Estimating person- to- person variability in VOC emissions from personal care products used during showering

An increasing fraction of volatile organic compounds (VOC) emissions come from the domestic use of solvents, contained within myriad commonplace consumer products. Emission rates are often poorly characterized and depend significantly on individual behavior and specific product formulation and usage. Time- concentration profiles of volatile organic compounds (VOCs) arising from the use of a representative selection of personal care products (PCPs) during showering are generated, and person- to- person variability in emissions calculated. A panel of 18 participants used a standardized set of products, dosages, and application times during showering in a controlled indoor bathroom setting. Proton transfer mass spectrometry was used to measure the in-room VOC evolution of limonene (representing the sum of monoterpenes), benzyl alcohol, and ethanol. The release of VOCs had reproducible patterns between users, but noticeable variations in absolute peak concentrations, despite identical amounts of material being used. The amounts of VOC emitted to air for one showering activity were as follows: limonene (1.77 mg ± 42%), benzyl alcohol (1.07 mg ± 41%), and ethanol (0.33 mg ± 78%). Real- world emissions to air were between 1.3 and 11 times lower than bottom- up estimates based on dynamic headspace measurements of product emissions rates, likely a result of PCPs being washed away before VOC evaporation could occur.


| INTRODUC TI ON
There is growing evidence that both aerosol and non-aerosol consumer products, including personal care products (PCPs) and household cleaning products (HCPs), contribute an increasing proportion of anthropogenic VOC emissions in high-income countries. The significance of these products has grown as historically dominant sources of VOCs such as road transport and fuel evaporation decline. 1 While atmospheric emissions of VOCs from fuels and vehicle exhaust have been well-characterized for many decades, both in terms of speciation and amount emitted, estimates of PCP emissions are only now becoming available. [2][3][4][5][6][7][8] The environmental and public health motivations to quantify and control VOCs from PCP and HCP sources are no different to other VOC emission sources. Their oxidation in the presence of NO x leads to the formation of tropospheric ozone, and they can form secondary organic aerosols (SOA), a component fraction of particulate matter. The impacts on health include, but are not limited to, respiratory and cardiovascular diseases, [9][10][11] along with several other conditions broadly characterized as "fragrance sensitivity" which includes the effects of both inhalation and dermal routes of exposure. 12 Symptoms of fragrance sensitivity include headaches, watery eyes, congestion, and contact dermatitis, which can lead to itching, swelling, and redness of the skin. These negative health effects are not limited to those with allergies, as they are not always triggered by an immune response.
A particular challenge associated with the quantification of VOCs from PCPs is that there is no common industry or regulatory standard for the disclosure of VOC ingredients or likely atmospheric emissions. VOCs can be classified in bulk terms, for example, as "parfum" or "fragrance," for reasons of intellectual property protection, but also labeling practicality, since many hundreds of VOCs may be used in a formulation. Steinemann (2009Steinemann ( , 2015 and Steinemann et al. (2011) report the range of volatile emissions found in consumer products, [13][14][15] which predominantly comprises of terpenoids and alcohols. Headspace speciation of VOCs in consumer products is a useful starting point for assessing possible emissions, but in isolation does not provide sufficient information to assess how much VOC might be released to air from PCPs based on human activity in the real world.
Yeoman et al (2020) 16 described laboratory-based atmospheric emission factors for seven commonly found VOCs in non-aerosol PCPs, two of these (limonene-representing the grouping of monoterpenes-and benzyl alcohol) being fragrance compounds. Of the VOCs released from the products studied, monoterpenes had the highest chemical potential for the formation of secondary products such as formaldehyde and SOA, dependant on the ingress of ozone from outdoors. 17 Limonene in particular has been reported previously by Carslaw and Shaw (2019) 18 to be one of the most relatively impactful VOCs on indoor chemistry due to its high potential for SOA and formaldehyde formation. [19][20][21][22][23][24][25][26][27] World Health Organization Guidelines 28 for Indoor Air Quality determine the exposure limit for formaldehyde to be 0.1 mg/m 3 (30-min average concentration) and name HCPs and cosmetics among indoor sources, along with textiles, insulating materials, and other consumer items.
While bottom-up estimates provide a standardized laboratory method for assessing the possible scale and composition of VOC emissions from individual products, they do not quantify the emissions variability arising from how individuals use those consumer products in the real world. There is likely to be variability based on amount of PCP used, duration and frequency of use, method of application, and so on. PCPs are predominantly an indoor VOC emission source, the bathroom being a location where they most commonly used, followed by the bedroom. 8 Showering is one activity, which for many people is a daily occurrence, that can include the use of a range of different products, and by extension is likely a significant component of daily VOC emissions from use of PCPs. There are several previous works describing exposure to VOCs from a range of consumer products, using both top-down and bottom-up approaches. These include product-use studies, 29-31 the use of modeling, 32,33 analysis of air samples, 34 direct analysis of consumer products themselves, 35 and combinations of these methods. 36 Despite these numerous previous works, there is no research specifically into the variability of VOC emissions from PCPs when in real-world use during specific activities such as showering. Known carcinogens and toxicants, such as trihalomethanes and chloroform, have already been identified as harmful compounds released during showering. 37,38 They are, however, contaminants and resulting reaction products of the water supply and are not a result of personal product choices or an individual's bathing habits. For consumer products specifically, there has been most emphasis in the research literature on quantifying real-world VOC emissions from domestic cleaning activities, potentially because in practical terms these are experiments that are somewhat easier to simulate, control, and measure. This is illustrated by Rossignol et al (2013), 39 where studies in an experimental house were used to identify and quantify VOCs emitted from a single HCP used in a real-life scenario.
The research presented here also takes a real-life approach to calculating emissions and concentrations of VOCs generated during showering across a cohort of volunteers using a single controlled showering facility. A common set of experimental parameters, for example, product types, dosages, duration, and ventilation were used, allowing an evaluation of the inherent variation in emissions between individuals based on their real-world behaviors. As previous work has measured simplified PCP compositions, we show here, through temporal profiles, the reproducibility between participants while those products are in-use.

