Formaldehyde Sorption and Desorption Characteristics of Gypsum

rn The sorption and subsequent desorption of form- aldehyde (CH20) vapor from unpainted gypsum wallboard have been investigated in environmental ...
0 downloads 0 Views 781KB Size
Formaldehyde Sorption and Desorption Characteristics of Gypsum Wallboard Thomas G. Matthews,

*vt

Alan R. Hawthorne,+and Cyril V. Thompsont:

Measurement Applications Group, Health and Safety Research Division, and Special Projects Group, Analytical Chemistry Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831

rn The sorption and subsequent desorption of formaldehyde (CH20)vapor from unpainted gypsum wallboard have been investigated in environmental chamber experiments conducted at 23 “C, 50% relative humidity, an air exchange to board loading ratio of 0.43 m/h, and CHzO concentrations ranging from 0 to 0.50 mg/m3. Both CH20 sorption and CHzO desorption processes are described by a three-parameter, single-exponential model with an exponential lifetime of 2.9 f 0.1 days. The storage capacity of gypsum board for CHzO vapor results in a time-dependent buffer to changes in CH20 vapor concentration surrounding the board but appears to cause only a weak, permanent loss mechanism for CH,O vapor. Prior to significant depletion of sorbed CH20,desorption rates from CH20-exposedgypsum board exhibit a linear dependence with negative slope on CHzO vapor concentration. Analogous CH20 emissions properties have been observed for pressed-wood products bonded with urea-formaldehyde resins.

Introduction The accurate modeling of time-dependent or steadystate pollutant concentrations in indoor environments depends on a quantitative knowledge of significant source and loss mechanisms (1). The experimental study and modeling of the sorption and desorption of indoor pollutants on common indoor surfaces (i.e., sink terms) has recently received increased attention as a potentially important pollutant loss term (2-5). For combustion products such as sulfur dioxide (SO,) and nitrogen dioxide (NO,), sink effects can be as important as air infiltration in removing indoor pollutants, particularly in tight dwellings (5). Sink rates for NO, and SOz in environmental chamber tests have been shown to be a function of relative humidity (RH), area and type of adsorbing surface, temperature, and air mixing (2-5). No significant secondary emissions of sorbed NOz or SOz have been observed. The potential sorption and desorption of formaldehyde (CHzO) vapor on indoor surfaces has received little attention in comparison to an extensive body of research concerning the CH,O emission properties of consumer and construction products. Formaldehyde emissions from pressed-wood products, urea-formaldehyde foam insulation (UFFI), and fibrous glass insulation have been shown to be strongly dependent on environmental parameters and the CHzO concentration in the air surrounding the product t Measurement Applications Group. *Special Projects Group.

0013-936X/87/0921-0629$01.50/0

(6-9). The CHzO emission rates (CH,O ER) of these products demonstrate a linear dependence (with negative slope) on CHzO concentration, This is consistent with a one-dimensional treatment of Fick’s Law (7, a), where D (m2/h)/L (m) is expressed as a simple mass transport coefficient K (m/h):

CHzO ER [mg/(m2 h)] = -K (m/h) X [CH20](mg/m3)

+ constant

(1)

Thus, the modeled CHzO emission rates maximize at zero CH20concentration and decrease linearly to a net emission rate of zero at some equilibrium CHzO concentration. If the CH,O concentration is raised beyond the equilibrium level with the addition of new CHzO sources, extrapolation of the model to higher CHzO concentrations indicates that sorption of CHzO by the original “emitter” is expected to occur. Experimental results of individual and paired emitter tests in small-scale environmental chambers by Pickrel et al. (10) and Godish et al. (11) are consistent with this theory. Paired source combinations of pressed-wood products, insulation, carpeting, and UFFI typically yielded CH20 concentrations less than the sum of the concentrations for the individual emitters. In addition, Godish found that the CH20emission strength of many weaker emitters in paired source combinations was enhanced following brief exposure to a stronger emitter (11). This suggests a temporary sorptive mechanism for CHzO followed by reemission of CHzO at reduced CHzO concentrations. We have studied the time and CH20 concentration dependence of potential CH20sorption and desorption from unpainted gypsum wallboard, which is an extremely weak CH20 emitter, in small-scale environmental chambers. Gypsum board is of great interest as an indoor sink material for CH20because of its typically large surface area and water content. Although gypsum wallboard is normally painted indoors, this should not strongly affect its CHzO sorption/desorption potential, because common latex paint has been shown to be ineffective in reducing CH20 transport (12). Formaldehyde sorptive processes were studied directly by exposure of gypsum board to CH20 produced with a controlled CH,O generation apparatus rather than with a second and more strongly emitting test product. Subsequent to CH20 exposure, CHzO desorption rates were compared against essentially undetectable emission rates from the gypsum board prior to CHzO exposure. The primary goals were to investigate the interaction of gypsum board with CHzO vapor as (1) a time-dependent buffer to changes in CH20concentration and/or as (2) a permanent CHzO loss mechanism.

