DSpace at VNU: Determination of bisphenol A (BPA) by gas chromatography-mass spectrometry and H-1 NMR spectroscopy during curing of epoxy-amine resins

1356, Universite´ de Toulon et du Var, BP 132, 83957 La Garde Cedex, France b Faculte´ de Chimie – Laboratoire de Chimie analytique, Universite´ Nationale de Hanoı¨, Vietnam Received 6 A

POLYMER TESTING

Polymer Testing 25 (2006) 912–922

Analysis Method Determination of bisphenol A (BPA) by gas

during curing of epoxy–amine resins F.X Perrina, , Thi Minh Hanh Nguyena,b, Thi My Linh Tranb, J.L Verneta

a Laboratoire Mate´riaux a` Finalite´s Spe´cifiques (UPRES no 1356), Universite´ de Toulon et du Var, BP 132, 83957 La Garde Cedex, France

b Faculte´ de Chimie – Laboratoire de Chimie analytique, Universite´ Nationale de Hanoı¨, Vietnam

Received 6 April 2006; accepted 20 May 2006

Abstract

Two analytical methods were developed to characterize the fate of bisphenol A (BPA) added in low amounts (1–4 wt%)

in two different epoxy–amine–BPA systems, one based on a mixture of aliphatic amine curing agents, diethylene triamine (DETA) and poly(oxypropylene)diamine (Jeffamine) and the other on the less reactive diaminodiphenylsulphone (DDS) hardener The method involves a liquid–liquid extraction (LLE) procedure up to gelation Identification and quantification was performed by gas chromatography–mass spectrometry (GC–MS) and proton nuclear magnetic resonance (1H-NMR) spectroscopy The data obtained by the two methods are in agreement The main purpose of this study was to determine the extent of aryl-etherification in actual complex epoxy/amine formulations It serves notably to detect potential contamination of the atmosphere, water, and soil in the area surrounding where the unbound BPA is released

r2006 Elsevier Ltd All rights reserved

Keywords: Bisphenol A; Epoxy; Curing; GC–MS; NMR spectroscopy

1 Introduction

Epoxy resins are one of the most important

classes of thermosetting polymers The addition of

additives or modifiers (catalysts, diluents, fillers,

resinous modifiers, elastomeric modification, etc.),

to the two main ingredients is frequently adopted to

change the characteristics of the uncured resin

(rheology, curing kinetics, etc.), or the properties

of the final material (fracture behaviour for

example) Some formulations proposed in industry were found to contain small amounts of phenolic compounds mixed with the amine hardener compo-nent Phenolic compounds are hydrogen donor molecules and thus could theoretically catalyse the epoxy-amine reaction by forming a trimolecular complex facilitating the amine attack, as illustrated

inScheme 1 The same type of push-pull mechanism accounts for the autocatalytic nature of the curing process observed with epoxy/amine systems, which

is related to the accumulation of hydroxyl groups during curing [1,2]

The formation of ether links during epoxy/amine cure by reaction between an epoxide group and a

www.elsevier.com/locate/polytest

0142-9418/$ - see front matter r 2006 Elsevier Ltd All rights reserved.

Corresponding author Tel.: +33 494 142580;

fax: +33 494 142448.

E-mail address: perrin@univ-tln.fr (F.X Perrin).

hydroxyl group is significant only with certain

epoxy/amine systems and in certain conditions

(stoichiometric excess of epoxide groups, high cure

temperatures, etc.)[3] The etherification reaction is

also accelerated by tertiary amine (Scheme 2) and

hydroxyl groups (Scheme 3) [4] Both the tertiary

amine and hydroxy catalysis etherification proceed

through the formation of trimolecular complex, as

shown inSchemes 2 and 3, respectively.Schemes 2

and 3 suggest that phenol compounds in

epoxy–a-mine–phenol ternary systems can be incorporated in

the epoxy network through aryl ether bonds

(Schemes 2 and 3 with R1¼phenyl) Eventually,

they may also accelerate etherification of aliphatic

hydroxyl (present in the starting epoxy material or

formed during the epoxy–amine reaction) with

epoxide (Scheme 2 with R1¼epoxy chain and

R2¼phenyl)

