ARTICLE pubs.acs.org/est
C-History Method: Rapid Measurement of the Initial Emittable Concentration, Diffusion and Partition Coefficients for Formaldehyde and VOCs in Building Materials Jianyin Xiong,†,‡ Yuan Yao,‡ and Yinping Zhang‡,* † ‡
School of Mechanical and Vehicular Engineering, Beijing Institute of Technology, Beijing 100081, China Department of Building Science, Tsinghua University, Beijing 100084, China
bS Supporting Information ABSTRACT: The initial emittable concentration (Cm,0), the diffusion coefficient (Dm), and the material/air partition coefficient (K) are the three characteristic parameters influencing emissions of formaldehyde and volatile organic compounds (VOCs) from building materials or furniture. It is necessary to determine these parameters to understand emission characteristics and how to control them. In this paper we develop a new method, the C-history method for a closed chamber, to measure these three parameters. Compared to the available methods of determining the three parameters described in the literature, our approach has the following salient features: (1) the three parameters can be simultaneously obtained; (2) it is time-saving, generally taking less than 3 days for the cases studied (the available methods tend to need 728 days); (3) the maximum relative standard deviations of the measured Cm,0, Dm and K are 8.5%, 7.7%, and 9.8%, respectively, which are acceptable for engineering applications. The new method was validated by using the characteristic parameters determined in the closed chamber experiment to predict the observed emissions in a ventilated full scale chamber experiment, proving that the approach is reliable and convincing. Our new C-history method should prove useful for rapidly determining the parameters required to predict formaldehyde and VOC emissions from building materials as well as for furniture labeling.
’ INTRODUCTION Poor indoor air quality can partly be attributed to emissions of formaldehyde and volatile organic compounds (VOCs) from building materials. These emissions negatively affect people’s comfort, health, and productivity.1,2 Formaldehyde, which is regarded as a human carcinogen,3 is of particular concern in the indoor environment.4,5 The three characteristic parameters affecting formaldehyde and VOC emissions from building materials are the initial emittable concentration, the diffusion coefficient and the material/air partition coefficient.611 Knowledge of these parameters is required to evaluate and control formaldehyde and VOC emissions from building materials and furniture. Many methods have been proposed to measure these three characteristic parameters. For the diffusion coefficient, Haghighat et al.12 reviewed different measuring methods including the cup method, twin chamber method, and porosity method. They analyzed these methods and pointed out that the wet cup method may overestimate the diffusion coefficient due to the relatively high VOC concentration in the experiment; the twin chamber method ignores the boundary layer effect created by the forced air convection; the porosity test method is only applicable for r 2011 American Chemical Society
homogeneous materials; and that there can be a difference of up to 700% in the reported data for a given technique. As an improvement to the porosity method, a new model was proposed to calculate the diffusion coefficient based on a detailed mesostructure analysis of the porous building material.13 To measure the partition coefficient, a microbalance test system was used to determine this parameter for VOCs in vinyl flooring by introducing VOCs at a constant inlet concentration in the sample chamber until sorption equilibrium was reached.14 Tiffonnet et al.15 also used the sorption idea to determine the partition coefficient of acetone in three kinds of building materials by injecting a given amount of acetone into the closed chamber. Both of these methods needed multiple sorption process runs, and generally required a relatively long experimental time (more than 1 week). To measure the initial emittable concentration, Cox et al.16 developed a fluidized bed desorption method, however, the Received: January 24, 2011 Accepted: March 8, 2011 Revised: March 2, 2011 Published: March 23, 2011 3584
dx.doi.org/10.1021/es200277p | Environ. Sci. Technol. 2011, 45, 3584–3590
Environmental Science & Technology
ARTICLE
process inside the building material is one-dimensional and that the VOCs in the chamber are well mixed. These assumptions have been widely adopted in previous studies 69 and are further validated in the later text by the comparsion between the model prediction and experimenetal data. Based on these assumptions, the analytical solution describing VOC emission is20 Ca ðtÞ ¼
Figure 1. Schematic of VOC emission from building material in a closed chamber.
experimental system was quite complicated and there was some concern that the properties of the material might change when ground to a powder. Smith et al.17 presented an extraction method to measure the initial emittable concentration. The tests often took a long time (4 weeks) and in the last few cycles the uncertainty of the measurement was prone to increase due to the very low VOC concentrations in the chamber. To overcome these problems, a multiemission/flush regression method was proposed to simultaneously determine the initial emittable concentration and the partition coefficient.18 Although this method could reduce the experimental time from 4 weeks to 1 week, it was still relatively long and may be too uncertain for engineering applications. In addition, Wang and Zhang19 put forward a new method to simultaneously determine the initial emittable concentration, diffusion coefficient, and partition coefficient. However, the relative error of the peak concentration detection could reach 20% due to the measurement errors of INNOVA-1312 for the target VOC and of the peak VOC concentration in the chamber after injecting VOC caused by that the INNOVA-1312 sampling time tended not to be the peak occurring time. In addition, the experimental time of this method also approached 1 week. From this analysis it is evident that a quick measurement method of the three characteristic parameters suitable for engineering applications is still needed. The purpose of this paper is to propose an effective method to rapidly and simultaneously determine the initial emittable concentration, diffusion coefficient, and partition coefficient of formaldehyde and VOCs in building materials.
