Determination of Carbonyl Compounds in Different Work Environments: Comparison between LC-UV/DAD and LC–MS/MS Detection Methods

There were two analytical methods for the determination of 12 carbonyl compounds (CCs) by using liquid chromatography (LC) coupled with mass spectrometry (MS/MS) and diode array detector (UV/DAD) that were developed and applied to 52 samples that were collected in 10 workplaces. Linearity (0.996 < R2 < 0.999), intra-day repeatability (0.7 < RSD% < 10), and inter-day repeatability (5 < RSD% < 16) were acceptable for both techniques, but the highest sensibility of the MS/MS method allowed us to correctly quantify 98% of the samples (versus 32% by UV/DAD). The comparison of the concentrations that were obtained by quantifying the same sample with both techniques showed good agreement for acetaldehyde and formaldehyde (0.1 < % deviation < 30) but much higher for the less abundant congeners. In real samples, formaldehyde was the most abundant congener (concentrations between 2.7 and 77 µg m−3), followed by acetaldehyde (concentrations between 1.5 and 79 µg m−3) and butyraldehyde (concentrations between 0.4 and 13 µg m−3). In all the beauty salon samples, instead, the most abundant congener was acetaldehyde (concentrations between 19 and 79 µg m−3), probably associated with the use of beauty products. Principal components analysis (PCA) confirms the ubiquitous character of formaldehyde and highlights the influence of minority CCs on different workplaces.


Introduction
Carbonyl compounds (CCs), such as aldehydes and ketones, are ubiquitous pollutants that are among the most widespread in the environment [1,2]. Airborne CCs are primarily emitted by anthropogenic sources, such as the incomplete combustion of fossil fuels or biomass burning activities [3]. When released into the environment, CCs play an important role in the atmospheric chemistry as the precursors to several species are involved in photochemical smog (e.g., free radicals, ozone and peroxyacyl nitrates) [4].
In the last years, CCs have received more normative, scientific, and public attention both for their active role in atmospheric chemistry and for their association with several adverse effects on human health [5]. Among CCs, formaldehyde and acetaldehyde have particular health concerns; indeed, formaldehyde is a well-known sensory irritant compound, especially for sensitive individuals and it has been associated with allergies and negative outcomes for the respiratory system, while acetaldehyde can irritate the skin, eyes, and nose [6]. Besides, the International Agency for Research on Cancer (IARC) classified formaldehyde and acetaldehyde as human carcinogens (Group 1) and possibly carcinogenic to humans (Group 2B), respectively [7][8][9]. Indoor environments increase the risk of adverse health effects that are related to the exposure to formaldehyde and acetaldehyde because it is well-known that the indoor concentrations of pollutants can be two to five times higher than the respective outdoor concentrations [10,11]. In addition, the quality of indoor air is of particular importance because the general population spends the majority of the time in indoor environments, both residential or non-residential [12]. Particularly relevant is the exposure of workers that are employed in the manufacture of formaldehyde or formaldehyde-containing products, healthcare professionals, firefighters, beauticians and hairdressers, printing-rooms workers, professionals of gross anatomy and pathology laboratories, veterinarians, and embalmers [13]. Given the ubiquitous nature of formaldehyde, the World Health Organization (WHO) indicated that occupational exposure occurs in all workplaces [5], mainly through the inhalation route, but also through the dermal route [14]. Acetaldehyde is the other majority aldehyde that was revealed in the workplace, whose presence is related to the use of cosmetic products, combustion appliances, cigarette smoking, and special adhesives [15,16]. In addition to formaldehyde and acetaldehyde, other CCs that are associated with different emission sources may be present in work environments [17]. For example, crotonaldehyde is synthesized and used in the field of industry [18], propionaldehyde is emitted by housing and furniture, and butyraldehyde, hexanal, heptanal octanal, and nonanal are associated with the use of solvents, paints, perfumes, and flavoring agents [15]. Several studies focused their attention on the appearance of adverse effects on human health due to combined exposure to different CCs [19,20], that can lead both to the formation of DNA adducts and to DNA damaging [21].
Due to their impact on human health, a simple, fast, and highly effective collection sampling method is required for CCs routine determination. Generally, the analytical approaches are limited to formaldehyde and acetaldehyde determination; however, given the potential toxicity of other CCs and the synergic effects deriving from the exposure to mixture of CCs, an analytical procedure for simultaneously measuring the levels of the most common airborne CCs would be very useful. The aims of this study were: 1. to optimize an analytical procedure useful for determining the concentrations of 12 CCs; and 2. to study the distribution of 12 CCs in ten different work environments. For this purpose, the air levels of these 12 CCs were measured in ten different work environments among those that were recognized as determining potential exposure to these compounds: six wards of four different Italian hospitals, a copy shop, a beauty salon, a pharmacy, and a chemistry laboratory. The sampling was carried out by portable sampling pumps and adsorbing and derivatizing cartridges, then treated and analyzed by liquid chromatography (LC) coupled both to a spectroscopic detector (diode array detector, DAD) and tandem mass spectrometer (MS/MS) in order to optimize the two different detection methods. The performances and the results of the two detection methods were compared to define their applicability in different exposure scenarios.