Currently, our knowledge about the reactivity of

phenol compounds with epoxy and epoxy/amine

systems is mainly based on the pioneering work

by Shechter and Wynstra[5]and Shechter et al.[6]

The catalytic effect of phenol for the alcohol–

epoxide etherification [5]and for the amine–epoxy

addition [6] were demonstrated These studies

were performed on soluble model compounds to

permit the use of conventional analytical methods

to elucidate the reaction mechanism The danger

with this approach is that the reaction rates and the

reaction paths observed with these

monofun-ctional compounds may be different from the

results obtained on real polyfunctional systems

For instance, the addition of phenol through aryl

ether bonds was found to strongly depend on the

presence of a base catalyst [6] Without a catalyst,

phenol reacts with epoxy at a temperature of 200 1C

or higher [5], while the reaction of phenol with

epoxide proceeded at a reasonable rate at 50 1C

in phenol–phenyl glycidyl ether–diethylamine (or

N-methyl aniline) mixtures[6] In this context, one

of the principle uncertainties arises from the fate of

a phenol compound added to a given polyfunctional epoxy–amine system In other words, does the hindered amine in polyfunctional epoxy–amine systems exert any catalytic effect toward the aryl etherification? It also appears fundamental to know

if phenol compound is chemically bound into the cross-linked network or if it is only physically bound (hydrogen bonds) for two main reasons: (1) the way phenol is incorporated into the cured resin may affect the physical and mechanical properties

of the final material in a different way; (2) the toxicological activity of phenols should be strictly considered if a significant amount of phenol remains not chemically bound to the network At the present time, we found that some commercial formulations contain low amounts of bisphenol A (4,40 -isopro-pylidenediphenol) (BPA) added in the hardener component The low amount of BPA in these formulations (1–4 wt%) suggest that BPA mainly serves here as a catalyst for the amine–epoxy reaction As far back as 1936, Doods and Lawson noted oestrogenic activity of BPA [7] More recently, BPA was found to be leached from a polycarbonate flask[8] It was also found in saliva collected from patients who were treated with a DGEBA-based dental sealant[9] In both cases, the

O

CH CH2

H O R2

O R1 H

Scheme 2 Trimolecular intermediate showing the catalytic effect

of a hydroxy compound on the etherification reaction (R 1 ¼ or 6¼R 2 ).

O

CH CH2

O R1

H NR3

Scheme 3 Trimolecular intermediate showing the catalytic effect

of a tertiary amine on the etherification reaction.

HN

CH

H2C O H O

R1

R2

Scheme 1 Trimolecular intermediate amine–epoxy–phenol

showing the catalytic effect of a phenol compound on the

amine–epoxy reaction.

leaching of BPA was ascribed to the hydrolysis of

the polymer

In this paper, we developed analytical tools to

investigate the fate of low amounts of BPA

(1–4 wt%) in two formulations of different

reactiv-ity: an epoxy–aliphatic amine–BPA commercial

formulation and a less reactive diglycidyl ether of

bisphenol A (DGEBA)–diaminodiphenylsulphone

(DDS)–BPA formulation We will also propose a

methodology for the determination of epoxide

equivalent weight (EEW) and component

composi-tion in epoxy resins based on DGEBA/DGEBF

mixtures using1H NMR spectroscopy

2 Experimental

2.1 Materials

The epoxy resin was a DGEBA–diglycidyl ether

of bisphenol F (DGEBF) mixture (see the next

section for the characterization data) supplied by

REA Industrie (France) DGEBF is used as a

functional diluent to prepare solventless epoxy

formulations Three different amine hardeners were

tested: a diethylene triamine

(DETA)/polyoxypro-pylene amine/BPA commercial hardener (CH)

having an amino hydrogen equivalent weight

(AEW) of 59 g, jeffamine (Aldrich) with an AEW

of 100 g and 4,40-diaminodiphenylsulfone (Aldrich,

97%) with an AEW of 62 g

BPA and naphtalene were supplied by Aldrich

Water was purified by means of a water purification

system All other solvents were of analytical reagent

grade (Aldrich) except dichloromethane used for the

LLE of BPA (Fluka, 95%)