0 1987 American Chemical Society

Environ. Sci. Technol., Vol. 21, No. 7, 1987

629

~

I I I

CH201NJECT10N SYSTEM

I

sorption (8) or desorption (d)

Is

I

---

I

!I

0.125 I

1 I

I

I

I

-1

CEA INSTRUMENT

I

a COMPUTER

I

I

Figure 1. Experimental apparatus for CH,O sorption and desorption studies. The symbols -, --, S, and D denote gas connections, electrical connectlons, connections used only in sorption experiments, and connections used only in desorption experlments, respectively.

Experimental Methods The experimental apparatus was designed to measure the temporal dependence of CH20sorption and desorption processes of gypsum wallboard at approximately 23 "C, 50% RH, and variable CH20concentration conditions (13). A dual environmental chamber system, shown in Figure 1, allowed repetitive, sequential measurements of CH20 concentration in air exiting an empty reference chamber, the gypsum board exposure chamber, and a CH20 zero filter. Both chambers were operated at the same temperature, RH, and air exchange rates and were fed from the same air supply. Mixing was enhanced in both chambers with small fans. During sorption experiments, CH20 vapor from a generation system, consisting of dilute formalin solution injected into a heated airstream (14),was fed into the air supply and split equally between both chambers. All CH20 measurements were performed in 15-min segments on a modified CEA Instruments Model 555 analyzer (13,15).Sampling of the two environmental chambers and filtered room air (i.e., MSA acid gases-organic and formaldehyde vapor) and data aquisition from the CEA instrument were microprocessor-controlled. Both chambers were identically constructed with a Teflon lining to minimize residual sinks inside the chambers. Sequential (i.e,, comparative) measurements between the gypsum board exposure and empty reference chambers in both sorption and desorption experiments further minimized the impact of residual sinks inside the chambers and allowed the monitoring of CH20 concentrations generated during sorption experiments. The air exchange rate [ N (h-l)] and loading [L,surface area (m2)/air volume (m3)] of the gypsum board used inside the exposure chamber for time-dependent sorption and desorption experiments were chosen from considerations of consumer use and experimental limitations. A simple empirical model of a single indoor compartment was considered for selection of product loading and air exchange conditions. The compartment was assumed to measure 4 X 4 X 2.5 m with an air exchange rate of 0.5 h-'. Three walls and the ceiling were assumed to be gypsum board. The resulting area and loading of gypsum board 630

selected CHzO concnl mg/m3 initial final 0 0 0.125 0.125 0.50 0.225 0.125

QAS VALVES

I------

~~

Table I. Time-Dependent CHzO Sorption and Desorption Experiments

FILTER E D

Environ. Sci. Technol., Vol. 21, No. 7, 1987

0.50 0.125 0.25 0.25 0 0 0.63 0.063

duration of experiment, days 210

28 0.8 1.9

0.1-0.2 3.9 0.9 0.9

'Due to temporary malfunctions of the CHzO generation apparatus, the 27-day, 0-0.50 mg/m3 sorption data set was divided up into three segments of approximately 2-, 13-, and 8-day duration. No data were taken on a total of 4 days during the malfunction.

