RESIN HYDROLYSIS AND MECHANISMS OF FORMALDEHYDE RELEASE
FROM BONDED WOOD PRODUCTS
George E. Myers
Research Chemist
Forest Products Laboratory
One Gifford Pinchot Drive
Madison, WI 53705-2398
Abstract
In this paper I examine the
available evidence for and against the
influence of urea-formaldehyde (UF)
resin hydrolysis on formaldehyde emission from UF-bonded wood products.
That evidence includes literature
reports from 1950 through 1984 as well
as unpublished results from my own
recent studies. I have considered
three related aspects: (a) the
chemistry and hydrolytic stability of
UF and phenol-formaldehyde (PF) materials and rates of formaldehyde
liberation from these materials,
(b) the chemistry and hydrolytic stability of formaldehyde-wood and UFwood reaction products and rates of
formaldehyde liberation from these
products, and (c) the formaldehyde
liberation behavior of UF and PF particleboards. The primary conclusions
are: (a) In an acid-catalyzed UF
board, formaldehyde can exist in a
wide variety of states, including dissolved methylene glycol monomer and
oligomers, parafonn, hexa, chemically
bonded UF resin states, chemically
bonded UF-wood states, cellulose hemiformals and formals. Each of those
states is a potential source of formaldehyde emission by evaporation
(methylene glycol) or initial hydrolysis. At present, we cannot quantify
the relative contributions of these
states over time. (b) In a basecatalyzed PF board, formaldehyde
states may include methylene glycol
monomer and oligomer, chemically
bonded PF resin states, chemically
bonded PF-wood states, and cellulose
hemiformals. Emission sources
apparently include methylene glycol,
cellulose hemiformals, and possibly
phenolic methylols. (c) Diffusion
processes very likely exert a major
influence on panel emission rates and
may involve movement of methylene
glycol in the wood's moisture or of
gaseous formaldehyde within the board
or within the board-air interface.
Introduction
General Background
Formaldehyde emission from some
urea-formaldehyde (UF) bonded wood
products has been recognized for a
number of years as a potential source
of indoor air pollution, leading to
inhabitant discomfort and possibly to
health problems. In response to a
need expressed by industry representatives for an independent evaluation
and summation of data from diverse
sources, I have prepared a series of
six critical reviews of the literature
on different technical and technological aspects of this problem. This
paper is the last of those critiques
and concerns the extent to which resin
hydrolysis in the cured board contributes to formaldehyde emission from the
board. As with the other critiques
(75-79), this one is based on a bibliography (80)1 derived from several
sources; it covers the period from
1950 through 1984. In addition, this
critique also includes considerable
unpublished data from my own work.
Note that a useful earlier review of
some related aspects of this hydrolysis question is available (64).
Over the past decade or so, great
progress has been made (21,63,106) in
improving the formaldehyde emission
from wood products such as particleboard, hardwood plywood paneling, and
medium-density fiberboard. Beneficial
steps have included reducing the
Formaldehyde-to-urea (F/U) mole ratio
impregnating the wood furnish
(substrate) with a formaldehyde
scavenger having hindered access to
the UF adhesive (78), and treating
boards with formaldehyde scavengers
and/or barrier coatings after manufacture (79). For example, many
plants in Europe now produce particleboard, that meets the German E-l
standard recommending large test
chamber formaldehyde levels of
1
Copies of the bibliography may be
In: Christiansen, Alfred W.; Gillespie, Robert;
requested from the Formaldehyde
Myers, George E.; River, Bryan H., eds. Wood
Institute, 1075 Central Park Avenue,
adhesives in 1985: Status and needs.
Scarsdale, NY 10583.
Proceedings 47344 of a conference; 1985
May 14-16; Madison, WI. Madison, WI: Forest 119
Products Research Society; 1986: 119-156.
<0.1 ppm (29). The United States wood
products industry is now producing
particleboard and hardwood plywood
paneling that meet the recently
imposed Housing and Urban Development
product standards (20) aimed at maintaining formaldehyde levels in new
mobile homes < 0.4 ppm.
Despite this practical progress,
great uncertainty still exists as to
the precise mechanism by which formaldehyde is held within a board and
slowly released as a gas to the atmosphere. Historically, the emission
potential of a board has been thought
to be governed, particularly in a
board's early life, by the board's
so-called "free" formaldehyde content
(82). "Free" formaldehyde is presumed
to derive from the excess formaldehyde
present in the UF resin. It exists in
some ill-defined, relatively loosely
bound states within the board, states
whose stabilities are sensitive to
temperature and humidity. At high
resin F/U ratios, the "free" formaldehyde content and board emission rate
fall rapidly after pressing and later
decrease more slowly. The "free"
formaldehyde content and board emission rate are lower after pressing
when using resins with F/U ratios
approaching 1.0, and they decrease
more slowly with time. What has never
been clear, however, is whether actual
UF resin hydrolysis, with attendant
formaldehyde production, causes a significant amount of the board's emission, and if so, at what point in the
board's life that occurs.
The question of the contribution
of UF resin hydrolysis to board emission is not a trivial one. If resin
hydrolysis contributes significantly
to emission, then, in principle, the
board would retain the potential to
emit during its useful life, in contrast to the situation in which all
the emission results from "free"
formaldehyde. In the former case,
efforts to minimize emission must be
directed toward resin stabilization
and/or to ensuring that incorporated
formaldehyde scavengers retain their
effectiveness at low formaldehyde
activities for the board's entire
useful life. Another consequence of
continued resin hydrolysis is possible
limits on the durability of UF bonded
products; in this case improvement may
be expected from more stable resins.
Objective and Approach of Paper
The overall objective of this
paper is to define the extent to which
board formaldehyde emission is controlled by resin hydrolysis or other
processes. To that end, I present the
results of an examination of the relevant literature plus the results of
recent experiments carried out at the
Forest Products Laboratory (FPL). The
literature review was not intended to
be an encyclopedic presentation of
literature results, but rather a brief
summary of primarily those findings
that I consider most relevant to the
issue. Experimental details of the
recent FPL testing are in the Appendix
along with explanations of calculation
procedures.
In addition to UF systems, this
paper includes a more abbreviated
treatment of phenol-formaldehyde (PF)
systems. There are two reasons for
this inclusion. First, PF-bonded
boards have been considered as one
possible lower-emitting substitute for
UF boards, and they thus provide a
useful baseline system for comparison.
Second, the PF chemistry is very different from that of the UF system and
comparisons between the two may yield
insights into their emission
mechanisms.
I have divided the overall presentation into three related aspects,
each addressing specific questions as
outlined in the following:
Part 1. Formaldehyde-urea and
formaldehyde-phenol states.
Do UF systems hydrolyze to produce
formaldehyde?
b. Are hydrolysis rates sufficient to
account for formaldehyde emission
rates from UF-bonded boards under
reasonable exposure conditions
(e.g., 20-40°C, 30-80%, relative
humidity (RH), pH 2-4)?
c. What are the answers to the above
when applied to phenol-formaldehyde systems in alkali?
a.
Part 2. Formaldehyde-cellulose and
resin-cellulose states.
a. What chemical products (chemical
states) are formed when wood is
exposed to formaldehyde and/or
resin molecules under the temperature and acidity conditions of
adhesive bond formation?
b. Are those states susceptible to
hydrolysis, thereby liberating
formaldehyde?
c. Are hydrolysis rates of those
states sufficient to account for
board emission rates?
120
Part 3.
boards.
a.
Formaldehyde emission from
(5)
Can we clearly distinguish formaldehyde emitted from boards via
cured resin hydrolysis mechanisms
from that emitted via other likely
mechanisms?
(6)
Part 1. Formaldehyde-Urea and
Formaldehyde-Phenol States
Chemical Background
Formaldehyde-urea states. That
UF materials hydrolyze to produce
formaldehyde is incontrovertible.
Over the past 30 years, investigators
have extensively examined the structure of low molecular weight UF compounds and the physical chemistry of
their formation and degradation in
aqueous systems. Earlier work by
de Jong and de Jonge (17-19),
Landquist (54,55) and others (16,25,
100) used classical solution techniques to study the kinetics and
equilibria of formation and hydrolysis. Subsequently, chromatographic
(24,53,56) and nuclear magnetic
resonance (NMR) (27,49,81,90,98,107)
techniques were applied. In effect,
many of the reactions that form UF
products may be formally regarded as
reversible condensation processes that
eliminate water in the forward direction and, therefore, involve hydrolysis in the reverse direction. For
some of the lowest molecular weight
products, for example, the reversible
reactions are:
(1)
(2)
(3)
(4)
Elevated temperature and neutral
or acid pH lead to polymer formation
primarily via forward reactions (3)
through (6), the last being responsible for branching and the possibility
of chain crosslinking. Resin cure is
normally conducted at temperatures
above 120°C and pH below 5 and is
presumed to leave the following
moieties (formaldehyde states) that
are subject to hydrolysis, with direct
or eventual liberation of formaldehyde
(R = H or -CH2):
Because most, if not all, of the
reactions leading to these formaldehyde states are reversible, the use of
acid catalysts to hasten adhesive bond
cure unfortunately increases the rates
of hydrolysis and formaldehyde liberation. Some evidence also exists for
acid-catalyzed hydrolysis of the amide
C-N bond (2,110), in which case the
product is not formaldehyde but an
amine and CO,. However, the extent of
this amide hydrolysis has never been
made clear.
As partial justification for the
above statements and to illustrate the
intensive investigation of the UF system, I will briefly summarize two
studies. Slonim and coworkers (98)
employed C-13 NMR to measure the
equilibrium constants (K) at pH 7 and
ambient temperature for several
formation/hydrolysis reactions,
including five reactions to produce
the mono, di, and tri methylolureas.
Corrected for differences in the number of reactive sites per molecule,
the K values in L/mole are: 350 for
formation of monomethylolurea (UF)
121
from urea (U) and formaldehyde (F);
300 for N,N’-dimethylolurea (FUF)
from UF and F; 60 for trimethylolurea
(FUF2) from UF and F; and 40 for FUF,
from FUF and F. At equilibrium,
therefore, species concentrations
decrease as the amide hydrogens on
urea are replaced by formaldehyde, and
the effect is particularly strong for
replacing the second hydrogen on a
given nitrogen. Similarly, the K for
Reaction 4 is approximately 2 and that
for formation/hydrolysis of the hemiformal of monomethylolurea is approximately 13, indicating even less
tendency to form those species.
In contrast to Slonim et al’s
work, Braun and coworkers (8) studied
the acid-catalyzed hydrolysis of carefully characterized methylene-diurea
(UFU). In this case the initial
hydrolysis products (Reaction 3) are
urea and monomethylolurea, the latter
then degrading to urea and CH2O
(Reaction 1). At 80°C and pH 5, about
20 percent degradation to CH2O is
observed after 3 hours, whereas at
pH 3 an equilibrium state is reached
at about 40 percent degradation in
only 30 minutes. Applying de Jong’s
activation energy of 19.5 Kcal/mole-K
(15) to the pH 3 results at 80°C
yields a degradation rate of approximately 0.5 percent per hour at 25°C
and pH 3. Thus, even the methylene
link, customarily presumed to be quite
stable in cured resins, will degrade
at a significant rate when dissolved
in aqueous acid at 25°C.