| Shower facility
A single shower facility was used for all experiments located in the Wolfson Atmospheric Chemistry Laboratories, Chemistry Department, University of York. As this study was focused on quantifying VOCs emitted, the shower facility chosen had no windows to

Practical Implications
• Showering is a common activity that can use multiple personal care products; each event is seen to release milligram quantifies of VOCs such as limonene, benzyl alcohol, and ethanol, and this can perturb transient indoor concentrations.
• Within a shower room, the amount of VOC emitted varies widely between different users even if the raw materials and timing of their use are carefully controlled for.
• A personal care product may not emit all its available VOC content to air when used because of solubility effects and because of limited time for volatilization before being washed away. The temperature and humidity within the room were measured using a HM1500LF probe (TE Connectivity). Participants were issued with pre-measured doses of commonly available PCPs selected from the general range available in UK supermarkets in 2019. A face wash, shampoo, conditioner, shower gel, moisturizer, and aerosol deodorants (male and female equivalents) were selected for the participants to use. All fragranced wash-off products were "citrus" based, with the expectation they would contain limonene, which was adopted as an easy to measure tracer of emissions. Participants were given the choice of two deodorants, which although differed in scent, had the same bulk VOC propellant. Each product to be used was pre-weighed in advance and is summarized in Table 1 below.

| VOC sampling
Concentrations of selected VOCs were measured using an Ionicon (GmbH, Innsbruck, Austria) high-sensitivity Proton-transferreaction mass-spectrometer (PTR-MS). This instrument has three Varian turbo-molecular pumps and a stainless-steel ringed drift tube (9.6 cm). The instrument has been described elsewhere 40-42 ; therefore, only a brief description of the instrument set-up will be included here.
Air for analysis by PTR-MS was sampled from the shower facil-  Table 2.
Mixing ratios were then determined using the instrumentspecific transmission coefficients and reaction rate constants (k) taken from the LabSyft kinetic library, which are taken from Wang, Spanel, and Smith (2003), 43  the products and how long for. They were given 3 min between the use of conditioner and moisturizer to turn the shower off and dry.  The shower data from all 18 participants are presented as temporal profiles for each compound (Figures 1-3) where each colored line represents one of the 18 participants. Mixing ratios are presented on the left-hand y-axis, and a concentration in mg/m 3 on the right-hand y-axis.
In order to assess the overall amount of VOCs emitted from fixed amounts of products, we consider the concentration over one 15min shower activity. Data for individual participants are presented in Supporting Information 4, with Figure 4 displaying the variation between participants, and a summary in Table 3.
Generally, the temporal pattern of concentrations is consistent between participants and the concentrations measured are broadly compatible (same order of magnitude) as a bottom-up estimate of likely in-room emissions modeled in Yeoman et al (2020). 16 In Table 4 Briefly, it was found that lower VOC concentrations were detected when the participants spent a longer period rinsing the products.
This explained how small deviations in how products are used can yield significant differences in emissions, and likely accounts for some of the variation in participant data. This is reflected in the spread in the interquartile range, 0.89 mg, which implies that emissions in real-world settings for controlled amounts could be estimated to within a factor of around two.
Although this is a relatively large source of uncertainty in emissions, it is small compared to the variability associated with the total amount of product used by individual consumers, the frequency of use, or indeed product to product formulation differences. It would suggest that to narrow further the uncertainties in PCP emissions it is the overall consumption and content of VOCs that would benefit from additional study, in advance of further data on variability in use between individuals.

