
REVISED TECHNICAL REPORT 

Measuring Formaldehyde Emissions from Low Emitting Hardwood Plywood Panels under Different Conditions of Temperature and Relative Humidity 
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                 prepared for
                                       
                     U.S. Environmental Protection Agency
                   Office of Pollution Prevention and Toxics
                     National Program Chemicals Division 
                                       
                                       
                                     under
                                       
                             Contract EP-W-09-024
                             Work Assignment 3-03
                                       
                                       
                                       
                                       
                                       
                                       
                                      by
                                       
                                   Battelle
                                Columbus, Ohio
                                       
	
                                       
                                       
                                 April 8, 2013


Acknowledgments

The work described in this report was conducted by Battelle under the direction of the U.S. Environmental Protection Agency through Work Assignments 2-07 and 3-03 of EPA Contract EP-W-09-024.  

Contents

Acknowledgements  	ii
List of Abbreviations  	v
Chapter 1  Executive Summary	1
Chapter 2  Background	3
Chapter 3  Test Design and Procedures	5
3.1  Project Goals and Experimental Design	5
3.1.1	Large/Small Chamber Test Comparison	5
3.1.2	Small Chamber Emissions Testing	6
3.2	Test Materials	8
3.3  Experimental Methods	8
3.3.1.	Large Chamber Tests	8
3.3.2	Small Chamber Tests	8
3.3.3	Chromotropic Acid Analysis	9
Chapter 4  Quality Assurance/Quality Control	10
4.1  Chamber Test Conditions	10
4.2  Chromotropic Acid Calibration	10
4.3  Chamber HCHO Background	11
Chapter 5  Statistical Methods	12
5.1  Large/Small Chamber Test Comparison	12
5.2  Small Chamber Emission Tests	12
Chapter 6  Test Results	13
6.1  Large/Small Chamber Comparison	13
6.2  Small Chamber Emissions Tests	14
6.2.1	Effect of Resin Type on HCHO Emissions	15
6.2.2	Effect of Temperature on HCHO Emissions	18
6.2.3	Effect of Relative Humidity on HCHO Emissions	18
6.2.4	Characterization of Measurement Variability Inherent to Small Chamber Testing	............................................................................................................................. 19
Chapter 7  Summary and Conclusions	20
Chapter 8  References	22
Appendix A   	24
Appendix  B	29



Tables

Table E-1.  Comparison of HCHO Emissions of Three Hardwood Plywood Panel Types at Lowest and Highest Temperature and Relative Humidity Conditions	2
Table 1. Test Matrix and Order of Testing for Small Chamber Tests.	7
Table 2.  Results of Large/Small Chamber Comparison with ULEF-UF HWPW	13
Table 3. Test Conditions and HCHO Results from Small Chamber Tests	14
Table 4.  Estimates and Standard Errors from Regression Model	17
Table 5.  Summary of Temperature and Relative Humidity Effects on Formaldehyde  Emissions from Three Types of Hardwood Plywood[a]	17
Table 6.  Comparison of HCHO Emissions of Three Hardwood Plywood Panel Types at Lowest and Highest Temperature and Relative Humidity Conditions	21




Figures

Figure 1.  HCHO Emissions vs. RH for T=25 °C and T=30 °C	15

List of Abbreviations

ASTM		ASTM International, formerly American Society for Testing and Materials
ATCM		Airborne Toxic Control Measure
°C		degrees Celcius 
CARB		California Air Resources Board
EPA		U.S. Environmental Protection Agency
HCHO		formaldehyde
HWPW	hardwood plywood
m[2]/m[3]		meters squared/meters cubed
MDF		medium density fiberboard
g		micrograms
g/mL		micrograms per milliliter 
mL		milliliters
MUF		melamine-urea-formaldehyde
NAF		no added formaldehyde
NAF-acrylic	no-added formaldehyde  -  acrylic resin
PB		particleboard
pMDI		polymeric diphenylmethane diisocyanate 
ppm		parts per million 
RH		relative humidity
SD		standard deviation
T		temperature
TPC		third party certifier
TSCA		Toxic Substances Control Act
ULEF		ultra low-emitting formaldehyde
ULEF-UF	ultra low-emitting formaldehyde-urea-formaldehyde 
ULEF-PF	ultra low-emitting formaldehyde- phenol-formaldehyde 



Chapter 1
Executive Summary


In July of 2010, the Formaldehyde Standards for Composite Wood Products Act, or Title VI of the Toxic Substances Control Act (TSCA), was signed into law.  The statute establishes formaldehyde emission standards for hardwood plywood (HWPW), medium density fiberboard (MDF), and particleboard (PB) sold, supplied, offered for sale, or manufactured in the United States.  These emission standards are identical to the Phase 2 standards established by the California Air Resources Board's (CARB) Airborne Toxic Control Measure (ATCM), which applies to products offered for sale, supplied, or manufactured for use in California.  TSCA Title VI requires EPA to promulgate regulations implementing the emission standards as well as a number of other provisions.  The statute also specifically directs EPA to consider provisions for ultra low-emitting formaldehyde (ULEF) resins and no-added formaldehyde (NAF) resins.

Both TSCA Title VI and the CARB ATCM require formaldehyde emission testing to be conducted quarterly using the ASTM, International (formerly American Society for Testing and Materials (ASTM)) E1333 large chamber method or ASTM D6007 small chamber method.  The CARB ATCM requires manufacturers of the regulated wood products to demonstrate compliance with the emission standards by having their product emission tests and quality control processes certified by a third-party certifier (TPC). Under the CARB ATCM, manufacturers who use NAF and ULEF resins may qualify for reduced testing requirements and/or an exemption from TPC certification requirements, with renewal of that status required every two years.  Under both CARB ATCM and TSCA Title VI definitions, a NAF composite wood product cannot incorporate a resin made with formaldehyde, and a ULEF composite wood product is one made from resins that may contain formaldehyde, but emit it at particularly low levels as specified in the definitions.  

Recent testing has suggested that some ULEF composite wood products may have significantly increased formaldehyde emissions when the temperature (T) and/or relative humidity (RH) are higher than the standard conditions (i.e., 25 degrees Celsius (°C) and 50 %RH) required in the ASTM E1333 and D6007 test methods.  Consequently, EPA initiated testing to assess the effect of elevated T and RH conditions on formaldehyde emissions from NAF and ULEF composite wood products.  Specifically, formaldehyde emissions from HWPW panels made with NAF-acrylic, ULEF-PF (phenol-formaldehyde), and ULEF-UF (urea-formaldehyde) resins were determined using standard methods by a CARB-approved TPC laboratory at several T and RH conditions, with sufficient repetition to assess the consistency of the test results.  In addition to testing at the standard T/RH conditions, each HWPW type (i.e., HWPW made with the three different resins) was tested at 25 °C with 70 and 85 %RH, and at 30 °C with 50, 70, and 85 %RH.  At each T/RH condition three emission tests were conducted on each HWPW type simultaneously using the ASTM small chamber (D6007) method.  The test results were evaluated to determine the effect of T and RH on the formaldehyde emissions of HWPW panels made with each of the three resins.  

The analysis of test data showed that formaldehyde emissions from the ULEF-UF panels increased with increasing T and RH, and that this relationship was statistically significant with respect to the panels selected for the experiment (see Appendix for further details on the statistical analyses).  For the NAF-acrylic panels, the analysis showed a small increase in formaldehyde emissions with increasing T and RH, but this relationship was not statistically significant with respect to the panels selected for the experiment.  Formaldehyde emissions for the ULEF-PF panels decreased slightly with increased T and RH, and this relationship was found to be statistically significant with respect to the panels selected for the experiment.  For each resin type, testing at all T and RH conditions was conducted with samples from the same panels, so that any effects specific to the panels (e.g., location in stack, manufacturer's lot) apply equally to all observations across T and RH.  Time (i.e., product aging during the sequence of tests) is also unlikely to have influenced the results, since the T and RH conditions used in testing were in random order.  
  
Table E-1 illustrates the results of this study.  For each panel type used in the study, Table E-1 reports the estimated formaldehyde (HCHO) emission at the standard testing conditions (25 °C/50 %RH) and at the highest T and RH conditions (30 °C/85 %RH), based on the statistical analysis of all test results for that panel type.  Table E-1 also shows the change in HCHO emissions between the standard testing conditions and the highest conditions.  
 
Table E-1.  Comparison of Estimated HCHO Emissions of Three Hardwood Plywood Panel Types at Lowest and Highest Temperature and Relative Humidity Conditions 
HWPW 
Panel Type
  HCHO Emission
  at 25 °C/50 %RH 
  (ppm)
  HCHO Emission 
  at 30 °C/85 %RH
  (ppm)
 Change
 (ppm)
ULEF-UF
  0.0471
  0.1230
 0.0759
ULEF-PF
  0.0356
  0.0333
 -0.0023
NAF-acrylic
  0.0238
  0.0290
 0.0052
  Note: TSCA Title VI requires no test result higher than 0.05 ppm (under standard T and  RH conditions) for HWPW made with ULEF resins to qualify for a TPC exemption.
  
Table E-1 shows that the combined effect of T and RH resulted in an increase in formaldehyde emission of the ULEF-UF panels by 0.0759 parts per million (ppm) between the standard conditions of 25 °C/50 %RH and the test conditions of 30 °C/85 %RH.  The HCHO emissions from the ULEF-PF panels decreased by 0.0023 ppm between the standard conditions and 30 °C/85 %RH.  The HCHO emissions from the NAF-acrylic panels increased by 0.0052 ppm between the standard conditions and the test conditions of 30 °C/85 %RH.

