Synthesis of water soluble hyperbranched polyurethanes using selective activation of AB2 monomers

Article abstract of DOI:10.1021/ma0489218

The synthesis of a novel water-soluble hyperbranched polyurethane is described. The synthetic strategy involves a new approach to hyperbranched polymerization through the selective activation of the A functionality of the unprotected of AB₂ monomer. 1,1′-Carbonyl diimidazole is added at room temperature to an amino-diol monomer with the corresponding selective formation of an active carboxamide and no detectable reaction at the hydroxyl B functional groups. The active monomer self-condenses to form hyperbranched polyurethanes with measured degrees of branching equaling 0.6 and M_a up to 11 000 Da. A series of model reactions has been conducted to confirm the polymerization mechanism, and ideally branched model molecules have been used to assign the ¹⁵N NMR spectra of the hyperbranched polymers. Reaction conditions have been varied, and the factors affecting the polymerization are discussed.

Full text of DOI:10.1021/ma0489218

9418
Macromolecules 2004, 37, 9418-9430
Synthesis of Water Soluble Hyperbranched Polyurethanes Using
Selective Activation of AB Monomers
2
Steve P. Rannard,* Nicola J. Davis, and Ian Herbert
Unilever Research and Development Port Sunlight, Quarry Road East, Bebington,
Wirral, CH63 3JW, United Kingdom
Received June 1, 2004; Revised Manuscript Received October 1, 2004
ABSTRACT: The synthesis of a novel water-soluble hyperbranched polyurethane is described. The
synthetic strategy involves a new approach to hyperbranched polymerization through the selective
activation of the A functionality of the unprotected of AB monomer. 1,1′-Carbonyl diimidazole is added
2
at room temperature to an amino-diol monomer with the corresponding selective formation of an active
carboxamide and no detectable reaction at the hydroxyl B functional groups. The active monomer self-
condenses to form hyperbranched polyurethanes with measured degrees of branching equaling 0.6 and
n
M up to 11 000 Da. A series of model reactions has been conducted to confirm the polymerization
1
5
mechanism, and ideally branched model molecules have been used to assign the N NMR spectra of the
hyperbranched polymers. Reaction conditions have been varied, and the factors affecting the polymeri-
zation are discussed.
Introduction
monomer. This approach was similar to our earlier
reported synthesis of polycarbonate dendrimers,13 where
the selectivity of imidazole carboxylic esters was im-
portant in directing reaction at the primary hydroxyls
of a triol containing one secondary and two primary
hydroxy groups.
The introduction of branching into linear polymers
has been the subject of intensive study by many groups
for the last two decades. Specifically, structures with
ideal, controlled branching such as dendrimers,1,2 den-
dronized linear hybrid materials containing dendrons
3
,4
Hyperbranched polyurethanes were also first reported
attached to linear polymer chains, and statistically
5
,6
14
branched or hyperbranched polymers have received
significant attention, and a number of chemistries have
been synthesized and reported.
by Spindler and Fr e´ chet using thermally labile blocked
isocyanates. The AB2 monomers contained a free benzyl
alcohol A functionality and two phenol carbamate B
groups that liberated isocyanate on refluxing in THF.
Kumar and Ramakrishnan15 have also used thermal
decomposition to generate active functionality and form
hyperbranched polyurethanes. In this case, 3,5-dihy-
droxy benzoyl azide, in the presence of catalyst at 110
°C, liberated the dihydroxy isocyanate that polymerized
in 95% yield. Azide chemistry has also been utilized by
Tang and co-workers16 to form alternating ether-
urethane hyperbranched polymers using a very similar
strategy. More recently, Bruchmann et al. have devel-
oped a commercially viable process using the selective
reaction of hindered isocyanates in a self-cross-linking
reaction to form hyperbranched polyurethanes on an
Highly branched polyurethanes are problematic to
synthesize due to the inherent difficulty in controlling
the reactive chemistry that is typically used during the
production of linear materials. Very few examples of
ideal branching in polyurethanes have been reported
7
with Spindler and Fr e´ chet being the first to demon-
strate the synthesis of aromatic urethane dendrimers,
with an alternating ether-urethane layer structure,
using a convergent growth approach. Alternating aro-
matic-aliphatic urethane dendrimers, also containing
ether links, have been synthesized by Taylor and
8
,9
Puapaiboon using a modified Curtius reaction. The
synthesis of aliphatic branched polyurethanes has been
10
17
published by Clark et al., but only the first generation
dendrimer was reported due to the complex divergent
synthesis used. Aliphatic dendrimers with alternating
urethane-urea links have been described up to the
second generation by Peerlings et al using a novel AB-
CD2 approach, directly from a bisisocyanate AB mono-
mer with differential reactivity.1
Very recently, we have reported the synthesis of the
first true aliphatic homodendrimers up to the fourth
generation, where the repeating structure is completely
conserved from the central branching amine through to
the surface groups.12 Our convergent growth strategy
relied upon the selective reaction of imidazole carboxylic
esters, derived from the reaction of 1,1′-carbonyl diimi-
dazole (CDI) and secondary alcohols, with the primary
amine functionality of a new diamino-alcohol AB2
industrial scale.
Our recent reports of reactions with carbonyl imida-
zole containing compounds have demonstrated the
18
19
highly selective formation of urethanes, carbonates,
18
and amides without the need to use protection/
deprotection strategies. These techniques have been
used successfully to synthesize aliphatic polyamide,
1
20
12
13
polyurethane, and polycarbonate dendrimers by
reacting CDI with the single focal point functionality
of a dendron in convergent growth strategies. The
introduction of the carbonyl imidazole group therefore
has been achieved with great success, in conditions
where the possibility of side reactions with other
functional groups has been removed. In this study, we
report the use of CDI in the selective activation of
unprotected AB2 monomers and their subsequent po-
lymerization to form a novel class of water-soluble
hyperbranched polyurethanes.
*
Corresponding author. E-mail: steven.rannard@unilever.com.
1
0.1021/ma0489218 CCC: $27.50 © 2004 American Chemical Society
Published on Web 11/17/2004