Formaldehyde-phenol states .
Whether these states undergo sisnificant hydrolysis at relevant exposure
conditions to liberate formaldehyde
is not clear. In an alkaline resole
system cured between 130 and 200°C,
the major formaldehyde states are
thought (48) to be residual methylols
and methylene bridges between phenolic
rings, e.g.,
(ML85 5504)
At temperatures not far above 130°C
and for short cure times, significant
amounts of methylene ether bridges may
also remain; e.g.,
(ML85 5505)
Formalin-phenol solutions
contain hemiformals of the type
φ - OCH2O(CH20)nH with n = 0 to 3 (49).
Similarly, formalin benzyl alcohol
solutions contain the hemiformals
HO - φ - CH2O(CH2O)nH, with n = 0 to
3. While such states probably will
not withstand typical alkaline cure
conditions, they may re-form after
cure of a high formaldehyde PF resin.
Finally, recent solid-state
C-13 NMR studies (59) have confirmed
the likelihood of other states, e.g.,
(ML85 5503)
Except for the hemiformals,
none of the above states is expected
to be highly, or even moderately,
unstable. For example, the durability of PF-bonded wood joints is
limited by the endurance of the wood,
not the resin (68). This recognized
stability and the generally much
lower formaldehyde emission from
PF-bonded products (28,74,94) have
exempted the PF system from any
intensive study of its inherent
tendency for formaldehyde liberation
under near ambient exposure conditions. However, 1.6 and 13.0 mole
percent formaldehyde was liberated
from o- and p-methylolphenols, respectively, after 6 hours in boiling
water (8), indicating the possibility
of reversion of the formaldehyde ring
substitution reaction. For cured PF
resins 3.6 percent of the resin weight
was lost as formaldehyde after about
122
5 hours in boiling water, and corresponding figures for 0.1N NH4OH and
(29) H2SO4 were 2.3 and 0.7 percent
Rosenberg studied the rate of
formaldehyde release from a cured PF
resin in 1N H2SO4 at 78°C and observed
a rapid initial release, followed by a
slower first-order kinetics region
(95); the first-order rate constant
decreased with increasing resin cure.
Summers found that cured novolak
resins can be solubilized by digestion
in 3 percent alkali at 280 to 320°C
for several minutes (105). My own
experiments on cured PF resin at more
realistic conditions (Fig. 1) showed
measurable amounts of liberated formaldehyde and a strong increase with
humidity.
Thus, the limited available data
indicate that formaldehyde liberation
from cured PF resins can indeed occur.
However, little is known about the
underlying chemical mechanisms.
Other formaldehyde states . In
addition to the above UF and PF formaldehyde states, two others must be
considered. Formaldehyde can undergo
reaction with itself to produce polymer such as paraform (Reaction 7), and
it may react with the ammonia that is
often present from UF cure catalysts
to yield hexamethylenetetramine (hexa,
Reaction 8). Both materials will
liberate formaldehyde by reversion
(hydrolysis).
(7)
(8)
Both the rate and extent of paraform
dissociation (hydrolysis) are strongly
pH dependent, being minimal between
pH 2.5 to 4.5 and increasing exponentially below and above that range
(114). Formation of paraform, therefore, seems to be much less likely in
an alkaline PF system than in a moderately acidic UF system. Although the
potential for hexa formation exists
only for UF systems, hexa is relatively unstable in acid; if formed
during UF cure, hexa may be relatively
shortlived in some UF systems.
Formaldehyde Liberation Rates
I conclude that formaldehyde can
indeed be produced by hydrolysis of UF
and other (paraform, hexa) states and
by some unknown mechanism from PF
systems. The question next arises
whether known hydrolysis rates of UF
and PF systems are sufficient to
account for a significant portion of
observed or permissible formaldehyde
emission rates from particleboards.
For example, if formaldehyde were
known to be liberated by hydrolysis
of all UF states at rates far below
permissible board emission rates,
then we could presumably ignore UF
resin hydrolysis as a potential source
of board emissions.
To attempt an answer to this
question I transformed existing data
on hydrolysis rates of model compounds
and cured resins into formaldehyde
emission rates from a hypothetical
particleboard by the following steps:
1. Converting measured liberation rates for model chemicals and
resins into units of mg liberated
formaldehyde per g dry board per
hour.2 The calculation assumes a
board of 16 mm thickness and
0.65 g/cc density, with 7 percent
resin solids. It also assumes that
the influence of the board matrix on
formaldehyde liberation is limited to
the board's composition and geometry.
2. Selecting a standard state
for hypothetical UF boards of
(a) 25°C; (b) 50 percent RH (water
activity 0.5); and (c) pH 3.0 or the
actual acidity of bondline, solid
resin, or solid model compound. For
hypothetical PF boards the same
standard state was used except
for a pH of 10 or actual basicity
of bondline, etc.
3. Emphasizing data close to the
standard conditions and making
approximate corrections to those
conditions (parenthetic rate values
in Table 1) using factors from the
actual reference whenever possible.
4. Defining maximum allowable
board emission rates of 9 x 10-5 mg/g
board-hr for the HUD large chamber
standard of 0.3 ppm (20) and 2 x
10-5 mg/g board-hr for the German E-1
(29) of 0.1 ppm (see Appendix 3C for
method of calculation).
Table 1 summarizes some transformed hydrolysis and formaldehyde
liberation rate information from the
literature and from my own recent
experiments (Figs. 1,3,5,7).2,3
Values corrected to standard conditions are in parentheses. Before
examining those transformed data,
2
See Appendix 3 for methods of
calculation.
3
See Appendix 1a for methods of
measurement.
123
however, some limitations should be
recognized.
1. Many of the corrections to
the specified standard conditions are
quite approximate. In addition, the
selection of pH 3 as standard for
UF's and pH 10 as standard for PF's
is a somewhat arbitrary compromise
because of variations in wood acidity
and catalyst levels among boards, plus
uncertainty about correlations between
measured board or resin
pH's and
actual bondline PH.4 Thus it would be
best to regard most of the corrected
rate values (parenthetic values) as
significant within perhaps one order
of magnitude.
2. In many cases, reported
first-order kinetic rate constants
were determined from measurements
during early stages of hydrolysis.
Because my calculated liberation rates
assume the same rate constant throughout the material's degradation course,
some rates calculated over a longer
time may be less correct. Similarly,
most, if not all, of the experiments
were conducted under conditions in
which the formaldehyde vapor pressure
or concentration was maintained at
very low levels. Consequently formaldehyde liberation rates were not
affected by back-reactions and were
maximized. In a board, however, it is
conceivable that these maximum
hydrolysis rates are not achieved due
to locally high formaldehyde concentrations. Moreover, actual board
emission rates may be greatly
lessened by formaldehyde-wood interactions and diffusion effects. By the
same token, the degrading material
will then act as a formaldehyde source
for longer times.
3. As will be noted later,
hydrolysis rates for dissolved species
are very likely to be higher than for
analogous species present in an
insoluble, crosslinked resin.
With these constraints in mind,
examination of Table 1 and the relevant figures leads to the following
conclusions for the various systems.
4
For example, suppose a pH of 4.0 is
measured for a ground UF particleboard
as a 1:10 slurry in H2O (see
Appendix 1b). The same board conditioned at 50 percent RH would be in
contact with nearly 100-fold less H2O.
For an acid such as HCl, this might
mean the actual bondline pH at 50 percent RH is closer to 2.0. For acetic
acid, however, the corresponding
change might be more like pH 4 to 3.
UF Models.
follows:
Conclusions are as
1. Initial rates for pure compounds from solution data can be very
high relative to the board standards
of 9 x 10-5 or 2 x 10-5 mg CH2O/g-hr.
Even if the particular formaldehyde
state represented by a model compound
were present as only a fraction of the
resin, initial emission rates could be
well above the allowable levels. This
is illustrated in Figure 2 for -NCH2OH
assumed present to the extent of
10 percent of the resin's formaldehyde, for -NCH2N- present at 80 per
cent, and for -NCH2OCH2N- present at
10 percent.
2. At such high early rates,
however, formaldehyde sources are not
long lived and may not constitute an
important direct emission source after
a few weeks. On the other hand,
short-lived primary sources might
still result in longer term board
emission because their liberated formaldehyde could react with wood substance to yield secondary emission
sources. This necessitates, of
course, that liberation rates from
secondary sources are lower than from
primary sources due to inherent stability or to diffusional delays within
the board.
3. In contrast to the situation
where initial rates are very high, the
liberation rate from a moderately
stable material might be maintained at
rates above allowable standards for a
much longer time. Thus, intermediate
stability might be less desirable than
low or high stability.
4. The durability of UF-bonded
wood products may be limited by both
the inherent hydrolytic sensitivity of
the resin and bond rupture caused by
swelling and shrinkage stresses (see
discussion in Part 3). Both events,
however, may eventuate in formaldehyde
liberation. Consideration of durability should therefore allow us to
estimate some formaldehyde liberation
rates for comparison with model compound and resin data. Let us assume
that board failure occurs due to rupture of one chemical bond type which
liberates one molecule of formaldehyde. Consider two cases: (a) a
rather conservative case in which only
5 percent of the bonds rupture in
50 years, i.e., probable board durability greater than 50 years; and
(b) a much less conservative case in
which 30 percent of the bonds rupture
in 20 years, i.e., probable failure in
20 years or less. Case (a) leads to a
124
first-order rate constant of 3.3 x
l0-11 s-1 and a liberation rate, given
by the "high durability" curve in
Figure 2, below the E-1 allowable
line. Case (b) leads to a first-order
rate constant of 5.7 x 10-10 s-1 and a
liberation rate, given by the "low
durability" curve in Figure 2, above
the HUD allowable line. Note that
both rate constants are several orders
of magnitude smaller than those in
Table 1 for UF model compound
degradation.
5. The available data on formaldehyde liberation from paraform
(entry A7) and hexa (entry A8; see
also Ref. 114) are not directly relevant to the possible contribution of
these states to liberation from cured
resin samples. At pH 8, paraform dissociates and vaporizes completely
within 2 days (entry A7 and Fig. 3)
but at pH 2 to 3, its dissociation
rate may be one or two orders of magnitude less (114). Hexa, on the other
hand, will hydrolyze approximately an
order of magnitude faster at pH 2 to
3 (114) than at pH 8 (entry A8 and
Fig. 3). Data are needed on the
stability of hexa and paraform at pH 2
to 3.
6. Formaldehyde liberation rates
calculated from model compound degradations in aqueous solutions obviously
are more than enough to account for
short-term (weeks) board emission.