The results of this study showed that HWPW panels made with ULEF-UF resin had a greater response to T and RH than either the NAF-acrylic or the ULEF-PF panels.  Statistical tests showed that the response of HCHO emission to T and RH for the ULEF-UF panels used in the experiment was significantly different from that of the NAF-acrylic and ULEF-PF panels.  The observed emission behavior of the three different types of HWPW in this study is consistent with the nature of the resins used in their manufacture.  The minimal emissions and absence of T and RH effects with the NAF-acrylic panels reflect the absence of HCHO in the NAF resin, and the differences in emissions and T and RH effects for the ULEF-UF and ULEF-PF panels are consistent with the greater chemical stability of the PF resin relative to the UF resin against degradation by moisture.
  

Chapter 2
Background

 
In 2007, the California Air Resources Board (CARB) issued an Airborne Toxic Control Measure (ATCM)[1] to reduce formaldehyde (HCHO) emissions from hardwood plywood (HWPW), medium-density fiberboard (MDF), and particleboard (PB).  On July 7, 2010 the Formaldehyde Standards for Composite Wood Products Act, or Title VI of the Toxic Substances Control Act (TSCA), 15 U.S.C. 2697 was signed into law.[2]  The statute establishes HCHO emission standards for HWPW, MDF, and PB sold, supplied, offered for sale, or manufactured in the United States.  These emission standards are identical to the CARB ATCM Phase 2 standards.  TSCA Title VI also requires EPA to promulgate regulations that include provisions on labeling; chain of custody requirements; sell-through provisions; finished goods; third-party testing and certification; auditing and reporting of third-party certifiers; recordkeeping; enforcement; laminated products; and exceptions from regulatory requirements for products and components containing de minimis amounts of composite wood products.  The statute also specifically directs EPA to consider provisions for ultra low-emitting formaldehyde (ULEF) resins and no-added formaldehyde (NAF) resins.

TSCA Title VI and the CARB ATCM both define NAF-based and ULEF-based resins in terms of the composition of the resin system and the maximum HCHO emissions for composite wood products made with these resin systems.  In general, a NAF composite wood product cannot incorporate a resin formulated with HCHO.  A ULEF composite wood product is one made from resins that may contain HCHO, but that emit it at particularly low levels.  TSCA Title VI and the CARB ATCM define identical statutory maximum emissions for products made with NAF or ULEF resins. 

The CARB ATCM and TSCA Title VI require manufacturers of the regulated composite wood products to demonstrate compliance with the emission standards by emission testing, conducted quarterly using the ASTM, International (formerly American Society for Testing and Materials (ASTM)) E1333 large chamber method,[3] or the ASTM D6007 small chamber method[4] with a showing of equivalence with ASTM E1333.  Other methods can be used for quality control testing, but these methods must be shown to be equivalent or to correlate with the ASTM E1333 method.  Under the CARB ATCM, manufacturers who use NAF or ULEF resins may be provided with incentives, such as reduced testing and/or exemption from third-party certification requirements.

Recent testing[5][,6] has suggested that some composite wood products may have increased HCHO emissions when the temperature (T) and/or relative humidity (RH) are higher than the standard test conditions (i.e., 25 °C and 50 %RH) required in the ASTM test methods.[3,4]  Frihart et al.[5] tested HCHO emissions from small pieces of HWPW panels made with ULEF-UF resin and a soy-based NAF resin at 25 and 35 °C and RH conditions of 30, 75, and 100 %.  They reported increases in HCHO emissions from the ULEF-UF material at elevated T and RH conditions.  In that study HCHO emissions were determined using a modified test method, EN 717-3 (Flask Method), which is difficult to compare with ASTM E-1333.  In response to the Frihart et al. study, Riedlinger et al.[6] determined HCHO emissions from four types of PB panels (made with UF, PF, melamine-urea-formaldehyde (MUF), and polymeric diphenylmethane diisocyanate (pMDI)) resins, one of which (the PF product) was certified as a ULEF panel under the CARB ATCM.  Emissions were determined at multiple time points at both the standard T/RH conditions and at 30 °C/75 %RH using the small chamber test method (ASTM D6007)[4] and the Dynamic Microchamber Method[7] for up to 50 days.  Aspects of the testing confound comparisons of the data; for example, testing at the standard and elevated T/RH conditions was conducted in two different laboratories, using different sampling procedures and analytical methods, with sampling at different time points.[6]  Nonetheless, the study appears to show that HCHO emissions from panels made with all four resin types increased by factors of 2 to 3 under the elevated T/RH conditions.  Emissions from panels made with two non-UF resin types (i.e., PF and pMDI) never exceeded the numerical emission limit of 0.09 parts per million (ppm) for PB, even at elevated conditions, whereas emissions from panels made with the UF resins (i.e., UF and MUF) exceeded that numerical emission limit at elevated T and RH until about 20 or 25 days after the start of the testing.  

In addition to the recent studies discussed above, a number of older studies have demonstrated that UF resins have increased HCHO emissions under conditions of elevated temperature and humidity.  For example, a 1985 review article[8] analyzed data from numerous studies from 1960-1984 on the effects of temperature or humidity on HCHO emissions from UF bonded PB and HWPW.  This article concluded that HCHO emissions increased exponentially with increasing temperature.  The relationship between humidity and HCHO emissions was more complex and variable, but the author concluded that the relationship was approximately linear.

Because of the collective evidence that composite wood panels made with UF resins may have increased HCHO emissions under conditions of elevated temperature and humidity, EPA initiated this study as part of EPA's effort to consider whether or what type of ULEF provisions to include in regulations implementing TSCA Title VI.  The purpose of this study was to assess the effect of elevated T and RH conditions on HCHO emissions from NAF and ULEF composite wood products made with various resins.  This study was designed to use ASTM standard chamber test methods conducted by a CARB-approved TPC laboratory to determine HCHO emissions at several T and RH conditions with sufficient repetition to assess the consistency of test results.  Subsequent chapters of this report describe the test procedures, data quality, statistical analysis, and HCHO emission results from the testing.
Chapter 3
Test Design and Procedures
 

3.1  Experimental Design 

Veneer core HWPW panels manufactured using each of three different resin types, NAF-acrylic, ULEF-PF , and ULEF-UF, were tested.  The HCHO emissions of each type of HWPW were measured at three different RH conditions (i.e., 50, 70, and 85 %RH) at each of two different T values (i.e. 25 °C and 30 °C).  Thus the HCHO emissions were characterized at the standard conditions of 25 °C and 50 %RH required by the ASTM methods E1333 and D6007, and at higher conditions of T and/or RH.

The testing to characterize HCHO emissions was conducted by PFS Corporation (Cottage Grove, WI), a TPC testing laboratory approved by CARB (http://www.arb.ca.gov/toxics/compwood/listoftpcs.htm).  Testing of the HWPW panels made with three different resin types was carried out in small chambers according to the ASTM D6007 method, as described below.  A limited comparison was also conducted of the HCHO emission results from the D6007 small chamber method to those of the E1333 large chamber method.  That comparison was conducted with ULEF-UF HWPW panels at two T/RH conditions representing the extremes of the test matrix (i.e., 25 °C/50 %RH and 30 °C/85 %RH).  In all testing, gas-phase HCHO was quantified using the chromotropic acid method as described in the E1333 and D6007 methods.  Note that the HCHO emission results reported in this document are as measured at the applicable designated test conditions, i.e., no adjustment to standard conditions (e.g., using the Berge equations cited in the ASTM methods)[3,4] was made.

The results of this study were used to (1) determine whether HCHO emissions were statistically significantly affected by elevated T/RH conditions relative to emissions at the standard conditions of 25 °C/50 %RH, (2) determine whether HCHO emissions and any T/RH effects were statistically significantly different between HWPW panels of different resin types, and (3) characterize the measurement variability inherent to the small chamber testing.

3.1.1	Large/Small Chamber Test Comparison

The CARB ATCM and TSCA Title VI emission standards are based on the ASTM E1333 test method, and the CARB ATCM requires TPCs to demonstrate on an annual basis that HCHO emission results obtained using the ASTM D6007 method are equivalent to those obtained using the E1333 method.  Requirements for that equivalency demonstration include comparison of at least 10 sample sets, with each sample set including nine panel samples tested by the D6007 method in groups of three with the results averaged together, and inclusion of at least five samples sets in each of at least two emission ranges defined by CARB.[1]  The TPC used in this project successfully conducted such an annual equivalency test according to the CARB ATCM 
§93120.9 (a)(2)(B),[1] on July 31, 2012, shortly before the start of testing for this study.

An additional comparison of the large and small chamber methods was conducted in this study, by emission testing of HWPW made with ULEF-UF resin at standard conditions (T = 25°C, RH = 50 %), and at elevated T/RH conditions (T = 30 °C, RH = 85 %).  In this comparison, the procedures of the E1333 and D6007 methods were followed, with specific minor exceptions that are noted below and/or described in Section 3.3.  In using both methods, care was taken to maintain low chamber background HCHO concentrations.  To accurately determine HCHO background concentrations, the TPC laboratory prepared HCHO standards for the chromotropic acid method at lower concentrations than those specified in the ASTM methods,[3,4] and prepared a calibration curve that included those additional calibration points.  The ULEF-UF HWPW test material was conditioned for 7 days before testing in the large chamber as required by the E1333 method; the conditioning period was 48 hours in the D6007 small chamber testing, considerably longer than the minimum of 2 hours required for that method.  This extended conditioning period was chosen to assure that the HWPW was completely equilibrated to the elevated T and RH conditions.  