Macromolecules, Vol. 37, No. 25, 2004
Water Soluble Hyperbranched Polyurethanes 9419
Experimental Procedures
mixture was stirred at room temperature for 20 h forming
carboxamide 8 that was not isolated. 1-Ethylpropylamine (0.54
g, 6.20 mmol) was added to the mixture, and the reaction was
heated to 60 °C for 6 h and allowed to cool to room tempera-
ture. The reaction mixture was concentrated in vacuo, and the
Materials. All reagents and solvents (Aldrich Chemical Co.,
UK) were used as received. In all cases, the highest available
purity (usually >99%) of each reagent was used. HPLC grade
toluene (<0.03% v/v H
2
O) and HPLC grade THF (<0.02% v/v
residue was dissolved in CH
water, dried over anhydrous Na
in vacuo to give the urea 9 as a colorless gum (99.5%). Route
: 1-ethylpropylamine (0.54 g, 6.16 mmol) was added dropwise
2
Cl
2
, washed three times with
H
2
O) were used as solvents.
2 4
SO , filtered, and concentrated
Characterization. H, 13C, and 15N NMR measurements
1
were conducted using a Bruker 300 MHz spectrometer. Mass
spectrometry was conducted using either atmospheric pressure
ionization (API-MS) or time-of-flight-matrix assisted laser
desorption ionization (TOF-MALDI). API-MS and HPLC-MS
were conducted using flow injection analysis comprised of a
Hewlett-Packard 1050 series pump and autosampler linked
to a VG Platform II spectrometer. TOF-MALDI was achieved
using either a Kratos Kompact MALDI IV instrument operat-
ing in linear mode or a Bruker BIFLEX III TOF-MALDI
instrument operating with continuous gridless extraction in
the linear mode. All MALDI experiments were achieved using
a gentisic acid (DHB) matrix. HPLC-ELSD was conducted
using a Hewlett-Packard 1090 series chromatograph with an
ACS 750/14 evaporative light scattering detector (ELSD).
Synthesis of Imidazole Carboxamide of 1-Ethylpro-
pylamine (4). A solution of 1-ethylpropylamine (EPA, 3) (1.08
g, 12.33 mmol) in toluene (10 mL) was added dropwise to a
stirred mixture of 1,1′-carbonyl diimidazole (CDI) (2.00 g, 12.33
mmol) and toluene (50 mL). The reaction was stirred at room
temperature and became clear in less than 1 min. The clear
solution was left to stir overnight. The crystallized solid
imidazole that formed was removed by filtration, and the
filtrate was concentrated in vacuo. The residue was dissolved
2
to a stirred mixture of CDI (1.00 g, 6.16 mmol) in THF (50
mL). The reaction was stirred at room temperature for 20 h
forming carboxamide 4, which was not isolated. N-(3-Amino-
propyl)-diethanolamine 2 (1.00 g, 6.16 mmol) was added, and
the reaction mixture was stirred at room temperature for a
further 18 h. The reaction mixture was concentrated in vacuo,
and the residue was dissolved in CH
with water, dried over anhydrous Na
concentrated in vacuo to give urea 9 as a colorless gum (99.3%)
2
Cl
2
, washed three times
2
SO , filtered, and
4
1
Analysis for urea 9 was identical through routes 1 and 2:
NMR (CD OD): 1.09 (CH -CH ), 1.42-1.87 (-CH -), 2.78 (CH
N), 3.37 (CH sNHCdO), 3.67 (CH -CH-CH ), 3.85 (CH -OH),
.17 (NH-CO-NH); C NMR (CD OD): 11.29 (CH -CH ), 22.95
and 29.46 (-CH -), 39.91 (CH sNHCdO), 54.26 (CH N), 54.35
CH), 57.99 (NCH ), 61.10 (CH OH), 161.73 (NH-CO-NH).
Symmetric Urea of N-(3-Aminopropyl)-diethanol-
H
2
-
3
2
3
2
2
2
2
2
13
6
3
2
3
2
2
2
(
2
2
amine (10). N-(3-Aminopropyl)-diethanolamine 2 (1.62 g,
1
5
0.00 mmol) was added to a stirred mixture of CDI (0.81 g,
.00 mmol) in THF (50 mL). The mixture was stirred at room
temperature for 24 h and concentrated in vacuo. The crude
residue was analyzed by NMR spectroscopy for evidence of
1
carbonate, urethane, or urea formation: H NMR (CD
3
OD):
sNHCd
-OH), 5.92 (NH-CO-NH); C NMR (CD OD):
-CH -CH ), 40.18 (CH sNHCdO), 54.39 (CH N),
8.07 (NCH ), 61.05 (CH OH), 161.59 (NH-CO-NH). Signals
in CH
drous Na
carboxamide 4 as a colorless liquid (87.4%). H NMR (CDCl
2
Cl
2
, washed three times with water, dried over anhy-
1
.82 (CH -CH -CH2), 2.63-2.82 (NCH -), 3.35 (CH
2 2 2 2
SO , filtered, and concentrated in vacuo to give the
4
13
2
O), 3.82 (CH
2
3
1
3
):
2 3 2 3 2 2
.96 (CH -CH ), 1.57 (CH -CH ), 3.80 (CH -CH-CH ), 7.00
2
5
9.04 (CH
2
2
2
2
2
0
2
2
(
imidazole, CHdCHsNdCH), 7.71 (imidazole, CHdCHsNd
for the imidazole byproduct were also present.
13
CH), 8.21 (NH-CO-Imid), 8.32 (imidazole, NsCHdN);
C
t-
G1 Dendrons BOC Protected 1-[N,N-Bis(2-hydroxy-
NMR (CDCl
3 2 3 2 3 2
): 10.66 (CH -CH ), 27.43 (CH CH ), 54.98 (CH -
ethyl)amino]-2-propanol (12). The synthesis of the G1
CH-CH
2
), 116.95 (imidazole, CHdCHsNdCH), 129.40 (imi-
6,11
1
dendron has been previously reported.
Analytical data:
), 1.48 (C(CH ), 2.56-2.95
), 3.73 (CH(OH)), 4.12 (CH sOCd
O). 13C NMR (CDCl ) 20.36 (CH-CH ), 28.37 (C(CH ) ), 54.04
H
dazole, CHdCHsNdCH), 136.37 (imidazole, CHdCHsNd
CH), 149.43 (NH-CO-Imid); m/z (API-MS) 182.35 (MH ).
NMR (CDCl
3
): 1.11 (CH(OH)-CH
3
3 3
)
+
(N-CH ), 3.34 (CH(OH)CH
2
3
2
N,N′-Bis(1-ethylpropyl)urea (5). A solution of EPA 3 (2.15
g, 24.66 mmol) in toluene (10 mL) was added dropwise to a
stirred mixture of CDI (2.00 g, 12.33 mmol) and toluene (50
mL). The reaction was stirred at room temperature and
became clear in less than 1 min. The clear solution was left to
stir overnight. The crystallized solid imidazole that formed was
removed by filtration and dried (1.67 g, 24.53 mmol. 99.5%
yield of byproduct). The filtrate was concentrated in vacuo,
3
3
3 3
(NCH CH ), 63.86 (NCH CH), 64.40 (CH(OH)), 65.10 (CH -
2
2
2
2
OCdO) 82.73 (C(CH ) ), 154.19 (CdO). 15N NMR (CDCl ,
3
3
3
formamide standard) -91.0 (N(R) ); m/z (API-MS) 364.63
3
+
+
(MH
G2 Dendron (13). The synthesis of the G2 dendron has
been previously reported. Analytical data: H NMR (CDCl3)
1.16 (CH(OH)-CH ), 1.29 (CH-CH ), 1.55 (C(CH ) ), 2.52-2.92
), 386.58 (MNa ).
6
1
3
3
3 3
and the residue was dissolved in CH
with water, dried over anhydrous Na
concentrated in vacuo to give the urea 5 as a colorless solid
2
Cl
2
, washed three times
SO , filtered, and
2 2
(N-CH ), 4.01-4.23 (CH(OH) and CH sOCdO), 4.84 (CHs
OCdO). 13C NMR (CDCl ) 20.16 (CH-CH ), 21.98 (CH CH-
3 3 3
2
4
(OH)), 30.00 (C(CH ) ), 55.83 (NCH CH ) 61.95 (NCH CH),
3
3
2
2
2
1
(
43.7%). H NMR (CDCl
3
): 0.94 (CH
-CH-CH ), 4.96 (NH-CO-NH); C NMR
): 10.57 (CH -CH ), 28.38 (CH CH ), 52.92 (CH -CH-
), 159.26 (NH-CO-NH); m/z (API-MS) 201.47 (MH ).
2
-CH
3
), 1.29-1.61 (CH
2
-
67.14 (CH OCdO), 75.82 (CHOCdO), 84.08 (C(CH ) ), 155.69
2
3 3
1
3
15
CH
3
), 3.54 (CH
2
2
(CdO), 157.01 (CdO). N NMR (CDCl , formamide standard)
3
(
CDCl
3
2
3
2
3
2
+
-91.0 (N(R) focal point), -91.5 (N(R) terminal groups). m/z
3
3
+
CH
2
(API-MS) 942.84 (MH ), 964.43 (MNa ).
+
Cyclic Carbamate of 2-Amino-2-methyl-1,3-propanediol
7). 2-Amino-2-methyl-1,3-propanediol 1 (1.00 g, 9.5 mmol) was
G0 DendrimersTris(urethane) (14). The general syn-
10
(
thesis of the GO dendrimer 14 has been previously reported.
1
added dropwise over 15 min to a cooled (0 °C), stirred mixture
of CDI (1.55 g, 9.5 mmol) and THF (30 mL). The mixture was
stirred at 0 °C for a further 4 h and then allowed to warm to
room temperature. The clear reaction mixture was concen-
3 3 3
Analytical data: H NMR (CDCl ) 1.55 (C(CH ) ), 2.52 (NH-
13
CH ), 3.15 (N-CH ), 5.38 (CdONH) C NMR (CDCl ) 28.68
2
2
3
(C(CH ) ), 38.78 (NHCH CH ) 54.47 (NCH ), 79.38 (C(CH ) ),
3
3
2
2
2
3 3
1
5
156.68 (CdO), 157.01 (CdO). N NMR (CDCl , formamide
3
trated in vacuo, and the residue was dissolved in CH
washed three times with water, dried over anhydrous Na
filtered, and concentrated in vacuo to give the cyclized car-
2
Cl
2
,
,
standard) -34.1 (NHC(dO)O),-86.8 (N(R) ). m/z (API-MS)
3
+
+
2
SO
4
450.64 (MH ), 472.36 (MNa ).
G1 DendrimersCarbonate-Urethane Layer Copoly-
mer (15). A mixture of G1 dendron 12 (3.00 g, 8.23 mmol),
CDI (2.01 g, 12.40 mmol), and KOH (0.2 g, 3.60 mmol) in
toluene (20 mL) was heated at 60 °C for 1 h. Tris(2-aminoethyl)
amine (0.40 g, 2.74 mmol) was added to the reaction, and the
reaction was heated for a further 6 h. The reaction mixture
was concentrated in vacuo, and the residue was dissolved in
1
bamate as a colorless liquid (99.7%). H NMR (CDCl
CR -CH ), 3.52 (CH -OH), 4.02 (CHH-O), 4.41 (CHH-O-);
NMR (CDCl ): 22.74 (CR -CH ), 59.48 (CR ), 67.18 (CH -OH),
3.46 (CH -O), 160.52 (CdO).
Asymmetric Urea of 1-Ethylpropylamine and N-(3-
Aminopropyl)-diethanolamine (9). The asymmetric urea
was synthesized through two different routes; route 1: a
3
): 1.21
1
3
(
3
3
2
C
3
3
3
3
2
7
2
9
CH
Na
dendrimer 15 as a colorless gum (93.1%) H NMR (CDCl
1.21 (CH-CH ), 1.47 (C(CH ), 2.45-2.95 (HN-CH ), 3.18 (N-
2
Cl
2
, washed three times with water, dried over anhydrous
solution of N-(3-aminopropyl)-diethanolamine 2 (1.00 g, 6.16
mmol) in THF (20 mL) was added dropwise to a stirred
mixture of CDI (1.00 g, 6.16 mmol) in THF (30 mL). The
2
SO
4
, filtered, and concentrated in vacuo to give the
1
3
):
3
3
)
3
2