But the dropoff with time is much too
rapid to account for longer term board
emission and is inconsistent with
durability considerations; moreover,
the rates are well above those
observed for cured resins. I therefore conclude that such homogeneous
solution data are not quantitatively
applicable to the heterogeneous solid
state/aqueous liquid/air system that
exists with actual cured resins or
bonded wood. This conclusion perhaps
especially applies to methylene links
(NCH2N) which should constitute a
major part of the cross-linked network
backbone and whose formaldehyde liberation rates (Table 1 and Fig. 2) are
much too high. In an actual cured
resin, many of these links may exist
within crystallites (104) or tightly
crosslinked non-polar regions and are
thus relatively inaccessible to water
and acid. Moreover, methylene links
with tertiary nitrogens are presumed
to be less hydrolytically sensitive
than those with secondary nitrogens
(64,73).
Therefore, and not unexpectedly,
this re-examination of UF model
compound solution hydrolysis data
confirms the qualitative possibility
of significant formaldehyde liberation
from cured resins and, thereby, the
necessity for further quantitative
consideration of liberation rates by
cured resins under realistic
conditions.
Cured UF Resins . The following
comments are primarily based on
entries B7 through B8 of Table 1
because those liberation experiments
most nearly approximated conditions in
actual bondlines (see column 2 and
Appendix 1 for experimental details)
and because the liberation rates were
directly obtained from slopes of
experimental data plots, not from rate
constants that are assumed to apply
throughout the degradation process.
1. Liberation rates cover a wide
range, corrected rate values at one
day differing by a factor of twenty.
Liberation rate is decreased
by lowering F/U ratio (entry B7a
versus B8a and B7e versus B8c), by
lowering humidity (entry B7a versus
B7b, and Fig. 4) and by extending
cure (entry B7a versus B7c, B8a versus
B8b).
3. It is possible that a major
portion of the liberated formaldehyde
might be due not to hydrolysis of
chemical states but to removal of
physically sorbed formaldehyde at
rates controlled perhaps by diffusion
within the resin particles. However,
in my view several findings argue in
favor of hydrolysis as the major
source:
(a) Experiments B4 through B8 all
employed -80 mesh (<l80 µm) powders, and other experiments
demonstrated little or no influence of particle size on liberation rates by fractions below
that size (73).
(b) Very strong increases in liberation rate occurred with
increasing humidity (entry B7a
versus B7b, and Fig. 4).
(c) Only a small decrease in rate
resulted from vacuum pumping at
ambient temperature for 17 days
(entry B7a versus B7d). The
observed decrease could be due to
removal of some physically sorbed
formaldehyde or to dissociation
of some chemical formaldehyde
states, e.g., methylols or paraform.
(d) Significant liberation rates
remained after lengthy post-cure
under vacuum (entry B7a versus
B7c and B8a versus B8b).
(e) Lengthy toluene elution
(Appendix 1e) of vacuum-pumped
UF and PF resins at a ambient
125
temperature (Fig. 5) removed
amounts of formaldehyde that
were significant fractions of
the respective Perforator
(Appendix 1c) values (about 40%
for the UF and 20% for the
PF) but were much smaller fractions of the amounts liberated
by exposure to 80 percent RH
(about 7% for both resins).
Elution from the PF resin hardly
changed between 10 and 50 days
while that from the UF resin continued to increase at a slow rate
after 50 days. In either case
the elution may be removing
physically sorbed formaldehyde,
dissolving a formaldehyde compound (paraform ?) or causing,
particularly in the UF, dissociation of chemical states by
imposing a very low concentration
of formaldehyde.
4. Cure in the presence of only
25 percent wood flour leads to significantly greater liberation rates
compared to neat resin (entry B7e
versus B7f, B8c versus B8d, and
Fig. 6). Moreover, the resin/wood
composites maintain higher rates for
longer times, with the consequence
that estimated times to reach the
allowable rates are greater than for
neat resin. It is not clear whether
the wood interferes with resin cure or
acts as a formaldehyde sink during
cure and as a secondary source thereafter. However, Perforator values (in
mg CH2O/g resin) are 3.5 and 3.3 for
F/U 1.6 resin and resin/wood samples
(entries B7e and B7f) respectively,
and 1.5 and 1.2 for F/U 1.2 resin and
resin/wood samples (entries B8c and
B8), respectively. The lower Perforator values-- and nearly identical
pH's--for the resin/wood composites
may indicate that cure interference
did not occur; if so, the increased
liberation in the presence of wood
must be due to hydrolysis of formaldehyde-wood states that are resistant
to the Perforator conditions (boiling
toluene, 2 hours).
5. By integrating the rates, one
can estimate the fractional loss of
formaldehyde that a given sample has
suffered by the time its liberation
rate reaches the allowable levels.
The F/U 1.6 resin (entry B7a), for
example, would lose about 18 percent
of its original formaldehyde before it
reached the allowable level, whereas
the F/U 1.2 resin (entry B8a) would
lose about 6 percent. Such levels may
or may not be significant for the
integrity of a bonded product,
depending upon the relative contributions to the loss of formaldehyde by
nonstructural moieties (hexa, paraform, -N(CH2O)nOH) and by structural
moieties (-NCH2N- and -NCH2OCH2N- in
effective network chains). Moreover,
because of these varied sources, it
seems unlikely that current spectroscopic techniques can adequately
define the structural changes that
may occur in similarly aged resins
(12,43,57).
6 Liberation rates transformed
into hypothetical board emission rates
(Fig. 7) are significantly above the
board allowable levels and fall rather
rapidly, the F/U 1.2 sample reaching
the HUD standard in approximately
1 month. These high liberation rates
contradict the general experience that
boards made with low F/U resins
(< 1.2) can and do meet the HUD
standard (32,75). This discrepancy
possibly results from greater hydrolysis rates of neat resins due to
absence of a wood sink for excess
acid and/or slower formaldehyde emission from real boards due to diffusional restrictions. Whatever the
details, cured UF resins clearly have
the potential for significant contribution to board emission.
To summarize this section on UF
resins, the salient points to be noted
are:
1. Hydrolysis of chemically
bonded formaldehyde states in neat
cured UF resins is, in my view,
responsible for the major portion of
liberated formaldehyde from those
materials.
2. Resin hydrolysis is potentially a significant contributing
mechanism towards board emission.
3. Data on neat resins do not
permit us to quantify resin contributions towards board emission.
Phenol-Formaldehyde (PF) Models
and Cured Resins. In contrast to UF
resins, phenol-formaldehyde resins are
normally-cured under alkaline conditions for somewhat longer times and at
higher temperatures. The more
rigorous cure conditions plus the
inherently greater hydrolytic stability of the linkages present in PFs
result in a cured product whose potential formaldehyde liberation rate
should be much less than that of a UF
system. In fact, the general experience with PF particleboard shows that
formaldehyde emission levels are as
little as one-tenth of the levels of
126
UF boards made from medium F/U ratio
resins (28,74,94).
Whether the limited model compound data (Table 1, entries C1 to C3)
agree with this conclusion is unclear
because the completeness of degradation during the rather short exposures
was not described in the original
references and because the conditions
were far removed from our standard
condition for PF resins in boards
(25C/50% RH/inherent alkalinity or
pH 10). However, the data do indicate
a significant potential for formaldehyde liberation. For example, if this
degree of degradation were spread over
50 to 500 days at standard conditions
with a cured resin that still contained 10 percent of its original
methylol groups, the corresponding
formaldehyde liberation rate would be
approximately 2 x 10 to
2 x 10-5 mg/g bd-hr, i.e., close to
the allowable values. (The assumption
of 10 percent residual methylols may
be conservative in view of Fyfe
et al's [35] finding that after
160 minutes at 180°C, the residual
CH2/CH2OH ratio in a PF resin was
only 2.9.) The PF resin data of
Freeman and Kreibich (33, entries Dla,
Dlb) were also obtained at rather
extreme conditions but indicate a
finite amount of liberated formaldehyde. Here, spreading that amount
over 50 to 500 days yields liberation
rates in the range of 10-4 to
10-3 mg/g bd-hr.
My own experiments on a cured PF
resin (entries D2a, b, c, and Fig. 1)
were much closer to standard conditions and warrant the following
comments:
1. A small, but finite, formaldehyde liberation occurs. At 80 percent RH, the rate for the PF
(entry D2a) after 1 day was approximately a seventh of that for the F/U
1.6 UF resin (entry B7d) but nearly
equalled that for the F/U 1.2 UF resin
(entry B8a). The initial corrected
rate was well above allowable levels
but dropped below them within 2 to
6 weeks (Fig. 7).
Liberation rate is strongly
decreased by lower humidity and
extended cure (Fig. 1 and entries D2).
3. Similar to studies on the UF
resins, several findings argue against
the likelihood that all of the
liberated formaldehyde derives from
physically adsorbed material. These
findings include retention of finite
liberation rates after 20 days of
vacuum pumping (entry D2a) or after
exposure to 200°C for 30 minutes in
vacuum (entry D2c); the very strong
influence of humidity (entry D2a versus entry D2b); and the elution by
toluene of only a small fraction of
the formaldehyde liberated at high
humidity (Fig.5).
4. Hexa is not a formaldehyde
source here because no ammonia was
present. Paraform seems an unlikely
source because the alkaline pH should
greatly increase its hydrolysis rate
above that in entry A6 where paraform
was totally vaporized within 2 days.
Solid state carbon-13 NMR spectra on
sample D2a showed no detectable formaldehyde bands (58), although Fyfe
et al. (35) reported trace amounts
still present in similar spectra after
curing 160 minutes at 180°C.
5. In view of point 4 and the
reported formaldehyde liberation from
methylolphenols (entries C1, C2), it
is possible that residual phenol
methylols account for observed formaldehyde from this PF resin. However,
we cannot rule out the possibility of
liberation due to degradation of
residual methylene ether linkages
between phenols.
Conclusions--Part 1
On the basis of information in
the literature plus the results of my
own experiments, I offer the following
conclusions regarding resin contributions to board emission.
1. There is little doubt that
most, perhaps all, formaldehyde states
in a cured UF resin are susceptible to
hydrolytic degradation.
2. Most of the hydrolysis reactions in UF resins are reversible but
can ultimately lead to formaldehyde
liberation if allowed to proceed.
3. Other sources of formaldehyde
liberation may coexist with a UF
resin, e.g., paraform and hexamethylenetetramine. Present data are insufficient to define the extent to which
these materials may be responsible for
observed formaldehyde liberation rates
from cured UF resins and boards, and
this reservation should be borne in
mind when considering the succeeding
remarks.
4. Initial formaldehyde liberation rates observed for neat UF resins
that are cured at conditions comparable to particleboard press conditions can be significantly above
levels that are allowable for particleboard under current emission
standards (HUD, 0.3 ppm; German E-1,
0.1 ppm).
127
5. The fact that boards manufactured with similar UF resins
(F/U 1.2) can meet the HUD standard
indicates that the wood system
strongly mitigates the influence of
resin hydrolysis on board emission.