Instrumentation
LC analyses were carried out on an HPLC 1260 Infinity II system (Agilent Technologies Italy S.p.A. Milan, Italy) that was equipped with high pressure mixing and an autosampler Agilent G7129A. The MS/MS determinations were performed by using a QTRAP 5500 mass spectrometer with an ESI source, operating in MRM (multiple reaction monitoring) mode (AB SCIEX S.r.l. Forster City, CA, USA). Analyses were achieved in negative mode. The DAD determinations were performed by UV-DAD 1290 DAD FS (Agilent Technologies Italy S.p.A. Milan, Italy), set at 360 nm. The used measuring cell had an optical path of 1 mm and a volume of 1 µL.

Sampling Sites and Strategy
The indoor carbonyl levels were measured in ten different workplaces, selected based on a recent systematic review [13]. The general information and characteristics of each sampling are reported in Table 1. Briefly, for each working environment, four to six sample pumps were used both as personal samplers, in order to control employees' exposure, and as environmental samplers, in order to estimate indoor contamination. In every sampling site, the assessment of CCs' background levels was made by placing a sampling pump outside and near the monitoring site. Air samples were collected into the dualbed cartridges at a flow rate of 0.14 L min −1 . Sampling was carried out during normal working hours, with sampling time ranging from 51 to 406 min ( Table 1). The chosen flow rate and sampling times ensure that the collected CCs would not consume more than 30% of the DNPH that was coated on the cartridges; this was checked for every sample by LC-UV/DAD analysis at 360 nm and subsequent quantification of unreacted DNPH (Supplementary Material S1). A preliminary test was performed in order to assess that the chosen sampling flow rate did not consume more than 30% of DNPH. Analysis of the blank samples (i.e., 5 unsampled cartridges) was performed to verify possible contaminations during extraction procedure. The portable pumps' flow rates were checked before and after every sampling campaign using a calibrated flowmeter. The sampled DNPH-coated cartridges were stored in a refrigerator at +4 • C until the analysis, carried out within two weeks from the sampling campaign.

Sample Analysis
The 12 analyzed carbonyl compounds (Fo-DNPH, Ace-DNPH, Pro-DNPH, Cro-DNPH, Bu-DNPH, Cy-DNPH, Val-DNPH, Hex-DNPH, Hep-DNPH, Oct-DNPH, Non-DNPH, Dec-DNPH) were eluted from the DNPH-coated cartridge with 4.0 mL of ACN, with an elution rate of 1 mL min −1 as suggested by the manufacturer of the cartridges. The eluate was filtered and then injected in the LC system and detected by using both spectrometric and spectroscopic detectors.

Optimization of the Analytical Determinations
Both for the LC-MS/MS and LC-UV/DAD determinations, the calibration curves were constructed by using an external standard method, in the range 1.56-100 µg L −1 for MS/MS and 8-1000 µg L −1 for UV/DAD. For the quantification of real samples, the external standard quantification method was used after verifying that there was no matrix effect for both the determination techniques. Limits of detection (LODs) and quantification (LOQs) were calculated as the minimum concentration of the analyte that produces a signal-to-noise ratio greater than 3 and 10, respectively. Intra-day reproducibility, evaluated as the relative standard deviation (RSD%), was calculated on 10 repeated injections in the same day at three concentration levels: Inter-day reproducibility, instead, was obtained by performing triplicate analysis of the same standard solution for seven consecutive days. The result was expressed as RSD%.