For the additional comparison of large (E1333) and small (D6007) chamber methods, a single sample of the ULEF-UF HWPW was tested for HCHO emission in the large chamber at each of the two test conditions noted above.  Duplicate air samples were drawn from the large chamber for determination of the HCHO concentration using the chromotropic acid method, and the average of those duplicate sample results was reported as the HCHO emission result.  For the small chamber comparison tests, three small specimens (3 in x 7 in) were cut from the interior of each of three separate 4 feet x 8 feet panels of the ULEF-UF HWPW panels.  The nine resulting specimens (three specimens from each of the three panels) were then conditioned together at the test conditions of T and RH.  After conditioning, the nine specimens were distributed into three identical small test chambers so that each chamber contained one specimen from each of the 4 feet x 8 feet panels.  Thus, each chamber contained identical sets of test specimens.  The three small chamber tests were carried out simultaneously.  Triplicate air samples were drawn from each of the three small chambers for determination of the HCHO concentration using the chromotropic acid method, and the average of the triplicate sample results from each chamber was reported as the HCHO emission result from that chamber.  

3.1.2	Small Chamber Emissions Testing

Small chamber (D6007) emissions testing was performed to characterize HCHO emission from each of the three types of HWPW panels (i.e., HWPW panels made with three different resins) at each of the six specific T/RH conditions.  For each combination of HWPW panel type and T/RH condition, triplicate tests were conducted simultaneously in three small test chambers. Thus with triplicate testing of three HWPW panel types at two T levels and three RH levels, a total of 3 x 3 x 2 x 3 = 54 discrete small chamber tests were conducted, 18 on each of the three HWPW panel types.  Cutting of test specimens, sample conditioning, and HCHO sampling for the small chamber tests were performed as described in Section 3.1.1.

The test matrix for the small chamber testing is shown in Table 1.  For each day of testing, Table 1 shows the HWPW panel type tested as well as the nominal T and RH test conditions.  Table 1 also includes a running number of all the tests (up to 54 total), and indicates the triplicate tests of a single HWPW panel that were run on a single test day. Table 1 shows the small chamber tests 

Table 1. Test Matrix and Order of Testing for Small Chamber Tests.
Test Day
Test #
HWPW 
Resin Type
Temperature
Relative Humidity
Replicate #
Test Day
Test #
HWPW
Resin Type
Temperature
Relative Humidity
Replicate #
1
1
NAF-acrylic
25 °C
50 %
1
10
28
NAF-acrylic
25 °C
70 %
1

2
NAF-acrylic
25 °C
50 %
2

29
NAF-acrylic
25 °C
70 %
2

3
NAF-acrylic
25 °C
50 %
3

30
NAF-acrylic
25 °C
70 %
3
2
4
ULEF-PF
25 °C
50 %
1
11
31
ULEF-PF
25 °C
85 %
1

5
ULEF-PF
25 °C
50 %
2

32
ULEF-PF
25 °C
85 %
2

6
ULEF-PF
25 °C
50 %
3

33
ULEF-PF
25 °C
85 %
3
3
7
ULEF-PF
30 °C
70 %
1
12
34
ULEF-PF
30 °C
85 %
1

8
ULEF-PF
30 °C
70 %
2

35
ULEF-PF
30 °C
85 %
2

9
ULEF-PF
30 °C
70 %
3

36
ULEF-PF
30 °C
85 %
3
4
10
NAF-acrylic
30 °C
70 %
1
13
37
ULEF-UF
25 °C
50 %
1

11
NAF-acrylic
30 °C
70 %
2

38
ULEF-UF
25 °C
50 %
2

12
NAF-acrylic
30 °C
70 %
3

39
ULEF-UF
25 °C
50 %
3
5
13
ULEF-PF
30 °C
50 %
1
14
40
ULEF-UF
30 °C
70 %
1

14
ULEF-PF
30 °C
50 %
2

41
ULEF-UF
30 °C
70 %
2

15
ULEF-PF
30 °C
50 %
3

42
ULEF-UF
30 °C
70 %
3
6
16
NAF-acrylic
30 °C
50 %
1
15
43
ULEF-UF
30 °C
85 %
1

17
NAF-acrylic
30 °C
50 %
2

44
ULEF-UF
30 °C
85 %
2

18
NAF-acrylic
30 °C
50 %
3

45
ULEF-UF
30 °C
85 %
3
7
19
NAF-acrylic
30 °C
85 %
1
16
46
ULEF-UF
30 °C
50 %
1

20
NAF-acrylic
30 °C
85 %
2

47
ULEF-UF
30 °C
50 %
2

21
NAF-acrylic
30 °C
85 %
3

48
ULEF-UF
30 °C
50 %
3
8
22
NAF-acrylic
25 °C
85 %
1
17
49
ULEF-UF
25 °C
85 %
1

23
NAF-acrylic
25 °C
85 %
2

50
ULEF-UF
25 °C
85 %
2

24
NAF-acrylic
25 °C
85 %
3

51
ULEF-UF
25 °C
85 %
3
9
25
ULEF-PF
25 °C
70 %
1
18
52
ULEF-UF
25 °C
70 %
1

26
ULEF-PF
25 °C
70 %
2

53
ULEF-UF
25 °C
70 %
2

27
ULEF-PF
25 °C
70 %
3

54
ULEF-UF
25 °C
70 %
3

in the chronological order in which they were conducted; due to product availability, testing with the ULEF-UF panels was conducted after testing with the NAF-acrylic and ULEF-PF panels had been completed.

3.2	Test Materials

The HWPW panels used for testing were labeled as NAF or ULEF as per the CARB ATCM and were obtained from major manufacturers who were known to the test laboratory or to EPA as potential sources of such products.  Each manufacturer identified the resin type used in their product.  All three panel types were 5-ply veneer core HWPW; the NAF-acrylic and ULEF-PF panels were 9 millimeters thick, and the ULEF-UF panels were 0.5 inch thick. All materials were obtained as stacks of 4 feet x 8 feet panels, sandwiched between additional panels of the same material and wrapped in plastic film.  Panels used for testing were taken from the interior of the stack.  All panels and cut test pieces were kept wrapped in storage at ambient conditions at the TPC laboratory until withdrawn for conditioning at test conditions. 

ULEF-UF panels from two different manufacturers were tested because the first panels tested showed emissions in initial small chamber tests that did not meet the ULEF emission standards.  The emission testing results from those panels are described in Appendix B.  ULEF-UF HWPW panels were obtained from a different manufacturer and were subjected to the full matrix of testing; the results from those ULEF-UF panels are presented in the body of this report.
   

3.3  Experimental Methods

3.3.1.	Large Chamber Tests 

With the exception of the elevated T and RH condition used in the comparison to the small chamber test, the large chamber test procedure (ASTM E1333) was used as published,[3] with a product loading ratio of 0.43 meters squared/meters cubed (m[2]/m[3]) as designated in the method for HWPW panels.  The large chamber volume was 22.9 m[3] (810 cubic feet) with an air exchange rate of 0.5/hour.  The  T and RH in the test chamber were monitored and recorded at 5-minute intervals throughout the large chamber tests, and those data are summarized in Section 4 of this report.

3.3.2	Small Chamber Tests

With the exception of the elevated T and RH conditions used in most tests, the small chamber test procedure (ASTM D6007) was used as published,[4] with an area-specific air flow rate (N/L) of 1.173 meters/hour, as designated for HWPW panels.  This equates to a 0.5/hour air exchange rate with the small chamber volume of  0.196 m[3] (6.92 cubic feet).  The T and RH in the test chambers were monitored and recorded at 1-minute or 5-minute intervals throughout the small chamber tests, and those data are summarized in Section 4 of this report.  A conditioning time of 48 hours was used to ensure that test materials were equilibrated to the elevated T and RH conditions used in testing. 

The small chamber tests shown in Table 1 were conducted using three identical small chambers, so that the three tests conducted on each test day were performed simultaneously.  Test specimens for small chamber testing were cut from large HWPW panels as described in Section 3.1.1.  All nine specimens to be tested on a single day (e.g., the NAF-acrylic specimens on Test Day 1 in Table 1) were conditioned together in a single conditioning chamber at the planned test conditions of T and RH, and upon completion of conditioning, identical sets of three test specimens were placed into each of the three small test chambers.  The HWPW specimens remained in the small test chambers until a steady state HCHO concentration was reached, i.e., until approximately 2.5 hours after the start of the small chamber test.

3.3.3	Chromotropic Acid Analysis

The TPC used the chromotropic acid method to quantify HCHO as specified in the ASTM E1333 and D6007 test methods, but the calibration curve was extended to lower concentrations to assure accurate determination of HCHO emission from the panels.  Specifically, standards were prepared with smaller amounts of HCHO Standard Solution B than the minimum 0.1 milliliter (mL) volume stated in the E1333 and D6007 methods.[3,4]  (Standard Solution B contains  10 micrograms per milliliter [ug/mL] HCHO, as specified in the ASTM methods.)  The calibration results for the chromotropic acid method are summarized in Section 4 of this report.

Chapter 4
Quality Assurance/Quality Control


This chapter summarizes the Quality Assurance/Quality Control efforts and resulting data quality in this study.  The following sections describe the quality of data relevant to the HCHO emission results shown in Section 6 of this report.