9420 Rannard et al.
Macromolecules, Vol. 37, No. 25, 2004
CH
2
) 4.12 (CH sOCdO), 4.84 (CH
2
3
-CH). 13C NMR (CDCl
NH), 53.97 (NCH
), 153.75
, formamide
)+, -89.7 (N(R) ).
m/z (API-MS) 1314.60 (MH ), 1336.70 (MNa ).
3
)
Scheme 1
1
8.53 (CH-CH ), 27.98 (C(CH ), 39.17 (CH
3
3
)
3
2
2
),
6
0.14 (NCH , core), 65.29 (CH OCdO), 82.10 (C(CH )
2
2
3 3
15
(
OC(dO)O), 156.69 (NHC(dO)O). N NMR (CDCl
3
standard) -36.2 (NHC(dO)O), -86.9 (N(R)
3
3
+
Hyperbranched Polyurethane SynthesissGeneral Pro-
cedure. A solution of N-(3-Aminopropyl)-diethanolamine (3.00
g, 18.49 mmol) in THF (30 mL) was added dropwise to a
solution of CDI (3.00 g, 18.49 mmol) in THF (45 mL) and
stirred at room temperature for 20 h. The reaction mixture
was concentrated in vacuo to leave a residue that was heated
at 90 °C for 3 h (optionally, a base such as KOH may be added
prior to solvent removal). After being cooled to 40 °C, THF
branched, linear, and cyclic structures that may be
formed and the predominance of analytical techniques
that have been developed and optimized for linear
polymers. To investigate the use of CDI in hyper-
branched polymer synthesis, a series of model reactions
have been conducted and a number of model materials
(
60 mL) was added to the residue and heated at 60 °C for a
further 1.5 h. After being cooled to room temperature, the
solvent was decanted, and the residue was dried to give the
1
polymer as a pale colorless sticky material. H NMR (D
2
O):
2
1
.57 (CH
NR ), 3.59 (CH
O): 25.99 (CH
NHCdO), 51.74 and 52.25 (CH
CH OH), 62.62 (CH OC(dO)NH), 158.19 (O-CO-NH).
NMR (D
N(R) dendritic fragment), -81.00 (N(C
fragment), -81.24 (NR(C OH) terminal fragment) BF )
.6. m/z (TOF-MALDI) repeat unit 188.23 Da.
2
-CH
2
-CH
-OH), 4.05 (CH
-CH -CH ), 38.27 (H
N), 55.34 (NCH
2
), 2.32-2.86 (N-CH
2
), 3.05 (NH(CH
2
)
2
CH -
1
13
1
3
have been synthesized to aid assignment of H, C, and
2
2
2
sOC(dO)NH); C NMR
15
N NMR spectra during characterization.
(
D
2
2
2
2
2
N-CH
2
), 38.89 (CH
CH
2
s
2
2
2
OH), 58.89
Model Reactions 1: Urea and Urethane Forma-
1
5
(
2
2
N
tion using CDI. The reaction of a primary amine with
CDI, and subsequent carboxamide formation, was stud-
ied to ascertain differences between amine and alcohol
reactivity. Alcohols require heating and base catalysis
to aid the formation of imidazole carboxylic esters.19 To
evaluate primary amine reactivity, a branched amine,
2
O), formamide standard) -30.38 (NHC(dO)O), -80.68
(
3
2 4 2
H OH)(R) linear
2
H
4
2
0
Results and Discussion
1
-ethylpropylamine 3 (EPA), was added slowly to a
Monomer Selection. Previously, we have synthe-
sized urethane containing materials by reacting an
alcohol with CDI, to form an imidazole carboxylic ester,
followed by subsequent reaction of the intermediate
with a primary amine. Suitable AB2 monomers for
polyurethane formation using CDI should therefore
contain both primary amines and primary alcohols. Two
commercially available monomers were therefore chosen
for this study, 2-amino-2-methyl-1,3-propanediol 1 and
N-(3-aminopropyl)-diethanolamine 2 (ADPE).
suspension of CDI in toluene at ambient temperature
with gentle stirring (Scheme 1). The reaction became
completely clear after less than 1 min, indicating rapid
consumption of CDI as it is not soluble in toluene. The
reaction was exothermic and was left to cool overnight
with stirring. If the reaction is conducted on a 1:1 (EPA/
CDI) molar ratio, the reaction forms the isolated car-
boxamide 4 in high recovered yield (87.4%). Purification
of the reaction was accomplished through filtering the
crystallized imidazole followed by removal of solvent by
rotary evaporation, being redissolved in dichloromethane,
and being washed with water. Confirmation of carboxa-
1
2
Model Studies. 1,1′-Carbonyl diimidazole (CDI) is
a well-known coupling agent that may be used to form
carbonates, ureas, amides, urethanes, and esters. To
maximize the success of coupling reactions using CDI,
the formation of the carbonyl imidazole containing
reactive intermediates is achieved by simply stirring an
amine, acid, or alcohol with CDI in an appropriate
solvent with heating and base if necessary. The second
reactant, amine or alcohol, may be added directly to the
reaction without isolation or purification of the inter-
mediate, and especially in the case of acid imidazolides,
it is beneficial to do this as the intermediates are often
hydrolytically unstable. However, the synthesis of hy-
perbranched polymers using conventional AB2 mono-
mers necessarily requires the presence of the two
functional groups that will be coupled to give the final
polymer. CDI will only be of use for hyperbranched
polymer synthesis if it can selectively activate either the
A or B groups of the monomer in the presence of the
other.
Ideally branched dendrimers are laborious and rela-
tively difficult to synthesize and purify by any of the
reported multistep procedures, but the characterization
of each stage of growth, either divergent or convergent,
is straightforward due to the need to produce only a
single structure at each step. In contrast, hyper-
branched polymers are usually produced using a single-
pot reaction and often require little purification. The
characterization of hyperbranched polymers, however,
is particularly difficult due to the number of possible
1
13
mide formation was accomplished through H and
C
nuclear magnetic resonance (NMR) spectroscopy (CDCl3),
with a characteristic carboxamide carbonyl signal at
1
3
1
49.43 ppm ( C NMR), the shift of the methine proton
1
signal to 3.80 ppm ( H NMR), the presence of three
distinct imidazole ring proton singlets at 7.00, 7.71, and
8
.32 ppm, and a doublet at 8.21 ppm corresponding to
the carboxamide NH proton. Any free imidazole byprod-
1
uct can be identified as it only exhibits H NMR signals
for two imidazole ring proton environments at 7.10 and
7.69 ppm. This is due to the rapid proton exchange
between the two ring nitrogens that leads to the
delocalization of the CdN double bond and the subse-
quent equivalence of the protons associated with the Cd
C double bond.
Increasing the ratio of EPA/CDI to 2:1, following the
same synthetic procedure, gave the symmetric urea 5
at room temperature with a recovered yield of 43.7%.
The imidazole that crystallizes during cooling was
collected, dried, and weighed and indicated a 99.5%
formation of the insoluble byproduct suggesting that
much of the symmetrical urea had been lost during
purification. Confirmation of urea formation was also
1
13
accomplished through H and C NMR spectroscopy
(CDCl3), with a characteristic carbonyl signal at 159.26
13
ppm ( C NMR), movement of the methine proton signal
1
to 3.54 ppm ( H NMR), and the appearance of a doublet
at 4.96 ppm corresponding to the two urea NH protons.

