Model Development and Calibration - ACS Publications - American ...

20 downloads 0 Views 1MB Size Report
Aug 1, 2016 - Based on SPME: Model Development and Calibration. Jianping Cao,. †,‡. Jianyin Xiong,*,§. Lixin Wang,. ∥. Ying Xu,. ⊥ and Yinping Zhang.
Article pubs.acs.org/est

Transient Method for Determining Indoor Chemical Concentrations Based on SPME: Model Development and Calibration Jianping Cao,†,‡ Jianyin Xiong,*,§ Lixin Wang,∥ Ying Xu,⊥ and Yinping Zhang†,‡ †

Department of Building Science, Tsinghua University, Beijing 100084, China Beijing Key Laboratory of Indoor Air Quality Evaluation and Control, Beijing 100084, China § School of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, China ∥ School of Environment and Energy Engineering, Beijing University of Civil Engineering and Architecture, Beijing 100044, China ⊥ Department of Civil, Architectural and Environmental Engineering, The University of Texas at Austin, Austin, Texas 78712-1094, United States ‡

S Supporting Information *

ABSTRACT: Solid-phase microextraction (SPME) is regarded as a nonexhaustive sampling technique with a smaller extraction volume and a shorter extraction time than traditional sampling techniques and is hence widely used. The SPME sampling process is affected by the convection or diffusion effect along the coating surface, but this factor has seldom been studied. This paper derives an analytical model to characterize SPME sampling for semivolatile organic compounds (SVOCs) as well as for volatile organic compounds (VOCs) by considering the surface mass transfer process. Using this model, the chemical concentrations in a sample matrix can be conveniently calculated. In addition, the model can be used to determine the characteristic parameters (partition coefficient and diffusion coefficient) for typical SPME chemical samplings (SPME calibration). Experiments using SPME samplings of two typical SVOCs, dibutyl phthalate (DBP) in sealed chamber and di(2-ethylhexyl) phthalate (DEHP) in ventilated chamber, were performed to measure the two characteristic parameters. The experimental results demonstrated the effectiveness of the model and calibration method. Experimental data from the literature (VOCs sampled by SPME) were used to further validate the model. This study should prove useful for relatively rapid quantification of concentrations of different chemicals in various circumstances with SPME.



INTRODUCTION Emissions of volatile organic compounds (VOCs) and semivolatile organic compounds (SVOCs) from building materials and consumer products contribute to poor indoor air quality, adversely affecting people’s comfort, health, and productivity.1−3 Exposure to certain VOCs and SVOCs can result in serious health problems (e.g., external malformations, reproductive disorders), sick building syndrome (SBS), and elevated risks of asthma, allergies, and cancer.4−7 For these reasons, a method to rapidly quantify these chemical pollutants in the indoor environment is required so that exposure levels and associated health risks can be evaluated. Knowledge of techniques to monitor levels of chemical pollutants in various environments is currently an important area of interest.8 The development of highly specific and sensitive instruments, e.g., gas chromatography/mass spectrometry (GC/MS) and high-performance liquid chromatography (HPLC), has greatly improved the reliability and accuracy of chemical quantifications. These days the main obstacles faced by analytical chemists include the operational complexity of sample preparation and the inconvenience of introducing extracted components to analytical instruments.9 Typical techniques of sample preparation include the Tenax-TA © 2016 American Chemical Society

sorbent technique, the polyurethane foam (PUF) sorption technique, the 2,4-dinitrophenylhydrazine (DNPH) technique, and the solid-phase microextraction (SPME) technique.10−14 Only the SPME method is defined as a nonexhaustive sampling technique where a very small sample volume is used in the extraction phase relative to the sample volume. This feature allows for convenient monitoring of the investigated system since sampling causes minimal perturbation to the sample.15 The extraction time of the SPME technique is much less than that of Tenax-TA sorbent, PUF, or DNPH techniques due to the small extraction volume, and this merit becomes significant for sampling SVOCs. In addition, SPME is quite applicable for sealed chamber tests (it can simplify the experimental system and improve the rapidity of measuring the characteristic parameter of SVOC emissions16−18) because of the features of SPME mentioned above.16 The SPME sampling technique has therefore gained in popularity since its invention.12,19 Received: Revised: Accepted: Published: 9452

March 16, 2016 July 26, 2016 August 1, 2016 August 1, 2016 DOI: 10.1021/acs.est.6b01328 Environ. Sci. Technol. 2016, 50, 9452−9459

Article

Environmental Science & Technology

many situations the time of reaching equilibrium is very long, and in some situations it may not be feasible to reach extraction equilibrium. Those have led to the development of the kinetic calibration method, which considers the diffusion-controlled process inside the coating, and a kinetic model relating the amount of chemical extracted to the extraction time is derived, with two model parameters (the coating/sample partition coefficient, K; and the diffusion coefficient of the chemical inside the coating, Dm) that need to be measured.22,24,25 It should be pointed out that the existing kinetic calibration method seldom considers the convection effect on the SPME coating surface when there is relative movement between SPME and the sample matrix. In some cases, a convective boundary layer thickness is introduced, but a linear concentration distribution inside the layer is assumed.24 This assumption results in significant deviations since the real concentration distributions are ignored. Prior studies have shown that, for chemical emissions from building materials and consumer products, the convective mass transfer process played a significant role on the emission characteristics, especially for SVOCs.26−28 Given that SPME sampling (i.e., sorption process) is the inverse process of chemical emissions, the convective mass transfer process will also significantly affect the SPME sampling. For some scenarios, SPME is used to measure the concentration of target analyte in static mode (both the sample matrix and SPME are static). Under these conditions, the concentration gradient of target analyte between the sample matrix and SPME coating surface also needs to be considered. Therefore, ignoring the effect of mass transfer between the sample matrix and SPME coating surface during sampling may lead to model prediction error and error in the measurement of parameters in the kinetic model. Moreover, in SVOC sampling, the SPME mainly focuses on quantification of liquid samples, and it has seldom been reported to quantify gaseous samples due to the difficulties associated with low gas-phase SVOC concentrations as well as the ubiquitous SVOC contamination in laboratories. Taking the above problems into consideration, the objectives of this paper are to (1) establish an analytical model to characterize the SPME sampling by taking into account the mass transfer process between the sample matrix and SPME coating surface and (2) determine the characteristic parameters (partition and diffusion coefficients) for typical SPME-chemical samplings, particularly for gas-phase SVOCs.

