Article pubs.acs.org/est
Cite This: Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Volatile Organic Compound Emissions from Polyurethane Mattresses under Variable Environmental Conditions Kira Oz, Bareket Merav, Sabach Sara, and Dubowski Yael* Civil and Environmental Engineering, Technion, Israel Institute of Technology, Haifa 3200003, Israel
Downloaded via BUFFALO STATE on July 17, 2019 at 04:52:41 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
S Supporting Information *
ABSTRACT: Sleeping microenvironment (SME), is characterized by higher temperature, humidity, and CO2 concentration. Emission of Volatile Organic Compounds (VOC) in SME is important considering the long duration people spend there with high proximity between their respiration inlets and potential emission sources, such as bedding material. This study concentrates on the influence of SME conditions on VOC emissions from polyurethane mattresses, and provides first approximation for inhalation exposure during sleep, based on measured emissions. Eight types of polyurethane mattresses were tested in a parallel continues-flow chamber system, to compare between VOC emission under different temperature, relative humidity, and CO2 concentrations. Contribution of mattress covers to emission fluxes was also examined. Eighteen VOCs were quantified with fluxes ranging from 10−4 to 10−1 mg/(h·m−2). Under sleeping conditions VOC emissions increased significantly. Elevated heat seems to be the major contributor to the enhanced emissions, compared to elevated relative humidity and CO2 concentration. Exposure levels estimated for sleeping child/infant indicate that SME can be a significant contributor to VOC exposure, yielding concerning exposure levels for few compounds. Furthermore, the present study demonstrates the strong dependency of sleeping person exposure on air exchange rate between his breathing zone and bedroom air (λBZ).
1. INTRODUCTION Exposure to anthropogenic volatile organic compounds (VOC) and semi-volatile organic compounds (SVOC) is unavoidable due to the abundance of emission sources. Connections to cancer,1 lung function and asthma,2−4 and developmental issues5 are only part of the potential risks that were found in connection with some VOC exposure. Elevated levels of VOC and SVOC are found in our atmosphere due to industrial, transportation-related, and volatile chemical products. However, the indoor environment in general and the sleeping microenvironment (SME) in particular are considered to be a significant source for VOC exposure.6,7 SME may expose us to biological (bacteria, fungi, etc.), physical (particulate matter), and chemical (VOCs, flame retardants, plasticizers, etc.) pollutants.8 The special interest in SME with regard to VOC exposure arises from four main factors: (1) the long duration we spend in this environment over 30% of our time for youth and adults, and up to >50% for infants,8 (2) presence of many VOC emission sources in this contained environment, such as mattresses, beddings, bed frame, and other bedroom furniture,8 (3) typically low ventilation of bedrooms during nighttime (often 97%. 2.2. Experimental Setup. Figure 1 shows the schematics of the experimental system. The setup consisted of two parallel
(1) Experimental setting 1 (ES-1): Cover-free mattresses (i.e., without their cloth cover) under full SME conditions (T = 36 ± 1 °C, RH = 90 ± 5%, [CO2] = 1000 ± 500 ppm) vs control environment (T = 20 ± 1 °C, RH < 5%, [CO2] < 0.5 ppm). (2) Experimental setting 2 (ES-2): Cover-free mattresses under heated environment (T = 36 ± 1 °C, RH < 5%, [CO2] < 0.5 ppm) vs humid and CO2 enriched atmosphere (T = 20 ± 1 °C, RH = 90 ± 5%, [CO2] = 1000 ± 500 ppm). B
DOI: 10.1021/acs.est.9b01557 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
