breath (76%), depending on the application. About half of the poll participants (48%), employed both off-line and on-line analysis, while the rest split almost equally in using exclusively only one of them (28% for off-line and 24% for on-line). Polymer (Nalophan, Teflon or Tedlar) and aluminum bags were more popular for sample storage respect to the use of traps such as SPME and NTME (37% vs 42% respectively).

1.5 Breath Sampling

1.5.1 Mixed or alveolar breath fraction

Depending on the specific application, breath samples can be collected in different ways:

• Mixed expiratory or total breath sampling;

• Controlled sampling (i.e. sampling over a pre-determined time after the start); • Alveolar sampling.

Mixed breath samples contain the whole expired air, including the air contained in the upper airways (anatomic dead space). This collection method is common due to simplicity but has a

major drawback in the uncontrolled dilution of the alveolar breath with dead space air [67]. Controlled sampling may be a better compromise, in terms of lower dilution and presence of contaminants than a mixed sample [68]. This can be achieved using a pCO2 or temperature

sensor.

At last, the most effective method to reduce contaminants and the dilution of compounds is the sampling of the only alveolar fraction, which contains most of the chemical information on blood composition. Literature reports that the concentration of endogenous substances in the alveolar air is generally 1-3 times higher than in the mixed expiratory samples [67]. To date, different methods have been evaluated to determine when the alveolar plateau is reached during each expiration, like the monitoring of CO2 partial pressure or the breath temperature [64]. A

respiratory cycle can be divided in two distinct moments, the inspiration and the expiration, which can also be divided into three different phases. These phases can be monitored by measuring the CO2 partial pressure (pCO2). As shown in Figure 7, during the inspiration the

pCO2 value is the same as in the ambient air (≈ 0.3 mmHg). As the expiration starts, this level

does not change significantly, because the dead space contains the same amount of CO2. When

the upper airways are emptied (end of phase I), the pCO2 starts to rise (phase II) when alveolar

air mixes with dead space air. This growth is maintained until a plateau is reached, where all the alveolar air is emitted (phase III) [64, 69].

The collection of the alveolar fraction is related to the measurement of the pCO2 during

inspiration and exhalation. Different approaches have been provided to achieve this result. In 2001, Schubert et al. proposed a CO2 controlled breath sampling device for monitor

mechanically ventilated patient. Their prototype was composed of a fast responding absorption mainstream CO2 analyzer, which supplied data to a processing unit (PU). Whenever the level

of CO2 raised above the set threshold, the PU opened or closed a two-way valve diverting breath

flow to an absorbing tube, which was packed with 80 mg of activated charcoal [70].

In 2006 Birken et al., manually collected alveolar fraction by using glass syringe and monitoring pCO2 by a fast mainstream sensor [71]. The collection of alveolar fraction can be

also performed by using a device in which a sampling valve is automatically triggered according to the actual pCO2 value measured during the respiratory cycle [72, 73].

Several types of breath sampling device are already available in the market. One example is the QuinTron® breath sampler (Campro Scientific GmbH, Germany), which consists of a T-shaped connector with two one-way valve as outlets. Each valve contains a silicon disk that opens the valve when pressure exceeds a threshold. Two bags are connected to the device for collecting the sample: the first contains up to 500 mL of air and it’s usually used as discard bag, while the other has a volume of 750 mL. As the subject starts to exhale, the first valve will open so that dead space and the mixed air will be collected in the first bag. After the filling of the bag, pressure will start to increase again until the second valve will open, letting the alveolar air in the second bag.

Another common device is the Bio-VOC® (Markes International, UK). Usually a healthy subject will exhale about 4 L in a forced expiration, and this device is made to retain only the last 100-150 mL of this volume. The Bio-VOC® is also built with an easy plugin for sorbent tubes, so that the VOCs contained in the alveolar sample could be transferred into sorbent tubes. Although the manufacturer states that this device can collect only the alveolar fraction, even if there’s no evidence of control on the subject’s breathing or its CO2 level. Devices that can

collect breath sample from one single exhalation should be avoided since this sample is not representative, as recently discussed by Lourenco and coworkers [64].

In 2008, Miekisch et al. developed a breath sampler consisting of a disposable mouthpiece, connected to a collection device (Tedlar bags or gas-tight syringe) with some T-shaped connector, and equipped with a CO2 infrared sensor. The exhaled air flows inside the connector

to reach the collection device. Thanks to the infrared sensor, which displays on a monitor in real time the pCO2, an operator can manually open or close the valve to collect the sample.