The structure of an SPME system is optimized to facilitate speed, convenience of use, and sensitivity. The system comprises a thin stainless steel (SS) needle and a sampling fiber attached to the SS.20 The sampling fiber consists of a cylindrical fused silica fiber with a coating surrounding it. The schematic of the SPME sampling fiber for chemical (or analyte) sampling is shown in Figure 1a.20 During the sampling process,

Figure 1. Schematic of SPME for chemical sampling: (a) SPME sampling in liquid or gas phase; (b) flow across a flat plate (simplified SPME sampling).



ANALYTICAL MODEL AND CALIBRATION METHOD If in a SPME the thickness of the coating (e.g., 7 μm, designated as L) is much smaller than the radius of the fused silica (e.g., 55 μm, designated as R), the cylindrical coating can be unfolded into a flat plate, as shown in Figure 1b.29 Detailed discussion about the error introduced by unfolding the cylindrical coating into a flat plate is presented in section S1 of the Supporting Information (SI), showing that this error is less than 5% when L is less than 7 μm (for R = 55 μm). It is assumed that the coating is uniform, concentration of target analyte in the sample matrix is constant, and that the chemical diffusion process inside the coating is one-dimensional. According to the mass transfer mechanism, the controlling equation can be written as

the sampling fiber is exposed to the sample matrix and the chemical (or analyte) is sorbed (or extracted) by the coating surrounding the fiber. The fused silica is generally assumed to be impenetrable to chemicals.21 The SPME needs to be calibrated to facilitate its application. The calibration method involves two procedures: the first is to establish a model for characterizing the SPME samplings; the second is to determine the parameters involved in the model. Equilibrium and kinetic calibration methods are the two most commonly used methods for SPME calibration.22 In the equilibrium calibration method, measurements are made after partition equilibrium has been reached between the coating and the target chemical. This method establishes a linear relationship between the amount of chemical in the SPME coating and the constant chemical concentration in the sample matrix, with an equilibrium partition coefficient that needs to be determined.22,23 In

∂Cm ∂ 2C = Dm 2m ∂t ∂x 9453

(1) DOI: 10.1021/acs.est.6b01328 Environ. Sci. Technol. 2016, 50, 9452−9459

Article

Environmental Science & Technology

Comparing this equation with eq 3, it is easy to find that hm can be expressed as (here, hm is treated as an “equivalent” mass transfer coefficient)

where Cm is the concentration of chemicals in the coating, μg/ m3, Dm is the diffusion coefficient of the chemicals in the coating, m2/s, t is the time, s, and x is the distance to the coating/silica interface, m. Since the fused silica is generally assumed to be impenetrable to the chemicals,21 there is no mass flux at the coating/silica interface. The boundary condition at the fused silica surface can then be written as ∂Cm = 0, ∂x

x=0

hm = Da S /A

where S is the shape factor (commonly used in the field of twodimensional mass or heat transfer) for a finite cylinder in an infinite medium with uniform concentration, m.29 In this way, eq 5 can be rewritten in the same form of eq 3. In the case of cylindrical SPME coating in a uniform SVOC concentration, S can be estimated by29,34

(2)

⎡ 1+ S = 4πL 1 − γ 2 /ln⎢ ⎢⎣ 1 −

At the coating surface exposed to the sample matrix (gas or liquid phase) of chemicals, there is a concentration gradient between the coating surface and the sample matrix. It should be noted that this concentration gradient (it will cause convective or diffusive mass transfer effect) is seldom considered in the traditional models, but it does in fact significantly affect the coating extraction characteristics. In the SPME sampling process, the amount extracted by the coating is generally very small and will thus not influence the concentration of chemicals in the sample matrix.22,30 If the measured media moves around SPME or SPME shakes during sampling (e.g., SPME sampling in ventilated chamber), the boundary condition at the coating surface can be expressed as ⎛ C ⎞ ∂C Dm m = hm⎜C in − m ⎟ , ⎝ K ⎠ ∂x

x=L

Initially there are no chemicals in the coating, i.e. Cm(x , t ) = 0,

⎛ C ⎞ ∂Cm = Da S⎜C in − m ⎟ ⎝ K ⎠ ∂x

t = 0,

0≤x≤L

(8)

The amount of chemicals extracted onto the coating can be expressed as M (t ) =

∫0

t

DmA

∂Cm ∂x

dt (9)

x=L

where M(t) is the amount of chemicals extracted onto the coating in extraction time t, μg. For eqs 1−9, the amount of chemicals extracted onto the coating can be derived using a Laplace transform