Macromolecules, Vol. 37, No. 25, 2004
Water Soluble Hyperbranched Polyurethanes 9421
Scheme 2
The formation of a urethane from 4 was demonstrated
through the addition of 1-butanol to 4 with a catalytic
amount of KOH followed by heating at 60 °C for 6 h.
After purification, urethane 6 was readily isolated (58%)
Treatment of 2 in a similar procedure, however, gave
the asymmetric urea 9 in high yield after purification
(99.3%) (Scheme 2). The crude product mixture was
analyzed in detail using 13C and H NMR spectroscopy
(CD3OD). A single carbonyl signal corresponding to urea
formation was observed at 161.73 ppm, and no detect-
able carbonyl or proton signals corresponding to carbon-
ate or urethane formation could be seen. To ensure
correct assignment of the NMR spectra of 9, it was also
synthesized in high yield (99.5%) through reaction of
EPA 3 with CDI followed by addition of 2 to the
resulting carboxamide. Again, a single carbonyl signal
at 161.73 ppm was observed, confirming the previous
assignment. To check that symmetric urea formation
was not hindered or unfavorable when using 2, a
reaction containing a 2:1 ratio of 2 to CDI was conducted
at room temperature for 2 h, and the crude reaction
mixture was evaporated and analyzed using 13C and H
NMR (CD3OD). The crude mixture showed exclusive
formation of the symmetric urea 10, with no evidence
of residual carboxamide or unwanted urethane synthe-
sis. The carbonyl region of the 13C NMR spectrum
1
1
8
and characterized as previously described, with a
1
3
characteristic C NMR (CDCl3) carbonyl resonance at
1
57.14 ppm.
The reaction of carboxamides to give ureas and
urethanes may proceed via the elimination of imidazole
and subsequent isocyanate formation (i.e., the carboxa-
mide is a blocked isocyanate). The recovery and puri-
fication of carboxamide 4 does suggest that these
materials are relatively stable, however, and the forma-
tion of the isocyanate intermediate therefore may
require specific conditions such as the presence of base
2
2
or heating to induce reactivity.
Model Reactions 2: Selective Reaction of CDI
with the Primary Amines of Amino Alcohols.
Having established the reactivity of a model primary
amine with CDI and subsequent urethane formation,
the selective generation of the carboxamides of 1 and 2
was studied. The carboxamides are generated in the
presence of two primary hydroxyl groups, and the
prevention of urethane formation will be extremely
difficult during isolation and purification. By reacting
the carboxamides that are formed with EPA, we aimed
to trap them as the asymmetric urea and confirm the
selective reaction of CDI with the primary amine
functional group alone. Carbonate and urethane forma-
tion are indicative of the reaction of CDI with the
primary hydroxyl groups to form an imidazole carboxylic
ester and subsequent reaction with another hydroxyl
group or EPA. First, 1 was added dropwise to a
suspension of CDI in THF at 0 °C and stirred for 4 h.
EPA was added to the mixture, and the reaction was
allowed to warm to room temperature. After purifica-
tion, a cyclized urethane 7 was isolated in very high
yield (99.7%). This result was also achieved when 1 was
added to CDI and allowed to stir without added EPA.
The cyclization of 1 was not entirely unexpected as we
have seen cyclic carbonate formation when reacting 1,2-
1
1
showed a single signal at 161.59 ppm, while the H
NMR spectrum showed accurate integrals confirming
the structure, a single signal for the urea NH protons
at 5.89 ppm, and the presence of free imidazole. The
synthesis of 10 through this facile procedure dramati-
cally demonstrates the selectivity of carboxamide for-
mation and the subsequent selectivity of the carboxa-
mide for reaction with amines in the presence of
hydroxyl groups.
The selectivity and reactivity of the carboxamide of 2
suggests potential problems in the use of this AB2
monomer in hyperbranched polyurethane synthesis
using CDI. The CDI/APDE stoichiometry must be kept
as close to 1:1 as possible to ensure a very high
conversion of amine to carboxamide. If residual amine
is present after CDI consumption, the symmetrical urea
10 will be formed, leading to the uncontrolled introduc-
tion of a core terminator into the polymerization and
low molecular weight polymers being formed. Excess
CDI, however, would guarantee complete carboxamide
formation but could also lead to carbonate groups in the
final polymer through the unwanted coupling of hy-
droxyls with the excess CDI.
1
9
diols in previous studies. The formation of a cyclic
urethane is probably also due to the 1,2-substitution of
the amine and alcohol favoring ring closure after car-
boxamide formation.