were 46 m2 and 1.15 m-l, respectively. The air exchange/product loading ratio [NIL (m/h)J for the model compartment was 0.43 m/h. Due to experimental limitations with small-scale chamber experiments and the CH20 generation system (23), the same N I L value was used but with 3-fold larger N and L levels. For example, the airflow rate, which would be 1.7 L/min for a N of 0.5 h-l, was raised to 5.15 L/min for a N of 1.5 h-l. The increased airflow rate provided a large excess of air over the 1.0 L/min sampling rate of the CEA Instrument, reduced potential wall effects inside the chamber, and minimized experimental problems with generating CHzO vapor at low airflow rates. The area of gypsum board was increased from 0.24 m2 (i.e., L of 1.15 m-l) to 0.72 m2 (i.e., L of 3.45 rn-'). Six unpainted gypsum board specimens (Le., 0.24 X 0.25 m) were used and exposed to CH20vapor on both sides. The specimen edges were sealed with paraffin to restrict CH20 exposure to primarily the faces of the gypsum board. Several sorption and desorption experiments were performed to elucidate the time-dependent CH20 sink behavior of the gypsum board (see Table I). Month-long sorption experiments were performed by exposing gypsum board specimens (previously conditioned for > 2 weeks at 95% of the generated CHzO concentration measured inside the reference chamber. Similar to the sorption experiments, the 4-day desorption series, starting at about 0.225 mg/m3, consisted of sequential measurements of the reference and

gypsum board exposure chambers performed daily for 3-6 h. The 1-day desorption series, starting at approximately 0.125 mg/m3, consisted of continuous CH20 measurements. For both the 1- and 4-day desorption series, measurements were initiated by simply disconnecting the CH20generation system from the chamber air supply; the gypsum board loading and air exchange rates were held constant. The desorption experiments of a few hours duration, starting at approximately 0.50 mg/m3, required a different protocol because CH20 measurements at several different N I L ratios were desired. The loading of gypsum board (Le., 0.72 or 0.24 m-l) was fixed prior to conditioning at approximately 0.50 mg/m3 CH20. The desorption experiment was initiated by disconnecting the CH2O generation apparatus and adjusting the airflow rate through the gypsum board exposure chamber for a prescribed air exchange rate. Formaldehyde concentration measurements taken during the first three air exchanges of the chamber were not included in subsequent data analysis. To minimize depletion of the sorbed CH20,measurements were continued for only 3 h before restarting the 0.50 mg/m3 generation system. Results and Discussion Due to the temporally sequential nature of the CH20 concentration measurements of the exhaust from the reference and gypsum board exposure chambers, a simultaneous comparison of the CHzO concentration data for the two chambers was not possible. As a result, the CHzO concentration inside the reference chamber (during measurement periods of the gypsum board exposure chamber) was interpolated from a second-order, linear model of the reference chamber data. A weighted interpolation program (16)was used for the 0.8-1.9-day sorption and desorption data sets where typically >25 data points for both the reference and gypsum board exposure chambers were taken in a consecutive fashion. An unweighted fit (i-e., average) was used for smaller data sets, such as the individual 3-6-h data segments, that were taken every 1-3 days during the 27- and 28-day sorption experiments. The results of the 1-28-day sorption and desorption measurements have been fitted to two- and three-parameter, single-exponential models (illustrated in Figure 2) with nonlinear regression analyses (17). Although exponential sorption and desorption processes may represent the physical behavior of the experimental system, the precise form of the models is empirically derived. The time-dependent CHzO sorption and desorption models are respectively R = Z (A - Z)(1 (2)

+

and

R’ = (1 - Z’)e-t/.’ (3) where R and R’represent unitless comparisons (i.e., ratios) of the CHzO concentrations inside the gypsum board exposure/reference chambers described below, Z and 2’ = offsets for R describing rapid changes in CH20 concentration during the first few air exchanges of the environmental chambers in the sorption and desorption experiments, respectively, A = the modeled, maximum value of R achieved at time infinity in a sorption experiment, and r and r’ = the modeled, exponential rise time and decay time of the sorption and desarption processes, respectively. The quantity 1- A represents the permanent loss of CHzO ie the sorption experiments. The physical basis of Z and Z’is uncertain but may represent limitations in the mixing of air within the test chambers, that is, a short circuiting