As a consequence, quantitative statements about the actual contribution
of UF resin hydrolysis to board emission cannot be made, although it seems
likely that contribution can be significant.
6. One cured PF resin exhibits
an initial formaldehyde liberation
rate approximately equal to that from
a UF resin with F/U 1.2. Phenol
methylol and/or methylene ether links
may be the sources of liberated formaldehyde. Statements similar to those
in point 5 may also be relevant here.
Part 2. Formaldehyde-Cellulose and
Resin-Cellulose States
During hot pressing of an acidcatalyzed UF particleboard, some resin
species, formaldehyde, and acid leave
the bondline and penetrate into and
interact with the wood substrate to
yield formaldehyde states in addition
to those in the cured resin bondline.
Similarly, in an alkaline-catalyzed PF
board, additional states can result
from penetration of resin, formaldehyde, and base into the wood. These
states within the wood constitute, or
contribute to, what has been called
"free" formaldehyde. What we must
examine is the nature of these states,
their hydrolytic stability, and their
likely contribution to both short and
long term formaldehyde emissions compared to those from the resins
themselves.
To that end, I present abbreviated reviews of the chemistry of formaldehyde-wood and resin-wood reactions
and of selected literature and FPL
data to illustrate possible formaldehyde rates from the presumed formaldehyde states. As in Part 1, my intention is not to offer a comprehensive
review but merely to provide evidence
for some likely interactions in a
bonded wood board and for the possible
emission consequences of these interactions.
Chemical Background
Acid-Catalyzed FormaldehydeCellulose Reactions . The acidcatalyzed reactions of cellulose with
formaldehyde and also with a wide
variety of N-methylolureas have been
intensively investigated, primarily as
a consequence of their potential for
imparting creaseproof properties to
cotton textiles (13,93,114). Less
intensive studies have also been
carried out into the acid-catalyzed
reactions between formaldehyde and
wood or paper because of interest in
the resultant dimensional stabilization (70,103,114,115).
Within the wood substance the
formaldehyde may conceivably exist
as a solution of monomer (presumably
as HOCH2OH) or oligomer, or even as
insoluble paraform, with the monomer
and oligomer being highly H-bonded
to
water and wood substance.5 That
strong interactions occur in the
formaldehyde-water-wood system is
evidenced by slow release of formaldehyde from impregnated veneer (51) and
by wood density changes (44). In
addition, however, it is known that
formaldehyde will react with cellulosic hydroxyls at elevated temperatures
in acid to yield a hemiformal (4,5,34,
83,99,110,111,114):
(9)
The hemiformal in turn can react with
another cellulose hydroxyl to give a
formal crosslink:
(10)
More generally, a polyformal crosslink
may form via direct reaction of a
5
A 5 percent loss of a UF resin's
formaldehyde into the wood 2 corresponds
approximately to 100 mg CH O per
100 g dry wood. At 50 percent RH this
would lead to a nominal formaldehyde
concentration in water of about 1 percent. Since the water by definition
has thermodynamic activity of only
0.5, however, this nominal 1 percent
formaldehyde concentration may have an
effectively higher activity. Note
also that even at 5 percent concentration only about 20 percent of the
formaldehyde exists as oligomer or
polymer at equilibrium (114), and the
equilibrium probably would be achieved
very slowly in a board.
128
formaldehyde oligomer with cellulose
or by stepwise build-up from the hemiformal; the overall effect is:
(11)
(12)
Reactions (9) to (12) apparently take
place primarily with secondary cellulose hydroxyls (45,113). Andrews has
reviewed the evidence for celluloseformaldehyde reactions (4).
On the basis of reported amounts
of formaldehyde that can be bound (not
removed by neutral water rinsing) by
cotton, it seems likely that formaldehyde could be bound by the wood substance during pressing of a UF board
in amounts that are potentially significant for emission. For example,
1.4 percent of bound formaldehyde
resulted from treating cotton with
6.5 percent formaldehyde in water at
pH 2.2 and subsequentlyo curing for
about 4 minutes at 105 C (67). Bound
quantities increase with formaldehyde
concentration, with greater acidity,
and with higher cure temperature
(67,114). At 1.4 percent, the bound
formaldehyde would tie up only a small
fraction (<5 mole %) of the secondary
hydroxyls present in the cellulose of
a particleboard. Moreover, that
amount of cellulose binding capacity
perhaps would not be needed since it
corresponds to 10 to 20 percent of
the UF resin's total formaldehyde content; i.e., probably more than a low
F/U resin could afford to lose without
undesirable consequences for cure and
for board properties.
The combined formaldehyde-acid
permeation into the wood during
pressing could also be expected to
produce methylolation of the lignin
(96). To what extent this reaction
will compete with the above formaldehyde-cellulose reaction is not
apparent. In any event, qualitative
or quantitative information on possible rates of formaldehyde liberation
from aromatic methylols is not
available.
Base-Catalyzed FormaldehydeCellulose Reactions . We are concerned
here with base-catalyzed PF systems
and the situation is less clear than
it is for the acid-catalyzed UF system. Treatment of wood or paper with
formaldehyde under either neutral or
alkaline conditions reportedly causes
little or no change in chemical or
physical properties, in distinct contrast to treatment under acid conditions (103,114). This is taken as
evidence that only cellulose hemiformals (Reactions 9 or 11) are
produced without further reaction
between the hemiformal and cellulose
to provide formal crosslinks (103,
114). Studies of reactions between
formaldehyde and simple alcohols
verify the absence of formals in base
(14,114). With regard to reaction
with lignin, the situation is similar
to that in acid.
"Uncatalyzed, Ambient"
Formaldehyde-Cellulose Reactions . At
their natural acidities, wood and
other cellulosics sorb formaldehyde
and can then act as formaldehyde
emitters. Almost certainly, the formaldehyde state resulting from such
sorption is not cellulose formal
because the conditions are too mild
and because the formaldehyde is easily
removed by water (114 and Fig. 8).
Other alternatives include paraform,
cellulose hemiformal, and/or water
solutions of methylene glycol and
oligomers, but available data do not
allow an unequivocal choice among
these. Recent NMR experiments, for
example, indicate little hemiformal
formation between formaldehyde and
cellobiose or ethylene glycol in water
(65), but Kamath et al. found the heat
of sorption of formaldehyde on cellulose at 25°C to 100°C to equal the
heat of formation of simple hemiformals (46), and we know that lower
alcohols readily form hemiformals with
formaldehyde (114). Within a postproduction particleboard, some degree
of dynamic cellulose hemlformal
formation/hydrolysis may take place
due to the presence of acid (UF) or
base (PF) and the cellulosic degradation products resulting from hot
pressing (89).
Acid-catalyzed N-methylolureaCellulose Reactions.
That reactions
of the type shown in equation (13) do
occur between cellulose and N-methylolureas has been amply demonstrated
and, in fact, studied in detail by
those interested in textile creaseproofing (4,6,13,66,84,85,91,102,114).
129
(13)
Here, R1, R2, and R3 may be H- or
-CH2-. The reactions are reversible
and acid-catalyzed, and their rates
and equilibrium positions are affected
by the structures of R1, R2, and R3.
If any of the R's possesses an Nmethylol group, then additional reactions with other cellulose hydroxyls
may occur, leading to cellulose crosslinking. Because R1, R2, and R3 will
possess N-methylols and/or active N-H,
bridging between substrate and the
resin matrix is even more likely.
One illustration will suffice to
demonstrate the possible extent of
bonding to cellulose under conditions
that approximate those during UF particleboard hot pressing (100 to 160°C,
3 to 6 minutes, acid catalysis).
Cotton was treated with an aqueous
solution of N, N'-dimethylolurea at
NH4Cl concentrations from 0 to 2 per
cent, subsequently dried at 115°C, and
heated at 150°C for 10 minutes (89).
The amount of nonwashable material
bound to the cotton varied from 2 to
6 percent as the NH4Cl concentration
increased. Whether such chemical
reaction does indeed occur between UF
adhesive and wood substrate during
bond formation has not been definitely
established. However, there is every
reason to expect numerous such resincellulose reactions within the resinwood interphase region, and we need to
consider their possible contributions
to board formaldehyde emission.
Base-Catalyzed PF Resin-Cellulose
Reactions. As with the UF resins,
evidence for chemical bonding between
PF resins and wood components during
board pressing is not definitive.
However, Allan and Neogi showed reaction between lignin and a PF model
compound under alkaline conditions and
elevated temperature (1). Thus, there
is good reason to expect the formation
of methylene and/or methylene ether
links between PF and lignin molecules;
such states should possess comparable
stability to the analogous ones in the
resin itself, however, and do not require explicit consideration. I have
also chosen not to consider possible
formals produced by reaction between
phenolic methylols and cellulose
hydroxyls because those are relatively
unlikely in strong base and because I
am not aware of any relevant data.
Conclusions . In view of the discussion thus far in Part 2, I conclude
that the following events are sufficiently probable and important to
warrant examining their consequences
for formaldehyde emission from boards:
1. Some undefined portion of a
UF resin's formaldehyde (as hydrated
monomer or oligomer) will permeate
into the wood during hot pressing and,
catalyzed by acid from the resin
and/or from the wood itself, will produce bound hemiformal and formal
states by reaction with cellulose.
2. UF resin species will penetrate the wood to a lesser extent but
can also undergo acid-catalyzed reaction with cellulose to form
N-methylolethers.
3.
During hot pressing of a PF
board,formaldehyde and alkali will
penetrate into the wood and react with
cellulose hydroxyls to yield bound
hemiformal states but probably not
formals.
4. Some of the formaldehyde that
penetrates into the wood during hot
pressing of a UF board may homopolymerize, but this is unlikely in an
alkaline PF board.
5. After board production, some
cellulose hemiformal may be forming
and hydrolyzing within the board.
Formaldehyde Liberation Rates
Many of the reactions discussed
in Part 2 (reactions 9-13) are reversible condensations that produce water
as well as the particular formaldehyde
states. Therefore, each of those
formaldehyde states is hydrolyzable,
with the potential for formaldehyde
liberation. Moreover, as with most of
the reactions in UF systems, the rates
of these forward and reverse reactions
are strongly influenced by acidity
(84,85,110).
Table 2 and Figure 2 present
selected data illustrating the stability and formaldehyde liberating
tendencies of several celluloseformaldehyde compounds or states.
As with the data in Part 1, the
cellulose-related values have also
been transformed to liberation rates
from a hypothetical particleboard
and,
in some cases, have been corrected to
standard conditions. They are subject, therefore, to the same limitations discussed in Part 1.
Within these constraints, the
Table 2 values plus other observations
in the literature lead to the following comments.