2.2 Methods

The samples were prepared by adding

appro-priate amounts of an amine/BPA solution or BPA

powder to the DGEBA/DGEBF resin either at

room temperature (aliphatic amine-based epoxy) or

at 120 1C (DDS-based epoxy) The specified

quan-tities for the epoxy, amine hardener and BPA

correspond to their molar equivalent of glycidyl

ether, hydrogen amines, hydroxy functionalities,

respectively After intimate mixing, the viscous

mixture was cast in a teflon mould and isothermally

cured at a temperature related to the reactivity of

the hardener (not greater than 60 and 200 1C,

respectively, for aliphatic and aromatic amines)

Samples (5 g) were taken at definite times during

curing, introduced into an Erlenmayer flask and cooled quickly to room temperature while placing the flask in a ice bath

2.2.1 LLE procedure (before gelation) About 50 ml of an aqueous solution containing sodium hydroxide (0.3 M) and sodium chloride (100 g/l) was added to a 50 ml ether solution of 5 g

of the epoxy/amine/BPA sample After vigorous stirring for 10 min at room temperature, the aqueous phase was extracted and treated with 10% hydrochloric acid to obtain an acidic pH (
1–2) solution NaCl was added to the acidic solution to obtain a 200 g/l NaCl concentration Then, the aqueous extract was extracted three times with 50 ml of dichloromethane by vigorous shaking for 10 min The extracts were treated with 1 g of magnesium sulphate and then concentrated to near dryness on a rotary evaporator under a gentle flow

of N2 2.2.2 Solid– liquid extraction (SLE) (after gelation)

About 1–5 g of the epoxy/amine/BPA gelled material was ground to fine powder at the tempera-ture of liquid nitrogen using a freezer mill SPEX Certiprep 6750 The powder was extracted twice with 20 ml of methanol by shaking overnight The combined extracts (40 ml) were either concentrated (for NMR analysis) or used to prepare methanol solutions for GC/MS analysis

2.2.3 1H NMR analysis

1

H NMR spectra were measured at room temperature on a Brucker AV400 spectrometer operating at 400 MHz, using deuterated dimethyl-sulphoxide (DMSO-d6) as a solvent and naphtalene

as an internal standard

2.2.4 GC-MS analysis Chromatographic measurements were performed with a gas chromatograph HP 6890 equipped with a mass spectrometer HP 6890 The carrier gas was helium and the flow rate was 1 ml/min The column was a fused silica capillary (30 m  0.25 mm i.d  0.25 mm film thickness) The temperature of the injection port was 280 1C using a split ratio of 40/1 The oven temperature was held at 150 1C for

2 min, then elevated to 280 1C at 30 1C/min and finally held at 280 1C for 2 min The interface temperature was 300 1C and the energy of ionizing electron was 70 eV The MS was operated in scan

mode with a scan range from m/z 45 to 300.

Quantification was done by an internal standard

method using naphthalene as an internal standard

Monitored ions were m/z (molecular ion of

naphthalene) and 228.1 (molecular ion of BPA)

3 Results and discussion

3.1 Characterization of the DGEBA/DGEBF

commercial mixture

The DGEBA/DGEBF commercial mixture was

characterized by 1H NMR spectroscopy Fig 1

illustrates the 1H-NMR spectrum of the DGEBA/

DGEBF mixture All the resonance signals have

been attributed to the corresponding protons of the

DGEBA and DGEBF structures

Recently, the 1H NMR was found to constitute

an alternative to the classical titration methods to

determine the EEW of low molecular weight

DGEBA resins[10] Here, the problem is somewhat

complicated by the presence in the resin of two different epoxies of very close chemical structure The average molecular weight Mn of the DGEBA/ DGEBF resin may be defined as

Mn¼x1Mn 1þx2Mn 2

where x1and x2are mole number of DGEBA and DGEBF contained in 1 g of mixture, respectively, and Mn 1 and Mn 2 are the average molecular weight

of DGEBA and DGEBF, respectively

Mn 1 and Mn 2 are related to the polymerization degree n1and n2by the relation:

Mn1¼284n1þ340, (2)

Mn2¼256n2þ312 (3) Considering the idealized structures in Fig 1

where one epoxy group terminates each end of the DGEBA and DGEBF resins, the average EEW is one-half the average molecular weight of the

ppm (t1)

0.0 1.0

2.0 3.0

4.0 5.0

6.0 7.0

0 50000 10000 15000 20000 25000

Fig 1 1 H NMR spectrum of the DGEBA/DGEBF commercial mixture.

