The objective of this study was to evaluate the suitability of five breath collection variants for onsite exhaled breath sampling from dairy cows connected to SPE cartridges to capture the VOCs from the collected exhaled breath.

The number of detected GC-MS peaks from the unused, cleaned sampling materials were as follows (mean ± standard deviation): carbon filter: 14 ± 2.83, over-mask: 28 ± 4.24, Teflon tube: 3 ± 0.7, and HR-XAW polymer: 4 ± 2.1. These results indicate that the materials used, particularly the over-mask, released some VOCs. Therefore, it was important to correct the measured VOCs using the corresponding background VOCs. In addition to VOC emissions from the sampling materials themselves, it is also crucial to assess possible VOC deposits within the sampling devices, which could lead to VOC carryover from one sample to another. To investigate this, data from experiment 1 were used to compare the number of detected GC-MS peaks in barn air samples collected either before (background air) or after (post-breath-collection air) sampling exhaled breath from an animal (table 1). The difference—i.e., the subtraction of background air peaks from those in post-breath-collection air—provides an estimate of VOCs possibly originating from deposits in the sampling system (potential VOC-deposits), representing a potential risk of VOC carryover between samples.

Table 1. The total number of GC-MS peaks in the barn air before (background air) and after collecting exhaled breath from three individual cows (post-breath-collection air) as well as the resulting calculated potential VOC deposits within the different breath collection devices (mean ± coefficient of variation).

 Breath collection devices MaskNGreenFeedUGreenFeedFBackground air1194 ± 0.011230 ± 0.011232 ± 0.01Post-breath-collection air1183 ± 0.011240 ± 0.031270 ± 0.03Potential VOC deposits (Δ Post-breath-collection air–background air)118 ± 0.17200 ± 0.53149 ± 0.17

MaskN: tight-fitting face mask, GreenFeedU: the GreenFeed system with air sampling conducted before the filter, GreenFeedF: GreenFeed system with air sampling conducted after the filter.

The GC-MS peak counts differed between the three sampling devices. A high number of peaks in the background air was detected using all three breath collection devices with the highest number of peaks (mean ± standard deviation) was detected using the variants of the GreenFeed system (GreenFeedF: 1232 ± 5.6; GreenFeedU: 1230 ± 2.5), followed by MaskN (1194 ± 10.2). The high number of peaks detected using GreenFeed can be attributed to the GreenFeed being a more open system (ratio exhaled air: surrounding air 1:40). This may possibly result in sampling more background VOCs compared to MaskN, which provides less contaminated sampling of VOCs by the accumulation of exhaled breath in MaskN. An alternative reason might be that the GreenFeed system involves offering bait feed to the animals, which possibly contaminates exhaled breath samples with feed VOCs.

The post-breath collection air sample contained similar numbers of peaks to the background air samples, but many of its peaks were not present in the background air (table 1). Those potential VOC-deposits (Δ post-breath-collection air–background air) consisted mainly of esters (21.3%), alcohols (12.2%), alkanes (11.2%), alkenes (10.7%), ketones (10.2%), ethers (8.24%), amines (5.57%), azoles (5.33%), and carboxylic acids (2.55%). This was particularly pronounced for GreenFeedU samples, in which 16.3% of the detected peaks were not present in background samples, followed by GreenFeedF (12.1%) and MaskN (9.88%) samples. These potential VOC deposits on the material of the sampling devices could lead to VOC carryover between cows. Particularly susceptible to VOC deposits are porous materials, filters, plastics, untreated metals, materials with large surface areas, and areas with the presence of dust [21]. A potentially more pronounced VOC carryover using the GreenFeed system may be related to its larger surface area compared to the MaskN system. This includes the feed trough from which air is drawn, which is comparable to MaskN samples in terms of surface exposure. The internal components of the GreenFeed system, including the air filter through which the air is subsequently transported, present potential surface areas for more VOC adsorption. Additionally, a significant proportion of GreenFeed system surfaces are inaccessible for cleaning with water and drying.

As mentioned earlier, VOCs released from bait feed could contribute to an increased number and concentration of VOCs potentially depositing in the GreenFeed filter and being released again later. Such VOC deposition or release could alter the VOC profile in the airflow downstream of the filter, potentially changing the composition of the sampled VOCs before (GreenFeedU) and after filtration (GreenFeedF). We did not identify the 467 (118 + 200 + 149) peaks, which can likely be considered deposited VOCs. However, all the post-breath-collection VOCs were present in exhaled breath samples after correcting for background air peaks [17], but in the latter samples in at least 1.5 times greater concentrations. Despite their relatively low concentrations, VOC deposition on the sampling material may have increased the concentrations of these VOCs in the subsequent sample.