(3)



sin qn

M(t ) = Mequ − 2Mequ ∑

q 2 /sin n=1 n

−2 2 qn t

qn + qn cos qn

e−DmL

(10)

where Mequ (= KCinVm) is the equilibrium amount of chemicals extracted onto the coating, μg, Vm is the volume of the coating, m3, Vm = AL, and qn are the positive roots of the following equations:

(4)

qn tan qn =

where Sh (= hmd/Da) is the Sherwood number, Re (= ud/v) is the Reynolds number, Sc (= v/Da) is the Schmidt number, u is the velocity over the coating surface, m/s, v is the kinematic viscosity of the sample matrix, m2/s, d is the diameter of the coating, m (d = 2R + 2L), Da is the diffusion coefficient of chemicals in the sample matrix, m2/s, and C and n are parameters related to the Reynolds number, which are given the values of 0.989 and 0.33, respectively. These values are appropriate for cross-flow over a cylinder with Re in the range of 0.4−4 and Sc over 0.7.29,31 Da can be estimated using empirical correlations32 or obtained from measurement results in the literature.33 As mentioned previously, SPME is sometimes used to measure the concentration of target analyte in static mode (e.g., SPME sampling in sealed chamber). In these scenarios, eq 4 is no longer applicable since Re approaches zero, and the mass transfer of target analyte from the measured media to SPME is controlled by molecular diffusion. The boundary condition at x = L, i.e., eq 3, should be represented as29,34 DmA

⎤ 1 − γ2 ⎥ with γ = d /H 1 − γ 2 ⎥⎦ (7)

where Cin is the concentration of chemicals in the sample matrix (liquid or gas phase), μg/m3, K is the coating/sample partition coefficient (or distribution coefficient), dimensionless, L is the thickness of the coating, m, and hm is the convective mass transfer coefficient across the coating surface, m/s, which can be calculated by the following empirical correlation (for cross-flow over a cylinder)29,31

Sh = C RenSc1/3

(6)

hm L DmK

(n = 1, 2, ...)

(11)

The terms in the infinite exponential series of eq 10 decay very fast as time increases. Thus, if the sampling time is sufficiently long, only the first term (n = 1) is significant. This means ⎛ ⎞ sin q1 −2 2 M(t ) = Mequ⎜⎜1 − 2 2 e−DmL q1 t ⎟⎟ q1 /sin q1 + q1 cos q1 ⎝ ⎠ = Mequ(1 − αe−βt )

(12)

DmL−2q21.

where α = 2 sin q1 + q1 cos q1); β = Detailed calculation (see SI, section S2) indicates that when t ≥ 0.12L2/Dm the relative deviation is less than 5% when eq 12 is applied as a substitute for eq 10. Thus, the condition for simplifying eq 7 into eq 12 is t ≥ 0.12L2/Dm (0.12L2/Dm is referred to as tmin). The analytical model, eq 10 or 12, establishes the relationship between the quantity of chemicals extracted onto the coating, and the extraction time when using SPME for chemical sampling. When determining chemical concentrations in the sample matrix (i.e, Cin) based on SPME, we can measure the quantity of chemicals adsorbed by SPME fiber coatings (M), Cin can then be determined using eq 12, if the q1/(q21/sin

(5)

where A is the exposed area of the coating, m2, A = 2π(R + L) H, and H is the length of SPME coating. 9454

DOI: 10.1021/acs.est.6b01328 Environ. Sci. Technol. 2016, 50, 9452−9459

Article

Environmental Science & Technology characteristic parameters (Dm and K) are known. Therefore, in routine application, we need to first determine Dm and K for the target chemical (calibration of SPME). Eq 12 can also be applied to address the calibration problem (determine Dm and K) if the time-dependent SPME adsorbed chemical amounts are measured. The calibration method provides a basis for quantifying the concentration of chemicals in various circumstances with SPME. Therefore, this paper mainly focuses on the determination of the characteristic parameters (Dm and K) for different SPME-chemical combinations, which is closely related to the adsorption/desorption properties of chemicals on the SPME fiber. Once the extraction amount (M) of a target chemical is measured at a series of intervals (>tmin), Mequ, α, and β can be obtained by fitting eq 12 to these measured points. Then K and Dm can be determined by following these steps: (1) Solve the following equation for q1 (positive root): sin 2 q1 q12 + q1 sin q1 cos q1

=

α 2

The schematic of the experimental system for SPME calibration is shown in Figure 2. Figure 2a is the experimental

(13)

(2) Calculate Dm: Dm = βL2/q21 (3) Calculate hm using eq 4 or 6, and then determine K according to eq 11, or K=

hm L Dmq1 tan q1

(14)

(4) Calculate tmin (= 0.12L2/Dm) with the obtained Dm. If the shortest sampling time is less than tmin, eliminate the sampling data before tmin and then repeat steps 1−4 with the remaining data; otherwise output Dm and K. Although the error introduced by unfolding the cylindrical coating into a flat plate can be large when L is larger than 7 μm (as shown in Figure S2), the analysis in section S4 demonstrates that the model (i.e., eq 12) is also effective when the coating thickness is comparable to the radius of the fused silica (55 μm for commonly used SPME35).

Figure 2. Experimental system for SPME calibration: (a) for DBP, (b) for DEHP.



