1. Introduction
As stated by Water Framework Directive 2000/60/CEE (WFD), contaminated water needs to decrease the concentration of its chemicals with time until reaching a total recovery of a good chemical status, indicating a minimum anthropogenic impact. Under this legislation scope, long-term levels of the concentration of contaminants based on different uses of water cannot be defined.
The new policies derived from WFD need the development of sustainable water management solutions in order to satisfy the demand of the uses of water from people and economic agents (agricultural, domestic, urban, industrial and recreational uses). The application of legislation to waters affected by chemical contamination (e.g., extracted groundwater) could ban its use, but the quality of the impacted water could be good enough for less stringent transitory uses under the scope of admissible risk for human health. In this situation, risk assessment methodologies could be used in order to establish protective reference levels (RLs) [
1].
Risk-based soil management is a known methodology able to assess the contamination of a site and defines remediation goals using chemical risk assessment as a base. The methodology includes the exposure assessment, toxicity assessment and target carcinogenic risk and hazard indexes [
1,
2].
As a part of this methodology, protective levels for the safe use of water (RLs) could be derived by considering a multiple pathway exposure that includes the oral pathway (direct ingestion of water or ingestion of vegetables irrigated with water), dermal pathways (direct dermal contact with water) and the inhalation pathway (inhalation of VOCs in air coming from dissolved VOCs in water) [
3,
4].
As the calculation of RLs needs a concentration referring to water and the inhalation pathway refers to air, the volatilization factors (VFs) must be introduced and have an important role in the RLs of VOCs.
For soil contamination, there are standard guidelines for risk assessment, where subsurface soil-air, superficial soil-air or groundwater-air VFs have been calculated [
5] (pp. 23–30). These VFs assume a constant concentration of the contaminant in water with time, equilibrium partitioning and steady-state transport, no loss of chemicals and steady-state and well-mixed conditions. The calculation of theses VFs start from the equilibrium concentration of the contaminant in the soil-gas phase, and models based on diffusion and advection allow obtaining a vertical flux of contaminants. By applying a mass balance in a hypothetical box or an enclosed space, the calculation of the inhalation concentration in outdoor or indoor scenarios is finally obtained.
In the case of VFs from water to air, there is a lack of these official compilation guidelines to apply for risk assessment studies, but it could be developed with the same approach. The main difference to consider is that for these water-air pathways, the transport of VOCs from water across the liquid and gas films and, finally, to air must be integrated to model the overall mass transfer coefficient of these contaminants.
Looking at the water to air pathways that are more common in risk assessment, flat and spherical geometries could be chosen to define the transference of VOCs from flat surfaces (e.g., pools, bathrooms, water extended in soil) [
6] (pp. 887–943) or droplets (sprinkler irrigation, shower) [
7] (pp. G1–G9).
Several models for the calculation of mass transfer coefficients in water could be applied to calculate these VFs [
6] (pp. 908–909). The film model states that the thickness of the limiting layer is independent of the substance. The surface renewal model defines a constant time of renewal for all volatile compounds. The boundary layer model, which is the most used model, introduces the Schmidt number. In the case of air mass transfer, models similar to the boundary layer model could be postulated.
Once the overall mass-transfer has been obtained, the second part of the VFs modeling could be developed in the same way as that for soil contamination. Starting from the fluxes or the emission with time of contaminants, box models [
5] (p. 35), [
6] (pp. 945–981) could be applied to finally obtain VFs.
A final aspect to consider is the sensitivity of VFs linked to the parameters used to calculate it. Previous works have been focused mainly on several parameters of irrigation [
8], but the role of volatility is crucial when considering new VOCs and must be investigated.
This paper develops these mentioned theoretical aspects and applies them to a specific case study that was used to calculate RLs in Catalonia [
4] to obtain VFs for several VOCs and to study their volatilization behavior as a function of the volatility (
KH) of the compounds.
3. Results and Discussion
3.1. Mass-Transfer Coefficients
Figure 1 shows the values of
Koverall as a function of
KH for the studied contaminants by using the models in
Table 1.
KH for relevant contaminants is marked in red with small letters detailed in the caption of
Figure 1. BTEX refers to benzene, toluene, ethylbenzene and xylene, all with similar volatility.
Figure 1.
Overall mass-transfer coefficients as a function of KH. a-Naphthalene, b-hexachlorobenzene, c-chloroform-dichloromethane, d-BTEX (benzene, toluene, ethylbenzene and xylene), e-trichloroethylene-tetrachloroethylene, f-carbon tetrachloride.
Figure 1.
Overall mass-transfer coefficients as a function of KH. a-Naphthalene, b-hexachlorobenzene, c-chloroform-dichloromethane, d-BTEX (benzene, toluene, ethylbenzene and xylene), e-trichloroethylene-tetrachloroethylene, f-carbon tetrachloride.