4.1  Chamber Test Conditions	

The T and RH were controlled in each of the chamber tests to achieve the test conditions shown in Table 1, with the goal of maintaining chamber T and RH within +- 1 °C and +- 4 %RH, respectively, of the nominal conditions as stated in the ASTM E1333 and D6007 methods.[3][,4]  To this end, the chamber T and RH were measured at 1-minute or 5-minute intervals over the entire duration of each chamber test.  The mean chamber T and mean chamber RH met the respective 
+- 1 °C and +- 4 %RH goals in all the large and small chamber tests.  Moreover, during each chamber test, the test conditions were closely controlled as demonstrated by the fact that the standard deviation of the individual recorded T values was 0.2 °C or less in 39 of the 54 small chamber tests, and never exceeded 0.5 °C.  Similarly, the standard deviation of the individual recorded RH values met the  +- 4 %RH target in 51 of the 54 small chamber tests, with the only exceptions occurring at elevated RH conditions.  Even when controlling at elevated RH conditions, the standard deviation of the individual RH values never exceeded 5.3 %RH.  In all cases the actual mean values of T and RH, rather than the nominal values, were used in the statistical analysis of T and RH effects on HCHO emissions.

Inspection of the individual T and RH values recorded in each test similarly showed close control of test conditions.  All individual T and RH readings were within the target tolerances noted above at all times for the large chamber test conducted at 25 °C/50 %RH, and for all but one five minute reading (in which the T was just slightly below the target range) for the large chamber test conducted at 30 °C/85 %RH.  Of the 3,257 total individual T readings and 3,257 total individual RH readings recorded during the 54 small chamber tests, only 115 of the T readings differed by more than 1 °C from the nominal test T, and only 251 of the RH readings differed by more than 4 %RH from the nominal test RH.  The individual T or RH readings that exceeded the target tolerances occurred primarily at the most elevated test conditions, or at the beginning of a chamber test when the chamber atmosphere had not yet fully recovered from the perturbation caused by opening the chamber to insert the test specimens.    

4.2  Chromotropic Acid Calibration

As is standard practice for the TPC, a single chromotropic acid calibration curve relating absorbance to sample HCHO content was prepared at the beginning of this study in August 2012, and was applied to all test data.  The calibration curve included a method blank and eight calibration points based on HCHO amounts ranging from 0.019 to 0.372 micrograms (ug) in the calibration standards.  The best fit linear regression calibration curve had the form (micrograms HCHO) = 17.548 x (absorbance) - 0.0028 ug, with a coefficient of determination (r[2]) of 0.9912.  The slope of this calibration curve was within approximately 10 % of previous calibrations performed by the TPC for projects between March and July 2012, indicating that the response of the laboratory spectrophotometer and the preparation of the calibration standards were consistent over time.  

Fresh sodium bisulfite solution was prepared at least once per week, and the chromotropic acid reagent solution was prepared fresh on each day of testing.  The absorbance of the sodium bisulfite + chromotropic acid reagent (or method blank) was measured on each test day as a quality control check.  The difference between the absorbance readings of the method blanks and that of distilled water was always less than 0.006 absorbance units (equivalent to less than 0.014 ppm HCHO in the gas phase).

4.3  Chamber HCHO Background

Care was taken to maintain low chamber background levels of HCHO during testing, because of the relatively low emissions expected from the HWPW panels being tested.  The chamber background HCHO concentration was measured using the same chromotropic acid sampling and analysis procedures used to determine product HCHO emissions, by sampling of the chamber before the products were introduced into the chamber.  

In the two large chamber tests conducted with the ULEF-UF HWPW for comparison to the corresponding small chamber results (Sections 3.1.1 and 6.1) the large chamber background HCHO was measured to be less than 0.010 ppm.  In small chamber tests conducted to assess T and RH effects on emissions, the mean (+- standard deviation) of the small chamber background HCHO values was 0.020 (+- 0.009) ppm, based on 22 measurements.  A considerable portion of the variability in the small chamber background resulted from two small chamber tests in which the background exceeded 0.03 ppm.  Excluding those two tests, the chamber HCHO background averaged 0.017 (+- 0.005) ppm, based on 20 measurements, with a maximum value of 0.026 ppm and a minimum value of 0.009 ppm.  




Chapter 5
Statistical Methods

As noted in Section 3.1, the results of this study were used to (1) determine whether HCHO emissions were statistically significantly different under different T and RH conditions, (2) determine whether HCHO emissions were statistically significantly different between HWPW made with different resin types, and (3) calculate the measurement variability inherent to the small chamber testing.  This chapter summarizes the data analysis used to meet these objectives.

5.1  Large/Small Chamber Test Comparison

Two one-sample comparison tests were performed using HWPW with ULEF-UF resin, one at standard T/RH conditions (T = 25 °C, RH = 50 %), and one at elevated T/RH conditions (T = 30 °C, RH = 85 %).  The data from these tests are reported in Section 6.1, but due to the small size of the data set no formal statistical analysis of these data was conducted.

5.2  Small Chamber Emission Tests

A regression model with interactions was used to model the dependence of HCHO emissions from HWPW on T, RH, and resin type.  The response variable was HCHO emissions.  Temperature and RH were continuous predictors in the model.  Resin type was a qualitative variable with three classes (NAF-acrylic, ULEF-PF, and ULEF-UF).  The model included an interaction between resin type and T and between resin type and RH.  The interaction terms in the model allowed for testing whether the response to T and RH differed by resin type and, if so, whether the effects of T and RH were statistically significant for each resin type.

Details on the statistical models, hypothesis tests, and limitations to the statistical methods can be found in the Appendix.

Chapter 6
Test Results
     
6.1  Large/Small Chamber Comparison

Table 2 shows the HCHO emission results obtained by testing with the ULEF-UF HWPW in the large (E1333) and small (D6007) chambers.  For each of the two test conditions used in this comparison, Table 2 shows the nominal and actual test conditions of T and RH, the individual and mean HCHO results (and standard deviations (SDs)) from the duplicate samples in the large chamber, and the individual and mean HCHO results (and SDs) from the triplicate samples in each of the three small chambers.

Table 2.  Results of Large/Small Chamber Comparison with ULEF-UF HWPW
 Nominal Test Conditions
(°C/ %RH)
Chamber
Actual Test Conditions
(°C/ %RH)
Individual HCHO
Sample Results[a]
(ppm)
Mean HCHO Sample Results
(ppm)[b]
25 / 50 
Large
25.6 / 51.1 
0.051, 0.051
0.051 (+- 0.0)

Small[c]
25.2 / 49.9
25.4 / 49.8
25.4 / 50.0
0.049, 0.044, 0.054
0.042, 0.044, 0.052
0.052, 0.044, 0.042
0.049 (+-0.005)
0.046 (+- 0.005)
0.046 (+- 0.005)
30 / 85 
Large
29.3 / 85.1
0.084, 0.092
0.088 (+- 0.006)

Small[c]
30.2 / 84.4
30.2 / 84.7
30.4 / 84.4
0.111, 0.123, 0.111
0.108, 0.108, 0.116
0.136, 0.133, 0.133
0.115 (+- 0.007)
0.111 (+- 0.005)
0.134 (+- 0.002)
a: Results shown for duplicate samples in large chamber and for triplicate samples in each small chamber. Results are not corrected to standard conditions.
b: Average (+- standard deviation)
c: Results are from three test chambers used simultaneously.  

Table 2 shows that the large and small chamber test results agreed closely in testing at the standard conditions of 25 °C and 50 %RH.  This observation is consistent with the fact noted in Section 3.1.1 that the TPC laboratory successfully conducted annual CARB equivalency testing shortly before the start of this study.  Note that the test results shown in Table 2 have not been adjusted to account for small differences between the actual and standard test conditions.  If the large chamber result of 0.051 ppm obtained at 25.6 °C and 51.1 %RH (Table 2) were adjusted to standard conditions by means of the Berge equations cited in the ASTM test methods[3][,4] the large chamber result would be 0.047 ppm.  This result is consistent with the CARB approval of the UF HWPW as a ULEF product.

Table 2 also shows that the small chamber results at 30 °C and 85 %RH ranged from about 25 to 50 % higher than the large chamber result at those test conditions.  The reason for this difference is unclear.  However, both the large and small chamber tests showed HCHO emissions that exceeded those at the standard conditions of 25 °C and 50 %RH.  

6.2  Small Chamber Emission Tests

Table 3 summarizes the actual test conditions and HCHO emission results from the 54 small chamber tests conducted on the HWPW panels made with different resins.  Shown in this table is the resin type, the average T and RH, and the HCHO emission value in ppm for each small chamber test, and the mean and SD of the three small chamber results in each triplicate set.  Table 3 lists the tests in the order conducted (i.e., in the same order as in Table 1).  The consistency of the triplicate results at each test condition is illustrated by the SD values in Table 3, which are <= 0.006 ppm in most tests, with values of 0.010 to 0.012 ppm in tests with the ULEF-UF panels at elevated T and/or RH.