Obviously, the major drawback of this device is the operator itself, since the variability of collection procedure depends on the ability of operator [65].

Di Francesco et al. proposed a CO2 controlled device for breath sampling in 2007. This

device was equipped with both a CO2 infrared sensor and a flow meter [74]. A dedicated

software acquired the parameter from both instruments and controlled a solenoid valve to sample different fractions of the expired air. In 2015, the same group developed an improved version of this prototype that could sample also the dead space air [73].As before, this result was achieved thanks to the measurement of the pCO2 and the airflow during the respiratory

cycle. A Nalophan bag was connected to the end of this device to be filled with the selected breath fraction. This approach allowed to sample in a single or multiple exhalations for a higher flexibility. The small dimension of this device allowed an easy transportation directly to the patients, which may be unable to reach the laboratory. The two devices described used a polymer bag as collection device [73].

De Silva et al. instead developed a prototype able to adsorb VOCs into an SPME fiber having 65 µm PDSM/DVB fused silica. This device consisted of two parts, an alveolar air separator and a VOC extractor. The separator identified the beginning of the alveolar fraction emission based on the volume of air expired, and direct this fraction to the VOC extractor. This one directed the VOCs to the SPME fiber, which was then inserted in the injector of a GC-MS system for the analysis. In the last few years, another technique for the collection of air samples was developed based on a needle packed with a sorbent to collect the analytes. Due to the shape and the fact that works as a SPME, this technique has the name of Needle Trap Micro Extraction (NTME) [75].

Mieth et al. developed a multibed needle trap device for breath sampling. The aim of their work was to facilitate the use of NTME in trace gas analysis, by increasing the adsorption capacity of the Needle Trap Devices (NTDs) [76]. TO achieve the aim, they proposed an innovative NTD packed with three different adsorbent material, precisely Tenax (35/60 mesh), Carbopack X (60/80 mesh) and Carboxen 1000 (60/80 mesh). These needles were then used to pierce a septum connected to the sampling device. Here, as in the previous example, the breath fraction was controlled using two ways valve and CO2 sensor to measure the pressure.

1.5.2 Nose vs mouth sample collection

Many VOCs have endogenous origins, produced in the airways, in the oral cavity or in the gut by bacteria; moreover, they can be also emitted from mucus, saliva and aerosols created in the respiratory tract. Smith et al. developed a sampling device able to determine the concentration of VOCs in both mouth and nose, using SIFT-MS as technique for analysis [77,

be partially systemic and mouth generated. At last, compounds of interest like acetone, methanol and isoprene showed similar profiled for both nose and mouth samples, suggesting a systemic origin. Thus, data obtained from both methods should bring the same results. A recent study from Khalid et al. [80] suggest that this is false. In fact, in their work they extracted the VOCs, produced from Gram-negative and Gram-positive bacteria, using an SPME technique. Gram-negative bacteria seem to produce more volatile sulfur compounds and amines, whereas Gram-positive were correlated to the production of acids, hydrocarbons, alcohols and aldehydes. In addition, a poor oral hygiene can lead to the production of ammonia from urea or ethanol from sugars, increasing the VOCs level in mouth [81].

1.5.3 Single or multiple breathes

Another question to answer when developing a method for the analysis of breath samples is how many respiratory cycles are needed. Sampling one single breath is a simple and fast technique but only a limited volume can be sampled. The composition of single breath may vary considerably depending on the respiratory rate and on the depth of breathing, for this reason sampling over multiple breaths might be more representative, if respiration is controlled.

Chapter 2

Breath analysis: analytical approaches to monitor VOCs

2.1 Thermal desorption technique to analyze VOCs

Analytical thermal desorption, most commonly referred as Thermal Desorption (TD), is an analytical technique used to concentrate VOCs in gas streams, prior to injection into a gas chromatograph. The history of TD can be traced back to the mid-1970s. Scientists struggling with the limitations of conventional GC sample preparation methods, began to experiment by packing standard GC injector liners with sorbent material. These sorbent-packed injector liners were used to sample a fixed volume of air or gas and were then dropped quickly into the GC inlet for desorption and transfer of analytes to the analytical column. The limitations of these primitive adaptations of conventional GC injectors are many and obvious (air ingress, volatile losses, variability, contamination from the outer surfaces of the liner, etc…), but the fact that it was attempted at all demonstrated the need for this technology. Another early incarnation of thermal desorption was in purge-and-trap technology. The US Environmental Protection Agency (EPA) first developed purge-and-trap/GC-based test methods to measure volatile organic compounds in drinking water in the late 1970s in response to several serious environmental incidents. The purge-and trap method developed by EPA at that time relied on volatiles being sparged from the water in a stream of pure nitrogen and trapped on a sorbent tube/trap. This was subsequently heated in a reverse stream of carrier gas to thermally desorb the organic chemicals of interest and transfer them to the GC analytical system in a standard TD-type procedure.