9422 Rannard et al.
Macromolecules, Vol. 37, No. 25, 2004
15N NMR spectroscopy was either carried out in D2O
2) or CDCl3 (11-15) using formamide as reference, 27
(
µs (90°) excitation pulses, and a relaxation delay of 10s.
APDE 2 was analyzed using various 15N NMR condi-
tions including no proton decoupling, broadband decou-
pling, inverse-gated proton decoupling, and gated proton
decoupling. In all cases, signals arising from the primary
and tertiary amines were present at -74 and - 87 ppm,
respectively, and broadband decoupling was chosen as
the standard condition for analysis of further samples.
The carbonate dendrons 12 and 13 were synthesized
using 11 as previously reported and analyzed to
investigate the effect of branching on the 15N shift of
the tertiary nitrogen atom. A shift of 0.5 ppm from
-90.5 to - 91 ppm was observed when the two primary
alcohols of 11 were reacted. Forming the second-
generation dendron showed no change in the shift for
the tertiary nitrogen at the focus of the wedge (-91
ppm), but a further shift in the nitrogen at the surface
branch was seen (-91.5 ppm). As there are no urethane
links within 12 and 13, the urethane 14 and urethane-
carbonate copolymer 15 were also synthesized following
reported methods12 and analyzed to determine the
position of the urethane nitrogen signal and to ascertain
whether the presence of urethane groups would further
2
Figure 1. AB monomers chosen for hyperbranched polyure-
thane study.
Model Materials: 15N NMR Investigation of
Model Branched Materials. CDI has therefore been
shown to react selectively with APDE 2 to form the
carboxamide in the presence of the primary hydroxyl
groups. Self-condensation of the active intermediate 8
would lead to a highly branched polyurethane, but char-
acterization of this material will be hampered by the
distribution of structures and molecular weights that
13
are generated during this uncontrolled reaction. 15
N
NMR is the most appropriate technique to determine
the degree of branching of these hyperbranched poly-
urethanes and has been shown to be effective in previ-
2
1
15
ous studies. To aid the assignment of the N NMR
spectra, a number of model materials have been syn-
thesized and analyzed, along with APDE 2 and 1-[N,N-
bis(2-hydroxyethyl)amino]-2-propanol 11 (Figure 2).
Figure 2. Model materials and the assignment of the broadband proton decoupled 15N NMR signals (relative to formamide).

Macromolecules, Vol. 37, No. 25, 2004
Water Soluble Hyperbranched Polyurethanes 9423
Scheme 3
shift the 15N signal for the tertiary nitrogen of the
branching unit. Compound 15 is derived from the
reaction of 12 with tris(2-aminoethyl) amine, and any
further shift changes of the tertiary nitrogen of the
dendron 12 from -91.5 ppm in 13 can be related directly
to the presence of the urethane group. The urethane NH
signal was observed at -34.1 ppm and was confirmed
stirring, the second step involves the removal of solvent,
usually by rotary evaporation, followed by a melt
polymerization of the residue at 90 °C for several hours
(Scheme 3). Purification is achieved by stirring at 60
°C in THF to remove the imidazole byproduct. These
hyperbranched polyurethanes are very water soluble
and dissolved rapidly at neutral pH.
Degree of Branching Determination. The degree
of branching in hyperbranched polymers is often con-
sidered the most fundamental factor that may be
determined, and it is crucial to the confirmation of the
hyperbranched structure and the nature of the polym-
erization. Deviations from the statistical degree of
branching (DB ) 0.49) for AB2 monomers suggests that
the mechanism is encouraging the branching process,
but values lower than the theoretical value suggest that
branching during polymerization is hindered and a
significant fraction of linear segments are produced in
the structure.
1
5
as the NH through N DEPT experiments. Compound
1
4 showed a single peak corresponding to the central
tertiary nitrogen atom at -86.8 ppm, which is signifi-
cantly shifted from any of the nitrogen signals of 11-
1
3 (possibly due to hydrogen bonding) but very similar
to that of the tertiary nitrogen of 2. Interestingly, the
1
5
copolymer 15 showed two distinct N signals for the
two different tertiary nitrogen environments (-86.8
ppm ) tris urethane core, -89.7 ppm ) urethane-bis
carbonate surface). The difference between the shifts of
the surface nitrogens of 13 and 15 of approximately 2
ppm suggests that the presence of urethane groups has
15
The study of model materials by 15N NMR spectros-
copy showed significant differences between branched
nitrogen environments, and a study of hyperbranched
polyurethanes was conducted using room-temperature
overnight signal acquisition in D2O (50% v/v solution
of polymer), broadband proton decoupling, and a pulse
width of 15 µs and a 3 s relaxation delay (formamide
standard). Under these conditions, two signals were
observed at -30.38 ppm (assigned as the urethane -NH)
and -81.00 ppm (Figure 3 ; low molecular weight
sample: Mn (NMR) ) 920; DP ) 5; conversion (calcd)
) 73.6%).
Expansion of the signal at -81.00 ppm showed three
distinct peaks at -80.68, -81.00, and -81.24 ppm. This
grouping of signals was assigned using guidance from
the model molecules. The tertiary amines of the termi-
nal groups of structures 13 and 15 have 15N NMR
a significant effect on the position of the N signal. The
-5 ppm difference between these peaks and the core
3
nitrogen of 14 and 15 suggests that the number of
urethanes in proximity to the nitrogen also has a
significant effect.
This investigation of model compounds led us to
15
expect measurable N NMR differences for the tertiary
nitrogens present within the linear, dendritic, and
surface groups of the hyperbranched polyurethanes
derived from 2.
Synthesis and Characterization of Hyper-
branched Polyurethanes Derived from APDE 2.
The procedure for APDE 2 polymerization is a facile
two-step process. The first step involves carboxamide
synthesis and can be achieved through the dropwise
addition of a THF solution of 2 to a stirred suspension
of CDI in THF at room temperature. After continued