130
Cellulose - CH2O Models and
alkaline conditions (103,114). ThereCompounds . Conclusions are as follows: fore, we should expect their contribution to PF board emission to be
1. The corrected acid-catalyzed
slight. On the other hand, cellulose
hemiformals apparently can be produced
formaldehyde liberation rates from
cellulose formal models are within an
under alkaline conditions and are more
order of magnitude of those for actual stable to hydrolysis in base than in
cellulose-formals (entries Al to A3
acid (114). Quite possibly, then,
versus B2b and B3b). The latter
cellulose hemiformals account for some
behavior may approach that of similar
of the small but finite formaldehyde
states in a board since both situaemission by PF boards.
tions involve nonhomogeneous systems,
Cellulose-UF Resin States . The
solid phase in contact with
cellulose-UF reaction products of conliquid.
cern are N-methylolurea ethers whose
2. Inter-cellulose polyformal
hydrolysis yields first the correbridges hydrolyze in acid 3 to 4 times
sponding N-methylolurea (reverse of
faster than bridges composed of single
Reaction 13), which then can decompose
formal links (entry A2 versus Al and
into the substituted urea and CH2O
B3b versus B2b).
(reverse of modified Reaction 1). It
3. The corrected cellulose
has been clearly established that the
formal liberation rates are an order
rate constants for the first step are
of magnitude less than initial rates
generally much greater than the rate
from realistically-cured UF resins
constants for the second step (3,85,
(c.f. entries B7a, c and B8a, c in
110). (This, of course, is just the
Table 1). Cellulose formals are,
reverse of the situation with cellutherefore, relatively stable states.
lose formals and hemiformals.) There4. Assuming that during board
fore, formaldehyde liberation rates
pressing 2 percent of the resin's
from the cellulose-UF reaction
formaldehyde yields cellulose formals,
products are controlled not only by
subsequent hypothetical board emission
those products (ethers) but also by
rates at 25°C/50 percent RH/pH3 are
their N-methylol daughters. On a
approximated by curve (d) in Figure 2.
relative basis over time, liberation
The rate is just below the E-l allowrates will initially be zero, will
able for at least a year, indicating
increase as the daughter concentrathat cellulose formals might constition builds up, and will eventually
tute a significant source of formaldedecrease as both parent and daughter
hyde emission from UF boards.
become depleted. This is illustrated
5. Although quantitative inforin entry D and Figure 2 (curve e) for
mation is lacking on hydrolysis rates
the ether resulting from the reaction
of proven cellulose hemiformals, their
between cellulose and dimethgloldihyrates are generally assumed to be much
droxyethyleneurea (DMoDHEU).6 Alfaster than hydrolysis rates of celluthough this particular material is not
lose formals. For example, investigaa realistic component of a conventors of textile cross-linking almost
tional UF wood adhesive, it is one of
universally assume that hot water or
the few N-methylol compounds for which
alkaline washes after cure will remove
relevant, self-consistent data are
cellulose hemiformals but leave celluavailable. Assuming that only 1 perlose formals intact (entries B1, 2)
cent of the resin's formaldehyde is
(45,113). Experience with the acetals
involved in such Cell-OCH-N bonds, the
and hemiacetals formed from simple
hypothetical formaldehyde liberation
alcohols also lends credence to
rate at 25°C/50 percent RH/pH 3
greater hydrolysis rates of hemiforexceeds the HUD allowable level for
mals relative to formals (14). Thus,
most of the first 100 days (curve e,
cellulose hemiformals formed during
Fig. 2). Although the relevance of
pressing of a UF board should hydrothis particular calculation can be
lyze rapidly and they may account for
questioned, significant formaldehyde
the early high emission of UF boards,
liberation emission by this mechanism
particularly where high F/U resins
is at least conceivable.
will have contributed greater amounts
of formaldehyde to the wood substance.
6. Under the alkaline conditions
6
of interest for a PF board, we could
This discussion and the rate calculaexpect cellulose formals to hydrolyze
tions assume no methylols existed prior
at rates comparable to those at pH 2
to the hydrolysis experiment.
to 4 (compare entries B3c). But, as
noted earlier, such groups apparently
do not form to any great extent under
131
Conclusions--Parts 1 and 2
It is well to emphasize once
again the semiquantitative nature and
even dubious applicability of some of
the rate calculations presented in
Parts 1 and 2. Nevertheless, I feel
that the following overall conclusions
are warranted from these guidelines.
Formaldehyde States in Cured UF
Resins . Conclusions are as follows:
1. Most of the formaldehyde
emitted by cured UF resins results
from hydrolysis of chemically bonded
formaldehyde states. All such states
are susceptible to hydrolysis and most
of the hydrolysis reactions produce
formaidehyde.
Initial formaldehyde liberation rates from LJF resins (F/U 1.2 to
1.6) that have been cured under
realistic conditions but with little
or no wood can be several-fold greater
than allowable rates based on the HUD
particleboard emission standard. This
implies that wood alters the resin
cure state or its pH or that wood
mitigates the rate of formaldehyde
loss.
3. Paraform and hexa hydrolysis
may be significant contributors to
emission but data are not yet sufficient to prove or disprove this point.
Formaldehyde States in Cured PF
Resins . Conclusions are as follows:
1. One cured PF resin exhibits
an initial formaldehyde rate approximately equal to that from a UF resin
with F/U 1.2. Degradation of phenol
methylol groups and/or of methylene
ether links presently appears the most
likely mechanism for this formaldehyde
production.
Formaldehyde States With
Cellulose. Conclusions are as
follows:
1. Small amounts of reaction
between UF resin methylols and cellulose to yield N-methylolethers are
likely. These states are very susceptible to hydrolysis and may well
contribute to board emission in early
stages and somewhat lengthen the time
for emission rate to fall below the
standard levels.
2. Cellulose formals are also
likely hydrolyzable products in an
acid-catalyzed UF board. These appear
to be relatively, stable states, however, and probably need to involve
several percent of a resin's formaldehyde before making a substantial
contribution to overall board emission. But their stability also
implies that their contribution to
emission may be long-lasting. Cellulose formals are unlikely in an
alkaline-cured PF board.
3. Cellulose hemiformals will
probably also be present in both an
acid-catalyzed UF board and an
alkaline-cured PF board, particularly
with high formaldehyde resins. Such
hemiformals, as well as N-methylol
hemiformals, are quite unstable in
acid, and these states may account for
much of the initial, rapidly falling
emission from boards made with high
F/U resins. Hemiformals hydrolyze
more slowly in base but may contribute
significantly to the normally low
emission from PF boards.
Part 3. Studies on Wood Systems
Here we need to examine the evidence available from studies on actual
wood systems that relates to the
mechanisms controlling formaldehyde
emission from bonded wood products.
Factors to be considered include:
(a) the degree to which formaldehyde
emission rate from wood systems is
controlled by diffusion processes,
(b) the contribution of resin hydrolysis to emission rate, and (c) the contribution of formaldehyde-wood states
to emission rate.
In the following section, I first
briefly summarize the reported evidence regarding diffusion control and
resin hydrolysis in actual bonded
products. I then present and discuss
some of my own recent experiments on
wood systems that attempted to shed
additional light on the questions of
resin hydrolysis and the mechanism of
formaldehyde emission.
Literature Evidence for
Diffusion Control
Although published evidence is
sparse, there is little doubt that
diffusion processes can play an
important role in board emission.
Some of the more critical findings
are as follows:
1. Particleboard emits two to
three times less formaldehyde after
conditioning than do exposed core
surfaces (38).
2. Emissions are higher from
board edges than from board faces
(several studies, including 11).
3. Emission levels are
decreased at higher board density
(11,60) and at lower board porosity
(11).
132
4. Ventilation rate and board
loading effects on emission levels in
chambers can be quantitatively
described by equations that are based
upon the assumption that diffusion
across a board-air interface layer
governs the emission rate (76). At
sufficiently high ventilation rates,
the dependence on ventilation rate
disappears and formaldehyde loss is
governed by within-board processes
(47).
It therefore appears that formaldehyde emission rate from a given
large panel may be controlled by diffusion either in the board-air interface or within the board or by chemical processes within the board. Which
of these predominates depends upon the
board's age, composition, physical
structure, and exposure conditions.
NaHCO3 to neutralize the acid cure
catalyst, which would otherwise
catalyze resin hydrolysis (40).
Literature Evidence for Resin
Hydrolysis in Actual Boards
3. no change in modulus or strength
of cured neat UF resin films during
humidity cycling, i.e., when no
swelling/shrinking substrate is
present (22);
Despite the rather massive
literature on formaldehyde emission
from UF-bonded wood products, evidence for a direct causal relationship
between resin hydrolysis and formaldehyde emission from bonded products is
almost nonexistent. Indeed, evidence
in the literature that UF resin
hydrolysis actually does occur in a
board arises primarily from studies
of whether the limited durability of
UF-bonded wood products is caused by
resin hydrolysis or by a particular
susceptibility of UF resin-wood bonds
to rupture from swelling/shrinkage
stresses.
That UF resin hydrolysis does
occur in boards is strongly indicated
by the following:
1. higher rates of strength loss for
UF boards and joints compared to those
made with other adhesives (phenolics,
isocyanates, melamines) during aging
at constant temperature/humidity,
particularly at high temperature/
humidity (7,83,92);
2. decrease in board modulus of rupture (MOR) but not in internal bond
after spraying only the surface mat
with water prior to pressing (92);
3. increase in solubility of cured
resin in both UF-bonded particleboard
and UF-bonded Perlite (nonswelling
volcanic glass) board during aging
(37);
4. decrease in loss of strength
during constant temperature/humidity
aging of plywood after soaking in
Evidence that the lower durability of UF-bonded products can also
be brought about by swelling/shrinkage
stresses in a board includes the
following:
1. faster strength losses for UF
boards than for others (phenolic, isocyanate, melamine) during cyclic
humidity/temperature aging, where
swelling/shrinkage stresses can be
high (23,37,62,71,88,97,116);
2. greater internal bond (IB) loss
and thickness swelling increase with
UF particleboard than with a UF Perlite
board (36,37);
4. increase in thickness swelling of
boards with low F/U resins both before
and after cyclic weathering (61),
accompanied by the postulate (22) that
low F/U resins are more brittle than
high F/U resins;
5. decreased strength loss on boiling
plywood bonded with UF resins containing polyfunctional ureas which are
postulated to produce more flexible
binder networks (41);
6. accelerated aging under stress of
UF joints relative to PF joints (50).
Finally, several studies have
yielded results whose interpretation
is less clear-cut:
1. far greater cumulative amounts of
formaldehyde emitted by boards than
can be accounted for by their Perforator (see Appendix 1c) values (74),
which have often been presumed to
measure primarily non-resin formaldehyde. Unfortunately, it will be shown
later that the Perforator value does
not necessarily measure all formaldehyde-wood states or only non-resin
formaldehyde.
