Mechanism of trialkylborane promoted adhesion to low surface energy plastics

Article abstract of DOI:10.1021/ma040095f

Excellent adhesion to low surface energy substrates such as polypropylene, polyethylene, poly(vinyl difluoride), and poly(tetrafluoroethylene) is obtained with acrylic polymerization initiated by trialkylboranes at room temperature and without need for surface pretreatment. The mechanism of adhesion is a consequence of a series of radical processes resulting from the initial oxidation of the trialkylborane followed by the production of alkoxy and alkyl radicals. This paper will elucidate the mechanism of adhesion using both experiment and theory.

Full text of DOI:10.1021/ma040095f

7974
Macromolecules 2004, 37, 7974-7978
Mechanism of Trialkylborane Promoted Adhesion to Low Surface
Energy Plastics
Ma r k F . Son n en sch ein ,* Steven P . Webb, P a tr ick E. Ka stl, Da n iel J . Ar r iola ,
Ben ja m in L. Wen d t, a n d Da n iel R. Ha r r in gton
Dow Chemical Company, Corporate Research and Development, Midland, Michigan 48674
Nelson G. Ron d a n
Dow Chemical Co., Corporate Research and Development, Freeport, Texas 77541
Received May 27, 2004; Revised Manuscript Received J uly 23, 2004
ABSTRACT: Excellent adhesion to low surface energy substrates such as polypropylene, polyethylene,
poly(vinyl difluoride), and poly(tetrafluoroethylene) is obtained with acrylic polymerization initiated by
trialkylboranes at room temperature and without need for surface pretreatment. The mechanism of
adhesion is a consequence of a series of radical processes resulting from the initial oxidation of the
trialkylborane followed by the production of alkoxy and alkyl radicals. This paper will elucidate the
mechanism of adhesion using both experiment and theory.
In tr od u ction
Exp er im en ta l Section
Polyolefins such as polyethylene, polypropylene, and
their copolymers are useful in the fabrication of many
objects including automotive components, footwear,
toys, and household appliances. These polymers are
inexpensive and lightweight and can be tailored to
achieve a wide range of properties. Unfortunately, the
inherent low surface energy of polyolefins inhibits
adhesion to most adhesive and coating formulations1
which tend to contain polar materials such as ure-
thanes, acrylics, and epoxies.2 Consequently, efforts
have been directed toward pretreating the surface of
these low surface energy materials in order to achieve
satisfactory adhesion. Such pretreatments include vapor
cleaning, acid treating, priming, corona discharge treat-
ing, or plasma treating.3-11 Other efforts to improve
adhesion have been directed to the use of adhesion
promoters.12 Unfortunately, such techniques are often
cumbersome, costly, and environmentally unsound.
Recently, we13 and others14-17 have disclosed new
technology that provides very strong adhesion to poly-
olefins and fluorinated polymers at room temperature
and without the need for surface pretreatment. Adhe-
sion to these plastic substrates is promoted by trialkyl-
borane catalysts that produce radical initiators at room
temperature. Previous work had speculated that tri-
alkylboranes produced radical initiators through an SN1
insertion of oxygen to the empty p orbital of boron with
the concomitant production of an alkyl radical and
peroxy radical. However, these radicals are usually
insufficiently energetic for the kind of chemistry that
might result in adhesion to olefin substrates. In this
article we provide evidence supporting a proposed
mechanism involving insertion of oxygen between the
boron and an alkyl group of the trialkylborane, followed
by subsequent homolysis of the peroxide bond to form
alkoxy radicals. Alkoxy radicals may then abstract
hydrogen from an olefin surface. Our conclusions are
based on experiments with model compounds and
results from theoretical modeling.
Organoboranes, 2,5,10,14-tetramethylpentadecane (pris-
tane), heptadecane, and maleic anhydride were obtained from
Aldrich Chemical (Milwaukee, WI). Grafting of the maleic
anhydride was performed by combining maleic anhydride and
pristane or heptadecane in a glass vial in equimolar amounts.