DGEBA/DGEBF mixture:

EEW ¼x1ð284n1þ340Þ þ x2ð256n2þ312Þ

2ðx1þx2Þ (4) EEW can be determined measuring the integrals

I1, I2 and I3 corresponding, respectively, to the

aromatic protons (peaks a and b), the epoxide

group protons (peaks e, f and g) and the benzylic

methyl protons (peak h)

Indeed, the idealized structures shown inFig 1give

R1 ¼I1

I2¼

8 x½ 1ðn1þ1Þ þ x2ðn2þ1Þ

6ðx1þx2Þ , (5)

R2 ¼I 3

I 2¼

6x1ðn1 þ 1Þ

Substituting Eq (6) into Eq (5) and re-arranging,

we get

R3 ¼3I14I3

4I2 ¼

x2ðn2þ1Þ

ðx1þx2Þ (7)

Eq (4) can be re-arranged to be expressed as

function of the integral ratios R1and R2(Eqs (5) and

(6)):

EEW ¼ 96R1 þ 14R2 þ 28 (8)

The EEW of the DGEBA/DGEBF commercial

mixture calculated from Eq (8) was 171.271.2 (value

at 95% confidence interval determined from five

different aliquots)

An EEW of 17472.7 (value at 95% confidence

interval determined from five chemical analysis) was

found by acid titration using the pyridinium

chloride–pyridine method [11] The data found by

1

H NMR spectroscopy are in good agreement to

those obtained from the classical chemical titration

The average polymerization degree, ¯n, can be

defined by the following equation:

¯n ¼x1n1þx2n2

Eq (8) can be re-written in order to reveal the

expression of the integral ratio R1:

¯n ¼½x1ðn1þ1Þ þ x2ðn2þ1Þ  ðx1þx2Þ

ðx1þx2Þ (10) Therefore,

The average polymerization degree was estimated

by Eq (11):

¯n ¼ 0:08

The low value of ¯n agrees with the low intensity of the multiplet at d ¼ 4.1 ppm (peak i ascribed to the aliphatic carbinol methine) and confirms that the epoxy resin contains a small fraction of oligomers (n1and n240)

The mole number x1and x2are related with the weight fraction of DGEBA, y1, and DGEBF, y2, as follows:

x1¼ y1

x2¼ y2

Re-arranging Eqs (12) and (13), we get

x1½284ðn1þ1Þ þ 56 ¼y1, (14)

x2½256ðn2þ1Þ þ 56 ¼y2 (15) Substituting Eq (6) into Eq (14) and Eq (7) into

Eq (15), the following equations can be obtained: 284R2ðx1þx2Þ þ56x1¼y1, (16) 256R3ðx1þx2Þ þ56x2¼y2 (17) Addition of Eqs (16) and (17) gives

ðx1þx2Þ ¼ 1

284R2þ256R3þ56. (18) Besides, from Eqs (6) and (7), it is clear that

x1oR2ðx1þx2Þ, (19)

x2oR3ðx1þx2Þ (20) Substituting the inequality Eqs (19) and (20) in Eqs (16) and (17), then, considering Eq (18) we get:

1  312R3 284R2þ256R3þ56oy1o 340R2

284R2þ256R3þ56.