deviation of a set of seven method blanks. Former approach was used for compounds that showed no signal at all in blanks runs. Analytical blanks for thermal desorption method included seven Tenax-GR tubes analyzed similarly to experimental samples. Analytical blanks for aldehyde consisted of fresh DNPH-cartridges eluted and extracted like experimental samples. Aldehydes’ MDL were determined as lowest concentration used in calibration procedure since linearity was achieved even for the lowest concentration. MDL values for all target compounds are presented in Table S1 of the Supporting Information (SI). 2.5. Estimation of Air Concentrations and Exposure Dose in SME. Bedroom air concentration ([Ci]R, [μg/m3]) of a pollutant emitted from mattress can be calculated using massbalance eq (eq 1), assuming chemical stability and a well-mixed bedroom:
(3) Experimental setting 3 (ES-3): Mattresses with vs without their original covers, both under full SME conditions (T = 36 ± 1 °C, RH = 90 ± 5%, [CO2] = 1000 ± 500 ppm). These comparisons were designed to examine the influence of SME conditions on VOCs emissions as well as to isolate which environmental factor (i.e., temperature or relative humidity and CO2 concentration) has the dominant impact. Additionally, the mattresses were examined with and without their original cover to infer what is the influence of the latter on VOCs emissions. The results are presented in two measures, emission fluxes and changes in emission fluxes as a function of environmental parameters. The emission fluxes are defined as the mass of a VOC collected at the reactor exit, divided by time of measurement (∼3 h) and geometric surface area of the mattress exposed to the air flow. 2.3. Measurement and Analysis. VOCs collected on Tenax-GR tubes were extracted and analyzed using a thermal desorption (ATD650, PerkinElmer)−gas chromatography mass spectrometer (Clarus 680, PerkinElmer) system. Thermal desorption was carried out in 280 °C for 20 min. The GC oven temperature was kept at 40 °C for 2 min and then increased at 7 °C/min up to 220 °C followed by 25 °C/min ramping up to 300 °C, where it remained for 5 min. Ions were collected in the range of 35−400 m/z. The GC column used was DB5, 30m, 0.25 mm ID, and 0.25 μm df (PerkinElmer). High purity He (99.999%) was used as carrier gas at flow rate of 1 mL/min. VOCs identification was carried out using Automated Mass Spectral Deconvolution and Identification software (AMDIS, NIST). Compounds quantification was done using external calibration curves for each material of interest using their analytical standards. Each preconditioned calibration tube was loaded with a 1 μL of standard solution at the desired concentration and an inert Helium gas (>99.999 purity) was purged through it for 5 min at 50 mL/min flow to purge solvent from the tube. Compounds presented in this study were identified with very high probability (probability >40%, match and R-match >800), and demonstrated significantly higher concentrations (at least 1 order of magnitude) compared to a blank sample of the empty glass cylinder measured under the same conditions of the mattress experiment. Aldehyde identification and quantification were done after derivatization using the DNPH cartridges following EPA TO11A standard method. The DNPH cartridges were extracted with 5 mL HPLC grade Acetonitrile (Mallinckrodt). Twenty μL samples from the extracts were injected into an LCMS (Agilent, A1200) equipped with Hypersil GOLD 150 × 4.6 mm2 column (Thermo Scientific). The quantification process was done using an external calibration curve with analytical standards of the derivatized aldehydes in Acetonitrile. 2.4. QA/QC. Sampling duration and air flow rates through the tube were set according to preliminary experiments that tested emissions under different flow rates (80 and 160 mL/ min) and durations (3 and 6 h). The results indicated that 3 h sampling at 80 mL/min were enough to quantify fluxes of VOCs of interest. Sorbent breakthrough was tested by putting two tubes in line (6 h run at 80 mL/min). Amount of VOCs of interest in second tube were negligible (i.e., at least an order of magnitude lower than in first tube). Method Detection Limits (MDLs) where determined following two established approaches:25 as the lowest concentration with linear response in calibration procedure, or by the mean blank concentration plus 4.5 times the standard