’ METHODS A schematic of a building material placed in a closed chamber (a chamber with an air exchange rate that is close to zero) with double surface emission is shown in Figure 1. Both surfaces of the building material are exposed to the air, and all edges are sealed. Considering that the emission characteristics from both surfaces are identical due to symmetry, we can treat the emission process as singe-sided emission from half of the building material. We assume that the building material is uniform, the VOC diffusion
¥ sin q Cm , 0 β n Dm L2 q2n t þ 2Cm, 0 β e Kβ þ 1 n ¼ 1 qn A n
∑
ð1Þ
where, Ca is the chamber VOC concentration, μg/m3; Cm,0 is the initial emittable concentration of VOCs in the building material, μg/m3; Dm is the diffusion coefficient of VOCs in the building material, m2/s; K is the partition coefficient between the building material and air; Bim is the Biot number for heat transfer (=hmL/ Dm); L is the half thickness of the building material, m; hm is the convective mass transfer coefficient, m/s;. An = 1 (Kβq2nKBi1 m þ1) cos qn (1 þ 2KBim )qn sin qn, n = 1,2,...; β = AL/V; A is the surface area of the building material (including both surfaces), m2; V is the volume of the chamber, m3; qn are the positive roots of tan qn 1 ðn ¼ 1, 2, :::Þ ¼ 2 1 qn KBim Kβ qn
ð2Þ
When the emission process reaches the equilibrium state, the equilibrium chamber VOC concentration, Cequ, can be represented as Cequ ¼
Cm, 0 β Kβ þ 1
ð3Þ
Combining eqs 1 and 3, we get ¥ sin q Cequ Ca ðtÞ n Dm L2 q2n t ¼ 2ðKβ þ 1Þ e q A Cequ n¼1 n n
∑
ð4Þ
For the infinite exponential series of eq 4, the terms decay very fast. Therefore, for a sufficiently long emission time, t, only the first term (n = 1) is significant (this is further analyzed in the “time interval” part). This means: ! ! Cequ Ca ðtÞ 2ðKβ þ 1Þsin q1 2 2 ¼ Dm L q1 t þ ln ln q1 A 1 Cequ ð5Þ where, q1 is the first root of eq 2; A1 is the first term of An. If we define Cequ Ca(t) as the excess concentration and (Cequ Ca(t))/Cequ as the dimensionless excess concentration, then eq 5 means that the logarithm of the dimensionless excess concentration is in a linear relationship with emission time. We denote the slope and intercept as SL and INT, respectively, yielding SL ¼ Dm L2 q21 2ðKβ þ 1Þsin q1 INT ¼ ln q1 A 1
ð6Þ !
Equation 5 can then be rewritten as ! Cequ Ca ðtÞ ¼ SL 3 t þ INT ln Cequ 3585
ð7Þ
ð8Þ
dx.doi.org/10.1021/es200277p |Environ. Sci. Technol. 2011, 45, 3584–3590
Environmental Science & Technology Therefore, if the chamber VOC concentration is treated as the form of the logarithm of dimensionless excess concentration in eq 8, the SL and INT can be obtained by doing linear curve fitting. The two parameters Dm and K can be obtained directly because they are functions of SL and INT, and we have two equations with two unknown parameters. Combining eq 3 and the known value of K, Cm,0 can be calculated. Since the main characteristic of this method is to apply the concentration history of VOCs in a closed chamber, we have called this method the C-history method for a closed chamber. It should be pointed out that neither the analytical model nor the C-history method consider the chemical state of the VOCs in the building materials. The detailed procedure for determining Dm, K, and Cm,0 is included in page S1 and Figure S1 of the Supporting Information (SI).