EXPERIMENTS Dibutyl phthalate (DBP) and di(2-ethylhexyl) phthalate (DEHP) are widely used as plasticizers and are the main SVOCs emitted from polyvinyl chloride (PVC) products.36−38 In addition, the gas-phase concentrations of DBP and DEHP have seldom been sampled by SPME. For these reasons, we chose DBP and DEHP as the target pollutants for the SPME calibration experiments. The experiments for DBP and DEHP were conducted in sealed and ventilated chambers, respectively. It should be noted that a sealed chamber is not very common for SVOC emission tests in prior studies. The reason is that the traditional sampling methods (e.g., Tenax-TA tube and PUF sampling methods) are not suitable for a sealed chamber since they need to extract significant amounts of pollutants from the chamber (especially for multiple samplings) but the quantity of pollutants in sealed chamber is limited. However, the SPME is a nonexhaustive sampling technique with a smaller extraction volume and a shorter extraction time, and thus is quite applicable for sealed chamber as mentioned in the Introduction. Using two kinds of chambers aims to examining the adaptation of SPME calibration in different test conditions, thereby extending the application range of SPME in routine analysis.

system designed for DBP. Pure DBP (purchased from SigmaAldrich Co. LLC, purity 99%, product ID. 524980-500 mL) in a petri dish (without cover) was put inside a 30 L chamber designed for VOC emission tests,39 where the air temperature was controlled using a water bath (25 ± 0.5 °C). During the experiment, the 30 L chamber was sealed. Once the concentration of DBP in the chamber reached equilibrium (about 2 days according to our observations), the gas-phase DBP was sampled by SPME. Several SPMEs were inserted into the chamber and adsorbed the DBP for different lengths of time, so as to get the time-dependent SPME extraction curve for calibration. After sampling, the surfaces of the stainless steel rod of SPME, which would also adsorb DBP (and other chemicals), were wiped using medical cotton wool soaked with dichloromethane (CH2Cl2). The SPME was finally analyzed using gas chromatography-flame ionization detection (GC-FID, Agilent Technologies 7890A GC system equipped with a flame ionization detector) as follows: inserted SPME (the whole sampling fiber and part of stainless steel rod) into the injection port (280 °C) of the GC; allowed the coating of sampling fiber to desorb for 15 min; cooled to room temperature. 9455

DOI: 10.1021/acs.est.6b01328 Environ. Sci. Technol. 2016, 50, 9452−9459

Article

Environmental Science & Technology

extraction amount for different model parameters. Figure S5a of the SI shows the results of sensitivity analysis of hm under a typical condition (the baseline parameters are set as K = 1 × 108, Dm = 1 × 10−11 m2/h, hm = 0.01 m/s, and Cin = 1 μg/m3). This figure reveals that hm has a substantial impact on the extraction rate of SPME, and the smaller the value of hm, the greater the impact. For SPME in these experiments, hm is calculated to be 3.36 × 10−2 m/s for the DEHP experimental system (ventilated chamber) by virtue of eq 4 (with Re = 0.414, Da = 3.37 mm2/s,33 and Sc = 4.72). While for the DBP system (sealed chamber), since there is no bulk air movement around the SPME during the whole experiment, hm (the “equivalent” mass transfer coefficient) should be calculated with eqs 6 and 7. In this way, hm is calculated to be 1.33 × 10−2 m/s (with Da = 4.21 mm2/s33) for the DBP system. These values (3.36 × 10−2 m/s and 1.33 × 10−2 m/s) lie within the simulated range of hm shown in Figure S5a, implying that it is very necessary to take hm into account in the models for characterizing SPME samplings. Therefore, the analytical model derived in this study can be regarded as a significant improvement on previous model studies that generally ignore the impact of hm. Results of sensitivity analysis of Dm and K are shown in SI, parts b and c, respectively, of Figure S5. It indicates that K has a significant impact on the extraction amount of SPME. With the increase of K, the equilibrium extraction amount (Mequ) increases and the time required to reach equilibrium increases as well. The influence of Dm on the extraction process is not as significant as that of K. Dm mainly takes effects in the initial extraction period (the extraction rate increases with the increase of Dm) and does not influence the equilibrium extraction amount of SPME. Determination of the Characteristic Parameters. The SPME experiment for DBP lasted for 26 h, while for DEHP experiment took 141 h. Using the measured time-dependent extraction amount, the simplified analytical model (eq 12) was used to perform nonlinear curve fitting. The characteristic parameters (Dm and K) were obtained using the procedure described at the end of the section Analytical Model and Calibration Method. OriginPro 8 (OriginLab Corporation) was employed for curve fitting. The fitted curves for the two SVOCs as well as the R2 for regression are shown in Figure 3, and the determined parameters are listed in Table 1. According to the ASTM Standard D5157-97,41 a correlation coefficient (R) of 0.9 or greater can be regarded as generally indicative of adequate model performance. In this case, all R2 are greater than 0.97, implying high regression precision. The two characteristic parameters (Dm and K) determined by the simplified model (eq 12) can be substituted into the complete analytical model (eq 10) to calculate the extraction amount of the SPME. We can then compare the simulated results with the experimental data, as shown in Figure 3. The comparison can be regarded as a preliminary validation of the calibration method. Figure 3 indicates that the simulated results agree well with the measured data. In addition, the simulated results from the complete analytical model (eq 10) and from the simplified model (eq 12) are almost the same when the time is longer than the application condition (tmin) of eq 12. The accord between the results from the two models and the experimental data demonstrates that the measured characteristic parameters are reliable. From the simulated results shown in Figure 3, we can see that the extraction amount approaches equilibrium after 26 h for DBP, while for DEHP the equilibrium is still not reached. These results imply that the calibration process can be terminated before the sorption of