d[Ci ]R E = λR − O × ([Ci ]out − [Ci ]R ) + dt VR
(1)
−1
where (λR−O) is AER with outdoor air [h ], E is the emission rate (= measured emission flux × mattress area, [μg/h]), and VR is the room volume [m3]. For very small ARE, as often happens in bedrooms during nighttime,8 [Ci]R increases almost linearly with time (eq 2) and the corresponding inhalation dose (D, [μg]) can be expressed by eq 3 (assuming, for simplicity, Ct=0 = 0). [Ci ]R (t ) =
E (Δt ) + [Ci ]R (0) VR
t1
D(t ) = =
∫t 0
C R (t ) × In R × dt =
E × In R VR
(2) t1
∫t 0
t × dt
(Δt )2 E × In R × VR 2
(3)
where InR is the inhalation rate of the sleeping person; taken as 3 L/min for infants below 2 years old and 6 L/min for older children.26 Such estimation probably underestimates the inhaled dose as mixing within bedroom environment is often limited, resulting in concentration gradient between breathing zone (BZ) of the sleeping person and bulk room.12 eq 4 depicts a mass balance for an inert compound i in the BZ, considering also AER between BZ and bulk room (λBZ): d[Ci ]BZ E = λBZ × ([Ci ]R − [Ci ]BZ ) + BZ dt VBZ
(4)
3
where VBZ is the BZ volume [m ], EBZ is the emission rate of compound i into the BZ [μg/h], and [Ci]R is its concentration in the bulk room [μg/m3]. BZ of a sleeping person is often estimated to be a hemisphere with radius of 30−50 cm,8 which corresponds to an effective mattress emission area (ABZ) of 0.3−0.8 m2, and a BZ volume of 0.057−0.262 m3. Considering the large volume difference between bulk room and BZ (about 2 orders of magnitude), it can be assumed that [Ci]R is dominantly affected by emissions from the mattress located outside the BZ and not from air exchange with the BZ air. In such case, eq 2 can be substituted in eq 4 (eq 5), yielding after integration the expression for [Ci]BZ (eq 6): d[Ci ]BZ E ×t E − λBZ[Ci ]BZ = λBZ R + BZ dt VR VBZ C
(5)
DOI: 10.1021/acs.est.9b01557 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
a
>1 >1 1.4 × 10−2
acetaldehyde formaldehyde hexanal
D
2079 (21) 6633 (158) BDL
HDR
NA NA NA
476 (224) 640 (36) 1514 (72) 378 (89) 966 (131) 203 (40) 474 (63) 256 (71) 867 (167) 5374 (614) 8692 (1178) 106283 (32387) BDL BDL 371 (21)
ng/(m2·h)
MDY
1034 (526) 5257 (710) BDL
304 (167) 138 (105) 322 (39) 193 (70) 428 (202) 286 (172) 352 (201) 146 (64) 138 (116) 3654 (1293) 553 (245) 10646 (6023) BDL 608 (387) 698 (472)
ng/(m2·h)
MDC
3979 (2430) 8368 (980) BDL
443 (325) 282 (202) 1140 (714) 296 (141) 786 (477) 212 (61) 391 (175) 158 (99) 179 (79) 4130 (1942) 4760 (3904) 197617 (53913) BDL 159 (3) 2769 (1889)
ng/(m2·h)
LDI
BDL BDL 3527 (1718)
101 (26) 155 (105) 1140 (934) 299 (184) 862 (589) 234 (99) 350 (184) 163 (64) 545 (276) 22058 (15958) 6498 (3985) 26149 (21569) 10074 (5100) 95 (71) 1095 (797)
ng/(m2·h)
LDY
4410 (1488) 6408 (554) BDL
490 (279) 331 (239) 1018 (529) 330 (151) 1346 (456) 185 (48) 386 (165) 166 (63) 225 (52) 6459 (2155) 5085 (3212) 42472 (26682) BDL 451 (253) 1086 (602)
ng/(m2·h)
LDR
NA NA NA
1108 (128) 537 (47) 2707 (257) 735 (292) 1738 (175) 513 (210) 852 (128) 245 (35) 478 (198) 3220 (187) 9333 (1190) 84105 (29482) BDL 72 (45) 3803 (3103)
ng/(m2·h)
BDL- bellow detection limit. bNA- wasn’t tested with DNPH cartridges. cThe last three rows display compounds analyzed using TD-GC-MS and the three aldehydes analyzed using DNPH cartridges.