Table 3. Test Conditions and HCHO Results from Small Chamber Tests[a]
Resin Type
T (°C)
RH (%)
HCHO (ppm)
Mean
(+- SD)
Resin Type
T (°C)
RH (%)
HCHO (ppm)
Mean
(+- SD)
NAF-acrylic
25.4
51.4
0.027
0.024
(+- 0.003)
NAF-acrylic
25.2
70.7
0.019
0.025
(+- 0.006)
NAF-acrylic
25.7
52.8
0.022

NAF-acrylic
25.4
69.7
0.026

NAF-acrylic
26.0
50.0
0.023

NAF-acrylic
25.5
70.0
0.030

ULEF-PF
25.4
53.8
0.034
0.037
(+- 0.003)
ULEF-PF
25.3
84.9
0.023
0.025
(+- 0.004)
ULEF-PF
25.4
51.7
0.039

ULEF-PF
25.4
84.5
0.022

ULEF-PF
25.5
52.3
0.037

ULEF-PF
25.4
85.0
0.029

ULEF-PF
30.2
66.5
0.056
0.054
(+- 0.005)
ULEF-PF
30.2
83.8
0.023
0.025
(+- 0.003)
ULEF-PF
30.2
69.0
0.058

ULEF-PF
30.3
83.7
0.028

ULEF-PF
30.3
69.8
0.048

ULEF-PF
30.3
84.5
0.024

NAF-acrylic
30.2
69.7
0.034
0.031
(+- 0.004)
ULEF-UF
25.2
49.9
0.049
0.047
(+- 0.002)
NAF-acrylic
30.3
70.1
0.026

ULEF-UF
25.4
49.8
0.046

NAF-acrylic
30.4
69.2
0.033

ULEF-UF
25.4
50.0
0.046

ULEF-PF
30.2
49.6
0.042
0.046
(+- 0.003)
ULEF-UF
30.2
70.0
0.110
0.108
(+- 0.005)
ULEF-PF
30.2
50.1
0.048

ULEF-UF
30.2
69.6
0.111

ULEF-PF
30.3
50.0
0.048

ULEF-UF
30.4
69.2
0.102

NAF-acrylic
29.8
49.5
0.041
0.036
(+- 0.006)
ULEF-UF
30.2
84.4
0.115
0.120
(+- 0.012)
NAF-acrylic
30.1
49.5
0.038

ULEF-UF
30.2
84.7
0.111

NAF-acrylic
30.3
49.9
0.029

ULEF-UF
30.4
84.4
0.134

NAF-acrylic
30.2
83.0
0.032
0.032
(+- 0.0)
ULEF-UF
30.2
49.9
0.099
0.095
(+- 0.010)
NAF-acrylic
30.2
84.6
0.032

ULEF-UF
30.2
50.0
0.102

NAF-acrylic
30.4
84.7
0.032

ULEF-UF
30.3
49.8
0.083

NAF-acrylic 
25.2
84.9
0.013
0.015
(+- 0.004)
ULEF-UF
25.3
84.8
0.081
0.093
(+- 0.012)
NAF-acrylic
25.5
85.0
0.019

ULEF-UF
25.5
84.6
0.104

NAF-acrylic
25.4
84.9
0.012

ULEF-UF
25.4
85.3
0.095

ULEF-PF
25.3
70.2
0.023
0.020
(+- 0.004)
ULEF-UF
25.3
70.2
0.067
0.066
(+- 0.011)
ULEF-PF
25.4
70.0
0.016

ULEF-UF
25.4
69.8
0.076

ULEF-PF
25.4
67.7
0.020

ULEF-UF
25.4
69.2
0.054

[a] Air exchange rate in all small chamber tests = 0.5/hour

Figure 1 displays the HCHO emission results by RH at both 25 °C and 30 °C for all three types of HWPW (i.e., HWPW made with three different resins) in this study.  This figure incorporates the test results listed in Table 3, and illustrates clear trends in the emission results over the range of T and RH used in testing.  Figure 1 shows linear relationships with respect to T and RH for the panels made with ULEF-UF (green symbols and traces) and NAF-acrylic (blue symbols and traces) resin types.  For the ULEF-UF panel, the HCHO emissions at the highest T and RH conditions (30 °C/85 %RH) are approximately 0.075 ppm higher than the emissions at the standard conditions (25 °C/50 %RH).  A more complex relationship with respect to T and RH is observed for the ULEF-PF panel type (red symbols and traces in Figure 1).  For this panel type HCHO emissions are similar for both T levels at 50 %RH and at 85 %RH, but they differ considerably at 70 %RH.  It is not clear what is causing this type of relationship.  

              TSCA Title VI requires no test result higher than 0.05 ppm (under standard T and 
              RH conditions) for HWPW made with ULEF resins to qualify for a TPC exemption.

Figure 1.  HCHO emissions vs. RH for T=25 °C and T=30 °C.

6.2.1	Effect of Resin Type on HCHO Emissions

One objective of the study was to determine whether the panels made with the three resin types responded differently to elevated T and RH.  To address this question, we conducted three separate comparisons: ULEF-UF vs. NAF-acrylic, ULEF-UF vs. ULEF-PF, and NAF-acrylic vs. ULEF-PF.  For each pair of resin types, we conducted three hypothesis tests to determine whether there was a difference in the intercept, a difference in the coefficients corresponding to T, and a difference in the coefficients corresponding to RH.  In each hypothesis test, the null hypothesis of no difference (e.g., equal intercepts and equal slopes) was tested against the alternative of a difference (e.g., either the intercept differed or one of the slopes differed).  If at least one of the hypothesis tests for a pair of resins found a difference, we concluded that the responses were not identical for that pair of resins.

For the ULEF-UF vs. NAF-acrylic comparison, the intercept, the T coefficient, and the RH coefficient were all statistically different between the two resin types (all three p-values <0.001), indicating that the two panels used in this experiment behaved differently under the testing condtions.  Compared to the NAF-acrylic panels, the ULEF-UF panels emitted a higher level of HCHO overall (higher intercept) and responded differently to changes in T and RH (different slope parameters).  This is evident from Figure 1 in that all ULEF-UF results are greater than the corresponding NAF-acrylic results, and the slope of the fitted regression line for the ULEF-UF emission results was higher than that for the NAF-acrylic emission results.

For the ULEF-UF vs. ULEF-PF comparison, the intercept, the T coefficient, and the RH coefficient were all statistically different between the two resin types (all three p-values <0.001), indicating that the two panels behaved differently under the testing condtions.  Compared to the ULEF-PF panels, the ULEF-UF panels emitted a higher level of HCHO overall (higher intercept) and responded differently to changes in T and RH (different slope parameters).  This is evident in Figure 1 in that all ULEF-UF panel results are greater than the corresponding ULEF-PF panel results, and the slope of the fitted regression line for the ULEF-UF emission results was higher than that for the ULEF-PF emission results.

For the NAF-acrylic vs. ULEF-PF panel comparison, the intercept, the T coefficient, and the RH coefficient were not statistically different between the two resin types (all three p-values > 0.14), indicating that the two panels behaved similarly under the testing conditions.  This is evident in Figure 1 in that ULEF-PF panel results are mixed among NAF-acrylic panel results and that the fitted regression lines appear similar.

The estimated coefficients of the T and RH effects and the standard errors of those coefficients determined in the statistical analysis are summarized in Table 4, and the results are described in Sections 6.2.2 and 6.2.3.  Table 4 shows, for each panel type, the estimated coefficients for T and RH, the respective standard errors, and the p-value indicating the strength of the indicated statistical relationship.  

           Table 4.  Estimates and Standard Errors from Regression Model 
Effect
Panel Type
Estimated Coefficient
Standard Error
p-value[c]
T
ULEF-UF
0.0080[a]
0.00078[a]
< 0.0001

ULEF-PF
0.0029[a]
0.00078[a]
0.0006

NAF-acrylic
0.0024[a]
0.00080[a]
0.00434
RH
ULEF-UF
0.0010[b]
0.00013[b]
< 0.0001

ULEF-PF
-0.0005[b]
0.00014[b]
0.0010

NAF-acrylic
-0.0002[b]
0.00014[b]
0.1545
	a: Units of temperature coefficients and standard errors are ppm HCHO/°C.
	b: Units of relative humidity coefficients and standard errors are ppm HCHO/%RH.
	c:  Statistically significant effects have a p-value below 0.0033, a threshold that ensures an experimentwise 	error rate of less than α = 0.05.

Table 5 summarizes the cumulative effects of T and RH found in this study.  For all three panel types, Table 5 shows the statistically determined coefficients that quantify the change in HCHO emission for a unit change in T or RH over the range of conditions tested.  Table 5 also shows the resulting changes in HCHO emission (in ppm) for each resin type that occurred over the range of T and RH used in testing.  The changes in HCHO emission due to T and RH are shown separately, and the combined change resulting from both T and RH together is also shown. 

Table 5 shows that HCHO emission from ULEF-UF and ULEF-PF panel types increased with increasing T over the range of 25 to 30 °C used in testing.  However, the effect of T for the ULEF-UF panels was approximately three times as large as for the ULEF-PF panels, and amounted to an increase in HCHO emissions of approximately 0.008 ppm per °C increase in T over the indicated range.  Table 5 also shows that HCHO emissions from the ULEF-UF panels increased with increasing RH over the range of 50 to 85 %RH used in testing, at a rate of approximately 0.001 ppm per 1 %RH increase over that range.  