2. reduced rate of cured resin film
cracking by incorporating acid-reactive filler. Such materials will
decrease the acidity within the resin,
thereby decreasing hydrolysis; however
they also reduce the extent of resin
133
Formaldehyde Removal By Gas
Elution. These experiments involved
the continuous collection of formaldehyde removed by a controlled flow of
3. decreased strength loss of UF pargas over the wood samples. Variables
ticleboards by using less acidic cure
included time, gas flow rate, sample
catalyst (92) or by incorporating acid
scavengers (42). (Arguments here are
comminution, gas type, humidity, and
identical to those described above.)
adhesive type.
1. Comminution and flow rate
4. greater mat moisture content (MC)
effects. Elution rates were measured
from UF particleboard at two
yielded greater formaldehyde emission
during particleboard pressing (86) and geometries--e.g., shredded (85 pct <
after pressing (87). Plausible alter- 1 mm) and 25x25x16 mm pieces.
natives to resin hydrolysis stipulate
Shredding was conducted in a sealed
that greater mat MC facilitates formsystem so that no formaldehyde was
aldehyde movement to the board surface
lost during that operation. The
and/or that it enhances hydrolysis of
eluting gas was nitrogen at zero and
cellulose formals and hemiformals.
20 percent RH and at flow rates corresponding to 0.4 to 4.5 changes in gas
Overall, the available literature
volume per minute (NCM).
supports the prevalent view that the
Small effects of flow rate were
durability of UF-bonded wood products
found with dry nitrogen between 0.5
is governed by the susceptibility of
and 1.0 NCM but none with 20 percent
cured UF resin bonds to scission by
RH nitrogen between 0.8 and 4.5 NCM.
both hydrolysis and swell/shrink
Figure 8 compares results for pieces
stresses. Moreover, in either case
and shredded particleboards at two
formaldehyde is a probable ultimate
levels of Perforator (see Appendix 1c)
product. Furthermore, mechanical
values. Several points should be
stress enhances the rates of many
noted:
chemical reactions (9). In fact,
simplistic calculations based on form(a) Elution from shredded UF board
aldehyde liberated from bond ruptures
was only slightly faster than
that limit product service life led to
from the 25x25 mm pieces, and the
the emission curves (f) and (g) in
increase was consistent with
Figure 2; those curves at least indiobserved effects of the flow rate
cate the possibility that formaldehyde
difference (1.0 NCM for shredded
from swell/shrink stress rupture could
versus 0.5 for pieces). This
contribute significantly to total
similarity in elution rates indiemission.
cates that the rate-limiting step
Thus, qualitatively we might
in formaldehyde release in these
expect UF resin bond scission to be
experiments is not "macroone source of board formaldehyde
diffusion" within voids but
emission. However, the available
either "micro-diffusion" within
studies do not permit quantitative
the wood or an actual bond rupstatements about the relative magniture step.
tudes of that source compared to other
(b) No burst of liberated formaldesources, such as formaldehyde-wood
hyde was observed during
states, during board lifetime.
shredding of the 25x25x16 mm
pieces in any of the tests on
Recent FPL Studies
shredded UF board. Apparently,
no significant amount of formalTo shed additional light on the
dehyde existed as gas within
emission mechanism and the contribuvoids; i.e., all formaldehyde in
tion of resin hydrolysis to formaldethe board pieces was present in a
hyde emission, my recent experiments
physically dissolved or sorbed
have examined the liberation or
state or in a chemically reacted
extraction of formaldehyde from parstate. This is consistent with
ticleboards, from wood containing
point (a) and with the discussion
sorbed formaldehyde, and from cured
in earlier sections on the high
resins. Some of the resin data were
reactivity of formaldehyde with
discussed in Part 1; here, I present
water, urea, and wood components.
results from particleboard and
(c) The elution process was quite
formaldehyde-sorbed wood experiments
slow and had not reached any
in which rates of formaldehyde removal
obvious endpoint after 10 days,
were measured by three different proalthough the evolved formaldehyde
cedures (see Appendix 1 for experitotalled only about 20 to 30 permental details).
cent of that removed by the
cure, thereby decreasing brittleness
and tendency to crack (31).
134
2-hour toluene boiling in the
Perforator test. Obviously,
therefore, dry nitrogen is not an
acceptable eluant for readily
removing formaldehyde because of
the nonpolar nature of nitrogen
and the removal of water from the
board.
2. Other eluant gases . The
effectiveness of dry N2, CO and CO2, as
eluants (Fig. 9) was compared in brief
tests. The three gases provided no
differentiation between formaldehyde
states in UF board.
3. Gas moisture effects . As
expected, the influence of moisture
in the eluting nitrogen was very strong
(Fig. 10). Points to be noted here
are as follows:
(a) The observed absence of an endpoint to the dry gas elution from
UF board after 10 days (Fig. 8)
was extended to 40 days (Fig. 11).
(b) During about 15 days of elution
at 80 percent RH (Fig. 10), the
UF board sample lost an amount of
formaldehyde equal to approximately 80 percent of the original
Perforator value and the rate
showed no indication of slowing.
Similarly, at 20 percent RH a UF
particleboard lost formaldehyde
to the extent of about 50 percent
of the Perforator
value in
40 days.7 Clearly, moisture
in the eluting gas removed formaldehyde from states within the
board that were not affected by
the Perforator conditions
(toluene reflux, 2 hours).
Whether those states include
formaldehyde bonded to resin,
i.e., whether resin hydrolysis
occurs under the elution conditions, cannot be firmly stated.
However, the rapid liberation
rate noted earlier for cured
resin at high humidity (Fig. 4)
provides strong, indirect evidence for resin hydrolysis contributions to the observed board
losses at high humidities.
4. Resin effects. Elution
experiments were also performed on
PF-bonded particleboard and on
Southern pine chips (furnish without
resin) that had sorbed formaldehyde
via room temperature vapor phase
equilibration (see Appendix 1d and 2).
Points to be noted are as follows:
(a) The elution patterns from zero
to 20 percent RH for the PF
board (Fig. 12) were very similar
to those for the UF board. However, the formaldehyde losses for
the PF board were approximately
ten-fold less than for the UF,
and the PF losses at 20 percent
RH are likely to exceed the
Perforator value sooner than
in the case of the UF board.
(b) The elution patterns from zero
to 20 percent RH for the formaldehyde-sorbed furnish (Fig. 13)
were again similar to those for
the two board types, although
elution rates were faster, relative to the respective Perforator
values, for 8the furnish than for
the boards. Obviously, the
Perforator test does not measure
the total of all possible formaldehyde non-resin states, even
where those states are formed in
the absence of heat or resin cure
catalysts (furnish pH = 3.9).
Formaldehyde Liberated in
Weighing Bottle Test . Tests equivalent to those on ground resins
(Part 1) were conducted on ground UF
and PF particleboards and on ground
Southern pine that had first been
impregnated with tartaric acid solutions at pH 2 or 3, then vapor-sorbed
with formaldehyde, and finally either
aged at room temperature for 2 weeks
or heated 4 minutes at 160°C to model
board pressing conditions. Liberation
tests were run at 27°C and at both
33 percent and 80 percent RH on
-80 mesh (< 180 µm) materials and on
several particle sizes between 180 µm
and 62 µm. Points to be noted are as
follows:
(a) At 33 percent RH (Fig. 14) the
formaldehyde-sorbed wood virtually completed its loss of
formaldehyde after about 15 to
20 days, whereas the UF particleboard appeared to continue to
slowly liberate formaldehyde.
(The PF particleboard liberation
is an order of magnitude below
that of the UF particleboard and
possesses poor accuracy.)
Heating the formaldehyde-sorbed
wood has caused either a loss of
7
Perforator values for one UF
board were not increased by extending
the toluene reflux time beyond the
standard 2 hours.
8
Negligible amounts of formaldehyde were eluted from the same furnish
unexposed to formaldehyde.
135
formaldehyde or a stronger
bonding to the wood (perhaps
formals). Liberated amounts for
the formaldehyde-sorbed wood
equalled or slightly exceeded the
Perforator values, while the UF
board Perforator amount was
exceeded quite early.
(b)
(c)
At 80 percent RH (Fig. 15) the
above differences between UF
particleboard and formaldehyde-sorbed wood were magnified.
Liberation from the UF particleboard continued rapidly at
30 days while that from formaldehyde-sorbed wood became nearly
constant in about 5 days.
wood samples also liberated total
amounts that were close to their
Perforator values measured at
high moisture. Most of the formaldehyde in the formaldehydesorbed wood was, therefore, very
weakly bonded (perhaps hemiformal and methylene glycol)
although there may have been
small quantities that were
liberated with greater difficulty, particularly at pH 3
relative to pH 2. The UF board,
in contrast, apparently contained little of the very loosely
bound formaldehyde but contained
greater amounts of more strongly
bound formaldehyde, as would be
expected. The PF board liberation was again well below that of
the UF board and behaved similarly to the formaldehyde-sorbed
wood samples except for greatly
exceeding its Perforator value.
At 80 percent RH the UF and PF
boards exhibited no significant
particle size effects on liberation rates between particle sizes
of approximately 60 and 180 µm.
In that size range, therefore,
within-particle diffusion did not
influence liberation rate from
the boards.
Formaldehyde extracted in water .
Formaldehyde liberated during continuous exposure to water at pH 3 was also
measured on the same materials used
in the weighing bottle test. Very
dilute slurries of -80 mesh material
were held at 25°C in the presence of
sodium azide as bacterial inhibitor.
In 1 or 2 hours almost all removable
formaldehyde was extracted from the
formaldehyde-sorbed wood samples
(Fig. 16); the total amounts were
nearly identical to those at 80 percent RH and to the Perforator values.
However, liberation from the UF board
continued rapidly after 6 days and at
30 days far exceeded the amounts at
80 percent RH and the amounts from the
wood samples in water. Interestingly,
liberation from the PF board in water
also exceeded that at 80 percent RH
and may have occurred in two or more
stages; even the apparent initial
stage, however, was an order of magnitude greater than the Perforator
value.
Interpretation and Extrapolation
to Boards in Service
In this section, I offer an
analysis of the Part 3 results in
terms of the findings from Parts 1
and 2, and on that basis, speculate
about their implications for large
panel formaldehyde emission.
Interpretation for Comminuted
Systems. The similarities and differences noted in Part 3 for the kinetics
of formaldehyde removal from UF and PF
particleboards and from formaldehydesorbed wood are brought out more
clearly by plotting relative formaldehyde losses versus time. Loss ratios,
i.e., formaldehyde loss by any material divided by the UF board loss at
the same time, are shown in Figures 17
and 18; included in Figure 17 are
analogous ratios for resin data from
formaldehyde liberation (weighing
bottle test, Appendix la) and formaldehyde elution (Appendix 1d) experiments. Examination of the data leads
to the following additional comments:
1. Southern pine containing
formaldehyde that was sorbed at the
wood's natural pH or at pH 2 to 3
held the formaldehyde in a state that
was strongly retained at low humidity
but relatively labile at moderate to
high humidities. The formaldehyde was
nearly completely released, for
example, in 12 days at 33 percent RH
(Fig. 14) in 5 days at 80 percent RH
(Fig. 15) and in 0.2 days in pH 3
water (Fig. 16). The available
information (Part 2) indicates this
formaldehyde is present as monomer
(methylene glycol) or oligomer dissolved in the wood's moisture or
possibly as cellulose hemiformals.