The maleic anhydride was suspended in solution using a glass-
encased magnetic stirrer. An equimolar amount of organobo-
rane was withdrawn through a silicone rubber septum with a
syringe and a fine gauge needle. The needle tip was then
submerged under the surface of the pristane (or heptadecane)
maleic anyhdride suspension and slowly injected with moder-
ate agitation. Trialkylboranes, especially triethylborane (TEB),
are pyrophoric in contact with air, and special care must be
exercised in its handling. In general, we found that the use of
a fine gauge needle with a nitrogen gap at the syringe front
(from the trialkylborane blanket in the container) was ad-
equate to prevent mishap. Upon introduction of the trialkyl-
borane into the solution, the maleic anhydride over a period
of minutes completely dissolved into the pristane solution. The
solution formed a second phase. A denser phase was a dark
yellow viscous liquid, while the less viscous upper phase was
pale yellow. In general, the two phases contained the same
products but at different concentrations. Significant heat was
generated through the course of this transformation, but gram
scale quantities could still be handled safely. Substitution of
a borane ester such as diethylmethoxyborane (DEMOB) for
the trialkylborane resulted in no detectable heat generation
and no visual evidence for reaction.
Samples from the reaction mixture were diluted in aceto-
nitrile. Gas chromatography and mass spectrometry (GC/MS)
measurements were performed on an Aglilent 6890N GC
coupled to a Micromass GCT, SN CA095, time-of-flight MS
system operating in the electron impact (EI) and positive ion
chemical ionization (PCI) modes. Conditions were as follows:
column: 15 m × 0.25 mm i.d. × 0.25 µm film Rtx-5 SILMS;
temperatures: column: 35 °C/hold 5 min to 275 °C at 10 °C/
min; injector: 280 °C GC; reentrant: 280 °C; source: 180 °C
(EI), 120 °C (PCI); flow: column: 1.5 mL/min; helium split:
300:1; detector: MCP: 2700V mode: +TOFMS, CENT; resolu-
tion: 9000 (at m/z 614); electron energy: 70 eV (EI), 100 eV
(PCI); trap current: 150 µA (EI); scan: 25-400 amu (EI), 65-
800 amu (PCI); rate: 0.3 s per scan; lock mass 201.9609 C6F5-
Cl.
Solution polymerization of PMMA using organoborane
initiators was performed in toluene at a 50 vol % concentration.
* Corresponding author.
10.1021/ma040095f CCC: $27.50 © 2004 American Chemical Society
Published on Web 09/21/2004

Macromolecules, Vol. 37, No. 21, 2004
Trialkylborane Promoted Adhesion 7975
[R•] and assuming termination to occur primarily by
recombination, the molecular weight should follow the
simple form of eq 7 via eqs 1-3, where Mn is the
number-average molecular weight and I is the normal-
ized concentration of initiator species.
The mixture was mixed with a glass encased magnetic stirrer.
Organoborane at the indicated molar concentration was
introduced into the solution as before and the glass vial capped
to prevent evaporation of the acrylic monomer or solvent. Size
exclusion chromatography was performed on these solutions
using 0.01 g of the polymer solution diluted in 10 mL of THF
eluent. Chromatography was performed with a Waters model
510 at a nominal flow rate of 1 mL/min on an injected volume
of 50 µL. Detection was performed using a Kratos Spectroflow
773 UV absorption detector set at 260 nm. Data analysis was
performed using Polymer Laboratories Calibre GPC/SEC,
acquisition version 6.0 and reanalysis version 7.04. The
calibration was determined using narrow molecular weight
distribution polystyrene standards from Polymer Laboratories.
1
1
[I]
Mn
(7)
[O ]2[R B]2
x
2
3
For the mechanism of eq 4, the relevant reaction
producing initiator is eq 5. This path leads to a molec-
ular weight dependence on initiator level characterized
by eq 8.
Resu lts a n d Discu ssion
1
1
Mn
(8)
Durable adhesion to polyolefins can be obtained by
increasing the surface energy by one of several means
listed above or, alternatively, by directly grafting poly-
mer chains to the substrate surface.18-21 One approach
to this method is to utilize radical chemistry that can,
in principle, abstract hydrogen atoms and provide a
radical initiation site from which a radical chain reac-
tion can commence. In the past, there have been several
barriers to using radical chemistry to achieve this end.