As can be seen, as a general trend, VOCs with a KH above 10−1 have an asymptotic value of Koverall. Expression (1) shows that for very volatile compounds, the water mass-transfer limits the overall mass-transfer; thus, Koverall is approximately KL for most of the studied compounds.
As KL has the same model for flat geometry for outdoor and indoor scenarios, both types of values match. Most of the contaminants have also a similar Dw value, and thus, an asymptotic value of 5 × 10−6 m·s−1 could be used as Koverall to assess these scenarios. The outlier in the graphic of flat models is MTBE, probably due to an error of Dw in the database.
In the case of spherical geometry, the overall transference is higher than in the flat surface, due to the falling of the droplets. Asymptotical values are also close to KL, but the dependence now is due to M and values between 3 × 10−5 m·s−1 and 4 × 10−5 m·s−1.
3.2. Fraction of Volatilization
In the case of spherical droplets, the fraction of volatilization (
fV) for three groups of scenarios (E1B and E1C; a group formed by E2A, E2B, E3B, E4A and Scenario E3A has been calculated. These groups are formed because they share the same parameters than influence
fV (see
Table 4 and
Table 5). Results are shown in
Figure 2. As these scenarios have a limited
Koverall, the fraction of volatilization is also limited to a maximum, but never reaches one. Agricultural scenarios, where the time of travel has been considered as higher, show fractions around 60%–70% for volatiles with
KH greater than 10
−2.
Figure 2.
The fraction of volatilization for droplets as a function of KH.
Figure 2.
The fraction of volatilization for droplets as a function of KH.
In the case of showers and irrigation, where a dropping time from two meters (0.64 s) and a diameter of 1 mm (Scenario E3A) or 2 mm (Groups 2–4) has been considered, the fraction of volatilization is around 6% to 12%.
3.3. VFs for Flat Geometries
VFs for flat geometry (Scenarios E2B, E5A and E1C) were calculated with the expressions of
Table 2 and the data from
Table 4,
Table 5 and
Table 6.
In
Figure 3, VFs are plotted as a function of
KH and are compared with 1,000 times
KH, which would have the equivalent units of L·m
−3.
Figure 3.
VFs as a function of KH for flat geometry.
Figure 3.
VFs as a function of KH for flat geometry.
For indoor scenarios and
KH below 10
−2, VFs could be estimated as 1000
KH. For higher values of
KH, the full expression has to be employed, as asymptotic values are not yet reached. On the contrary, for outdoor scenarios, the asymptotic value (around 2 L·m
−3) is obtained for most of the compounds when
KH is higher than 10
−2. This concept of limited VFs for volatiles follows Andelman’s model [
15], which establishes a VF = 0.5 L·m
−3 when
KH exceeds 4 × 10
−4.
3.4. VFs for Spherical Geometries
Figure 4 shows the VFs values for the indoor scenarios (E1B, E2A, E2B, E3A and E4B). In all cases, for volatilities greater than 10
−2, asymptotical values (that correspond to the asymptotical values of
fV) were obtained. From the two indoor scenarios, the model without the renovation of air is more conservative, because
tsh is greater than the inverse of
fR. Values equivalent to
KH are only obtained when volatility is very low and for Scenarios E2A and E3A. The flow rate, the volume of the room and the renovation or time of the shower are, thus, the main parameters that will influence these VFs.
Figure 4.
VFs for indoor scenarios as a function of KH for spherical droplets geometry.
Figure 4.
VFs for indoor scenarios as a function of KH for spherical droplets geometry.
In the case of outdoor scenarios (E1C, E2B and E4A),
Figure 5 shows also steady values for high volatile compounds, but the values are very far from the K
H values, as the renovation in the outdoor scenario is very high.
Figure 5.
VFs for outdoor scenarios as a function of KH for spherical droplet geometry.
Figure 5.
VFs for outdoor scenarios as a function of KH for spherical droplet geometry.
The flow rate, the fraction of volatilization and the width of the zone are important parameters that influence VF, as can be seen in
Table 2.
3.5. Comparison with Experimental/Real VFs Values
The highest VFs correspond to Scenarios E2B and E5A for flat geometries and E2A and E3A for sphere (drop) geometries. In these scenarios, the role of VFs in RLs is very important, and thus, a review of these values is needed to check if values are overestimated or underestimated.
Table 7 summarizes some experimental VFs and f
v obtained from the reviewed references linked to HHRA scenarios [
16]. The table also includes VFs for chloroform obtained from the measured concentration in swimming pools reviewed in [
17]. References have been chosen with temperatures of water ranging from 20 to 30 °C in all cases. These temperatures are representative of annual exposure to water and are the temperatures of reference for the
KH given in
Table 6. In the table, n means the number of experiments performed in each reference to obtain the values.