5555 (3705) 6080 (284) BDL
523 (256) 680 (145) 1802 (996) 332 (92) 895 (319) 307 (89) 356 (74) 110 (26) 161 (87) 5084 (2155) 4298 (2945) 33273 (14937) BDL 2053 (1024) 2466 (2278)
2.0 × 10−1 1.3 × 10−1 3.8 × 10−2 1.3 × 10−2 1.2 × 10−2 6.3 × 10−3 2.7 × 10−3 2.3 × 10−3 1.4 × 10−3 5.0 × 10−4 4.0 × 10−4 4.0 × 10−5 1.2 × 10−5 6.9 × 10−6 5.5 × 10−7
hexane benzene toluene ethylbenzene p-xylene α-pinene 1,2,4-trimethylbenzene 1,4-dichlorobenzene octametyl cyclotetrasiloxane nonanal decamethyl cyclopentasiloxane 2-ethylhexanoic acid tris(1-chloro-2-propyl) phosphate (TCPP) butylated hydroxytoluene 2,6-di-tert-butyl-1,4-benzoquinone
150 (105) 307 (268) 536 (213) 274 (195) 385 (139) 222 (104) 309 (216) 120 (68) 110 (82) 3889 (1049) 531 (358) 34354 (21962) BDL 227 (30) 763 (451)
HDM ng/(m2·h)
HDY ng/(m2·h)
bar
mattress type
vapor pressure at 25 °C
Table 1. Emission Fluxes Averages (and Their Standard Deviations) of Different VOCs, Emitted from the Tested Mattresses at the SME Experiments (36°C, ∼90% RH, ∼1000 ppm of CO2)a,b,c
Environmental Science & Technology Article
DOI: 10.1021/acs.est.9b01557 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology [Ci ]BZ, t =
ÄÅ ÉÑ ÑÉ ÅÄ ER × t E ÑÑ E ÑÑ 1 ÅÅÅÅ E BZ 1 ÅÅÅÅ E BZ + − R ÑÑÑÑ − − R ÑÑÑÑ ÅÅ ÅÅ λBZ ÅÅÇÅ VBZ λBZ ÅÅÇÅ VBZ VR VR ÑÑÖÑ VR ÑÑÖÑ
exp(− λBZ × t )
of TCPP from this mattress was on the order of 10 000 ng/(m2· h). This value is much higher than TCPP emission flux previously measured at 23 °C from a mattress (12 ng/(m2·h)), but very similar to the flux observed from polyurethane upholstery foam (7700 ng/(m2·h)).13 Unfortunately, these authors provided no information regarding the mattress they tested and to the best of our knowledge there are no additional reports quantifying emission fluxes of TCPP from PUF substances (see recent review by ref 8). TCPP has reproductive and developmental human health effects, and possibly also carcinogenicity.31 Considering the amount of time infants spend sleeping and these potential adverse health effects of TCPP, its detection in the airflow above mattresses, with area-specific emission flux on the order of μg/(m2·h), raises a question on the benefit of adding such flame retardant to mattresses. Another compound which depicted higher emissions fluxes from the infant mattress than from the other tested mattresses was nonanal (average values of 22 vs 2000 ng/(m2·h) in the HDY mattress) probably results from the diversity of antioxidants used in PUF mattresses. In recent years there is growing concern regarding butylated hydroxytoluene wide usage (including in food and cosmetic products) due to few studies suggesting it has carcinogenic potential (e.g.,33). Recent studies have shown that butylated hydroxytoluene is a common contaminant in indoor dust.34,35 2,6-Ditert-butyl-1,4-benzoquinone, a transformation product of butylated hydroxytoluene,35 depicted emission fluxes on the range of 300−3000 ng/(m2·h), with even higher fluxes from the low density raw PUF (LDR). In opposition to the nondefinitive health effects of butylated hydroxytoluene, 2,6ditert-butyl-1,4-benzoquinone was proven to have adverse toxic effects even at quite low concentrations, causing internucleosomal DNA fragmentation through H2O2 and oxygen radical generation.36,37 It is worth noting that 2,6-ditert-butyl-1,4benzoquinone was often detected in house dust in both urban and rural areas in China.35 The two additional substituted aromatic compounds: 1,4-dichlorobenzene and 1,2,4-trimethylbenzene, depicted slightly lower emission fluxes, with average values around 170 and 400 ng/(m2·h), respectively. 1,4-