One such barrier is that carbon-centered radicals are
generally not energetic enough to achieve substantial
hydrogen abstraction. Alkoxy radicals are more suitable
for abstraction of hydrogen.22 However, generation of
alkoxy radicals generally requires decomposition of
peroxide bonds at elevated temperatures which usually
sets up an unfavorable competition between homopo-
lymerization of adhesive monomer and hydrogen ab-
straction/chain propagation.
[O ][R B]
[I]
x
x
2
3
The use of trialkylboranes for vinyl polymerizations
has been described in the literature and the sensitivity
of the polymerization to oxygen quantified.26
The very high room temperature reactivity of trialkyl-
boranes can be mitigated by reaction of the borane (a
Lewis acid) with a Lewis base.27 Primary amines have
typically been employed in this function and can produce
complexes with the trialkylboranes that are stable solids
or liquids at room temperature. When trialkylborane-
amine complexes are included in adhesive compositions
and the amine is subsequently deblocked from the
trialkylborane by reaction with an acid, an isocyanate,
or an aldehyde for instance, radical chemistry initiates
room temperature vinyl polymerization and chain graft-
ing to the polyolefin surface. It has previously been
reported that alkoxydialkylboranes are nearly as effec-
tive as trialkylboranes in promoting vinyl polymeriza-
tions.28 However, when an alkoxydialkylborane is used
in adhesive formulations, polymerization is observed,
but no adhesion is measured by any of several standard
adhesion tests.
Trialkylboranes are well-known generators of room
temperature radicals23 and can be pyrophoric when
exposed to the atmosphere, depending on the attached
alkyl groups.24 A proposed mechanism of borane oxida-
tion involves a diffusion-controlled reaction with oxy-
gen.25
The mechanism differences by which trialkylboranes
promote polymerization and adhesion while alkoxydi-
alkylboranes promote only polymerization provides a
basis for understanding these new adhesion promoting
catalysts and their mode of action. Measurement of
polymer molecular weight as a function of radical
initiator concentration is a conventional way of provid-
ing insight into the concentration and composition of
initiator species. Plots of 1/(number-average molecular
R3B + O2 f R2BO2• + R•
(1)
(2)
(3)
R• + O2 f RO2
•
RO2• + R3B f R2BOOR + R•
An alternative oxidation mechanism26 of the borane
proceeds with the oxygen inserting in a concerted
fashion between a boron-carbon bond, skipping the
generation of intermediate radical, resulting directly in
the boron-containing peroxide (first product in eq 3).
weight) vs [I] for the mechanism of eqs 1-3 or vs [I]
x
for the mechanism of eqs 4 and 5 should yield slopes
reflecting the efficiency of radical generation. Figure 1
shows that tributylborane (in this case) produces a much
greater number of radical initiators than DEMOB since
the slopes are quite different. Further, methyl meth-
acrylate polymerizations in toluene polymerized with
either tri-n-butylborane (TBB) or diethylmethoxyborane
yielded Mw/Mn of 2 ( 0.2, suggesting that the polym-
erization itself is unaffected by the initiating radicals
as once reported29 and that chain termination is not
limited to radical recombination. A comparison of the
data fit for the two plots does not lead to a clear
preference for either model, but the fit is marginally
better for the mechanism with a square root dependence
on [I] as predicted by eqs 4 and 5 (Figure 1B).
R3B + O2 f R2BOOR
(4)
(5)
R2BOOR f R2BO• + RO•
Applied to eqs 1-3, a steady-state assumption yields
eq 6 for the total radicals concentration:
krad[R3B]2[O2]2
[R•] )
(6)
kt
x
where krad is the overall effective rate constant for
production of initiator radicals and kt is the rate
constant for bimolecular chain termination. Given that
molecular weight theoretically varies as the inverse of
Experiments on model compounds were performed to
determine whether trialkylboranes and alkoxydialkyl-
boranes could produce similar products in controlled
experiments and whether the resultant products or

7976 Sonnenschein et al.
Macromolecules, Vol. 37, No. 21, 2004
RO• + R′-H f ROH + R′•
3R2BO• f R3B3O3 + 3R•
(9)
(10)
•
R• + nM f RMn
(11)
(12)
(13)
R′• + nM f R′Mn• (adhesion)
R′• + RMn• f R′MnR (adhesion)
where R′ represents an organic substrate and M is the
polymerizable monomer.