The first early commercial configurations of dedicated general-purpose thermal desorption technology were invariably based on desorption of a single tube or badge. The ‘Coker cooker’, designed by Environmental Monitoring Systems Ltd (UK) in the mid-1970s [82], was a popular example and accommodated samples or sorbents contained in ¼″ o.d. tubes. These early desorbers were very primitive by modern standards, typically offering only single-stage desorption and without any of the functions that would now be regarded as standard such as leak testing or pre-purging of air from the tube. However, within specific constraints (e.g. packed column only, stable compounds only, narrow concentration-and volatility ranges), they operated sufficiently well for routine applications such as workplace air monitoring in the

A large scale of this technique is dated back in 1985, when it was specified in the Record of Decision for the McKin Company Superfund site within the Royal River watershed in Maine [83]. In this report, it was reported a method for the collection of volatile organic compounds from a solid matrix, specifically from soil. This consist in a rotating cylindrical drum of about 2 m in diameter and 8.5 m in length. The soil excavated would be inserted inside this drum and then heated to a temperature of 200°C. During this process, the drum rotates at an approximatively speed of 6 revolutions per minutes. The exhaust gas was directed towards an air treatment system for the cleaning.

The thermal desorption is the most powerful and versatile of all gas chromatography (GC) sample introduction technologies. It is readily automated and serves to combine sampling/sample preparation, selective concentration and efficient GC injection in one fully-automated procedure. It is compatible with sampling and analysis of gas-phase organics trapped on sorbent media and allows concentration factors up to 106 to be comfortably achieved. It can also be used for direct gas extraction of volatiles and semi-volatiles from solid or liquid matrices. Thermal desorption is generally coupled with GC, either on its own or in conjunction with a mass spectrometer (GC/MS). However, it can also be used with alternative vapour-phase analytical options including sensors (e-nose technology). Thermal desorption also provides the basis for many other GC sampling procedures – most notably purge-and-trap, sorptive extraction, some forms of large-volume injection and headspace-trap. The development of thermal desorption was fundamentally driven by the limitations and complexity of conventional GC sample preparation methods (i.e. liquid extraction) [84].

The principle at the basis of the TD technique it’s a straightforward extension of gas chromatography. In both techniques, there’s a stationary phase which interact with the analytes. Then, when the sorbent tube is rapidly heated, analytes adsorbed on the stationary phase may reach the GC column by a stream of inert carrier gas. With this technique, it is also possible to concentrate the analytes in smaller and smaller volumes of gas, by repeating the process of deposition and desorption with sequential stages (Figure 8).

Figure 8. Figure reprinted from [83]. Illustration of the concentration potential of multi-stage TD.

Another advantage of the thermal desorption, is that is possible to quantitatively retain the target compounds during the trapping stages, whereas interferent compounds (e.g. water and oxygen) can be selectively purged out to vent, preserving the GC column.

2.1.1 Thermal Desorption advantages

Thermal desorption techniques bring many advantages when compared with other extraction method, like solvent extraction for example. The first and most obvious is that it is possible to transfer 100% of the retained analytes to the system of analysis. This result is not achievable with the solvent extraction, due to physical limit of the technique. Other advantages may involve the extraction/desorption efficiency, the reproducibility of extraction, the reduced interference, lower cost and the possibility of automation.

2.1.2 Extraction/Desorption Efficiency

Assuming appropriate selection of sampling and analytical conditions (sorbent, temperature and flow), it is usually very straightforward for TD methods to exceed 95% desorption efficiency [85]. This is possible because TD is a dynamic process, with gas continually purging

phase by the rising temperature. In contrast, typical solvent extraction procedures are static, with analytes partitioning between the sorbent, solvent and vapor (headspace) phases. This limits desorption efficiency. Standard methods for solvent extraction therefore typically specify only 75% recovery [86].