9424 Rannard et al.
Macromolecules, Vol. 37, No. 25, 2004
15
Figure 3. N NMR spectrum of the hyperbranched polyurethane of APDE and expansion of the tertiary amine signal to showing
1
n
branching. (Sample details: DP ) 5; M ( H NMR) ) 920; conversion (calculated) ) 73.6%.)
signals at significantly lower values than either the
linear equivalent structure (focal point) within 13 or the
dendritic equivalent structure (the core) of 15. This
suggests that the signal at -81.24 ppm of the hyper-
branched polyurethane is indicative of the terminal
units of the polymer. The dendritic core unit of 15 has
a much larger shift from the terminal groups (2.8 ppm)
than that shown between the terminal and the linear
focal group of 13 (only 0.5 ppm). This suggests that the
signal at -81.00 ppm may be assigned to the linear
units and the signal at -80.68 ppm to the fully branched
dendritic units. This assignment is also supported by
the approximate 1:2:2 ratio of the integrals of the three
signals (dendritic/linear/terminal) for the low molecular
weight (DP ) 5) sample shown in Figure 3 as it is not
possible to generate more dendritic than terminal
groups.
1H NMR Calculations of Conversion and Num-
ber Average Molecular Weight (Mn). The determi-
nation of conversion and number average molecular
1
weight (Mn) has been achieved using H NMR (D2O).
Spectra were acquired on crude samples to ensure
analysis of the full reaction mixture at the end of the
reaction. Imidazole was present during the analysis, but
the signals from this impurity do not interfere with
those assigned to the hyperbranched polyurethane.
1
There are four distinct signals in the H NMR spectra
that are unaffected by overlap with neighboring signals
1
or imidazole byproduct (Figure 4; Mn ( H NMR) ) 2980;
DP ) 16; conversion (calcd) ) 91%) and are indicative
of key features of the polymerization. The number of
amines present at the beginning of the reaction is easily
1
determined from the H NMR as it may calculated using
either of two distinct methylene signals at δ ) 1.57 and
3.04 ppm that stay significantly unchanged, and free
of overlap with other signals, after polymerization.
These are assigned to the central methylene of the
propyl chain between the primary and tertiary amines
(signal b, Figure 4) and the methylene closest to the
tertiary amine (signal d, Figure 4), respectively, and as
these are numerically equivalent, they can both be used
to minimize error during calculation. The number of
amines that have successfully reacted to form urethane
groups is also equal to the number of hydroxyl groups
that have reacted, and a distinct signal at δ ) 4.05 ppm
with no overlap has been assigned to the CH2-OCONH
group (signal a, Figure 4). Using the integrals of these
signals, it is possible to determine the extent of reaction
(or conversion) during each polymerization through a
Although the conditions that the 15N NMR spectra
acquired were not completely quantitative, the integra-
tion of the three tertiary amine signals repeatedly gave
a degree of branching approximately equal to 0.6 over
a series of samples. This suggests that the synthesis is
allowing more branching than would be expected from
the statistical reaction of an AB2 monomer although the
reason for this is unclear. The formation of the first
urethane at the terminal hydroxyl, however, may
activate the second hydroxyl group through hydrogen
bonding between the carbonyl group of the urethane and
the second -OH. This may enhance the nucleophilicty
of the remaining hydroxyl of linear units and favor
reaction with the focal carboxamide over terminal
groups.

Macromolecules, Vol. 37, No. 25, 2004
Water Soluble Hyperbranched Polyurethanes 9425
1
Figure 4. Crude H NMR spectrum (D
2
O) of the hyperbranched polyurethane of APDE. Assignments and repeat structures
1
used to calculate conversion and M
n
are also shown. (Sample details: DP ) 16; M
n
( H NMR) ) 2980; conversion (calculated) )
91%.)
simple ratio a/b (or a/d). In a typical reaction, it has been
possible to achieve conversions of up to 98%, although
we have found that the reaction conditions are impor-
tant to achieve high conversion as will be discussed
later.
specific chemistry that is present within the polymer
with an idealized repeat mass ) 187.23 Da (16, Figure
4). Compound 16 is not the traditional descriptive
structure used to represent hyperbranched polymers but
the structure required to determine the repeat unit
mass used above to calculate Mn. The function (DPn +
1) compensates for the number of protons that need to
be added to this idealized repeat structure at the
terminal groups, and the subtraction of 27 Da is a
correction for the lack of one carbonyl group at the
terminal monomer repeat unit. Using this approach, Mn
values up to 9000 Da have been determined from the
The end methylene of the propyl group closest to the
urethane nitrogen overlaps with other signals and
cannot be used directly for conversion or Mn calculation.
This is also true for the focal point methylene; therefore,
1
it is difficult to directly calculate Mn from the H NMR.
The number average degree of polymerization (DPn) is,
however, related to the average number of unreacted B
functional groups per polymer, through the simple
expression Bn ) DPn + 1. To calculate Bn, it is important
to know the total number B groups present (signal c,
Figure 4) and the number of polymer molecules formed
during the polymerization. Care must be taken when
integrating signal c as residual THF overlays with this
peak. The residual THF can be accurately calculated
from the signal at δ ) 1.8 ppm and used to accurately
determine the correction to the signal c integration. As
1
H NMR spectra of the polymers.
Mass Spectrometry and 13C NMR Investigations.
Matrix assisted laser desorption ionization spectrometry
(TOF-MALDI) and 13C NMR (D2O) have been used to
analyze the mechanism of the polymerization. The
model experiments showed no urea or carbonate forma-
tion when CDI was added to APDE and exclusive
formation of the carboxamide at the primary amine
group. In almost all cases, the hyperbranched polyure-
thanes that were formed showed a single carbonyl signal
in the 13C NMR spectra at δ ) 159.16 ppm, supporting
the mechanism described previously. In a number of
1
the H NMR signal for the focal point is hidden among
other signals, it is necessary to calculate the number of
unreacted focal amines that are theoretically present
and therefore the number of polymer molecules within
the sample. It is possible to do this for the polymers in
this study as the signals d and b (Figure 4) correspond
to the number of starting amines, and signal a corre-
sponds to the number of amines that have reacted. The
relationship DPn ) (c/(d - a)) - 1 can then be used to
give the number average degree of polymerization.
DPn can be easily converted into Mn through a simple
relationship Mn ) (mass of repeat × DPn) + (DPn + 1)
1
cases, however, although the H NMR has shown no
evidence of unwanted reactions, there have been extra
minor signals within the 13C NMR (D2O) spectra that
indicate urea (δ ) 161.08 ppm) or carbonate (δ ) 156.69
ppm) coupling. Carbonate coupling was very rare and
only seen at an extremely low concentration, but urea
coupling was seen repeatedly over a number of samples
(Figure 5).
Urea formation suggests that structures analogous to
10 are being formed in the early stages of the polym-
erization, probably during the formation of the activated
-
27 Da. This relationship is specific to the hyper-
branched polyurethanes of this study and reflects the

9426 Rannard et al.
Macromolecules, Vol. 37, No. 25, 2004
apparent excess of APDE and the formation of urea after
CDI consumption. Urea formation was not seen in all
polymerizations but was often eliminated when correc-
tions for CDI purity were used.
TOF-MALDI analysis of the samples also supported
the polymerization mechanism with a repeat unit of
1
88.23 Da, 16, which is clearly seen (Figure 6).
The spectra show masses up to approximately 6500
Da (DP ) 35 monomer units), but several distributions
can be seen. The strongest signals are derived from the
+
+
+
MH , MNa , and MK adducts where the mass ion is
derived from HBP-NH2 (i.e., the hyperbranched poly-
mer (HBP) with a primary amine focal group). A
distribution for the HBP-Imid (polymer with the car-
boxamide focal group) is also often seen but mainly as
+
the MK adduct. Interestingly, distributions of signals
corresponding to the MKH2 of the HBP-Imid, and
other doubly charged species (Figure 7), are regularly
seen suggesting masses >11 000 Da (DP approximately
+
1
3
Figure 5.
C NMR spectra (carbonyl region) of hyper-
6
0 monomer units). The formation of ureas or carbon-
branched polyurethanes showing urethane (a), urea (b), and
carbonate (c) signals.
ates generated through APDE coupling cannot be
analyzed via TOF-MALDI as the masses of polymers
linked through carbonate, urethane, or urea groups are
identical and the distribution of repeat units would be
the same. In several spectra, however, a grouping of
signals that did not fit any theoretical series of masses
was seen, even when considering the formation of cyclic
species. These signals had a repeating pattern corre-
sponding to increasing numbers of APDE monomers,
APDE 8. This would occur if the stoichiometry of the
reaction is not accurately maintained and there is either
an excess of APDE or the CDI being added is not of high
purity. A study of the purity of CDI was undertaken,
and a number of samples from different batches, as
received from the supplier, were studied using gas
chromatography coupled to mass spectrometry (GC-MS)
+
+
+
1
with subsequent H , Na , and K adducts, but were
approximately 57 Da lower than the spectrum of the
and H NMR. As CDI is water sensitive, the solvents
used for this study were also analyzed for trace water,
and calculations of purity were made using these
figures. The supplied CDI was very variable in purity
and in some cases was less than 90% when delivered.
Typical values of purity ranged from 99-92.5% with a
single impurity identified as imidazole, the nonvolatile
byproduct of CDI hydrolysis. This would account for the
+
MH series (Figure 7).
This appears to be a due to an impurity in the APDE
monomer that was also used as supplied. APDE is
presumably synthesized through the Michael addition
of acrylonitrile to diethanolamine followed by hydroge-
nation. Diethanolamine has a molecular weight that is
Figure 6. TOF-MALDI spectra of the hyperbranched polyurethane of APDE.