Since DEHP can be very easily adsorbed by the chamber surface (e.g., chamber surface/air partition coefficient (Ks) = 1500 m for DEHP while Ks = 60 m for DBP38), and the sorption area of the 30 L chamber surface is much larger than the emission area of the pure SVOC, the time required for the DEHP concentration in the 30 L chamber to reach equilibrium would be relatively long.17,40 For example, the DEHP concentration reached equilibrium after more than 150 days in a CLIMPAQ (with a volume of 51 L).17,40 To obviate this problem, we designed a new experimental system for DEHP, as shown in Figure 2b. The small chamber is made of stainless steel, with a volume of 1.8 L and interior surface area of 0.13 m2. Pure liquid DEHP (purchased from Sigma-Aldrich Co. LLC, purity ≥99%, product ID. D201154-500 mL) was put inside the small chamber. The temperature of the chamber was controlled by a water bath, fixed at 25.0 ± 0.5 °C. During the experiment, clean air with controlled relative humidity (50 ± 5%) was introduced into the chamber. The air flow was 90 mL/ min, and the diameters of the inlet and outlet tubes were both 6.0 mm. To reduce the effect of DEHP sorption on the chamber and tube, the surfaces of the chamber and tube were wiped with pure DEHP at the beginning of the experiment. After 2 days (the time required for DEHP concentration to reach equilibrium according to our observations), gas-phase DEHP was sampled by SPME at the chamber outlet. The procedure for analyzing the SPME sample is the same as described above for DBP. A six-point calibration curve for DBP or DEHP was obtained by following these steps. Pure DBP or DEHP was diluted to 1, 10, 60, 200, 500, and 1000 μg/mL (the solvent is CH2Cl2), and 1.0 μL of each dilution was injected into a GC. All samples (SPME samples and dilutions) were analyzed by GC-FID. The chromatographic column was an HP-5MS SemiVol (30 m × 0.25 mm × 0.50 μm). The carrier gas was He, with a flow rate of 50 mL/min, split, 10:1. The column temperature program was: 120 °C for 2 min; increased to 300 °C at a rate of 15 °C/ min; and held for 10 min (24 min in total). The temperature of the injection port and FID were both 280 °C. The calibration curve is demonstrated to be valid since R2 of the curve is greater than 0.99. SPMEs were purchased from Sigma-Aldrich Co. LLC. (Supelco Analytical, Cat. No. 57302). The coating was made of polydimethylsiloxane (PDMS, feasible for sampling of nonpolar SVOCs) with a thickness of 7.0 μm and a length of 1.0 cm. The diameter of the fused silica fiber was 110.0 μm. Before the experiments, each SPME was conditioned by heating it in a GC injection port at 280 °C for 5 min. The carrier gas was He, with a flow rate of 10 mL/min. After conditioning, the remained mass of DBP and DEHP in the coating should be below the limit of quantitation (LOQ, i.e., 1 ng) of the relevant GC-FID method.



RESULTS AND DISCUSSION Sensitivity Analysis. For SPME samplings, it is necessary to perform sensitivity analysis with the derived analytical model, so as to know the impact of some model parameters (hm, K, and Dm) especially hm on the extraction of chemicals from the sample into the coating. The SPME introduced in the Experiments section is used for this analysis. For common material−SVOC combinations, K is in the range of 105− 1011.13,28 Dm is estimated to be in the range of 10−14−10−10 m2/ h according to equation S3 in Cao et al.’s study.18 Based on these data and the analytical model, we can obtain the SPME 9456