Table 5.  Summary of Temperature and Relative Humidity Effects on Formaldehyde  Emissions from Three Types of Hardwood Plywood
Resin Type
Coefficient of 
T or RH Effect
Separate Effect on Emissions 
Combined Effect on Emissions[c]
(ppm HCHO)

T 
(ppm HCHO/°C)
RH 
(ppm HCHO/%RH)
T[a] 
 (ppm HCHO)
RH[b]
 (ppm HCHO)

ULEF-UF
0.0080
0.0010
0.0400
0.0359
0.0759
ULEF-PF
0.0029
-0.0005
0.0148
-0.0171
-0.0023
NAF-acrylic[d]
0.0024
-0.0002
0.0121
-0.0069
0.0052
a: Change in formaldehyde emission (in ppm) due to T between 25 and 30 °C.
b: Change in formaldehyde emission (in ppm) due to RH between 50 and 85 %RH. 
c: Total change in formaldehyde emission due to T and RH between 25 °C/50 %RH and 30 °C/85 %RH.
d: Neither T nor RH effects were statistically significant for NAF-acrylic. 


The combined effect of T and RH increased the HCHO emission of the ULEF-UF panels by approximately 0.076 ppm between the standard conditions of 25 °C/50 %RH and the most elevated test conditions of 30 °C/85 %RH, with the T and RH effects contributing roughly equally to this increase.  In contrast, the HCHO emissions from the ULEF-PF panels decreased slightly with increasing RH, as indicated by the negative RH coefficient in Table 5.  The combined effect of T and RH on formaldehyde emission of the ULEF-PF panels was close to zero, i.e., a calculated decrease in emission of approximately 0.002 ppm between the standard conditions of 25 °C/50 %RH and the most elevated test conditions of 30 °C/85 %RH.  The cause of the apparent reduction in formaldehyde emission with increasing RH for the ULEF-PF panels is not known, and was not investigated in this study.  Neither the T nor RH effect was statistically significant for the NAF-acrylic HWPW panels, and the combined effect of T and RH was close to zero, i.e. a calculated increase in emission of approximately 0.005 ppm between the standard conditions of 25 °C/50 %RH and the most elevated test conditions of 30 °C/85 %RH.

6.2.2	Effect of Temperature on HCHO Emissions

The effect of T on HCHO emissions for the ULEF-UF panels used in this study was statistically significant (p-value <0.0001), and the value of the T coefficient was greater than zero (≈ 0.008).  This T effect can be seen  in Figure 1 in that the 30 °C ULEF-UF emission results are greater than the corresponding 25 °C ULEF-UF emission results.  Therefore, it was concluded that HCHO emissions from the ULEF-UF panels used in this study increase with T, and that (based on the coefficient in Table 4) every 1 °C increase in T between 25 °C and 30 °C results in an increase in HCHO emission of approximately 0.008 ppm.

The effect of T on HCHO emissions for the ULEF-PF panels used in this study was statistically significant (p-value 0.0006) and the value of the T coefficient was greater than zero (≈ 0.003).  Therefore, it was concluded that HCHO emissions from the ULEF-PF panels used in this study increase with T at approximately 0.003 ppm per °C increase in T between 25 °C and 30 °C (based on the coefficient in Table 4).  This T effect can be seen in Figure 1 in that most of the 30 °C ULEF-PF emission results are greater than those of the corresponding 25 °C ULEF-PF samples.

The effect of T on HCHO emissions for the NAF-acrylic panels used in this study was not statistically significant (p-value 0.00434).  Although most of the emission results from the 30 °C NAF-acrylic panels are greater than the corresponding 25 °C NAF-acrylic emission results in Figure 1, these T differences are not statistically significant with respect to the panels used in this study.

6.2.3	Effect of Relative Humidity on HCHO Emissions

The effect of RH on HCHO emissions for the ULEF-UF panels used in this study was statistically significant (p-value <0.0001) and the value of the RH coefficient was greater than zero (≈ 0.001).  Therefore, it was concluded that HCHO emissions increase with increasing RH for the ULEF-UF panels used in this study at approximately 0.001 ppm for each 1 %RH increase between 50 and 85 %RH.  This RH effect is evident in Figure 1 in that HCHO emissions increase as RH increases for the ULEF-UF product.

The effect of RH on HCHO emissions for the ULEF-PF panels used in this study was statistically significant (p-value =0.0010) and the value of the RH coefficient was negative (-0.0005).  Therefore, it was concluded that HCHO emissions decrease with increasing RH for the ULEF-PF panels used in this study, at approximately 0.0005 ppm for each 1 %RH increase between 50 and 85 %RH.  As illustrated in Figure 1, the emission results from the ULEF-PF panels at 30 °C and 70 %RH are greater than those at both 50 and 85  RH.  It is not clear whether those results represent the T/RH behavior of the ULEF-PF material or are due to some variation in the testing.

The effect of RH on HCHO emissions for the NAF-acrylic panels used in this study was not statistically significant (p-value = 0.1545).  

6.2.4	Characterization of Measurement Variability Inherent to Small Chamber Testing

The variance of the model residuals, calculated using the equation presented in the Appendix, was 6.4958 x 10[-5].  The square root of this estimator is 0.00806 ppm, which is slightly lower than the variability estimate noted in the small chamber ASTM method documentation[4] (i.e., within-laboratory precision of approximately 0.01 to 0.02 ppm for small chamber tests).  It is therefore concluded that the within-laboratory variability in the small chamber tests conducted in this study is well within the variability expected, after accounting for the effects of the variations in T, RH, and resin type. 
Chapter 7
Summary and Conclusions

Hardwood plywood panels made with three different types of ULEF resins were tested to determine the effect on HCHO emissions of T and RH conditions other than the standard test conditions of 25 °C and 50 %RH .  The panels tested included HWPW made with NAF-acrylic, ULEF-PF, and ULEF-UF resins.  Testing was conducted using the small chamber method (ASTM D6007) under test conditions of 50, 70, and 85 %RH at each of two temperatures (25 and 30 °C).  Triplicate small chamber tests were performed on each HWPW-resin panel type at each of the six T/RH conditions.  The resulting HCHO emission results were analyzed with statistical models to assess the differences in emission behavior from HWPW panels made with the three resin types, the effects of T and RH  on emissions, and the variability of the small chamber test procedure.

Emission Behavior from HWPW Panels made with the Three Resin Types 
   * The study results show that the ULEF-UF panels differed significantly in emission behavior from both the NAF-acrylic and ULEF-PF panels.  
   * HCHO emissions from HWPW panels made with ULEF-UF resin were found to increase with T and RH.  The effects of both T and RH on emissions from ULEF-UF panels were large in relation to the TSCA Title VI limit of 0.05 ppm and statistically significant with respect to the panels used in the study.  
   * HCHO emissions from HWPW panels made with ULEF-PF resin increased with T but decreased with RH. The effects of both T and RH were small relative to the TSCA Title VI 0.05 ppm limit, but statistically significant with respect to the panels used in the study.  
   * The relationships between HCHO emissions and T and RH for HWPW panels made with NAF-acrylic resin were not statistically significant with respect to the panels used in the study.  

Quantified Effects of T and RH  on HCHO Emissions 
   * Statistical analyses were performed to determine coefficients that quantify the change in HCHO emissions for a unit change in T or RH over the range of conditions tested.  
   * For HWPW panels made with ULEF-UF resin, HCHO emissions increased by approximately 0.008 ppm per °C increase in T over the range of 25 to 30 °C, for a fixed RH.  HCHO emissions increased by approximately 0.001 ppm for each 1 %RH increase in RH over the range of 50 to 85 %RH, for a fixed T. The combined effect of T and RH was to increase HCHO emissions from the ULEF-UF panels used in this study by approximately 0.0759 ppm between standard test conditions (25 °C/50 %RH) and the most elevated conditions tested (30 °C/85 %RH). The effect of T on emissions from ULEF-UF panels evaluated for this study was approximately 3 times as great as for ULEF-PF or NAF-acrylic over the temperature range tested. 
   * For HWPW panels made with ULEF-PF resin, HCHO emissions increased by approximately 0.0029 ppm per °C increase in T over the range of 25 to 30 °C, for a fixed RH.  HCHO emissions decreased by approximately 0.0005 ppm for each 1 % increase in RH over the range of 50 to 85 %RH, for a fixed T. The combined effect of T and RH was to decrease HCHO emissions from the ULEF-PF panels used in this study by approximately 0.0023 ppm between standard test conditions (25 °C/50 %RH) and the most elevated conditions tested (30 °C/85 %RH). 
   * For HWPW panels made with NAF-acrylic resin, HCHO emissions increased by about 0.0052 ppm between standard test conditions (25 °C/50 %RH) and the most elevated conditions tested (30 °C/85 %RH), although neither the T nor the RH effect on emissions was statistically significant with respect to the panels used in the experiment.

Table 6 summarizes the results of this study.  For each panel type used in the study, Table 6 reports the HCHO emission at the standard testing conditions (25 °C/50 %RH) and at the highest T and RH conditions (30 °C/85 %RH), based on the statistical analysis of all test results for that panel type.  Table 6 also shows the change in HCHO emissions between the standard testing conditions and the highest conditions.  
 
Table 6.  Comparison of HCHO Emissions of Three Hardwood Plywood Panel Types at Lowest and Highest Temperature and Relative Humidity Conditions 
HWPW 
Panel Type
  HCHO Emission
  at 25 °C/50 %RH 
  (ppm)
  HCHO Emission 
  at 30 °C/85 %RH
  (ppm)
 Change
 (ppm)
ULEF-UF
  0.0471
  0.1230
 0.0759
ULEF-PF
  0.0356
  0.0333
 -0.0023
NAF-acrylic
  0.0238
  0.0290
 0.0052
  Note: TSCA Title VI requires no test result higher than 0.05 ppm (under standard T and  RH conditions) for HWPW made with ULEF resins to qualify for a TPC exemption.
  