However, with the large amount of alkyl radical
produced from these reactions it is also possible that
another pathway is available following eq 222
2RO2• f ROOOOR
(14)
(15)
ROOOOR f 2RO• + O2
the products of which would then proceed as described
above.
In this series of reactions, the alkoxy radicals are
proposed to be responsible for hydrogen abstraction from
the substrate, yielding a carbon-centered radical capable
of initiating polymerization (eq 12) or terminating a
growing chain (eq 13).
F igu r e 1. Inverse relationship between number-average
molecular weight and molar concentration of radical initiators
normalized to concentration of monomer. The fit of data is not
sufficiently different to distinguish between the alternative
oxidation mechanisms but is slightly better for the mechanism
described by eqs 4 and 5.
The disparity between mechanisms described by eqs
1-3 and eqs 4 and 5 is readily probed by modern
computational techniques. The results of this analysis
point to the direct insertion of oxygen between carbon
and boron (eq 4) being the favored mechanism (Figure
3B). The calculations were carried out using density
functional theory30,31 and the 6-31G* basis set (B3LYP//
6-31G*) contained in the SPARTAN program.33 The
peroxide formed via eq 3 or 4 should be stable given a
dissociation energy of ca. +29 kcal/mol, yet we observe
that the radicals formed, and the resulting adhesion
occur readily at room temperature.33 However, the
calculations reveal that initial formation of the peroxide
releases more than enough energy (-55 kcal/mol) to
subsequently homolyze the oxygen bond. The fact that
amine is not required to produce adhesion (acting as
an inducer of the homolysis reaction) suggests that the
release of energy is at least somewhat adiabatic. We also
found that formation of the trialkylboroxine (eq 10)
resulting in alkyl radical initiator is exothermic by 80
kcal/mol, consistent with the reaction products found
in Figure 2.
The same experiment performed with diethylmethox-
yborane (DEMOB) resulted in the TIC presented in
Figure 4. The product distributions with TEB and
DEMOB are strikingly different. The chromatogram
shows the maleic anhydride adduct to pristane is
formed, but only in trace amounts when catalyzed by
DEMOB. In the chromatogram the peak at 12.13 min
is unmodified pristane, and 13.1, 15.92, and 20.36 min
are pristane impurities (impurities in pristane were
unique from bottle to bottle but were alkanes). Most of
the detectable borane in the DEMOB system gets tied
up in the reaction of an ethylmethoxyboroxy radical with
MAH (8.52 min). Surprisingly, some of the borane is
again tied up in triethylboroxine (2 min) despite the fact
the B-O bond is more stable than the B-C bond.34 No
trialkylborane was detected in the DEMOB sample.
This suggests the reaction of an ethyl radical to dem-
product distributions were different. The compound
2,6,10,14-tetramethylpentadecane (“pristane”) is a use-
ful model for polypropylene, and maleic anhydride
(MAH) is a grafting vinyl monomer particularly useful
in this experiment since it will not homopolymerize.
Thus, the graft reaction of MAH and pristane results
in products with well-characterized structures and
molecular weights. When TEB is used to graft MAH to
pristane, the total ion chromatogram (TIC) of Figure 2
is observed with the resulting products determined by
mass spectrometry. Figure 2 shows that TEB is efficient
at grafting MAH to pristane, and there is a slight
preference for grafting to the 2-position relative to the
6-positionspresumably a result of steric availability. All
of the borane found in this system is accounted for in
the peak at 3.82 min as triethylboroxine. The large peak
at 13.97 min is unreacted pristane, and the peak at 9.02
min is the reaction product of MAH with two ethyl
radicals. The peak at 6.31 min represents the reaction
product of a single ethyl radical with MAH.
The presence of butanol has been observed in head-
space measurements of similar systems (with TBB),
confirming that alkoxy radicals are generated and are
available for abstraction from hydrocarbon substrates
such as pristane. Also indicating the presence of alkoxy
radicals is the presence of diesters of maleic acid
captured in the small peaks between 10 and 12 min,
materials that are not detected with DEMOB initiation.
The fact that only boroxine is present as a product of
borane reactions suggests that borane undergoes only
a single oxidation with O2 and that the cyclic structure
is preceded by homolysis of the peroxide structure.