’ EXPERIMENTAL SECTION For the purpose of studying the chamber volume adaptability of the C-history method, two kinds of environmental chambers, a small scale chamber (30 L) and a full scale chamber (30 m3) were used in the experiment. Formaldehyde, which is the major pollutant from wood-based boards, is selected as the target chemical pollutant. The small scale chamber experimental system is shown in SI Figure S2. The chamber is made of stainless steel, with a fan in the top to promote the mixing of the VOCs in the air. INNOVA-1312, a real-time gas VOC analyzer, was used to monitor the formaldehyde concentration and approach to equilibrium in the chamber. The INNOVA-1312 was calibrated in Denmark on April 20, 2010, and the experiment was carried out within one year of that time. In addtion, the measurement of formaldehyde was also calibrated by HPLC. Compensation for water vapor interference was included in the values returned by the system. Air samples were taken from the chamber, then went through the analyzer, and finally came back to the chamber, via Teflon tubes. A water-bath jacket was used to control the air temperature in the chamber. The full scale chamber experimental system was located in a 120 m3 room. There are no insulation layers outside this chamber except the bottom surface. Both the room and the chamber were air-conditioned separately. Formaldehyde was collected and analyzed according to MBTH (3-methyl-2-benzothiazolino-nehydrazone) spectrophotometry specified in GB/T 18204.26,21 which was different from that of the small scale chamber. For the closed chamber experiment, the “equilibrium state” is defined as the point where the relative deviation of the mean concentration of the target VOC during the subsequent hour to that during the preceding hour was less than 1.00%. Since wood-based boards are widely used for furniture, floor and decorative board in China, three kinds of medium density fiberboards (designated as MDF1, MDF2, and MDF3, respectively), two kinds of particle boards (designated as PB1 and PB2, respectively) and one block board (BB), were chosen for our experiments. Table 1 gives the geometrical dimensions of the tested building materials and the experimental conditions. The chamber relative humidity (RH) was maintained at 50 ( 5%. Four different temperatures were selected for the closed chamber experiments, with the purpose of studying the temperature adaptability of the C-history method. The MDF1, MDF2, PB1, and BB samples were tested in the small scale chamber under closed chamber conditions. The MDF3 and PB2 samples were tested in the full scale chamber under both closed chamber and
ARTICLE
Table 1. Experimental Conditions of the Tested Building Materials temperature material
(C)
MDF1 MDF2
33.0 ( 0.5 25.0 ( 0.5
MDF3
23.0 ( 0.5
PB1
27.0 ( 0.5
PB2
23.0 ( 0.5
BB
27.0 ( 0.5
chamber
length width
number
volume (m3) thickness (m m m) of pieces 0.03 0.03 30 0.03 30 0.03
0.1 0.1 0.0028 0.1 0.1 0.0038
2 1
1.225 1.025 0.004
4
0.1 0.1 0.0158
1
1.22 1.22 0.016
4
0.1 0.1 0.0159
1
Figure 2. Linear curve fitting applying eq 8 for MDF1.
ventilated chamber experimental conditions, for the purpose of using an independent ventilated chamber experiment to validate the measured characteristic parameters from the closed chamber experiment. The ventilated chamber experiments were carried out at 23 ( 0.5 C, 50 ( 5% RH, and 1/h air exchange rate.
’ RESULTS AND DISCUSSION The experimental time for the emission process of the six kinds of building materials in the closed chamber to reach equilibrium was generally less than 3 days, and the convective mass transfer coefficient is 0.0025 m/s for the cases studied. The result of applying eq 8 to the chamber formaldehyde concentration data followed by linear curve fitting for MDF1 is shown in Figure 2. The results for the other building materials are shown in detail in SI Figure S3. Table 2 lists the determined characteristic parameters (Cm,0, Dm, and K) and the square of the correlation coefficient (R2). According to ASTM Standard D515797,22 a correlation coefficient (R) of 0.9 or greater can be regarded as generally indicative of adequate model performance. For the cases studied, all the squares of the correlation coefficients are greater than 0.97, which implies a high regression accuracy. The standard deviation of the diffusion coefficient and the partition coefficient can be obtained by numerical calculation. The procedure is (1) add the standard deviation of the slope (defined as SDSL), and the standard deviation of the intercept (defined as SDINT) of the regression line to the slope and intercept, respectively; (2) use SL ( SDSL and INT ( SDINT and run the same program to obtain the updated diffusion and partition coefficients, which include the deviations; (3) compare the updated diffusion and partition coefficients with the original 3586
dx.doi.org/10.1021/es200277p |Environ. Sci. Technol. 2011, 45, 3584–3590
Environmental Science & Technology
ARTICLE
Table 2. Measured Characteristic Parameters Based on the C-History Method material
Cm,0 (μg/m3)
Dm (m2/s)
K
R2
MDF1
(1.91 ( 0.10) 107
(5.58 ( 0.25) 1011
(1.46 ( 0.09) 103
0.995
11
MDF2
(4.01 ( 0.34) 10
(2.72 ( 0.21) 10
(5.52 ( 0.53) 103
MDF3
0.976
(1.53 ( 0.12) 107
(9.25 ( 0.71) 1012
(5.94 ( 0.58) 103
0.992
PB1