To compare the suitability of the five sampling variants for the collection of exhaled VOCs from cows, the data from experiment 2 were used. The mean number of GC-MS peaks detected in exhaled breath samples from Exp2 was around 1218. The numbers are comparable with the number of peaks detected in our previous study [5] using MaskN to sample exhaled VOCs from dairy cows. The number of GC-MS peaks varied by breath collection variant and among the cows sampled (table 2). Breath samples collected with GreenFeedU contained the highest mean peak number (+11%–15% compared to the other variants), followed by MaskN, MaskF, GreenFeedF, and MaskV samples. The greater number of peaks in GreenFeedU (+13.93 ± 0.32%) compared to GreenFeedF samples suggests that exhaled VOCs may either remain attached to the GreenFeed filter or undergo a reduction in concentration, which, unsurprisingly, aligns with the intended function of a filter. As an alternative to air filtering, background VOCs may be reduced by supplying synthetic air as inhaled air for the animal, as demonstrated by the reduced number of peaks observed using MaskV. To some extent, supplying synthetic air permits the separation of the barn environment from the exhaled breath and the sampling process. The number of GC-MS peaks in exhaled breath samples was corrected using the respective background air samples to determine the exhaled VOCs. Specifically, VOCs were considered exhaled VOCs if their peak areas exceeded those of the background air by at least 50% [17] (tables 2 and 3).

Table 2. The total number of GC-MS peaks detected in the exhaled breath and background air samples from six cows using the five breath collection variants as well as exhaled breath peaks exceeding background air peaks by at least 50% as considered exhaled VOCs (mean ± coefficient of variation).

 Breath collection variantsNumber of peaks detectedMaskNMaskFMaskVGreenFeedUGreenFeedFExhaled breath samples1209 ± 0.081197 ± 0.061165 ± 0.071341 ± 0.311177 ± 0.08Background air samples1210 ± 0.011220 ± 0.031161 ± 0.011231 ± 0.011232 ± 0.01Exhaled VOCs512 ± 0.29539 ± 0.36596 ± 0.37541 ± 0.30516 ± 0.28

MaskN: tight-fitting face mask, MaskF: the face mask with the openings sealed using activated carbon filters, MaskV: the face mask covered with an over-mask ventilated with synthetic air supply for cow breathing, GreenFeedU: the GreenFeed system with air sampling conducted before the filter, GreenFeedF: GreenFeed system with air sampling conducted after the filter, exhaled VOCs: VOCs from exhaled breath samples were considered exhaled VOCs when they exceeded background air peaks by at least 50% [17].

Table 3. Number of exhaled volatile organic compounds (VOCs) from six dairy cows, collected using five different breath sampling variants, that exceeded background air peak levels by at least 50%. (mean ± coefficient of variation).

 Breath collection variantsChemical compound groupMaskNMaskFMaskVGreenFeedUGreenFeedFAldehydes9.0 ± 0.479.3 ± 0.6515.8 ± 0.4812.5 ± 0.489.8 ± 0.80Alcohols39.8 ± 0.5136.3 ± 0.3045.7 ± 0.3937.3 ± 0.2236.5 ± 0.32Alkanes68.8 ± 0.3471.3 ± 0.3979.0 ± 0.5561.3 ± 0.3859.3 ± 0.37Alkenes55.3 ± 0.5360.7 ± 0.5261.5 ± 0.5948.7 ± 0.5449.0 ± 0.37Alkynes1.8 ± 1.172.0 ± 1.003.2 ± 0.722.0 ± 0.701.5 ± 0.73Azoles13.3 ± 0.5316.0 ± 0.3417.8 ± 0.4916.5 ± 0.2114.8 ± 0.45Amides27.5 ± 0.6123.8 ± 0.4627.7 ± 0.4627.2 ± 0.4821.0 ± 0.59Amines28.5 ± 0.5222.2 ± 0.3426.2 ± 0.4025.0 ± 0.3523.0 ± 0.55Carboxylic acids13.7 ± 0.3410.7 ± 0.3512.7 ± 0.2612.8 ± 0.5514.8 ± 0.39Esters110.2 ± 0.27104.8 ± 0.23116.7 ± 0.28121.0 ± 0.22112.2 ± 0.22Ethers21.7 ± 0.2418.7 ± 0.2622.5 ± 0.4422.0 ± 0.3921.8 ± 0.27Ketones69.0 ± 0.3262.7 ± 0.2986.7 ± 0.3572.3 ± 0.3767.7 ± 0.28Nitriles8.2 ± 0.289.3 ± 0.489.0 ± 0.4110.7 ± 0.2410.3 ± 0.28Pyridines6.8 ± 0.548.3 ± 0.417.7 ± 0.577.0 ± 0.467.2 ± 0.57Others61.5 ± 0.3053.5 ± 0.3059.5 ± 0.3761.8 ± 0.2162.0 ± 0.30