(6)
Since empirical λBZ values are not available (to the best of our knowledge), we assumed as a first approximation of a “worst case scenario” it is driven only by inhalation rate, such that λBZ = In R/VBZ. In reality, additional parameter, such as thermal plume, will also affect it. The resulting exposure dose at any time t is given by eq 7. Again, for simplicity, [Ci]t=0 in bulk-room and in BZ were assumed to be zero. t2
Dt = In R +
∫t1
1 ijj E BZ jj λBZ jk VBZ
2 l o exp(− λBZt ) jij E BZ Er zyz o Er t z jj [C BZ]t dt = In R m + 2 o j V − V zz o 2Vr λ BZ r{ BZ k n o t2 E zyji 1 zyz| o zz} − r zzzzjjjjt − o t1 Vr {k λBZ z{o ~
(7)
It should be noted that in all the above calculations concentration of emitted pollutant are not assumed to reach steady state.
3. RESULTS AND DISCUSSION 3.1. VOCs Emission Fluxes. Emission fluxes under SME conditions were calculated for each mattress based on four different measurements for each. The presented results (Table 1) concentrate on 18 different VOCs that were identified at very high certainty with NIST mass spectral library, and later verified and quantified using external analytical standards. Surfacespecific emission rates for most of these compounds were of the same order of magnitude in all tested mattresses, despite the large confidence intervals due to heterogeneity of PUF mattresses and the limited number of repetitions. Studies that quantified VOCs emission rates from mattresses are quite limited8 and yet most VOCs detected in this study were also reported in previous studies on polyurethane mattresses.11,16,17 The highest emission fluxes in all tested mattresses were obtained for 2-ethylhexanoic acid, reaching in some cases above 190 μg/(m2·h). The observed flux range of 2-ethylhexanoic acid is well inside the range reported by Boor et al.11 for crib mattresses. 2-Ethylhexanoic acid is a common catalyst in polyurethane production27 and hence its presence is not surprising. According to currently available data 2-ethylhexanoic acid does not present an acute toxicity hazard, and has shown developmental effect in animal studies only at high doses (>100 mg/(kg(BW)·day)).28 Two additional carbonyls, formaldehyde and acetaldehyde, were also detected frequently with fluxes on the order of few μg/(m2·h). These aldehydes may originate from the PUF production process29 and/or from degradation of other organic precursors. In contrast to the above-mentioned carbonyls that were detected in emissions from almost all mattresses, high hexanal emissions were observed only from the infant crib mattress (LDI). It is possible that hexanal was used as an additive to the phosphate flame retardants,17 which were also detected in this mattress. Tris(1-chloro-2-propyl) phosphate (TCPP), is a flame retardant added to the highly flammable30 polyurethane mattresses to answer local regulations. In the present study, TCPP was detected only in emissions from the infant mattress (LDI), which by Israeli law is required to contain flame retardants. Chemical analysis of this LDI mattress indeed reviled the presence of TCPP and V6 flame retardants (Stapleton, unpublished data). Under SME conditions the flux E
DOI: 10.1021/acs.est.9b01557 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 2. Average changes in emission fluxes between: (a) full SME (EFSME; 36 ± 1 °C, 90 ± 5% RH, [CO2] ≈ 1000 ppm) vs control environment (EFcontrol; 20 ± 1 °C,