Statistical Elaboration
The statistical software R (R-project for statistical computing, Ver. 3.0, 32-bit) was used to perform principal component analysis (PCA) on the 12 individual CCs' congeners in order to illustrate the distribution pattern and the possible emission sources. Before the statistical analysis, the matrix of the data was transformed by column mean centering and row and column autoscaling in order to correct for different variable scaling and units.

Optimization of the Chromatographic Separation
To find the best chromatographic condition for CCs separation, we used a C18 Acclaim Carbonyl column, testing different elution gradients and flow rates. The best solution was a mobile phase consisting of (A) ammonium acetate 2 mM and (B) ACN at a flow rate of 0.8 mL min −1 . The gradient elution started from A:B 53:47 (v/v %) for 5 min and linearly increasing to 90% B in −5 min, then isocratic elution for 5 min. The initial elution conditions were then restored for 5 min. The total duration of the analysis was 20 min. The injection volume was 5 µL. The same optimized chromatographic separation was coupled both with MS/MS and UV/DAD detectors. The retention times of the selected analytes are reported in Table 2. Table 2. Optimized spectrometric conditions for each studied analyte. The letter "Q" indicates the ion that was chosen as quantifier, the letter "q" indicates the ion that was chosen as qualifier.

Optimization of the Spectrometric Determination and Data Quality
To choose the precursor and product ions that are useful to quantitative and qualitative determinations and to find the best parameters for ionic transmission in mass spectrometry, the infusion of standard solution of Ccs (10 µg L −1 in ACN) at a flow rate of 7 µL min −1 were carried out in negative mode. Nebulizer, turbo, and curtain gases were respectively set at 30, 55, and 60 psi; the temperature was set at 500 • C; and declustering and entrance potential were set at −60 and −10, respectively. The optimized parameters for each analyte are reported in Table 2, together with precursor and product ions.
The validation parameters of the LC-MS/MS optimized method are shown in Table 3. The method linearity in the investigated concentration range (1.56-100 µg L −1 ), expressed as R 2 , was between 0.999 and 0.996 for all the studied analytes. The values of intra-day repeatability (RSD%) calculated at the lowest level (L L , 0.20 µg L −1 ) ranged from 3.0% to 7.8%. The RSD% were found to be lower at higher concentrations (M L , 0.78 µg L −1 and H L , 20 µg L −1 ) with values between 1.5% and 3.3% for M L and between 1.2% and 2.7% for H L . The obtained results were much lower than those that were obtained by Chi et al. [22]. The intra-day repeatability values (RSD%) were between 8.6% and 15% for L L , between 6.3% and 14% for M L , and between 7.7% and 16% for H L . LODs and LOQs were comparable or lower than those that were found by Chi et al. [22] and acceptable for the analytical determinations. The analytical data that were considered for the optimization of the spectrometric determination have confirmed this method to be sensitive, precise, and accurate for the evaluation of CCs in indoor air. Supplementary Material S2 shows the LC-MS/MS chromatogram that was obtained by injecting 5 µL of aldehydes-DNPH standard solution at 0.2 µg L −1 .

Optimization of the Spectroscopic Determination and Data Quality
The method linearity in the investigated concentration range (8-1000 µg L −1 ), expressed as R 2 , was 0.999 for all the analytes that were under study, as reported in Table 4. Although the range of the investigated concentrations was rather wide, the results were comparable with those that were obtained by Feng et al. [23]. The values of intra-day repeatability (RSD%), calculated at lowest level (L L , 25 µg L −1 ), were between 2.3% and 7.8%. The RSD% were found to be lower at higher concentrations (M L , 62 µg L −1 and H L , 250 µg L −1 ) with values between 1.5% and 3.3% for M L and between 1.2% and 2.7% for H L ( Table 4). The intra-day repeatability values (RSD%) were between 8.6% and 15% for L L , between 6.3% and 14% for M L , and between 7.7% and 16% for H L ( Table 4). The values that were obtained both for intra-and inter-day repeatability were comparable with those that were obtained for MS/MS determination, despite the different concentrations of the three examined levels. As expected, LODs and LOQs were higher than those that were obtained by MS/MS analysis but comparable or lower than those that were obtained in other works [24,25]. Supplementary Material S3 shows the LC-UV/DAD chromatogram that was obtained by injecting 5 µL of aldehydes-DNPH standard solution at 25 µg L −1 . Table 4. LODs (µg L −1 ), LOQs (µg L −1 ), linearity (R 2 ), and the intra-and inter-day repeatability (RSD%) that were obtained from the LC-UV/DAD optimized method. L L , M L , and H L correspond respectively to 25, 62, and 250 µg L −1 .