DOI: 10.1021/acs.est.6b01328 Environ. Sci. Technol. 2016, 50, 9452−9459

Article

Environmental Science & Technology

The determined gas phase concentrations for DBP and DEHP are 488 and 5.58 μg/m3, respectively. In the Experiments section, pure SVOC liquids are used. Thus, for the sealed chamber, the Cin determined from the experiment should be the same as the saturated gas phase concentration (designated as ysat) of pure chemicals, i.e., ysat = Cin, while for the ventilated chamber ysat is a function of Cin, i.e., ysat = Cin(1 + Q/(hm,eAe)) (eq 5 of Liang and Xu38), where Q is the air flow rate of the chamber; hm,e is the convective mass transfer coefficient at the source surface; Ae is the emission area of the SVOC source. In this study, for DBP, ysat is equal to Cin because sealed chamber is used, i.e., ysat = 488 μg/m3; while for DEHP, ysat = 1.09Cin (Q/(hm,eAe) = 0.09 with Q = 90 mL/min, hm,e = 0.26 m/h,42 Ae = 0.13 m2), i.e., ysat = 6.08 μg/m3, as listed in Table 1. Recently, Liang and Xu38 designed a special chamber to measure phthalate emissions from building materials, in which the gas phase concentrations were measured by Tenax-TA tube sampling followed by thermal desorption (TD)-GC/MS. In their study, ysat of DBP and DEHP were measured by coating the interior chamber surfaces with pure DBP and DEHP liquids. According to Liang and Xu’s results,38 ysat are 464 μg/ m3 and 5.64 μg/m3 for DBP and DEHP, respectively. The ysat values in the literature are very similar to the measured values in this study, with relative deviations (RD) of no more than 7.8%, as indicated in Table 1. Although there is no direct comparison with traditional analytics (e.g., Tenax-TA tube sampling followed by TD-GC/MS) in our experiments, the consistency between our results of ysat and that in literature for the same compounds also provides convincing evidence that the application of SPME for SVOC samplings is appropriate and effective. Impact of Test Time on the Determined Parameters. As mentioned above, a major benefit of this method is that the process can be terminated before the sorption process of the SPME coating reaches equilibrium. It is therefore necessary to investigate the impact of test time on the determined characteristic parameters (Dm and K), so as to determine whether the test time can be further reduced. To this end, the values of Dm and K are obtained again by fitting eq 12 to the experimental data while excluding the longest test time (i.e., 26 h for DBP and 141 h for DEHP). For DBP, when the test time is shortened from 26 to 11.7 h, the obtained Dm decreases from 2.81 × 10−15 to 1.37 × 10−15 m2/s and K decreases from 3.20 × 107 to 2.93 × 107. For DEHP, when the test time is shortened from 141 to 97.5 h, the obtained Dm value decreases from 2.13 × 10−16 to 1.31 × 10−16 m2/s, while K increases from 6.49× 108 to 6.83 × 108. These results show that the deviation of K is very small (relative deviation is less than 10%) as a result of shortening the test time, while that of Dm is fairly large (the relative deviations are 38% and 51% for DEHP and DBP, respectively). By substituting the values of Dm and K obtained using the reduced sampling data into eq 9, the value of M at 26 h for DBP and 141 h for DEHP can be estimated. The relative deviations between the estimated M and measured M are quite small, i.e., 7.7% for DBP and 5.8% for DEHP. Such small deviations indicate that the results for the characteristic parameters obtained by shortening the test time are acceptable, despite the relative deviation of Dm being as large as 51%. This analysis indicates that the SPME extraction amount is not very sensitive to Dm especially for SVOC samples (consistent with the result of sensitivity analysis of Dm), implying that the extraction process is externally controlled for SVOCs. This result is consistent with a prior study28 where the emission

Figure 3. Fitted curves and comparison of SPME extraction amount for the simulated results based on the determined characteristic parameters and the experimental data: (a) DBP; (b) DEHP.

Table 1. Determined Characteristic Parameters by Virtue of the Model and Comparison of the Saturated Gas-Phase Concentrations of Pure Chemicals Measured in This Study with That Measured in the Literature chemicals

Dm (m2/s)

K (−)

ysat_measured (μg/m3)a

ysat_literature (μg/m3)b

RD (%)c

DBP DEHP

2.81 × 10−15 2.13 × 10−16

3.20 × 107 6.49 × 108

488 6.08

464 5.64

5.2 7.8

a

Measured in this study. bMeasured results of Liang and Xu.38 cRD is calculated by |ysat,measured − ysat,literature|/ysat,literature × 100%.

SVOC on to the SPME coating reaches equilibrium. This is a salient feature of the present method because it can save time, particularly for the calibration of chemicals with large vaules of K (e.g., DEHP). Measured Saturated Gas-Phase SVOC Concentration and Comparison with Literature. In the calibration method, the equilibrium extraction amount of chemicals onto the SPME coating (Mequ) can be obtained by virtue of nonlinear curve fitting, following which the steady-state concentration of SVOCs in the gas phase (Cin = Mequ/KVm) can be determined. 9457

DOI: 10.1021/acs.est.6b01328 Environ. Sci. Technol. 2016, 50, 9452−9459

Article

Environmental Science & Technology process of DEHP from vinyl flooring was found to be controlled by the external convection process. In this study, we found that no meaningful results could be obtained through nonlinear curve fitting by further shortening the test time. The primary reason for this is that there are not enough samplings. The relationship between the shortest test time, the characteristic parameters, the number of samplings and the time interval between contiguous measurements needs further investigation. In addition, systematic studies focusing on the impact of shortening test time on the accuracy of the determined parameters are also required. The previous sections demonstrate the effectiveness of the calibration method for DBP and DEHP (SVOCs) tests. There are many experimental data reported in the literature for other SPME-chemical combinations. To further validate the applicability of the calibration method for VOCs, we analyze data from two references43,44 as examples. Detailed analysis and results are described in SI, section S4 (using the calibration method to analyze data from the literature). The fitted curves of eq 12 together with the model predictions all agree well with the experimental data, implying that the proposed calibration method can be regarded as a general method both for SVOCs and VOCs (for the compounds studied). Limitations and Further Study. For the cases from the literature, the good results (SI, section S4) illustrate the feasibility of unfolding the cylindrical coating into a flat surface, even when the coating thickness and the silica radius are comparable. Such treatment reduces the complexity in the model development while still maintaining high precision (see details in SI, section S3; the solution of the model of an unfolded cylindrical coating is quite complicated). Nonetheless, the results of Dm should be interpreted with caution when the coating thickness is not much smaller than the radius of the fused silica, since under this condition the Dm obtained with the present method is an equivalent value (i.e., the diffusion pathway is in fact along the cylindrical wall rather than along the flat wall). The deviation between the determined Dm and the actual Dm may be dependent on R (radius of the fused silica), L (thickness of the coating), and K. Determination of the actual Dm of chemicals in the fiber coating requires a complete model, i.e., a model expressed in cylindrical coordinates (the structure of SPME is cylindrical) rather than in rectangular coordinates. The complete model is described in SI, section S1, and the analytical solution of this model is provided in SI, section S3 (eqs S9−S18). Applying eq S18, the actual Dm of chemicals in the cylindrical coating can be obtained with a similar procedure provided at the end of the section Analytical Model and Calibration Method. However, the solving process may require multifarious mathematical methods due to the complexity of the analytical solution, which is out of the scope of the present study. In some scenarios, we need to optimize the structure of the SPME for target chemical sampling to minimize the measurement error (e.g., optimize the thickness of SPME coating and select the optimum coating material). Under this condition, the actual Dm is requisite. Therefore, further study is necessary to solve the mathematical challenges for determining the actual Dm with the complete model. In addition, the present method requires to estimate the mass transfer coefficient (hm) with an empirical correlation, which may introduce some uncertainties to the determined characteristic parameters.45,46 Moreover, as discussed in Sensitivity Analysis, uncertainty in hm may also influence the precision of SPME in real field samplings. Development of