Trimerization of the singly oxidized material with the
concomitant production of alkyl radical completes the
sequence (vide infra). The results strongly indicate that
the initiation and propagation reactions cascading from
eq 3 or 5 are then

Macromolecules, Vol. 37, No. 21, 2004
Trialkylborane Promoted Adhesion 7977
F igu r e 2. TIC of products resulting from the reaction of pristane, MAH, and TEB.
clization to form the boroxine. In fact, the methyl radical
has been trapped by experiments using TEMPO as a
radical trapping agent.
Employing the same bond energy calculations as
before reveals that the dissociation energy of the
peroxide formed by insertion of oxygen between a B-C
bond of DEMOB is favored but substantially less (by
ca. 12 kcal/mol) than the same dissociation of the
peroxide formed from a trialkylborane (Figure 5). This
difference is presumably responsible for the fact that
the same reactions are available to DEMOB for grafting,
but the occurrence of such reactions is much less. The
virtually inert dialkoxyboranes are reduced an ad-
ditional 10 kcal/mol in energy change through peroxide
homolysis.
These experiments show that the unexpected room
temperature adhesion to polyolefin surfaces with no
surface pretreatment can be ascribed to previously well-
characterized radical processes that occur within a
fortuitous window of conditions. While the exact se-
quence by which radical fragments get to their eventual
products is not known with specificity, calculations
strongly suggest that the preferred mechanism of bo-
rane oxidation occurs initially with oxygen insertion.
The fragments themselves are reasonable byproducts
of the basic reactions described by the equations. The
implications of this result to other systems such as
peroxymerics,29 metal alkyls,35,36 and reactions with
cocatalysts29 achieving a similar result are clear.
F igu r e 3. Calculated energies for the reaction path from
trialkylborane and oxygen to radicals using two mechanisms
proposed in refs 25 and 26, respectively. A positive number
signifies an endothermic reaction while a negative number
corresponds to an exothermic process. The mechanism in B is
energetically favored.
ethylate DEMOB and form an ether (lost to the head-
space) and a diethylboroxy radical. The stepwise reac-
tion of this material with successive demethylated
DEMOB molecules ends with the intramolecular cy-

7978 Sonnenschein et al.
Macromolecules, Vol. 37, No. 21, 2004
Refer en ces a n d Notes
(1) Brewis, D. M.; Briggs, D. Polymer 1981, 22, 7-16.
(2) Pocius, A. V. Adhesion and Adhesives Technology; Hanser:
Munich, 1997; pp 183-214.
(3) Rauhut, H. W. Adhes. Age 1969, 12, 28-38.
(4) Devine, A. T.; Bodnar, M. J . Adhes. Age 1969, 12, 35-37.
(5) Clemens, R. J .; et al. Prog. Org. Coat. 1994, 24, 43-54.
(6) Badey, J . P.; et al. Int. J . Adhes. Adhes. 1996, 16, 173-178.
(7) Schonhorn, H.; Hansen, R. H. J . Appl. Polym. Sci. 1967, 11,
1461-1474.
(8) Westerdahl, C. W.; Hall, J . R.; Schramm, E. C.; Levi, D. W.
J . Colloid Interface Sci. 1974, 47, 610-620.
(9) Blais, P.; Carlsson, D. J .; Csullog, G. W.; Wiles, D. M. J .
Colloid Interface Sci. 1974, 47, 636-649.
(10) Schmitz, P. J .; Holubka, J . W. J . Adhes. 1995, 48, 137-148.
(11) Brewis, D. M.; Mathiewson, I.; Wolfsenberger, M. Int. J .
Adhes. Adhes. 1995, 15, 87-90.
(12) Yang, J .; Garton, A. J . Appl. Polym. Sci. 1993,48, 359-370.
(13) (a)Sonnenschein, M. F.; Webb, S. P.; Rondan, N. U.S. Patent
6,706,831, 2004. (b) Sonnenschein, M. F.; Webb, S. P.;
Rondan, N. U.S. Patent 6,713,578, 2004. (c) Sonnenschein,
M. F.; Webb, S. P.; Rondan, N. U.S. Patent 6,730,759, 2004.
(d) Webb, S. P.;. Sonnenschein, M. F. U.S. Patent 6,740,716,
2004. (e) Sonnenschein, M. F.; Webb, S. P.; Wendt, B.