Comparison between LC-MS/MS and LC-UV/DAD
The MS/MS detection method was compared with the UV/DAD detection at 360 nm method. All the real samples were analyzed by using both detection methods. Supplementary Material S4-S6 show the LC-MS/MS and LC-UV/DAD chromatograms that were obtained by injecting 5 µL of extracted real samples (PL5-doc, BS1-hair, and ParL4-lab). The results of three carbonyls (Fo-DNPH, Ace-DNPH, and Pro-DNPH) are listed in Table 5. Unfortunately, the comparison was possible only for three congeners because, due to the lower sensibility of the UV/DAD detection method, only lower molecular weight analytes were detected in more than 50% of the samples that were analyzed by using UV/DAD detector. Table 5 shows that for formaldehyde and acetaldehyde existed good agreement between the two detection methods with percentage deviations between 0.2 and 29% and 0.1 and 30%, respectively. For pro-DNPH, since the determined concentrations are much lower than those that were obtained for fo-DNPH and ace-DNPH, with values between 0.32 and 5 µg m −3 , small differences can cause a considerable deviation (percentage deviations between 1.2% and 50%). The obtained percentage deviations were in line with those that were found by Chi et al. [22]. The deviations of cy-DNPH and cro-DNPH were in the range of 443-2004% and 16576-1601%, respectively; the obtained results were not acceptable, and this trend could be explained by considering the lack of selectivity of the spectroscopic method. In real samples, in fact, there could be an interference that co-elutes with the target compounds causing an increase in the signal with a consequent overestimation of the concentrations when samples were analyzed by using UV/DAD detector. This problem was overcome with the use of a high selectivity spectrometer detector operating in MRM mode. Dure to the lower LODs (Tables 3 and 4), the quantification of low concentration's CCs in air samples (mainly medium and high molecular weight carbonyls) was much more accurate by using the MS/MS detector than by using UV/DAD. For this reason, the real results will be described by using the concentrations that were obtained by MS/MS analysis.   Table 6 reports the concentration of the 12 CCs that were assessed in ten different work environments. The average blank concentrations (Supplementary Material S7) were subtracted from sample concentrations. Analyzing the samples with LC-MS/MS technique, only 2.4% of the samples were <LOQ (Table 6); this percentage rises to 68% when the samples were analyzed by using LC-UV/DAD. As we can see from Table 6, the highest concentration of hexanal (13 µg m −3 ) was recorded in one of the samples that was collected in the beauty salon (BS5-bea) and associated with a personal sampler of a beautician. Unfortunately, a comparison with the literature is not possible, due to the lack of data. The highest concentrations of cyclohexanone, crotonaldehyde, heptanal, and octanal (2.7 µg m −3 , 1.8 µg m −3 , 3. µg m −3 , and 3.2 µg m −3 , respectively) were recorded in the sample HL1lab, collected in the histology laboratory (environmental sampler). In the same workplace, was also recorded the highest concentration of decanal (4.0 µg m −3 ) in a personal sampler (HL5-tech). The concentrations that were obtained in this study were comparable or much higher than those that were obtained by Lu et al. [26,27] in special rooms and different wards of Chinese hospitals. The highest concentrations of butyraldehyde, nonanal, propionaldehyde, and valeraldehyde (13 µg m −3 , 9.9 µg m −3 , 5.0 µg m −3 , and 8.0 µg m −3 , respectively) were recorded in the enclosure environment (HRL5-lab). Unfortunately, a comparison with the literature is not possible because, to our knowledge, no study was conducted in a similar work environment.  Formaldehyde and acetaldehyde concentrations in work environments' indoor air ranged from 2.7 to 77 µg m −3 and from 1.5 to 79 µg m −3 , respectively (Table 6). For all the sampling sites, the indoor concentrations of the two pollutants were higher than the corresponding outdoor values (Table 6). For all the sampling sites, the concentrations of formaldehyde were comparable or higher than the respective concentrations of acetaldehyde, except for the beauty salon. In all the five beauty salon's samples, in fact, the concentrations of acetaldehyde were much higher than the respective concentrations of formaldehyde (Figure 1). This trend was also found in the beauty salon of Tehran by Hadei et al. [10], and could be due to the composition of cosmetic products that are used in the beauty salon. Evtyugina et al. [28] measured CCs' concentrations in a beauty salon in Spain, finding formaldehyde and acetaldehyde mean concentrations of 12 and 9.0 µg m −3 , respectively. These values are much lower than those that were determined in this work (mean concentration of 24 and 52 µg m −3 for formaldehyde and acetaldehyde, respectively).