Macromolecules, Vol. 37, No. 25, 2004
Water Soluble Hyperbranched Polyurethanes 9427
1
3
+
+
Figure 7. Expansion of the TOF-MALDI spectra of samples showing strong urea carbonyl C signals (a, a′ ) MH , MNa ,
+
+
+
amine focal point (calcd: 2421.9, 2444.0; found: 2422.0, 2445.9); b, b′, b′′ ) MH , MNa , MK , diethanol amine containing species
2
+
(
calcd: 2364.9, 2387.0, 2404.1; found: 2365.2, 2388.3, 2404.1); c ) MKK , amine focal point (calcd: 2473.0; found: 2474.1); d )
2+ 2+
MHK , carboxamide focal point (calcd: 2501.0; found: 2501.8); e ) MNaK , diethanol containing species (calcd: 2530.6; found:
533.2).
2
5
7.09 Da lower than APDE 2, and this suggests that
formation of the carboxamide can also act as a base for
the urethane synthesis. In previous work, we have
employed KOH as a base to synthesize nonpolymeric
urethanes and carbonates, but the addition of a strongly
nucleophilic base may also hinder the reaction.
some polymer chains are terminated with or contain
diethanolamine residues 17 or 18 possibly linked to the
main polymer chain through an asymmetric urea link,
1
8. The reaction of CDI with primary amines occurs at
room temperature, whereas secondary amines require
heating to form the carboxamide.22 The primary amine
carboxamide of APDE, 8, that is formed during the
initial activation stages of the polymerization may
therefore react with the diethanolamine impurity re-
sulting in the capping of the growing chains. This may
either lead to a urethane linkage, which will not be
discernible by NMR, or a urea link that will be discern-
ible (Scheme 4).
A series of polymerizations heated at 90 °C for 3 h
with increasing KOH concentration (0-1 mol equivalent
based on APDE) were conducted, and the effect on M
n
1
(as measured by H NMR) was studied. In this series
of polymerizations, M varied dramatically from 5000
n
Da when no additional base was added to less than 500
Da at >0.7 mol equivalents of base (Figure 8a). Conver-
sion appears to vary linearly with base concentration
and ranged from > 97% when no extra base was added
to <20% with 1 equiv of base (Figure 8a). The additional
base therefore has a significant effect on the polymer-
ization mechanism possibly through the competitive
reaction of the hydroxide with the carboxamide focal
group or degradation of the urethane group during the
course of the polymerization. Another possibility is the
addition of water to the polymerization as the KOH
supplied has a significant amount of associated water
(approximately 15% w/w). This would potentially lead
to the degradation of the carboxamide and limit the
growth of the polymer.
This may be the most likely source of the urea
1
3
resonance that can be seen in the C NMR spectra of
some polymers and may explain why the urea resonance
was not always eliminated through correction for CDI
purity. The presence of end-capped material will also
lead to a decrease in the achievable molecular weight
of the polymerization. The level of diethanolamine
within the supplied sample was not easily determined,
and purification was not attempted.
Effect of Reaction Conditions on Conversion
and Mn: Reaction Time, Temperature, and aAd-
dition of Base. The synthesis of urethanes via the
reaction of a carboxamide and an alcohol requires the
nucleophilic attack of the hydroxyl, probably as an
alkoxide, at the carbonyl group. The addition of a base
to the reaction would be expected to aid the polymeri-
zation reaction. However, APDE is a basic monomer
containing a tertiary amine branching atom, and the
imidazole byproduct that is eliminated during the initial
A comparison of reactions with added KOH (0.1 mol
equiv) and polymerizations without added base was
conducted to understand the role of reaction tempera-
ture and length of reaction on conversion and Mn. In
all cases, the polymerizations with 0.1 mol equiv of KOH
achieved significantly lower molecular weights and
conversions. Reaction temperatures were varied from
60 to 145 °C, and the polymerizations were conducted

9428 Rannard et al.
Macromolecules, Vol. 37, No. 25, 2004
Figure 8. Effect of reaction conditions on the polymerization of APDE monomer and the formation of hyperbranched polyurethane.
Scheme 4
for 3 h as stated previously. Without added base, the
polymerization at 60 °C was unsuccessful, and almost
no conversion could be detected, but conversions of
almost 80% (with low Mn) were achieved when KOH
was added at this temperature. At temperatures >90
The effect of temperature on conversion does not seem
to be too dramatic in the absence of base, but a large
effect is seen when base in present. This may be
indicative of the hydroxide driven decomposition of the
carboxamide and subsequent regeneration of primary
amine at elevated temperatures, however, the carboxa-
mide appears to be temperature sensitive under the
conditions of the reaction with no added KOH. The
formation of polymer appears to be faster in the absence
of KOH with higher conversions being achieved at all
temperatures, although the detrimental effect of higher
°
C, however, the polymerizations without base showed
much higher conversions and Mn than those with added
KOH (Figure 8b). As the temperature of polymerization
is increased, the conversions and Mn achieved decrease
suggesting that either the polymer is not stable at these
higher temperatures or the carboxamide is decomposing.