Table 7.
Experimental indoor VFs and fV for VOCs.
Table 7.
Experimental indoor VFs and fV for VOCs.
| T (°C) | Model, Scenarios Representative | VF (L·m−3) | fV | n | Refs. |
---|
Acetone | 23–24 | Flat, E2B, E5A | 0.44–0.46 | 0.049–0.058 | 2 | [18] |
Toluene | 23–24 | Flat, E2B, E5A | 3.6–5 | 0.29–0.31 | 2 | [18] |
Ethylbenzene | 23–24 | Flat, E2B, E5A | 3.1–4.6 | 0.31–0.33 | 2 | [18] |
Chloroform | 20–30 | Flat, E2B, E5A | 1.4–21.4 | - | 4–70 | [17,19,20,21,22] |
Acetone | 21–22 | Drop, E2A, E3A | 0.83–1.3 | 0.063–0.093 | 4 | [23] |
Ethylbenzene | 21–23 | Drop, E2A, E3A | 1.5–4.8 | 0.58–0.63 | 4 | [23] |
Toluene | 21–24 | Drop, E2A, E3A | 4.1–9.1 | 0.58–0.64 | 4 | [23] |
Trichloroethylene | 21–27 | Drop, E2A, E3A | 15–88 | 0.44–0.57 | 4 | [24] |
Chloroform | 26–29 | Drop, E2A, E3A | 3.5–18 | 0.46–0.52 | 4 | [24] |
Trichloroethylene | 21–22 | Drop, E2A, E3A | 54–103 | 0.50–0.67 | 2 | [25] |
VFs from flat geometry scenarios (E2B and E5A) have been compared with reference [
18], which worked with the example of a bath with surface volatilization. In this reference, the ratio surface of water to the air flow-rate is similar to the E2B scenario and half of the E5A scenario. Other important sources to calculate real VFs are the concentrations of chloroform in air and water in swimming pools [
17,
19,
20,
21,
22]. Though the parameters behind these real values are unknown in these references, they offer a realistic approach for the expected VFs in recreational and industrial indoor pools.
Comparing VF from
Table 7 with the values of
Figure 3, it can be seen that acetone, toluene and ethylbenzene have been overestimated in our study, with values that are close to 1000
KH. This overestimation is less than ten times for acetone and around ten times for ethylbenzene and toluene. In the case of chloroform in pools, the range of VFs reviewed in
Table 7 is 2
–20 times below the calculated VFs and, thus, is also overestimated.
For scenarios with drop geometry in showers (E2A and E3A), several references in
Table 7 have been used. Based on the water flow-rate to air flow-rate ratio, [
23] is the most representative of E2A and E3A. Another reference [
24] covers scenarios that are from one- to 20-times more conservative than E2A and E3A, and [
25] uses scenarios three- and 20-times more conservative.
As can be seen in
Table 7, the fraction of volatilization measured in all cases is around 4
–5-times higher than the modeled in E2A and E3A (
Figure 2). This is probably due to the use of a travel time (
ttravel) equal to the flight time (0.64 s) instead, considering the flight time plus the time remaining in the floor of the shower. When this time is increased to 10 s (Scenario E1B), the fraction of volatilization reaches the values reported in
Table 7.
When comparing VFs, it could be seen that that acetone and toluene ethylbenzene in [
23] have been underestimated about 5
–10-times in our study. This underestimation is linked to the use of low travel time. In
Table 7, [
24] uses scenarios that are more conservative, but one experiment is similar to E2A and E3A with a VF of around 15 (L·m
−3) for trichloroethylene and chloroform, which means values 10-times higher than the values of the present study.
4. Conclusions
VFs of VOCs for several scenarios could be easily calculated applying the mass-transfer from water to air concept combined with flat and spherical geometries. The Koverall is limited to KL and also limits fv for all of the volatiles studied.
As a general rule, this situation implies that VFs for high-volatility VOCs reach a limit value that is a function of
KL and other parameters from
Table 2, excluding
KH. On the contrary, low volatility VOCs in indoor scenarios could reach values equivalent to
KH, and the parameters of
Table 2 have less relevance for defining VFs.
High VF values become relevant for RLs and for risk assessment case studies, as this means that inhalation is the most important pathway that contributes to the risk. In the present work, the comparison of the highest values of VFs and fV from the present work with experimental values from other references has shown that VFs for flat geometries are overestimated for Scenarios E2B and E5A and underestimated for E2A and E3A. In all of the modeled VFs, the values are around the same order of magnitude as the references with similar hypothesis. The underestimation is mainly due to the use of low travel times for the drop. This means that this parameter could be improved, including not only the flight time, but also the time remaining in the shower.