2. With PF board, the amount of
formaldehyde released within the time
scale of these experiments varied more
with humidity, and did not obviously
exist in only one state (cf. Figs. 11,
14-16). While a portion of the
removable formaldehyde very likely
exists in the same state(s) as in the
136
formaldehyde-sorbed wood, a major portion is more strongly held but still
sensitive to moisture. The latter
state perhaps is phenolic methylols.
3. The UF board undoubtedly contained some of the same moisturelabile states that were present in the
formaldehyde-sorbed wood, and these
account for some of the initially
rapid loss observed at 80 percent RH
(Fig. 15) and in water (Fig. 16). Up
to 20 percent RH the release pattern
from the UF board by nitrogen elution
was very similar to that from the PF
board and formaldehyde-sorbed wood,
indicating similar release mechanisms
from all three comminuted wood systems
under those conditions (Fig. 17). In
the other types of experiments at
higher humidities, however, the
release pattern from the comminuted UF
board clearly differed from the patterns observed in the other two wood
systems (Fig. 18). The continued
evolution of formaldehyde from the UF
board beyond the very early portion
and at rates increasing with humidity
strongly indicates extensive hydrolytic sources other than those present
in the PF and formaldehyde-sorbed
wood. Obviously, those additional
sources are most likely UF resin and
UF-wood states, with some possibility
of cellulose formals, as discussed in
Parts 1 and 2.
4. Point 3 suggests a similar
release mechanism for the shredded
boards and furnish particles during
nitrogen elution at 20 percent RH and
below (Fig. 17). This implies identical rate-limiting steps, which might
involve a chemical bond rupture or a
monomeric formaldehyde diffusion
process. The nature of the three systems dictates that a chemical process
most probably involves hydrolysis of
cellulose hemiformals. However, the
evidence for significant amounts of
that formaldehyde state to be present
is not clear-cut (Part 2). Since
small nitrogen flow rate effects were
observed (Fig. 8) in the range
employed in these experiments
(0.5 NCM), some control of elution
rate by gaseous formaldehyde diffusion
through the shredded or furnish
particle-gas interface (vaporization)
must have existed. Intraparticle diffusion limitations also seem likely at
these particle sizes (~100 to
1,000 µm), although particle size
effects were not observed in the high
humidity weighing-bottle tests with
particle sizes below 180 µm. Intraparticle diffusion presumably involves
methylene glycol, whose effective diffusion rate in the wood's water may
well be decreased by strong interactions with cellulosics during its
passage to the particle surface.
Implications For Formaldehyde
Emission From Large Panels . Much of
the above discussion should be
directly relevant to large panel
emission. If intraparticle diffusion
of methylene glycol is hindered under
some conditions with comminuted materials, it will obviously continue to
be hindered in an actual board. Moreover, gaseous diffusion through
particle-gas interfaces will be
greatly slowed in a large panel
because no eluting gas is present to
reduce the concentration gradient and
the interface layer thickness. In
addition, the diffusion path to the
panel surface will be tortuous, and
panel surface-air layer gaseous diffusion limitations may exist. Diffusion
effects are therefore very important
in panel emission rates.
For a given board composition and
structure, the presence of diffusion
limitations leads to lower emission
rates and somewhat higher internal
concentrations of dissolved methylene
glycol. The concentration increase
may be sufficient to slow the net production of formaldehyde via reversible
hydrolyses, thereby lowering and prolonging the emission contributions
from hydrolytic processes. Unfortunately, given the currently available
knowledge, we can only speculate about
which formaldehyde states in the board
may be responsible for emission at
various points in the board's life.
However, the water extraction data
(Fig. 16) suggest the possibility of
distinguishing between "loosely held"
formaldehyde (perhaps methylene glycol
monomer and oligomers and cellulose
hemiformal) and more firmly bonded
formaldehyde, the latter presumably
including hydrolytic sources (perhaps
UF, UF-wood, and cellulose formal).
The shape of the UF board curve in
Figure 16 indicates that from 20 to
40 mg of formaldehyde per 100 g of
board belongs in the "loosely held"
category. In addition, the Perforator
value for this board (11 mg/lOO g)
indicates that it should meet the HUD
and possibly the E-l standards,
leading to emission rates at standard
conditions between 2 x 10-5 and
9 x 10-5 mg per g board per hour
(Appendix 3c). If "loosely held"
formaldehyde is primarily responsible
for those emission rates, the time
required to dissipate those formaldehyde states is 3 to 6 months at the
HUD level and 1 to 2 years at the E-1
137
level. Continuing with the argument,
subsequently emitted formaldehyde
should derive from hydrolytic
processes. Obviously, additional
water extractions plus measurements of
actual emission rates on identical
boards would be needed to confirm this
approach towards distinguishing formaldehyde sources within boards.
Summary and Conclusions
This paper addresses the general
question of the extent to which resin
hydrolysis is responsible for formaldehyde emission from bonded wood
products. I have both reviewed
briefly the relevant literature and
presented original FPL data in three
areas: the chemistry and hydrolytic
stability of formaldehyde resins and
model compounds (Part 1); the reactions of formaldehyde and UF compounds
with wood components and the hydrolytic stability of their products
(Part 2); and formaldehyde emissions
from bonded wood products (Part 3).
Major Findings . The major
findings are as follows:
1. In an acid-catalyzed UFbonded board, formaldehyde can exist
in a wide variety of states. These
states may include dissolved methylene
glycol monomer and oligomers, paraform, hexa, chemically bonded UF resin
states, chemically bonded UF-wood
states (amidomethylene ethers with
cellulose), cellulose hemiformals,
and cellulose formals.
2. Each of these states is a
potential source of formaldehyde
emission by evaporation (methylene
glycol) or by initial hydrolysis (all
others). Unfortunately, we cannot
now provide a complete listing of
states in the order of their potential
importance as emission sources.
Clearly, however, some of the most
weakly held states would be methylene
glycol, cellulose hemiformal, amidomethylols, and cellulose amidomethylene ethers.
3. In a base-catalyzed PF-bonded
board, formaldehyde states may
include: methylene glycol monomer and
oligomers, chemically bonded PF resin
states, chemically bonded PF-wood
states, and cellulose hemiformals.
Emission sources apparently include
methylene glycol, cellulose hemiformals, and a PF resin state-possibly phenolic methylols.
4. In Southern pine containing
formaldehyde that was sorbed at room
temperature and at the wood's natural
pH or at pH 2 or 3, formaldehyde
states may include methylene glycol
monomer and oligomers and possibly
cellulose hemiformals. These are all
apparently readily removed from the
comminuted wood at 80 percent RH
(5 days) or in pH 3 water (0.2.day).
5. Diffusion processes can very
likely exert a major influence on
emission rates from large panels.
Depending upon board structure, composition, age, and exposure condition,
emission-limiting diffusion steps may
involve methylene glycol within the
board's water or gaseous formaldehyde
within the board or within the boardair interface.
Subsidiary Findings . The subsidiary findings were as follows:
1. Formaldehyde liberation from
cured neat resins (PF and UF) is much
greater than expected for those same
resins cured in a particleboard, indicating that the wood alters the resin
cure and/or the bondline pH or that
diffusion effects predominate in the
board.
2. A cured PF resin liberates
formaldehyde at significant rates that
increase with humidity.
3. The Perforator test measures
formaldehyde in states that are
present in cured neat PF and UF
resins, in boards made with both
resins, and in formaldehyde-sorbed
wood. In all but the last, the
Perforator values are much less than
the amounts removable by simple exposure to high humidity.
4. The limited durability of
UF-bonded wood products probably
results from the susceptibility of UF
resin and UF-wood bonds to chain
scission from both hydrolysis and
swell/shrink stresses. In either
case, formaldehyde is a possible
product.
Acknowledgment
This work was partially funded by
the Formaldehyde Institute. Vital
assistance was provided by two groups
at the Forest Products Laboratory--the
staff of the Library in literature
search and retrieval, and the Systems
and Automatic Data Processing group in
establishing and using a computerized
literature file. I am also greatly
indebted to members of the Technical
Committee of the Formaldehyde
Institute for advice and for supplying
materials. Professor James Koutsky,
University of Wisconsin-Madison
Chemical Engineering Department, and
several of his students aided this
effort with both advice and laboratory
138
aid. Much of the FPL data were
obtained by Ralph Schaeffer and
Jill Wennesheimer.
Appendix 1. Experimental Procedures
a. Formaldehyde liberation by
weighing bottle technique
Ground, sieved (-80 mesh or
smaller) powder was weighed (10 to
150 mg) into a glass weighing bottle
(40 mm dia. x 40 mm high), a small
glass cross placed on the bottom of
the container, a glass beaker (22 mm
dia. x 25 mm high) containing 5 ml of
sulfuric acid or salt solution placed
on the cross, and the bottle sealed
with a greased cap. The assembly was
then stored in a temperature chamber
(usually at 27°C) for a specified
period. The beaker was then removed
and replaced with a fresh solution.
For a given weighed sample, the solution was replaced no more than twice.
At each removal the solution was analyzed for formaldehyde, usually by the
chromotropic acid procedure.
Resin samples were vacuum-dried
before weighing into the bottles and
thereafter evacuated for various times
to minimize physically sorbed formaldehyde. Humidity in the sealed
bottles was controlled by the concentration of sulfuric acid or salt.
b. pH of cured resins and model
compounds
The ground sample (usually
-80 mesh) was shaken with distilled
water at a ratio of 1/10 in a capped
vial at least overnight.
The pH
of the supernatant was measured using
a combination electrode.
c. Perforator test
With particleboard the standard
procedure (27) was followed. About
100 g of 25 x 25 mm specimens were
refluxed in toluene for 2 hours with
continuous extraction of formaldehyde
into water. The water was then analyzed for formaldehyde concentration
by the acetylacetone fluorometric
method (72). For ground resins and
other materials, sample amounts were
adjusted to produce comparable formaldehyde concentrations.
d. Nitrogen elution of particleboard,
furnish, and cured resin
25 x 25 x 16 mm specimens were
placed on a wire screen inside a horizontal glass tube (30 mm diam. x
750 mm long). Smaller particle size
material was placed either in a similar vertical tube, with bottom gas
feed, or in a continuously shaken
Erlenmeyer flask, with gas feed via a
tube leading to the flask's bottom.
Entering gas was preconditioned by
passage through or over saturated salt
solutions at room temperature (23 ±
1oc). Exiting gas was continuously
scrubbed of formaldehyde by passage
through a series of impingers containing water and held in ice water.