Macromolecules, Vol. 37, No. 25, 2004
Water Soluble Hyperbranched Polyurethanes 9429
temperatures on Mn is more highly pronounced.
Scheme 5
A number of polymerizations were conducted at 90
C with either no added base or with 0.1 equiv of KOH
°
for increasing reaction times from 1 to 7.5 h (Figure 8c).
Mn decreases considerably with the extended reaction
time, both in the presence and absence of added base.
Conversion, however, appears to be relatively stable in
both cases, and only minor differences were observed
after 1 or 7.5 h. This suggests that there may be polymer
degradation or cyclization reactions that are liberating
low molecular weight species, although this would affect
conversion if these species contained amine functional-
1
13
ity. No evidence of cyclization was seen in the H or
C
NMR studies, and species corresponding to cyclized
structures could not be observed in the TOF-MALDI
spectra. It also is possible that intermolecular coupling
reactions are occurring. If these lead to the elimination
of low molecular weight branched oligomeric species
with amine focal functionality, the number of focal
groups would not change dramatically; therefore, the
conversion would not be greatly affected. The overall
molecular weight distribution would, however, broaden
as Mn would decrease and Mw would increase. Unfor-
tunately, reproducible aqueous GPC conditions could
not be achieved for these materials during our study,
and data describing Mw or polydispersity were not
obtained.
to be 292.33 Da based on complete reaction of all
hydroxyl groups). A repeating pattern of 24.01 Da,
however, was apparent, which suggests multiply charged
species.
Reactivity of Hydroxyl GroupssFormation of
Benzoic Acid Esters. The hyperbranched polyure-
thanes of APDE are very water soluble presumably due
to the large numbers of hydroxyl groups present com-
bined with the tertiary amine branching atom of the
repeat unit. The material should be readily modified
through reaction of the hydroxyl groups to produce a
range of materials, and this has been investigated
through the formation of benzoate esters in an initially
heterogeneous reaction in THF. Benzoic acid was treated
with CDI to form benzoic acid imidazolide, which can
be considered as a benzoyl chloride analogue. This was
achieved through the simple stepwise addition of CDI
to a solution of benzoic acid in THF. This reaction is
accompanied by vigorous effervescence as CO2 is liber-
ated. The solution was added to a sample of hyper-
branched polyurethane and stirred at 60 °C for several
hours after which time a homogeneous solution had
been formed, showing the solubility of the modified
material in organic solvent (Scheme 5).
Conclusions
In summary, we have described the novel synthesis
of a water-soluble hyperbranched polymer through the
first reported selective activation of an AB2 monomer.
The reaction of 1,1′-carbonyl diimidazole with primary
amines in the presence of primary alcohols has been
studied using a series of model reactions and has been
shown to be highly selective, forming an active carboxa-
mide at the primary amine of an amino-diol monomer,
leaving the primary alcohol groups unreacted. The
water-soluble hyperbranched polyurethanes that are
formed by the self-condensation of this active intermedi-
ate have been characterized using MALDI -TOF spec-
1
13
15
trometry and H, C, and N NMR spectroscopy, using
a number of model molecules to aid peak assignment.
Polymers with Mn up to approximately 11 000 Da with
degree of branching equaling 0.6 have been determined.
The modification of the hydroxyl terminal groups with
benzoic acid was conducted using mild conditions and
reached complete formation of benzoate ester as shown
by NMR spectroscopy.
After purification and solvent removal, the material
1
13
was analyzed by H and C NMR (d6-acetone) and
MALDI-TOF mass spectrometry. The resonance at δ )
.59 ppm ( H NMR), attributed to the two methylene
1
3
Acknowledgment. The authors thank Courtaulds
Corporate Technology, Coventry where the work was
conducted. The expertise of Tim Jenkins and Christine
Wallace at Courtaulds in chromatography/mass spec-
trometry and NMR spectroscopy and Amelia Jackson
at Kratos Ltd is gratefully acknowledged.
protons next to the OH end groups (NCH2CH2OH), had
disappeared from the spectrum and new signals at δ )
4
.25 ppm (corresponding to the methylene next to a
benzoate ester), and groups of aromatic protons at δ >
7
.10 ppm were observed. Resonances at δ ) 59.82 and
1
3
5
6.31 ppm ( C) corresponding to these methylene
Supporting Information Available: Selection of H, 13C,
1
carbons had also disappeared in the spectrum, and a
new resonance at δ ) 66.94 ppm attributed to the
methylene carbon adjacent to the benzoate ester was
observed. Carbonyl signals at δ ) 167.48 ppm were
observed, confirming the presence of the aromatic ester.
Reaction of the benzoic acid imidazolide with the amine
to form an aromatic amide was not observed.
MALDI-TOF analysis of the material was extremely
complicated, and the clear repeat unit of the modified
hyperbranched polyurethane could not be seen (expected
15
and N NMR spectra for compounds 2, 4, 5, 7, and 9-15. This
material is available free of charge via the Internet at http://
pubs.acs.org.
References and Notes
(1) Fr e´ chet, J. M. J. J. Polym. Sci. Part A 2003, 41, 3713.
(
2) Tomalia, D. A.; Fr e´ chet, J. M. J. J. Polym. Sci. Part A 2002,
40, 2719.
(
3) Malkoch, M.; Carlmark, A.; Woldegiorgis, A.; Hult, A.;
Malmstr o¨ m, E. E. Macromolecules 2004, 37, 322.

9430 Rannard et al.
Macromolecules, Vol. 37, No. 25, 2004
(
4) Santini, C. M. B.; Johnson, M. A.; Boedicker, J. Q.; Hatton,
T. A.; Hammond, P. T. J. Polym. Sci. Part A 2004, 42, 2784.
5) Voit, B. I. C. R. Chim. 2003, 6, 821.
(18) Rannard, S.; Davis, N. Org. Lett. 2000, 2(14), 2117.
(19) Rannard, S.; Davis, N. Org. Lett. 1999, 1(6), 933.
(20) Rannard, S.; Davis, N.; McFarland, H. Polym. Int. 2000, 49,
1002.
(
(
(
6) Voit, B. J. Polym. Sci. Part A 2000, 38, 2505.
7) Spindler, R.; Fr e´ chet, J. M. J. J. Chem. Soc., Perkin Trans.
(
21) Hobson, L. J.; Feast, W. J. Polymer 1999, 40, 1279.
22) During our studies of carboxamide formation, we have
observed that the addition of primary amines to stirred
suspensions of CDI in toluene at room temperature form clear
solutions very quickly. However, when secondary amines are
added at room temperature, the suspension remains un-
changed, and TLC analysis of the supernatant shows no
formation of carboxamde. When the suspensions containing
secondary amine are heated to 50 °C, they become clear very
quickly and produce carboxamide in moderate yields (>70%).
Interestingly, the carboxamides that are formed from second-
ary amines do not readily react with alcohols to form
urethanes, or with amines to form ureas, under the conditions
that we have studied. This suggests that the pathway leading
to possible isocyanate intermediate formation and subsequent
imidazole liberation may involve the carboxamide NH proton
that is not present in examples derived from secondary
amines.
1993, 1, 913.
(
(
(
8) Taylor, R. T.; Puapaiboon, U. Tetrahedron Lett. 1998, 39,
8005.
9) Puapaiboon, U.; Taylor, R. T.; Jai-nhuknan, J. Rapid Com-
mun. Mass Spectrom. 1999, 13, 516.
(
(
(
(
10) Clark, A. J.; Echenique, J.; Haddleton, D. M.; Straw, T. A.;
Taylor, P. C. J. Org. Chem. 2001, 66, 8687.
11) Peerlings, H. W. I.; Van Benthem, R. A. T. M.; Meijer, E. W.
J. Polym. Sci. Part A 2001, 39, 3112.
12) Feast, W. J.; Rannard, S. P.; Stoddart, A. Macromolecules
2003, 36, 9704.
13) Rannard, S. P.; Davis, N. J. J. Am. Chem. Soc. 2000, 122,
1729.
14) Spindler, R.; Fr e´ chet, J. M. J. Macromolecules 1993, 26, 4809.
15) Kumar, A.; Ramakrishnan, S. Chem. Commun. 1993, 1453.
16) Hong, L.; Cui, Y.; Wang, X.; Tang, X. J. Polym. Sci. Part A
002, 40, 344.
17) Bruchmann, B.; K o¨ niger, R.; Renz, H. Macromol. Symp. 2002,
87, 271.
1
(
(
(
2
(
1
MA0489218