| Concentrations profiles and links to VOC properties
There is link between VOC solubility in water and its concentration profile; the less soluble a compound is, the more defined and higher its concentration during showering. Limonene is the best example of this; it has a relatively high octanol/water partition coefficient and is released at times when they had finished following the showering instructions and that we were unable to monitor.
A compound's potential for dermal absorption through skin lipids may also be an influencing factor on concentration, and there is potential for all of the products used in this study to be dermally absorbed, even if this is just through hands while applying to the hair in the case of conditioner. Limonene is very effective at penetrating the skin, 48 and there is evidence it could be used as an enhancer for drug delivery for this reason. 49,50 Consequently, this may have a pronounced effect on the amount of limonene available for evaporation. Although also dermally absorbent, 51 benzyl alcohol does not have the drug delivery enhancement potential limonene does, which may explain why the bottom-down calculations for limonene have been overestimated to a much greater degree, as seen in Table 4.
The production of secondary pollutants is possible, formaldehyde in particular from the reaction of limonene with OH radicals, and subsequent unimolecular decomposition. 52  These data are informative in terms of the peak concentration of VOCs individuals may be exposed to during a single, common activity. For context, in 1998 the World Health Organization (WHO) reported no indication of inhalation risk from limonene due to limited data on the rate in which a harmful concentration can be reached on evaporation. 54 Although this paper does not attempt to address this specifically, it provides a possible timescale for reaching high concentrations during one activity. As previously described, fragrance sensitivity, and the health risks associated with it, can occur through routes other than that of inhalation, such as the dermal route. Contact dermatitis usually relates to direct application of a compound to the skin and is known to occur with limonene as it is oxidized. [55][56][57] However, if exposure levels are high enough in the gasphase, there may also be potential for a dermal reaction to be triggered, particularly to the eyes which can be especially sensitive. 58 As limonene is relatively unaffected by the presence of liquid water, it can be used as a "tracer" for the variability and uncertainty Although there remain considerable uncertainties in PCP emissions, and this field of work is in its infancy, it seems plausible that greatly improved domestic VOC emissions estimates could be TA B L E 4 Bottom-up and top-down estimates using emission factors calculated in Yeoman et al (2020), 16 product usage estimates found in Table 1   , and a correction factor for real-world use, accounting for the reality that only a fraction of the VOC content in a product is released to air when used. Although scaling up emissions from a very small study such as this carries with it large uncertainties, using a median emission of 1.8 mg limonene per showering activity, and assuming this activity is replicated by half the UK population each day, would lead to an annualized emissions of around 13 tonnes of limonene per year from showering.

ACK N OWLED G EM ENTS
AMY acknowledges support for a PhD studentship from NERC.
ACL and MS are supported by the National Centre for Atmospheric Science LTSS national capability program. MS also acknowledges financial and practical support from Syft Technologies (New Zealand).
AMY also acknowledges Matthew Thompson for his role in data acquisition.

CO N FLI C T O F I NTE R E S T
There are no conflicts of interest to disclose. The project complies with the ethical research approval processes at the University of York which require prior ethics assessment and independent approval at project conception, and are captured as part of the annual ethical approval process for the NCAS research program delivered at York.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/ina.12811.