Exhaled VOCs were identified at Level 3 (assigned to their respective compound classes based on mass spectral similarity [19, 20]), MaskN: tight-fitting face mask, MaskF: the face mask with the openings sealed using activated carbon filters, MaskV: the face mask covered with an over-mask ventilated with synthetic air supply for cow breathing, GreenFeed U: the GreenFeed system with air sampling conducted before the filter, GreenFeedF: GreenFeed system with air sampling conducted after the filter.

The exhaled VOCs were identified at Level 3 (table 3), meaning they were assigned to their respective compound classes based on mass spectral similarity [19, 20]. Overall, about 567 Level 3 exhaled VOCs of 15 chemical compound groups were identified, accounting for about 45.1% of all detected VOCs in exhaled breath samples. Esters (20.9%) were the most prevalent, followed by ketones (13.2%), alkanes (13.0%), alkenes (10.2%), alcohols (7.23%), amides (4.70%), amines (4.61%), ethers (3.94%), azoles (2.90%), carboxylic acids (2.39%), aldehydes (2.09%), nitriles (1.76%), pyridines (1.37%), and alkynes (0.39%) (table 3). The proportions of the most detected chemical compound groups were comparable to those reported by Eichinger et al., who used MaskN to sample exhaled VOCs from dairy cows [5]. However, in the present study, a higher number of amides (+81.6%), amines (+68.4%), carboxylic acids (+68.6%), esters (+72.6%), and ketones (+66.7%) were identified. These differences may be attributed to the use of XAW cartridges in the present study, which have a higher sensitivity to capture ketones. An alternative reason might be variations in metabolism, potentially influenced by differences in lactation stages, as in the present study cows in the dry period shortly before calving were used. After calving, energy expenditure is elevated due to the initiation of high milk production. However, the energy uptake through feed is incapable of meeting the energy demands, which consequently leads to catabolism of adipose tissue and elevated ketone body production [22].

All chemical compound groups were detectable in the exhaled breath samples from all six cows, regardless of the breath collection variant. However, the number of VOCs per chemical compound group showed large variations between animals (up to at least 50%). The number of Level 3 VOCs within a particular chemical compound group differed between the breath collection variants. Aldehydes varied most among the breath collection variant, with the highest number of VOCs collected by MaskV (+61%–69% compared to the other sampling variants) and GreenFeedU (+27%–39% compared to MaskN, MaskF and GreenFeedF). Furthermore, MaskV samples contained the highest number of ketones (+19%–38%), alcohols (+15%–26%), and alkanes (+11%–33%) compared to the other sampling variants. GreenFeedU exhaled breath samples exhibited the highest number of esters (+4%–15%) compared to the other sampling variants. In contrast, alkenes were primarily detected using MaskV (+11%–26%) and MaskF (+10%–25%) compared to the other breath sampling variants.

A total of 75 individual VOCs were detected and identified at Level 2 (match factor >80% and difference in the calculated RI of in maximum ± 15 of the reference RI [19, 20]) from the compound groups of alcohols, aldehydes, esters, ketones, phenols, pyridines, and terpenes (table 4; supplementary table S2) in exhaled breath samples. All 75 VOCs were present in all breath samples from all cows using all five breath collection variants. The concentrations of the exhaled VOCs exhibited considerable variation among cows (table 4; supplementary table S2). This may be explained by differences in metabolism and feeding [8, 9]: in Exp2, two cows were fed according to recommendations, and four cows had an energy-richer diet. The concentrations of the exhaled VOCs also differed between the breath collection variants, with concentrations varying up to 95% (e.g., for propyl propionate) between one variant and another. Across the 75 detected VOCs, the highest mean VOC concentrations were observed in the samples collected using MaskV, followed by GreenFeedU (concentrations around 11% lower than in MaskV samples), MaskF (−12%), GreenFeedF (−25%), and MaskN (−30%).