The number of impingers in series
varied with gas flow rate and scrubbing time, based on prior experiments
that had established conditions providing greater than 95 percent
scrubbing efficiency. The gas flow
was interrupted at intervals to allow
changing to a series of fresh impinger
solutions; the removed impinger solutions were analyzed separately or
in combination, usually with the
acetylacetone fluorometric method
(72). A variety of tests confirmed
that no significant formaldehyde
losses were caused by adsorption on
the polyethylene tubing or by leaks.
A number of analyses by both the
acetylacetone and chromotropic acid
methods showed no significant
differences.
Ground resin was eluted by nitrogen in a similar manner; the primary
exception was the use of only a few
grams held in a glass tube that contained sintered glass frits at both
ends.
e. Toluene elution of cured resin
About 1 g of vacuum-dried resin
(-80 mesh) was weighed into a glass
tube (16 mm diam. x 30 mm long) containing a medium porosity glass frit
and evacuated for a specified period
before being placed in the elution
column. Under a dry nitrogen atmosphere, toluene was allowed to drip
slowly (0.05-0.15 ml/min) through the
resin to the bottom of a water column
(-300 mm long), where each drop of
toluene bubbled back up through the
water and was then collected in a
separate flask. The toluene flow was
interrupted at intervals to change the
toluene collection flask and the water
in the extraction column. The toluene
in the flask was re-extracted with
water and that water and the column
water were analyzed for formaldehyde
by the acetylacetone method. All but
a few percent of the formaldehyde was
in the column water.
139
f. Water extraction of ground wood
or board
Approximately 0.4 g of ground
(-80 mesh) sample were placed in a
stoppered flask with 75 mL of water
made to pH 3 with HCl and containing
100 mg/L sodium azide as bacterial
inhibitor. The flasks were shaken at
25°C and 10 mL aliquots were removed
periodically by sucking through a
sintered glass filter. At each
removal, 10 mL of fresh liquid were
added to the flask through the filter;
each flask was sampled no more than
three times. Aliquots were analyzed
by the fluorometric acetylacetone
procedure (72).
Appendix 2. Materials
The two UF resins (F/U 1.2 and
1.6) were prepared using a hybrid
recipe agreed upon by three adhesive
manufacturers as representative of
current commercial procedures. The
synthesis began with UF concentrate
at F/U 4.8 to which urea was added in
three separate steps: the first at
pH 6.8 to 7.2 at 95°C, the second at
pH 6.2 to 6.4 at 85°C, and the third
at pH 7.0 to 7.5 at 60°C. Final pH
was adjusted to 7.5 to 7.8 and solids
content about 60 percent. The standard cure condition was 4 minutes at
160°C with 0.7 percent NH4Cl. The
cured material was vacuum-dried,
ground, sieved, further vacuum dried,
and stored cold over desiccant.
For the resin-wood composites,
the wood furnish was ground and sieved
to produce -80 mesh material which was
mixed with resin in a 1:4 wood:resin
solids ratio. NH4Cl catalyst was
added at 0.7 percent (resin solids
basis) and the mixture cured as a thin
film under nitrogen for 30 minutes at
100°c. Neat resin was treated similarly. The cured products were
treated as above.
The PF resin was a commercial
product with solids content of 51 percent, viscosity F by Gardner-Holt, and
pH 9.8. The standard cure condition
used was 3.5 minutes at 205C.
The UF particleboard was a commercial low emission product made with
a resin having an F/U ratio below 1.2.
The PF board was an experimental
industrial product, and the furnish
was standard industrial Southern pine
material.
Formaldehyde-sorbed Southern pine
furnish was prepared by allowing
furnish to equilibrate for several
days at room temperature over water
solutions of salts and formaldehyde,
with the salt serving to control
humidity. Formaldehyde-sorbed ground
Southern pine was similarly prepared
except for prior soaking in tartaric
acid solution (with sodium azide) at
pH 2 or 3.
Appendix 3. Methods of Calculation
Formaldehyde liberation rates on
hypothetical board basis
Liberation rate = R in units of
mg CH20 per g dry board per hour
Assumptions:
- 0.07 g resin solids per 100 g
dry board
- 16-mm thick board of density
0.64 g/cc
- emission from both sides of
large panel
- mean F/U = 1.4
- decompositions are all apparent
first order with rate constant
k in hour-1 except where rates
are measured directly from
slopes of experimental data
(1) Rate constant k measured on molar
basis--e.g., for a model compound.
(A1)
where
M = molecular weight of starting
compound
(2) Rate constant k measured on weight
basis--e.g., for cured resin.
(A2)
(3) Rate measured from slope of data
plot.
(A3)
where
S = slope of plot of cumulative
mg liberated CH2O per g
starting material versus time
in hours
(4) Rate R' from component of resin
present as resin weight fraction wi.
(A4)
140
b. Calculation of rate variation with
time, using rate constants .
c. Rates for particleboard emission
standards
(In all cases F = formaldehyde
and zero subscripts indicate initial
amounts.)
Assuming a steady state condition
for the concentration CS in ppm of
formaldehyde in air and an emission
rate ER from board in units of mg
CH2O per g dry board per hour:
(A10)
(A5)
where
K = constant for conversion of
units
N = ventilation rate in hours
L = board loading in m2 exposed
board area per m3 of air
space
(A6)
Literature Cited
(A7)
with
(A8)
(A9)
141
142
143
144
145
146
TABLE 1.--FORMALDEHYDE LIBERATION FROM MODEL COMPOUNDS AND RESINS EXPRESSED AS HYPOTHETICAL BOARD EMISSION RATES
TABLE 1.--FORMALDEHYDE LIBERATION FROM MODEL COMPOUNDS AND RESINS EXPRESSED AS HYPOTHETICAL BOARD EMISSION PATES--cont.
TABLE 1.--FORMALDEHYDE LIB&RATION FROM MODEL COMPOUNDS AND RESINS EXPRESSED AS HYPOTHETICAL BOARD EMSSION RATES--CONT.
TABLE 2.--FORMALDEHYDE LIBERATION FROM CH2O-CELLULOSE STATES EXPRESSED AS HYPOTHETICAL BOARD EMISSION RATES
TABLE 2.--FORMALDEHYDE LIBERATION FROM CH2O-CELLULOSE STATES EXPRESSED AS HYPOTHETICAL BOARD EMISSION RATES--cont.
Figure l.-- Formaldehyde liberation
from cured PF resin. Weighing bottle
test at 27C. Cured 3.5 min. 205C.
-80 mesh. Evacuated 17 days.
Figure 2. --Calculated CH2O liberation
rates for hypothetical board from data
on models. Corrected to 25C/pH 3.O/RH
50 pct. HUD and E-l lines are allowable rates. (a) -NCH2N- at w = 0.8;
(b) -NCH2OCH2N- at w = 0.1; (c)
-NCH2OH at w = 0.1; (d) Ce110(CH20)
Cell at w = 0.02; (e) Ce110CH2N- at
w = 0.1; (f) "High" durability case;
5 pct loss in 50 years; (g) "Low"
durability case; 30 pct loss in 20
years.
Figure 3.-- Formaldehyde liberation
rates for model compounds. Weighing
bottle test at 27C, 80 pct relative
humidity.
Figure 4.-- Formaldehyde liberation
from cured UF resin; F/U 1.6 Weighing
bottle test at 27C. Cured 4 min. 160C,
0.7 pct NH4Cl. Evacuated 17 days.
152
Figure 5. --Toluene elution of cured
resins. 0.05-0.15 ml/min. 25C. -80
mesh. Evacuated 3-4 weeks. Cured as
in Figs. 1 and 4.
Figure 6. --Formaldehyde liberation
from resin versus resin/wood. 80
pct relative humidity/27C. Cured
30 min. 130°C. -80 mesh. Evacuated 2 weeks. Resin/plywood at
3/1. P = Perforator value in
mg/g resin.
Figure 7. --Calculated liberation
rates for hypothetical board from
data on cured resins. PF as in Fig.
1. UF samples as in Fig. 6.
Figure 8.-- Particleboard elution by
dry nitrogen; sample geometry effects
(o ! shredded; 1.0 NCM;!
! " 25x25x16 mm
0.5 NCM; duplicate runs. P = Perforator value in mg/l00 g dry board,
measured on starting material at ~6
pct moisture content.
153
Figure 9. --Particleboard elution by
different dry gases. Differences
in the two curves due to different
flow rates and experimental configurations. P as in Fig. 8.
Figure 10. --Urea-formaldehyde particleboard elution by nitrogen; relative humidity (RH) effects. (0.4
NCM. P as in Fig.8.)
Figure 11.--Urea-formaldehyde particleboard elution by nitrogen at different relative humidities (RH).
(0.5 NCM. P as in Fig. 8.)
Figure 12. --Phenol-formaldehyde
particleboard elution by nitrogen
at different relative humidities
(RH). (0.5 NCM. P as in Fig. 8.)
154
Figure 13. --Elution of
formaldehyde-sorbed furnish by nitrogen at different relative humidities (RH). (0.5 NCM. P
as in Fig. 8.)
Figure 14. --Formaldehyde liberation from particleboards and CH2O-sorbed wood at 27°C and 33
percent relative humidity (RH); weighing bottle
test with -80 mesh materials. o Southern pine
impregnated with pH 2 tartaric acid and vaporequilibrated with CH2O/salt solution at ~50
pct RH; ∩ as before except heated 4 min. at
160°C after CH2O sorption; ! urea-formaldehyde
particleboard; phenol-formaldehyde particleboard, values approximate; P = Perforator
value at indicated moisture content [MC).
Figure 15. --Formaldehyde liberation
from particleboards and CH2O-sorbed
wood at 27°C and 80 percent relative
humidity (RH). Weighing bottle test
with -80 mesh materials. o Southern
pine impregnated with pH 2 tartaric
acid and vapor-equilibrated with
CH2O/salt solution at ~75 pct RH; #
as before except pH 3 tartaric acid;
! urea-formaldehyde particleboard;
phenol-formaldehyde particleboard, parentheses indicating approximate values; P and MC as in
Fig. 14.
Figure 16. --Formaldehyde liberation
in water at 25°C and pH 3 from particleboard and CH2O-sorbed wood; all
materials -80 mesh. Sodium azide in
water at 100 mg/L as preservative;
symbols and abbreviations as in Fig.
14.
155
Figure 17.--Formaldehyde loss ratios at 20 percent relative
humidity for various materials. Formaldehyde removed from
a material divided by that removed from urea-formaldehyde
(UF) particleboard. Board elution by nitrogen. Resin liberation by weighing bottle test. PF = phenol-formaldehyde.
Figure 18.--Formaldehyde loss ratios at 80 percent relative
humidity (RH) and in water. Loss ratio = CH2O liberated
relative to that from urea-formaldehyde (UF) particleboard
in same test. WB = weighing bottle test; PF = phenol-formaldehyde; aq = water extraction test at pH 3. All materials -80 mesh. Southern (So.) pine impregnated with pH
2 tartaric acid and CH2O vapor-sorbed.
156