The observed emission behavior of the three different types of HWPW used in this study is consistent with the nature of the resins used in their manufacture.  The minimal emission rates and absence of T and RH effects with the ULEF-NAF-acrylic panels reflect the absence of HCHO in the NAF resin.  The ULEF-UF and ULEF-PF panels are both made with resins containing HCHO.  However, the differences in their emission rates and T and RH effects are consistent with the greater chemical stability of the PF resin relative to the UF resin against degradation by moisture.  This stability makes PF resin wood products particularly suitable for external applications where exposure to variable environmental conditions may occur.[11][,1][2]

Variability of the Small Chamber Test Procedure 
   * Statistical analysis of the variability of the small chamber tests produced a variance estimate, the square root of which is 0.00806 ppm.  
   * This uncertainty value is slightly lower than the within-laboratory precision of stated in the ASTM small chamber method (approximately 0.01 to 0.02 ppm).[4]  
   * This result indicates that the within-laboratory test variability achieved in this study is well within the variability expected, after accounting for the effects of the variations in T, RH, and resin type. 




Chapter 8
References

   1. Airborne Toxic Control Measure to Reduce Formaldehyde Emissions from Composite Wood Products, Section 93120.  California Air Resources Board, 2007.  Available at http://www.arb.ca.gov/regact/2007/compwood07/fro-final.pdf.

   2. Formaldehyde Standards for Composite Wood Products Act, 111th Congressional Record S4891-92 (daily ed. June 14, 2010) and H4701-05 (daily ed. June 23, 2010), available at http://www.gpoaccess.gov/crecord/digest2010/d14JN101.html.

   3. Standard Test Method for Determining Formaldehyde Concentrations in Air and Emissions Rates from Wood Products Using a Large Chamber, ASTM E1333-96 (2002), American Society for Testing and Materials, 2002.  Available for purchase at http://enterprise.astm.org/filtrexx40.cgi?HISTORICAL/E1333-96R02.htm

   4. Standard Test Method for Determining Formaldehyde Concentrations in Air from Wood Products Using a Small-Scale Chamber, ASTM D6007-02 (2008), American Society for Testing and Materials, 2008.  Available for purchase at http://dx.doi.org/10.1520/D6007-02R08.

   5. Frihart, C. R., Wescott, J. M., Birkeland, M. J., and Gonner, K. M., Formaldehyde Emissions from ULEF- and  NAF-Bonded Commercial Hardwood Plywood as Influenced by Temperature and Relative Humidity. Published in the Proceedings of the International Convention of the Society of Wood Science and Technology and United Nations Economic Commission for Europe  -  Timber Committee. October 11-14, 2010, Geneva, Switzerland.

   6. Riedlinger, D., Martin, P., and Holloway, T., Particleboard Formaldehyde Emissions and Decay under Elevated Temperature and Humidity Conditions, study conducted for Arclin, Inc., March 28, 2012.

   7. Huber, C.W., Anderson, W.H., Vieira, J. W., and Liles, W.T., The Dynamic Microchamber Computer Integrated Formaldehyde Test System: User Manual, revised version, March 2007; available at www.gp-dmc.com.

   8. Myers, G.E., The effects of temperature and humidity on formaldehyde emission from UF-bonded boards: a literature critique, Forest Prod. J., 35(9):20-31, 1985.

   9. Dean, A. and Voss, D. (1999) Design and Analysis of Experiments.  New York: Springer.

   10.  Hsu, J. C. (1996) Multiple Comparisons: Theory and Methods. Boca Raton: Chapman & Hall/CRC.


   11. The Encyclopedia of Wood, ISBN-13: 978-1-60239-057-7, U.S. Department of Agriculture, Washington, D.C., 1999.

   12. Gardziella, A., Pilato, L.A., and Knop, A., Phenolic Resins: Chemistry, Applications, Standardization, Safety, and Ecology, 2nd Edition, ISBN 3-540-65517-4, Springer-Verlag, Berlin, 1999.  
         


Appendix A
Statistical Methodology

This appendix provides details on the statistical methodology used to analyze the data from the small chamber emission tests.

A.1 Overview

A regression model with interactions was used to model the dependence of HCHO emissions from HWPW on T, RH, and resin type.  Formaldehyde emissions, T, and RH are continuous variables in the model.  Resin type is a qualitative variable with three classes (NAF-acrylic, ULEF-PF, and ULEF-UF); therefore, two indicator variables were used in the model.  

The regression model with interactions assumes that after accounting for differences in conditioning and resin type, all measurements are independent of each other.  This is a reasonable assumption because all small chamber testing was done on samples from three HWPW panels of each resin type.  The experimental units can be viewed as independent samples where the "population" is the exact set of panels used in the experiment (three panels for each of three resin types).  Thus, the experiment satisfies the assumptions underlying the regression model.  All estimates of variability and tests of statistical significance apply to the variability due to sampling from the panels in the experiment and using different chambers for testing.  These results cannot be extended to the population of panels in current or future use that were not sampled as part of the experiment.  We also cannot assess panel-to-panel variability in the panel lots because no single HCHO measurement can be attributed to one specific panel.

Before the data were fit to the model, HCHO emissions were plotted against T and RH to determine whether or not a linear relationship was reasonable between the variables.  The assumption of a linear relationship between HCHO emissions and T and RH was found to be reasonable for ULEF-UF and NAF-acrylic panels.  The relationship between HCHO emissions and T and RH for the ULEF-PF panels was more complex, but a linear relationship was assumed because the source of the complex pattern was not clear, and there were too few experimental conditions to allow consideration of a more complex model.

Statistical tests, described below, were conducted to explore the relationships between HCHO emissions and resin type, T, and RH.  The analysis consists of 15 hypothesis tests on the various model parameters.  Consequently, the threshold significance level was adjusted for each test such that the overall experiment-wise significance level was α = 0.05.  The threshold statistical significance level was corrected using the Bonferroni method found in many statistical texts.[9][,10]  To apply this correction, each hypothesis was tested at a threshold significance level of 0.0033, which is equal to 0.05/15.  This threshold significance value gives an experiment-wise error rate of α = 0.05 adjusted for multiple comparisons.  

A.2 Statistical Models

The formulation of the model is as follows.  Let

Yi=HCHO emissions from HWPW from measurement i
Ti=T at measurement i
RHi=RH at measurement i
Zi1=1 if resin type at measurement i is NAF acrylic;0 otherwise, and
Zi2=1 if resin type at measurement i is ULEF PF;0 otherwise.

The regression model with interactions is then:

Yi=β0+β1Ti+ β2RHi +β3Zi1+β4Zi2+β5TiZi1+β6TiZi2+β7RHiZi1+β8RHiZi2+εi   (1)

Where it is assumed that the εi terms are independent and identically distributed N0,σ2 for i=1,...,n.

The response function for model (1) is then:

EY=β0+β1T+ β2RH +β3Z1+β4Z2+β5TZ1+β6TZ2+β7RHZ1+β8RHZ2

The response function can be simplified according to resin type.  For example, the response function for resin type ULEF-UF, for which Z1=0 and Z2=0, is as follows:

EY=β0+β1T+ β2RH
                                       
For resin type NAF-acrylic, Z1=1 and Z2=0, and the response function is:

EY=β0+β1T+ β2RH +β3+β5T+β7RH=β0+β3+(β1 +β5)T+ (β2+β7)RH

Finally, for resin type ULEF-PF, Z1=0 and Z2=1, the response function is:

EY=β0+β1T+ β2RH +β4+β6T+β8RH=β0+β4+(β1 +β6)T+ (β2+β8)RH 

Thus, interaction regression model (1) implies that each resin type has its own regression line, with different intercepts and slopes for the different resin types.  

One benefit to using the interaction regression model instead of separately fitting three regression models, one for each resin type, is that the former model uses 54 observations (see Sect. 3.1.2) as opposed to only 18 observations for each of the three individual models by resin type.  Since the interaction regression model has many more observations used in its fit compared the other models, it provides a better estimate of the error term (σ2), and therefore provides more powerful results.  Furthermore, a combined regression model facilitates comparing results across resin types, a key question of interest to EPA.

A.3 Hypothesis Tests

In order to investigate whether HCHO emissions were statistically significantly different between HWPW made with different resin types, three formal hypothesis tests were performed, one for each pair of resin types.