2.1.3 Reproducibility

Static partitioning systems such as most solvent extraction procedures are also subject to increased variability of analyte recovery depending on the nature of the compounds of interest and the presence of interferences. The desorption efficiency of methods specifying charcoal sample tubes with CS2 extraction for example, have been shown to fall as low as 20 or 30% for

polar compounds in the presence of water [87]. This uncertainty is particularly problematic for air monitoring methods or measurements of industrial VOC emissions, as the analyst may not be aware of field/sample conditions such as high water content; moreover, poor recovery may lead to significant under-reporting.

2.1.4 Interferents

Solvent interference can be a major consideration for liquid extraction methods. One of the reasons CS2 was originally selected as the preferred solvent for many charcoal-based air

sampling methods was that it gives little or no signal on a GC flame ionization detector. However, nowadays, with the preference for MS detection, this advantage no longer holds. Common concerns include masking of peaks of interest, signal quenching (for components co-eluting with the solvent) and baseline disturbances. All these solvent interference issues make peak integration difficult and more prone to error. Thanks to a selective purging, this is not a problem with thermal desorption technique. In fact, depending on the volatility of the compounds of interest, thermal desorption usually facilitates selective purging of sample interferences such as water or ethanol prior to analysis. Applications as diverse as monitoring VOC emissions from paint and characterizing the aroma of whisky benefit from the selection of sorbents that quantitatively retain compounds of interest while allowing water, and in the latter case, ethanol, to purge to vent. Selectivity is usually only possible for solvent extraction procedures when there is a very significant volatility difference between the compounds of interest and the interferences.

2.1.5 Cost, safety and automation

Vapor samplers designed for solvent extraction are invariably one-shot only. For example, the charcoal tubes traditionally used for industrial hygiene monitoring comprise glass tubes with drawn/sealed ends that are broken during sampling and analysis. Pre-packed TD tubes (glass, stainless or coated steel) are typically about 10 times more expensive than charcoal tubes, but they can then be re-used at least 100 times. They are also automatically cleaned by the TD process. This generally reduces the sampling costs of TD methods to roughly a tenth of equivalent solvent extraction procedures. Thanks to the minimal preparation required for thermal desorption, the analysis can be almost fully automated. In fact, once that the tube has been loaded with the sample, it is possible use an automatized thermal desorber, which comprehend a cold trap where the analytes are concentrated before reaching the GC system. In matter of safety, TD-GC/MS systems do not require any toxic solvents (such as CS2) and can

be generally installed without ventilation equipment or fume-hood. All that is needed is that all outlet point, including sample split lines, are configured with appropriate filters.

2.1.6 Sorbent Materials used for thermal desorption

The design of the sorbent tubes used for sample collection and thermal desorption is a major factor contributing to the overall performance of the equipment. The sorbent tubes are generally packed with layers of various sorbent materials, so that a broad range of compounds, may be trapped on an appropriate sorbent. The more volatile compounds break-through the initial layers of sorbents, but are trapped by succeeding layers. Each sorbent layer protects the next increasingly active layer, preventing a compound from being held so tenaciously that it cannot be desorbed quickly and completely during the heat cycle without degradation. Materials chosen for sorbent tubes may differ slightly from those used in a trap. In general, materials with optimum retention characteristics are chosen for the sorbent tube, whereas materials in the focusing trap have higher desorption efficiency. For example, silica gel would not be a good choice for sampling in an atmosphere saturated with water vapor due to its water retention characteristic, but it would be an excellent material in a focusing trap, providing good desorption efficiency for low-boiling polar compounds, since the water will be purged off the collection tube before the sample is transferred to the focusing trap [88].

Sorbent or combination of sorbents must be selected considering this generic rule: sufficiently ‘strong’ to retain target analytes during sampling/concentration, but weak enough to release them efficiently during the thermal desorption phase. Sorbent strength is usually

of common sorbent/sorbate combinations and describe how these values can be determined. Retention volumes are susceptible to temperature and are typically reported at 20 °C. As a (very) approximate rule, retention volumes halve for every 10 °C rise in temperature. The performance of strong sorbents such as carbonized molecular sieves (CMS) are adversely affected by high relative humidity (RH > 80%) as recorded in standard methods. The retention volumes of hydrophobic sorbents such as carbon blacks, TenaxTA and other porous polymers are much less sensitive to atmospheric humidity with negligible impact reported even up to 90% RH. [90]. A wide range of weak, medium and strong commercial sorbents are now available for air monitoring. Vapor-phase organics should be sampled using the weakest compatible sorbent, i.e. one that offers a practical/useful retention volume and quick, quantitative recovery during desorption and analysis [91]. In Table 1 are reported some of the sorbent commercially available along with their strength and characteristical features.