Real Samples
The highest concentrations of formaldehyde were detected in the hospital environments (sampling sites HL3, PL3, PL4, PL5, PL1, HL5, and HL2) where formalin solutions were used. Lower concentrations were detected in the hospital environments of gastroenterology and parasitology; in the first ward formaldehyde was not deliberately used and in the second ward a closed system was used to contain formaldehyde emissions. The maximum and minimum formaldehyde contamination levels that were detected in different hospital wards in this study were compared with the literature. The comparison showed that formaldehyde contamination levels in this work were much lower than those that were found in other Italian and international hospitals [29][30][31][32][33]. Concerning the acetaldehyde contamination levels, the values that were found in this work were comparable or much lower than the values that were found by Sousa et al. [6] and Lu et al. [26] in Brazilian and Chinese hospitals, respectively. For formaldehyde and acetaldehyde contamination levels in the copy shop, the concentrations that were found in this study were in line or lower than those that were found by Saraga et al. [34] in a printery in the center of Athens but higher than those that were found by Ho et al. [1] in a photocopy center at a Chinese university. Formaldehyde and acetaldehyde contamination levels in the selected pharmacy were in line with those that were found by Loh et al. [35] in eight different drug stores in the USA. Unfortunately, to our knowledge, there are no studies that were carried out in chemistry laboratories. However, the contamination levels that were found in this selected work environment were comparable with those that were found in the pharmacy and in some wards of the investigated hospitals (HpL and CA). Unfortunately, for the concentrations of the other CCs, a comparison with the literature is not possible given the lack of data. However, to further investigate the distribution patterns of target analytes in different workplaces, PCA has been conducted on the 12 individual CCs' congeners and 52 samples. The PCA results are summarized in the biplot that is shown in Figure 2, while scores and loadings are shown in Tables 7 and 8, respectively.       Two significant components (PC1 and PC2), accounting for 68.4% of the total variance, were obtained. The biplot in Figure 2 well separates four clusters of work environments, each characterized by the presence of different CCs. The first cluster that was identified on the central part of the biplot, was predominated by formaldehyde, which did not show high variability across the 44 samples. This is in agreement with the fact that formaldehyde is considered a ubiquitous indoor pollutant [36,37]. The second cluster, in the upper right of the biplot, is characterized by the presence of all the indoor beauty salon samples and three carbonyl compounds (acetaldehyde, propionaldehyde, and hexanal) that are probably associated with the use of beauty products. The third cluster to the right of the biplot is characterized by the presence of the two samples (one personal and the other environmental) that are related to the enclosure. This environment was characterized by the presence of valeraldehyde, nonanal, and butyraldehyde, maybe associated with the presence of living animals. The fourth cluster, at the bottom right of the biplot, is characterized by the presence of one environmental sample that was collected from a histology laboratory. This work environment is characterized by the presence of five carbonyl compounds (decanal, heptanal, octanal, cyclohexanone, and crotonaldehyde) that are probably associated with the use of chemicals for different purposes, such as cleaning and disinfection.

Conclusions
This study demonstrated that LC-MS/MS detection is a suitable method for the determination of carbonyl compounds in different work environments. Good selectivity, sensibility, linearity, and reproducibility were all obtained. The proposed method, compared with LC-UV/DAD detection, resulted to be the most suitable strategy for the analysis of CCs in work environments given the high sensibility and the lowest LODs. The application of the optimized method to real samples also achieved excellent results: only 2.4% of the samples resulted to be lower than LOQs. When the same samples were analyzed by using LC-UV/DAD, this percentage rises to 68%. Among most of the work environments, formaldehyde was the most abundant detected carbonyl, followed by acetaldehyde and butyraldehyde. On the contrary, the most abundant congener that was detected in the beauty salon was acetaldehyde; this trend could be attributed to the composition of cosmetic products that are used in beauty salon. The PCA that was performed on the dataset confirms the ubiquitous character of formaldehyde and highlights the different influence of minority CCs on different work environments.