(2.68 ( 0.22) 107
(5.52 ( 0.27) 1010
(1.64 ( 0.15) 103
0.993
PB2
(2.80 ( 0.21) 107
(4.16 ( 0.17) 1010
(4.23 ( 0.34) 103
0.991
BB
(4.19 ( 0.27) 106
(3.38 ( 0.22) 1010
(4.31 ( 0.40) 102
0.995
6
determined parameters, and obtain the standard deviation. Then, according to error propagation theory, the standard deviation of the initial emittable concentration, SDC, can be represented as follows based on eq 3: SDC Kβ SDK ¼ Kβ þ 1 3 K Cm, 0
ð9Þ
where, SDK is the standard deviation of the partition coefficient. The standard deviations obtained for the three characteristic parameters for the six kinds of building materials are also summarized in Table 2, for the convenience of comparison. Based on these data, the relative standard deviations, defined as SDC/Cm,0, SDD/Dm, SDK/K for the three characteristic parameters, respectively, can be obtained. It was calculated that the maximum relative standard deviations for the initial emittable concentration, diffusion coefficient and partition coefficient were 8.5% (MDF2), 7.7% (MDF2), and 9.8% (MDF3), respectively. In Table 2, it can be seen that MDF2 has the lowest correlation coefficient, which corresponds to the analysis from the relative standard deviation. The maximum relative standard deviation for the cases studied is relatively small, being generally less than 10%, which is acceptable for engineering application. Validation of the Measured Parameters. For the C-history method, the measurement of the three characteristic parameters is not obtained by direct nonlinear regression of experimental data to the analytical solution (eq 1), so we can substitute the measured Cm,0, Dm, and K into the analytical solution to calculate the chamber formaldehyde concentration, and then compare the calculated (simulated) results with the experimental data. This can be regarded as a preliminary validation of the C-history method. Figure 3 shows the comparison of the chamber formaldehyde concentration between the simulated results and the experimental data for MDF1, and SI Figure S4 shows the comparison for the other materials. The vertical bars in the curves are the fluctuations due to the uncertainty of the determined parameters. For the cases studied, the relative error of the experimental data is less than 3.6%, which is relatively small thus is not included in the figures. These figures show that most of the experimental data is located within the simulated emission curve, with some deviations for MDF2 in the initial 3 h. However, if the uncertainties of the measured parameters are considered, the experimental data lies almost within the range of fluctuations of the calculation for all the tested building materials, demonstrating that the measured characteristic parameters are accurate and reliable. Independent experiments to further validate the measured parameters need to be carried out. In the meantime we should consider whether the characteristic parameters derived from the closed chamber experiment can be used to predict the VOC
Figure 3. Comparison of simulated results with closed chamber experimental data for formaldehyde concentration of MDF1.
emission process in a real, ventilated chamber. We therefore performed a full scale experiment in a ventilated chamber for the MDF3 and PB2 samples. Based on the determined parameters from a closed chamber and the analytical results for a ventilated chamber,2325 the formaldehyde concentrations in the ventilated chamber were calculated and compared to the experimental data (Figure 4). For the tested experimental time (within 180 h), the simulated results are in very good agreement with the observed data, suggesting that the measured parameters in the closed chamber are reliable and can be used to estimate VOC emissons in real ventilated conditions. It should be pointed out that, for the MDF3, one duplicate experiment was also performed, and the determined parameters of Cm,0, Dm, and K are (1.31 ( 0.10) 107μg/m3, (10.10 ( 0.78) 1012 m2/s, (5.17 ( 0.50) 103, respectively. The relative error between the two experiments is less than 14.4%, which is relatively small and should be acceptable for engineering application. In addtion, we can compare the C-history method results with results obtained from the multiemission/flush regression method (MEFR) since the MEFR method considers the emission process in a closed chamber.26 The values obtained for Cm,0 and K using the C-history method and the MEFR method (the latter cannot measure Dm) for the same kind of medium density fiberboard are 1.31 107 μg/m3, 1.06 107 μg/m3 and 1.93 103, 1.56 103, respectively. The maximum relative deviation (RD) between these two methods is less than 24%, which is acceptable for engineering application. These results can also be taken as an independent validation of the C-history method. For the C-history method, it is necessary to determine whether the experimental time is long enough to reach equilibrium. Taking the formaldehyde emissions from MDF1 as an example, the simulated results for a longer emission time (4 days) based on 3587
dx.doi.org/10.1021/es200277p |Environ. Sci. Technol. 2011, 45, 3584–3590
Environmental Science & Technology
ARTICLE
Figure 5. Schematic of the time interval.
Generally, for most VOC-materials in the indoor environment, the following parameter ranges exist:27 20