Table 1. Data from [90]. Sorbent materials for thermal desorption.

Sorbent Strength Max. Temp. (°C) Features

Quartz Wool Very weak > 450 Very inert Non-water retentive Carbograph™ 2TD Carbopack™ C Weak > 450 Hydrophobic Minimal artefacts (< 0.1 ng)

Tenax TA Weak 350 Inert, Hydrophobic Low artefacts (< 1 ng)

Tenax GR Weak 350

Inert, Low breakthrough volume More hydrophobic than

Tenax TA

Carboxen 569 Strong > 450

Inert Minimal artefacts

(< 0.1 ng)

Carboxen 1003 Very Strong > 450

Inert, Not hydrophobic Minimal artefacts

(< 0.1 ng)

Molecular Sieve 5Å Very Strong > 400 High artefacts (≈ 10 ng) Significantly hydrophilic

2.2 Derivatization methods for analyze carbonyl compounds in air

The direct analysis of carbonyl, especially aldehydes, is challenging because of interactions between the compounds themselves. These interactions can lead to poor peak resolution and/or unsymmetrical peaks, which make proper peak identification difficult or impractical. In many cases, conversion to derivatized products will reduce the interaction interfering with analysis. Compounds that co-elute with or that are poorly resolved from other sample components can frequently be resolved if one or more of them is converted to an appropriate derivative [92].

A good derivatizing reagent and procedure should produce the desired chemical modification of the compound(s) of interest, and be reproducible, efficient, and nonhazardous [93].

Nowadays, the analysis of airborne VOCs (especially aldehydes and ketones) are conducted with HPLC or GC techniques. Both require a previous stage of derivatization, and this is

achieved with the use of 2,4-Dinitrophenylhydrazine (DNPH) or

O-2,3,4,5,6-Pentafluorobenzylhydroxylamine (PFBHA).

Figure 9 and Figure 10 shows the reaction mechanism between the derivatizing agent (i.e. DNPH and PFBHA) with a generic carbonyl compound.

Figure 9. Reaction mechanism between DNPH and a generic aldehyde/ketone.

As reported in Figure 9 and Figure 10, both agents react with aldehydes or ketones in a nucleophilic addition. The NH2 group present in both these molecules reacts with the carbonyl

group of the aldehydes or ketones, with the sequential loss of a water molecule, and the formation of a hydrazone (in the case of DNPH, Figure 9) or an oxime derivate (in the case of PFBHA, Figure 10).

2.2.1. Comparison between PFBHA and DNPH

Although both of these derivatizing agents are used, there are some difference between that need to be discussed when deciding to use one or another.

DNPH has a better application for HPLC technique, especially since it can be used with UV detector. That is useful if the laboratory is not equipped with an MS system. Also, it is possible to buy cartridge filled with this agent, for liquid samples. These devices are very simple to use. The only necessary operations is to load the sample, wait for the derivatization to reach completion (which is signaled by a change of color of the cartridge), and then extract the derivatized analytes with a solvent (usually MeOH).

Although those benefit, the drawbacks of this derivatizing agent are more important than the advantages. One major problem is the separation achieved with this technique, which doesn’t show any difference between the two isomers formed. Also at high temperature, the derivatized compounds are unstable and easily decompose, which doesn’t make them suitable for GC analysis [94].

On the contrary, derivatization with PFBHA is very fast even under mild reaction conditions. The oxime formed are also more stable than the hydrazone formed by DNPH, which allows them to be analyzed in a GC system [95]. Moreover, the separation of the oximes, within a capillary column, show more peak for each compound, representing both the isomer formed, and allowing to a more complete analysis of the sample.

2.2.2. UNICHIM Method

Nowadays, many authors have proposed new techniques for the analysis of VOCs in the air. Since that this topic has such important applications (i.e. the monitoring of the working ambient air), a standardized method is needed. In 2009, the Associazione per l'Unificazione del Settore dell'Industria Chimica (UNICHIM) proposed a SPME method which involved the use of Polydimethylsiloxane (PDSM) fibers. In their paper, the fiber needed to first be exposed at the headspace of a solution of o-2,3,4,5,6-Pentafluorobenzylhydroxilammine (PFBHA), a derivatizing agent, to collect the VOCs from the headspace of the sample [96].