Table 4. Exhaled volatile organic compounds (VOCs) from the as physiologically relevant considered chemical groups (aldehydes, alcohols, azoles, amides, amines, carboxylic acids, esters, ethers, ketones, nitriles, pyridines, sulfur containing compounds and terpenes) after subtraction of VOCs considered as of barn origin (<1.5 * blank peak area [17];) detected in all six dairy cows of experiment 2 using five different collection variants.

 Collection variant (GC-MS peak area) MaskNMaskFMaskVGreenFeedUGreenFeedFVolatile organic compoundsMeanCV (%)MeanCV (%)MeanCV (%)MeanCV (%)MeanCV (%)αα-4-trimethyl-cyclohexanemethanol4095 466144554 937712960 30368162 23870273 726150Benzyl alcohol215 677183891 200116694 633210291 952104308 686953,5,5-trimethyl-1-hexanol59 45311085 720128119 894153115 7398350 1011042,6-dimethyl-7-octen-2-ol785 55173941 985132344 196112345 75744196 8222012,6-dimethyl-2-octanol4665 148113951 503713525 8901051904 332841955 9401252,5-bis(1,1-dimethylethyl)-1,4-benzenediol92 57669110 09949109 5637682 12069143 852632-ethyl-1-hexanol64 090272516 326100476 22130682 8748571 662951-undecanol94 27283146 78671140 707129130 29771120 353841-dodecanol30 1937137 3635933 33312334 1776327 943801-decanol34 13516593 55278105 43720467 8638250 4101131-(2-methoxy-1-methylethoxy)-2-propanol134 805135161 96868238 787135151 83375123 84979(E)-3-hexen-1-ol275 317108280 29185197 35688604 189107363 19393Octanal163 70371178 35096208 31997250 13965211 39379Decanal133 17317431 57915991 31689173 25210549 9922184-methyl-benzaldehyde127 0549794 742100121 495111133 272107103 0831604-ethyl-benzaldehyde7917 107618934 549449340 122698430 345468790 349413-ethyl-benzaldehyde7917 118618933 933449340 960698431 196468790 394412,5-dimethylbenzaldehyde4248 027997818 764956952 761804287 066737210 844852,4-dimethyl-benzaldehyde9299 3625910 214 3844310 672 266639874 5524510 212 864412-methyl-benzaldehyde805 62957829 992671212 274781158 37656963 391592-methyl-3-phenyl-2-propenal15 25426077 56587112 00129522 8109217 9861062-ethyl-benzaldehyde7917 139618934 906449340 101698431 336468790 349411H-pyrrole-2-carboxaldehyde638 1541061136 345601051 576145543 13878557 94043(E,E)-2,4-hexadienal363 957133188 14694707 376106323 382110525 179105sec-butyl butyrate168 27793173 171105263 373149162 231110154 754129Propyl propionate131 1003052355 5181192551 846318297 774110201 02078n-propyl butyrate158 15793143 65072107 636105278 47490316 198104n-propyl benzoate40 1998021 9519140 21810230 9215231 47786n-hexyl acetate97 603170143 73669165 178171139 0796588 44497n-butyl butanoate90 342266207 35279128 68622529 4039730 052264Methyl salicylate22 52913740 5239137 85417851 7849247 08174Isopropyl myristate36 9218729 94612751 5687738 1279230 977107Isobornyl acetate90 43377140 60798166 15092113 54290112 37060Ethyl butanoate1154 3371752707 1641492719 9851872950 7101843178 443129Ethyl benzoate80 1976460 8584757 35173121 7484998 444114Ethyl 2-hydroxypropionate6534 141627991 283436017 204965961 149404354 65070Dimethyl butanedioate338 974126535 22797621 562192400 49398321 881125Butyl propionate126 543176194 982144225 261224148 81914593 162153Benzyl acetate30 3409339 2183846 08411324 2844023 915373-methyl-1-butyl butanoate50 4776912 10213915 5888927 85218632 3441083-hydroxy-2,2,4-trimethylpentyl isobutyrate165 963110101 627104216 882160243 579108174 1671762-ethoxyethyl acetate37 34413025 66926349 069173489 24513046 0582252-acetoxy-1-propanol206 87999263 006110341 835114280 574120282 432811-methoxy-2-propyl acetate163 140160526 28492541 571223341 622101282 810104α-isomethyl ionone21 5217966 01719230 973221131 51322460 03943Cyclohexanone73 519160198 48778186 76319987 0016681 37760Benzophenone10 14419326 0774127 01521210 9503311 03139Acetophenone423 0461501084 216811207 895207724 36285643 905746,6-dimethyl-, (1R)-bicyclo[3.1.1]heptan-2-one95 474113134 652138202 078132233 348103116 409141