novel methods that can eliminate the effect of hm on the determination of Dm and K warrants further investigation. It should be noted that whether the analytical model is applicable for the calibration of other extraction scenarios when using SPME for sampling is still unknown; thus, more experimental validation is needed.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.6b01328. Additonal discussions about conditions and procedures; Table S1; Figures S1−S5 (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel: +86 10 68914304. Fax: +86 10 68412865. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This research was supported by the National Natural Science Foundation of China (Grant Nos. 51476013 and 51136002). We thank Dr. Cong Liu of Tsinghua University for helpful discussions.



REFERENCES

(1) Kim, Y. M.; Harrad, S.; Harrison, R. M. Concentrations and sources of VOCs in urban domestic and public microenvironments. Environ. Sci. Technol. 2001, 35 (6), 997−1004. (2) Godish, T. Indoor Environmental Quality; CRC Press: Boca Raton, 2000. (3) Weschler, C. J. Changes in indoor pollutants since the 1950s. Atmos. Environ. 2009, 43 (1), 153−169. (4) Salthammer, T.; Mentese, S.; Marutzky, R. Formaldehyde in the indoor environment. Chem. Rev. 2010, 110 (4), 2536−2572. (5) Heudorf, U.; Mersch-Sundermann, V.; Angerer, J. Phthalates: toxicology and exposure. Int. J. Hyg. Environ. Health 2007, 210 (5), 623−634. (6) Hauser, R.; Meeker, J. D.; Duty, S.; Silva, M. J.; Calafat, A. M. Altered semen quality in relation to urinary concentrations of phthalate monoester and oxidative metabolites. Epidemiology 2006, 17 (6), 682−691. (7) Bornehag, C.-G.; Sundell, J.; Weschler, C. J.; Sigsgaard, T.; Lundgren, B.; Hasselgren, M.; Hägerhed-Engman, L. The association between asthma and allergic symptoms in children and phthalates in house dust: a nested case-control study. Environ. Health Persp. 2004, 112 (14), 1393−1397. (8) Walt, D. R. Miniature analytical methods for medical diagnostics. Science 2005, 308 (5719), 217−219. (9) Pawliszyn, J. Handbook of Solid Phase Microextraction; Elsevier: Amsterdam, 2011. (10) Sporstol, S.; Gjos, N.; Lichtenthaler, R. G.; Gustavsen, K. O.; Urdal, K.; Oreld, F.; Skei, J. Source identification of aromatic hydrocarbons in sediments using GC/MS. Environ. Sci. Technol. 1983, 17 (5), 282−286. (11) de Andrade, J. B.; Tanner, R. L. Determination of formaldehyde by HPLC as the DNPH derivative following high-volume air sampling onto bisulfite-coated cellulose filters. Atmos. Environ., Part A 1992, 26 (5), 819−825. (12) Arthur, C. L.; Pawliszyn, J. Solid phase microextraction with thermal desorption using fused silica optical fibers. Anal. Chem. 1990, 62 (19), 2145−2148. 9458