The oxime formed from the reaction of PFBHA with the carbonyls can be then easily analyzed with a GC/MS system. It is important though, to choose the optimal condition during the sample, to prevent the possibility of interferences. More specifically, the UNICHIM report state to pay attention to the following variables:

• Temperature; • Humidity; • Air Flow;

• Presence of ozone;

Chapter 3

Materials and Methods

3.1 Chemicals and Materials

Ketones C4-C9 kit (2-Butanone, 2-Pentanone, Pentanone, 2-Hexanone, 4-Heptanone, 3-Heptanone, 2-3-Heptanone, 3-Octanone, 2-Methylcyclohexanone, 3-Methylcyclohexanone, 4-Methylcyclohexanone, 2-Octanone, 5-Nonanone, 2-Nonanone) and branched Ketones C4-C8 kit (3-Methyl-2-butanone, 3,3-Dimethyl-2-butanone, 2-Methyl-3-pentanone, 4-Methyl-pentanone, 2,4-Dimethyl-3-4-Methyl-pentanone, Methyl-3-hexanone, 5-Methyl-hexanone, 2-Methyl-3-heptanone, 5-2-Methyl-3-heptanone, 2,6-Dimethyl-4-heptanone, 4-Methylpent-3-en-2-one, Acetophenone, Cyclopentanone, Cyclohexanone) were purchased from AccuStandard, Inc. Chemical Reference Standard (USA). Acetone, 3-Hydroxy-2-butanone, 2,3-butandione, Propanal, 2-Methylpropanal, Butanal, 3-Methylbutanal, 2-Methylbutanal, Pentanal, Hexanal, Heptanal, Octanal, Nonanal, Benzaldehyde, Pyruvaldehyde, Methacrolein, Isobutyraldehyde and Acrolein were purchased from Sigma Aldrich (Italy). Water at LC-MS grade (purity ≥99.9%) was purchased from Fluka (Italy).

Labeled Toluene-D8 and Acetone-D6 were purchased from ARMAR Chemical (Switzerland) at purity higher than of 99.8%.

O-2,3,4,5,6-Pentafluorobenzylhydroxylamine hydrochloride powder (purity >99%) was purchased from Alfa Aesar (Germany).

Medical Air (Hydrocarbon free, purity of 99.95%), Helium 5.5 IP and Nitrogen 5.0 IP were purchased from Sol Group Spa (Italy) and Rivoira (Italy), respectively. Each gas was further purified with a super clean filter purchased from Agilent Technologies (USA) to remove water, oxygen and hydrocarbon contaminants. Humidified medical air (RH 90%) was obtained by flowing medical air (500 mL/min) through a purge and trap glass system filled with 5 mL of fresh LC-MS water at 37 ± 1 °C.

Commercial Sorbent Tubes for Thermal Desorption (Stainless Steel, O.D. 6.4 mm, I.D. 5 mm, 89 mm length) packed with 250 mg of Tenax GR (60/80 mesh), adsorbed materials composed by 70% of 2-6-diphenylene-oxide polymer resin (Tenax TA) and for 30% of graphite, were purchased by Markes International (UK).

Disposable Nalophan bags were fabricated at the lowest (film) surface-to-(sample) volume ratio (S/V) following the procedure described by Ghimenti et al [97].

Briefly, a piece of the polymeric material (70 cm) was cut from the Nalophan tube and then a strip (8 cm) from one cut was folded in half to obtain a dead end in the orthogonal direction, starting from each border towards the middle of the bag, so that two series of superimposed 1 cm large creases were obtained. The resulting bundle of creases was finally folded in half and then tightened using a nylon cable tie. A simplified procedure was used for the other end of the Nalophan paring, as in this case the first and last steps were not performed and the two series of creases were tightened around a PTFE tube (O.D: 6 mm, I.D. 4mm, 60 mm length) connected to a stopcock (Nordival Srl, Italy) placing another nylon cable tie 2 cm from the bag end. Figure 11 shows our hand-made disposable Nalophan bag.

Figure 11. Deflated Nalophan bag (47 cm × 47 cm, S/V = 0.3 cm-1_{) composed of (a) Nalophan piece, (b) PTFE}

tube and (c) stopcock.

In the case of experiments carried out using humidified medical air, Nalophan bags (47 cm × 47 cm, S/V = 0.3 cm-1) made up of 2 layers were prepared by plugging together two different tubular Nalophan films. The bags were assembled following the procedure described above.

3.2 Preparation of standard mixtures

3.2.1 Liquid mixtures

A standard liquid mixture of Aldehydes was prepared by mixing 50 μL of each pure liquid compound, starting from the less (i.e. Acetaldehyde) to the highest volatile (i.e. Nonanal), in a