Testing the equality of the two response functions for ULEF-UF and NAF-acrylic involved the alternatives:

H0:  β3=β5=β7=0
H1:  not all βk in H0 equal zero

Testing the equality of the two response functions for ULEF-UF and ULEF-PF involved the alternatives:

H0:  β4=β6=β8=0
H1:  not all βk in H0 equal zero

If the two tests above both failed to reject the null hypothesis that the functions were equivalent (i.e., there is no evidence to contradict that β3=β5=β7=0 and β4=β6=β8=0), then there is no need to test the equality of the two response functions for ULEF-PF and NAF-acrylic, as that test would also fail to reject the null hypothesis that the response functions are equivalent.  If one of the two tests above failed to reject the null hypothesis and the other test rejected the null hypothesis (i.e., concluded that the response functions are different), then again there would be no need to test the equality of the two response functions for ULEF-PF and NAF-acrylic, as that test would also reject the null hypothesis and conclude that the response functions are not equivalent.  If both tests above reject the null hypothesis and conclude that the response functions are not equivalent, then a third test to test the equality of the two response functions for ULEF-PF and NAF-acrylic would be conducted using the following alternatives:

H0:  β3=β4and β5=β6 and β7=β8
H1:  at least one βk!=βj in H0

The results of these tests determine the regression line for each resin, and this in turn impacts the investigation of whether or not formaldehyde emissions are statistically significantly different under different T or RH conditions.  For example, if all three tests above fail to reject the null hypothesis, then this essentially means that there is not enough evidence to conclude a different regression line for each resin.  In other words, model (1) reduces to

Yi=β0+β1Ti+ β2RHi +εi

For this reduced model, testing whether or not HCHO emissions are statistically significantly different under different T conditions for any resin type involves the alternatives:

H0:  β1=0
H1:  β1!=0

Similarly for the reduced model, testing whether or not HCHO emissions are statistically significantly different under different RH conditions for any resin type involves the alternatives:

H0:  β2=0
H1:  β2!=0

However, if at least one of three response function equivalency tests between two resins rejects the null hypothesis, then model (1) may be reduced, but not to the extent just shown.  For example, if all three response function equivalency tests reject the null hypothesis, meaning that each resin type has a unique regression line, then as previously shown, the response function for resin type ULEF-UF (Z1=0 and Z2=0) is:

EY=β0+β1T+ β2RH
                                       
For resin type NAF-acrylic (Z1=1 and Z2=0) the response function is:

EY=β0+β3+(β1 +β5)T+ (β2+β7)RH

And for resin type ULEF-PF (Z1=0 and Z2=1) the response function is:

                      EY=β0+β4+(β1 +β6)T+ (β2+β8)RH

Thus, model (1) implies that each resin type has its own regression line, with different intercepts and slopes for the different resin types, so different hypothesis tests were conducted to determine the statistical significance of T and RH for each resin type.  Under this "full" model, testing whether or not HCHO emissions are statistically significantly different under different T conditions for ULEF-UF involves the alternatives:

H0:  β1=0
H1:  β1!=0

Similarly under this "full" model, testing whether or not HCHO emissions are statistically significantly different under different RH conditions for ULEF-UF involves the alternatives:

H0:  β2=0
H1:  β2!=0

Again under this "full" model, testing whether or not HCHO emissions are statistically significantly different under different T conditions for NAF-acrylic involves the alternatives:

H0:  β1+β5=0
H1:  β1+β5!=0

and testing whether or not HCHO emissions are statistically significantly different under different RH conditions for NAF-acrylic involves the alternatives:

H0:  β2+β7=0
H1:  β2+β7!=0

Finally under the "full" model, testing whether or not HCHO emissions are statistically significantly different under different T conditions for ULEF-PF involves the alternatives:

H0:  β1+β6=0
H1:  β1+β6!=0

and testing whether or not HCHO emissions are statistically significantly different under different RH conditions for ULEF-PF involves the alternatives:

H0:  β2+β8=0
H1:  β2+β8!=0

The tests of whether or not HCHO emissions are statistically significantly different under different T and RH conditions have been outlined for the cases where model (1) reduces completely to one regression line and where model (1) is not reduced at all (i.e., the "full" model).  However, depending on the data and the outcomes of the equivalency response function tests between two resin types, model (1) ended up being something between these two extremes, and statistical tests involving T and RH were modified accordingly.  That is, the analysis found that ULEF-UF behaved differently from ULEF-PF and NAF-acrylic, but ULEF-PF and NAF-acrylic behaved similarly to each other.  Based on their similarities, statistical tests could be modified to examine the effect of T/RH on the combined ULEF-PF/NAF-acrylic observations.  However, because of the chemical and physical differences between ULEF-PF and NAF-acrylic, all statistical tests were conducted assuming 3 separate resin types.

A.4 Characterizing Measurement Variability

To characterize the measurement variability inherent to small chamber testing, the following estimator was calculated:

S2=145i=154(Yi-Yi)2

where Yi is observation i and Yi is the predicted value for observation i on the appropriate fitted regression line.  S2 is an unbiased estimator for σ2, the variance of model (1).  The square root of this estimator (S) was compared to the within-laboratory precision estimate noted in the ASTM D6007 documentation[4] of approximately 0.01 to 0.02 ppm.

Appendix B
Emission Testing of First Batch of Commercial ULEF-UF Panels


B.1 Introduction

The first batch of ULEF-UF panels tested gave unexpectedly high emission testing results and therefore were not used in evaluating the effects of temperature and humidity on emissions.  This appendix describes these panels and the emission testing results.

B.2 Product Description

The first ULEF-UF panels tested were 5-ply veneer core HWPW, 0.5 inch thick and 4 ft by 8 ft in size.  The panels were manufactured on July 10, 2012, and shipped directly from the manufacturer to the testing laboratory, where they were received on July 16 as a stack of 10 panels wrapped in plastic sheet.  The panels were labeled as compliant with the CARB ATCM and "93120 TPC Exempt."  The panels were stored, cut to size for chamber testing, and equilibrated with T and RH conditions as described in section 3.1.1 of this study report.
                                       
B.3 Emission Testing

The HWPW was subjected to HCHO emission tests using both the ASTM large chamber (E1333) and small chamber (D6007) methods.  The tests were intended to provide a comparison of the HCHO emissions when determined under the same T and RH conditions by both the large and small chamber methods.  Table 1 summarizes the 13 tests conducted on the HWPW panels.  The tests in Table 1 are listed and numbered in chronological order, and the table shows the test date, the chamber size (large or small), the nominal and actual T and RH conditions, the HCHO emission in ppm, and notes concerning each test.  As described in section 3.1.1, the large chamber tests were conducted singly, whereas the small chamber tests were conducted as simultaneous triplicate tests.

Table B.1. Chamber tests conducted on a Commercial Veneer Core Hardwood Plywood 
                                  Test Number
                                     Date
                                    Chamber
                                 Nominal T/RH
                                    (°C/%)
                                 Actual T /RH
                                    (°C/%)
                                 HCHO Emission
                                     (ppm)
                                     Notes
                                       1
                                    8/2/12
                                   Large[1]
                                     25/50
                                   25.6/49.1
                                     0.080
Invalidated  -  48 hour conditioning
                                       2
                                    8/2/12
                                   Small[2]
                                     25/50
                                   25.3/50.0
0.117

                                       3
                                    8/2/12
Small
25/50
                                   25.1/50.0
0.110

                                       4
                                    8/2/12
Small
25/50
                                   25.5/50.2
0.110

                                       5
                                    8/8/12
                                     Large
                                     30/85
                                   30.0/85.1
                                     0.220
Invalidated  -  48 hour conditioning
                                       6
                                    8/14/12
Small
25/50
                                   25.5/51.0
0.123

                                       7
                                    8/14/12
Small
25/50
                                   25.1/50.1
0.120

                                       8
                                    8/14/12
Small
25/50
                                   25.4/50.9
0.120

                                       9
                                    8/17/12
Small
30/85
                                   30.0/90.2
0.166

                                      10
                                    8/17/12
Small
30/85
                                   30.1/86.9
0.162

                                      11
                                    8/17/12
Small
30/85
                                   30.6/92.2
0.155

                                      12
                                    8/24/12
                                     Large
                                     25/50
                                   26.1/50.5
                                     0.040
Repeat of Test 1  -  7 day conditioning
                                      13
                                    9/26/12
                                     Large
                                     30/85
                                   30.1/85.1
                                     0.199
Repeat of Test 5  -  7 day conditioning
[1]:  ASTM E1333
[2]:  ASTM D6007

As a comparison of large and small chamber results at the same T and RH conditions, it was intended that Test 1 would be compared to Tests 2 to 4, and that Test 5 would be compared to Tests 9 to 11.  Tests 6 to 8 were small chamber tests conducted as part of the planned T/RH test matrix, and are replicates of Tests 2 to 4. However, the HWPW panels were only conditioned for 48 hours prior to Tests 1 and 5, rather than for 7 days as required in the large chamber method, and consequently, those two tests were invalidated and were repeated as Tests 12 and 13, respectively.

Table 1 shows that after 7 days of conditioning at the standard T and RH conditions of 25 °C and 50 %RH (Test 12), the HWPW exhibited a HCHO emission of 0.040 ppm in the large chamber test, consistent with its designation as a ULEF product (the ULEF emission standard is 0.05 ppm).  However, in the corresponding small chamber tests (Tests 2 to 4 and 6 to 8) the HWPW exhibited HCHO emissions ranging from 0.110 to 0.123 ppm, substantially higher than both the large chamber result and the ULEF emission standard.  Samples for the small chamber tests were conditioned for 48 hours at the test conditions prior to testing.

Table 1 also shows that after 7 days of conditioning at the elevated T and RH conditions of 30 °C and 85 %RH (Test 13), the HWPW exhibited a HCHO emission of 0.199 ppm in the large chamber test.  In the corresponding small chamber tests (Tests 9 to 11) the HWPW exhibited HCHO emissions ranging from 0.155 to 0.166 ppm.  These results suggested that the HCHO emissions of the ULEF-UF HWPW increased at the elevated T and RH conditions relative to those at the standard T and RH conditions.  

Because of the unexpectedly high HCHO emissions of the HWPW in the small chamber tests at standard conditions (Tests 2 to 4 and 6 to 8), testing of this batch of panels was discontinued, and panels from another manufacturer were obtained.  The intent of this study was to assess the impact of elevated T and RH on HCHO emissions of panels that meet the NAF or ULEF definitions.  Because these HWPW panels did not meet the ULEF emission standard in the small chamber tests at the standard conditions, they were not used for evaluating the impact of elevated T and RH on formaldehyde emissions.  