L.; Harrington, D. R. US Patent 6,762,260, 2004.
(f) Sonnenschein, M. F.; Webb, S. P.; Wendt, B. L.;
Harrington, D. R. U.S. Patent 6,777,512.
(14) Skoultchi, M. M.; Merlo, N. V. U.S. Patent 5,106,928, 1992.
(15) Zharov, J . V.; Krasnov, J . N. U.S. Patent 5,539,070, 1996.
(16) Maandi, E. U.S. Patent 6,632,908, 2003.
(17) Kendall, J . L.; Righettini, R. F.; Abbey, K. J . U.S. Patent
6,646,076, 2004.
F igu r e 4. GC chromatogram of reaction product of DEMOB
with MAH and pristane. Reaction and measurement conditions
are the same as in Figure 2. Very little heat was generated by
the mixing of DEMOB into the pristane-MAH mixture,
however.
(18) Pesetskii, S. S.; J urkowski, B.; Makarenko, O. A J . Appl.
Polym. Sci. 2002, 86, 64-72.
(19) Yamamoto, K.; Tanaka, H.; Sakaguchi, A.; Shimada, S.
Polymer 2003, 44, 7661-7669.
(20) Bergbreiter, D. E.; et al. In Polymer Surfaces and Interfaces:
Characterization and Application; Mittal; Lee, K.-W., Eds.;
Van Nostrand: New York, 1997; pp 3-18.
(21) Bergbreiter, D. E.; Xu, G.-F.; Zapata, C. Macromolecules 1994,
27, 1597-1602.
(22) Ingold, K. U. In Free Radicals; Kochi, J ., Ed.; Wiley: New
York, 1973; Vol. 1, p 80.
(23) Ollivier, C.; Renaud, P. Chem. Rev. 2001, 101, 3415-3434.
(24) Ingold, K. U. In Free Radicals; Kochi, J ., Ed.; Wiley: New
York, 1973; Vol. 1, p 88.
(25) Brown, H. C.; Midland, M. M. Chem. Commun. 1971, 699-
700.
(26) Welch, F. J . J . Polym. Sci. 1962, 61, 243-252.
(27) Hoberg, H.; Gotz, V. J . Organomet. Chem. 1976, 118, C3-
C5.
(28) Furukawa, J .; Tsuruta, T; Imada, T.; Fukatani, H. Makromol.
Chem. 1959, 31, 122-139.
(29) Dudonov, V. A.; Grishin, D. F.; Aksenova, I. N Polym. Sci.
1993, 35, 1744-1746.
F igu r e 5. Energetics of the formation of radical products from
the reaction of oxygen with DEMOB assuming the energeti-
cally favored oxygen insertion mechanism of Figure 3B. The
reaction to products is favored but substantially less so than
the reaction of TEB.
(30) Kohn, W.; Sham, L. J . Phys. Rev. 1965, 140, A1133.
(31) Becke, A. D. J . Chem. Phys. 1993, 98, 1372.
(32) Spartan 2004, Wavefunction Inc., Irvine, CA.
(33) The reliability of these calculations is illustrated by compar-
ing the calculated dissociation energy of diethyl peroxide 36.5
kcal/mol with the experimental value of 37.9 ( 1 kcal/mol.
Cf.: Batt, L.; Christic, K.; Milne, R. T.; Summers, A. J . Int.
J . Chem. Kinet. 1974, 6, 813.
Ack n ow led gm en t. We acknowledge Dr. Ray Drum-
right for helpful discussions and Dr. David Meunier for
molecular weight measurements.
(34) Midland, M. M.; Brown, H. C. J . Am. Chem. Soc. 1971, 93,
1506-1509.
(35) Ryu, I.; Araki, F.; Minakata, S.; Komatsu, M. Tetrahedron
Lett. 1998, 39, 6335-6336.
Su p p or tin g In for m a tion Ava ila ble: A compendium of
TIC and mass spectral analyses based on fragmentation
patterns for all experiments mentioned in this article. This
material is available free of charge via the Internet at
http://pubs.acs.org.
(36) Bertrand, M. P.; Feray, L.; Nouguier, R.; Perfetti, P. J . Org.
Chem. 1999, 64, 9189-9193.
MA040095F