DESORPTION MODEL

R’

0

I 1

I

I

2

3

I 4

I 5

TIME (days)

Flgure 2. Illustration of CH,O sorption model (Le.,eq 2) and CH,O desorption model (Le., eq 3).

of a portion of the supply air directly to the exhaust of the chamber without physical contact with the surface of the gypsum board. The expressions for R and R ’ shown below include the product of (1) the ratio of the change in CHzO concentrations inside the gypsum board exposure ([CHzO] ) and reference ( [CH2Olref)chambers and (2) a normaEation constant. The CH20concentration data for both the exposure and reference chambers are expressed as a ratio of A concentrations to account for differences in starting concentrations and the potential span of CH20 concentrations in each chamber during the experiment. For example, some sorption experiments range from 0 to 0.125 or 0.5 mg/m3 while others range from nonzero levels (Le., 0.125 mg/m3) to 0.25 mg/m3. The normalization factor provides R and R’values of 1 and 0, respectively, in the absence of CHzO sinks in both the reference and gypsum board exposure chambers. R and R’ are calculated as

R=

[CH2OIgyp - [CHzOIgypo [CHzO]final- [CH,O],,p [CH,OIref - [CH2Olre~[CH20lfina1- [CH@lgyp~ (4)

and R‘=l-R

(5)

where [CH20Ifinal = the final CH20 concentration inside both measurement chambers in the absence of CH20 sinks (e.g., 0.50 mg/m3 for the 0-0.50 mg/m3 sorption series) and [CHzOl,o and [CH20],,p = the measured CH20concentrations in the gypsum board exposure and reference chambers at the start of a sorption or desorption series. Equations 2-5 represent simple models with explicit limitations. The mathematical form of eq 2 and 3 will probably depend to varying extent on temperature, RH, N , L , mixing inside of the chamber, and surface coatings on the gypsum board. Such model components cannot be experimentally evaluated from the results of these studies. In addition, the expressions for R and R ‘are defined only for time periods following the start of the sorption and desorption experiments, respectively. The study of the CHzO sorption and desorption processes closely approaching time zero is outside the scope of this work. Environ. Sci. Technol., Vol. 21, No. 7, 1987

631

Table 11. Model Coefficients and Asymptotic Standard Errors from Nonlinear Regression Analyses 1.0

model

Z

A

all sorption data all desorption data all sorption and desorption data

t , days

1

DESORPTION EXPERIMENTS 0

0.225-0 rnglrna

0 , 0 0 126-0.003 rnglrnS

0.938 f 0.009 0.148 f 0.011 3.07 f 0.13 0.212 f 0.014 2.55 f 0.31 0.924 f 0.006 0.146 f 0.009 2.89 i 0.10

12

I

0

0.5

1 1.0

1 1.5

1 2.0

I

I

I

2.5

3.0

3.6

0

TIME (days)

Flgure 4. Fit of

0

0-05Omglm' 0 - 0 125 rnglm'

Q

0.125-0.21 mglma. 0.8 d a y *

0

04

n 0.125-0.25 mg/me,

1.9 d a y s

i

0 1

0 0

1

8

12

18

20

24

18

i

TIME (day31

Figure 3. Fit of all CH,O

+ ( A - 0.14)[1 -

e(t-tcero)/T]

(6)

A 2 value of 0.14 was assumed on the basis of the final results of fitting all of the sorption data to eq 2 (see Table 11). Values for t,,,, of 0.8 f 0.3 and 17.6 f 0.6 days were obtained for the second and third segments of the 0-0.50 mg/m3 sorption data set. These t,,,, values were used to correct the elapsed time of the sorption data taken following the malfunctions of the generation apparatus so that these data could be modeled with the rest of the sorption data by eq 2. The statistical fit of all sorption measurement series to eq 2 and all desorption measurement series (excluding the 0.1-0.2-day desorption experiments) to eq 3 is shown in Figures 3 and 4, respectively. The model coefficients and asymptotic standard errors from the nonlinear regression analyses are given in Table 11. The