DOI: 10.1021/acs.est.6b01328 Environ. Sci. Technol. 2016, 50, 9452−9459

Article

Environmental Science & Technology (13) Liu, X. Y.; Guo, Z. S.; Roache, N. F. Experimental method development for estimating solid-phase diffusion coefficients and material/air partition coefficients of SVOCs. Atmos. Environ. 2014, 89, 76−84. (14) Bartkow, M. E.; Booij, K.; Kennedy, K. E.; Muller, J. F.; Hawker, D. W. Passive air sampling theory for semivolatile organic compounds. Chemosphere 2005, 60 (2), 170−176. (15) Heringa, M.; Hermens, J. Measurement of free concentrations using negligible depletion-solid phase microextraction (nd-SPME). TrAC, Trends Anal. Chem. 2003, 22 (9), 575−587. (16) Cao, J. P.; Zhang, X.; Little, J. C.; Zhang, Y. P. A SPME-based method for rapidly and accurately measuring the characteristic parameter for DEHP emitted from PVC floorings. Indoor Air 2016, DOI: 10.1111/ina.12312. (17) Liu, C.; Liu, Z.; Little, J. C.; Zhang, Y. P. Convenient, rapid and accurate measurement of SVOC emission characteristics in experimental chambers. PLoS One 2013, 8 (8), e72445. (18) Cao, J. P.; Weschler, C. J.; Luo, J. J.; Zhang, Y. P. Cm-history method, a novel approach to simultaneously measure source and sink parameters important for estimating indoor exposures to SVOCs. Environ. Sci. Technol. 2016, 50, 825−834. (19) Ouyang, G. F.; Pawliszyn, J. Recent developments in SPME for on-site analysis and monitoring. TrAC, Trends Anal. Chem. 2006, 25 (7), 692−703. (20) Pawliszyn, J.; Pawliszyn, B.; Pawliszyn, M. Solid phase microextraction (SPME). Chem. Educ. 1997, 2 (4), 1−7. (21) Pawliszyn, J. Theory of solid-phase microextraction. J. Chromatogr. Sci. 2000, 38 (7), 270−278. (22) Ouyang, G. F.; Pawliszyn, J. A critical review in calibration methods for solid-phase microextraction. Anal. Chim. Acta 2008, 627 (2), 184−197. (23) Zhang, X.; Es-haghi, A.; Musteata, F. M.; Ouyang, G.; Pawliszyn, J. Quantitative in vivo microsampling for pharmacokinetic studies based on an integrated solid-phase microextraction system. Anal. Chem. 2007, 79 (12), 4507−4513. (24) Ai, J. Solid phase microextraction for quantitative analysis in nonequilibrium situations. Anal. Chem. 1997, 69 (6), 1230−1236. (25) Ouyang, G. G.; Cui, S. F.; Qin, Z. P.; Pawliszyn, J. One-calibrant kinetic calibration for on-site water sampling with solid-phase microextraction. Anal. Chem. 2009, 81 (14), 5629−5636. (26) Xu, Y.; Zhang, Y. P. An improved mass transfer based model for analyzing VOC emissions from building materials. Atmos. Environ. 2003, 37 (18), 2497−2505. (27) Xiong, J. Y.; Liu, C.; Zhang, Y. P. A general analytical model for formaldehyde and VOC emission/sorption in single-layer building materials and its application in determining the characteristic parameters. Atmos. Environ. 2012, 47, 288−294. (28) Xu, Y.; Little, J. C. Predicting emissions of SVOCs from polymeric materials and their interaction with airborne particles. Environ. Sci. Technol. 2006, 40 (2), 456−461. (29) Incropera, F. P. Fundamentals of Heat and Mass Transfer; John Wiley & Sons: New York, 2011. (30) Louch, D.; Motlagh, S.; Pawliszyn, J. Dynamics of organic compound extraction from water using liquid-coated fused silica fibers. Anal. Chem. 1992, 64 (10), 1187−1199. (31) Yang, S. M.; Tao, W. Q. Heat Transfer Theory; Higher Education Press: Beijing, 2006 (In Chinese). (32) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmental Organic Chemistry; John Wiley & Sons: New York, 2005. (33) Lugg, G. Diffusion coefficients of some organic and other vapors in air. Anal. Chem. 1968, 40 (7), 1072−1077. (34) Rohsenow, W. M.; Hartnett, J. P.; Cho, Y. I. Handbook of Teat Transfer; McGraw-Hill: New York, 1998. (35) Shirey, R. E. In Solid Phase Microextraction: A Practical Guide, Wercinski, S. S., Ed.; Marcel Dekker: New York, 1999. (36) Wensing, M.; Uhde, E.; Salthammer, T. Plastics additives in the indoor environment–flame retardants and plasticizers. Sci. Total Environ. 2005, 339 (1−3), 19−40.

(37) Bornehag, C.-G.; Lundgren, B.; Weschler, C. J.; Sigsgaard, T.; Hagerhed-Engman, L.; Sundell, J. Phthalates in Indoor Dust and Their Association with Building Characteristics. Environ. Health Persp. 2005, 113 (10), 1399−1404. (38) Liang, Y. R.; Xu, Y. Improved method for measuring and characterizing phthalate emissions from building materials and its application to exposure assessment. Environ. Sci. Technol. 2014, 48 (8), 4475−4484. (39) Xiong, J. Y.; Yao, Y.; Zhang, Y. P. C-history method: rapid measurement of the initial emittable concentration, diffusion and partition coefficients for formaldehyde and VOCs in building materials. Environ. Sci. Technol. 2011, 45 (8), 3584−3590. (40) Clausen, P. A.; Hansen, V.; Gunnarsen, L.; Afshari, A.; Wolkoff, P. Emission of di-2-ethylhexyl phthalate from PVC flooring into air and uptake in dust: emission and sorption experiments in FLEC and CLIMPAQ. Environ. Sci. Technol. 2004, 38 (9), 2531−2537. (41) ASTM D5157-97. Standard Guide for Statistical Evaluation of Indoor Air Quality Models; American Society for Testing and Materials (ASTM Stardands): West Conshohocken, 2003. (42) Axley, J. W. Adsorption modelling for building contaminant dispersal analysis. Indoor Air 1991, 1 (2), 147−171. (43) Tuduri, L.; Desauziers, V.; Fanlo, J. L. Dynamic versus static sampling for the quantitative analysis of volatile organic compounds in air with polydimethylsiloxane−Carboxen solid-phase microextraction fibers. J. Chroma. A 2002, 963 (1), 49−56. (44) Bartelt, R. J.; Zilkowski, B. W. Nonequilbrium quantitation of volatiles in air streams by solid-phase microextraction. Anal. Chem. 1999, 71 (1), 92−101. (45) Holman, J. P. Heat Transfer, 9th ed.; McGraw-Hill: New York, 2002. (46) Xiong, J. Y.; Cao, J. P.; Zhang, Y. P. Early stage C-history method: Rapid and accurate determination of the key SVOC emission or sorption parameters of indoor materials. Build. Environ. 2016, 95, 314−321.

9459

DOI: 10.1021/acs.est.6b01328 Environ. Sci. Technol. 2016, 50, 9452−9459