(21)

Eq (21) is a general expression that can be used for estimating the interval of the weight fractions of DGEBA and DGEBF in mixtures of these compo-nents

For the commercial epoxy used in this study, the width of the interval was found to be very low: 0:34oy1o0:35

These data are in excellent agreement with the data given by the supplier: 65 wt% of DGEBF (and 35% of DGEBA)

3.2 Preconcentration procedure

A preconcentration procedure was essential to

avoid strong interferences of amine and epoxy

components during analysis We adopted a

liqui-d–liquid extraction method and a SLE method,

respectively, before and after the gelation point of

the mixture The optimization of the extraction

procedure was carried out on a stoechiometric

DGEBA/DGEBF–jeffamine–BPA (1 wt%) mixture

For the LLE procedure, BPA was extracted in two

steps: in a first step, BPA was totally removed as its

dissociated form in the alkaline aqueous phase The

relatively hydrosoluble amine hardener was found

to be present in great amounts in the aqueous phase

In the second step, the aqueous phase was acidified

to form the amine salt and, thus, to selectively

extract BPA in an organic solvent Ethyl ether,

n-hexane and dichloromethane were tested as

extraction solvents Dichloromethane was the most

efficient, giving nearly 100% recovery of BPA after

three successive extractions (see the next section for the determination of the extraction yield) We noted that the ionic strength of the aqueous phase was one

of the key parameters that affects both the yield of BPA recovery and the efficiency of the separation The ionic strength was adjusted using NaCl to obtain NaCl concentrations from 0 to 300 g/L A

100 g/L NaCl concentration in the aqueous alkaline phase at the first step of the LLE ensures that a major fraction of the epoxy resin remains in the ether phase Besides, a 200 g/L NaCl concentration

in the acidic aqueous phase at the second step of the LLE was found to be necessary to totally extract BPA in dichloromethane by a salting-out effect 3.3 Analytical determination of BPA

3.3.1 1H NMR spectroscopy: extraction yield

Fig 2 shows a typical 1H NMR spectrum of a complex epoxy–amine–BPA mixture (Fig 2(a)) together with the spectrum of the organic phase

naphtalene

biphenol A

ppm (t1)

1.0 2.0

3.0 4.0

5.0 6.0

7.0 8.0

9.0

DMSO H20

x

Ha Hao

Hc

Hd

(a)

(b)

Fig 2 1 H NMR spectrum of a stoechiometric DGEBA/DGEBF–jeffamine–BPA (1 wt%) mixture before (a) and after (b) the LLE procedure.

Fig 3 Typical chromatogram of a LLE residue (shown here is a DGEBA/DGEBF–jeffamine–BPA(1%) mixture after 2 h at room temperature): (1) naphthalene, and (2) BPA.

Fig 4 Mass spectra obtained in scan mode for BPA.

(Fig 2(b)) after the LLE procedure The1H NMR

spectrum of Fig 2(b) is dominated by the proton

signals of BPA (aromatic doublets at 6.64 and

6.98 ppm and methyl singlet at 1.51 ppm), while the

proton signals of BPA are barely perceptible in

Fig 2(a) This reveals the efficiency of the

separa-tion, although minor amounts of the epoxy resin

(small peaks at 7.05 and 6.77 ppm) are still present

in the organic extract after extraction

The wt% recovery of BPA was determined by

adding naphthalene as an internal standard to the

organic residue obtained from the LLE procedure

The two doublets at 7.5 and 7.9 ppm are the

absorption of the aromatic protons of the standard

The wt% of BPA in the epoxy/amine mixture is

calculated by comparing the integral of the

four aromatic protons of BPA at 6.98 ppm (Ia) with

the integral of the four aromatic protons of

naphthalene at 7.9 ppm (Iao):

%BPA ¼W N

W R

I a

Iao

MBPA

MN

where WN is the weight of naphthalene, WR is the weight of resin and MBPAand MNare the molecular weight of BPA and naphthalene, respectively The calculated wt% of BPA from Eq (22) was compared to the theoretical amount of BPA in different epoxy–amine–BPA mixtures of known composition (0.5–2 wt% BPA) and just prepared

at room temperature: the extraction yield was systematically higher than 95%

The same methodology was followed to deter-mine the efficiency of SLE from a stoichiometric epoxy/CM blend cured at 20 1C for 14 days The extraction yield of BPA was found to be near 100%

as could be forecast from the low curie temperature

0 0.2 0.4 0.6 0.8

0.2 0.4 0.6 0.8

1

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

time (min)

GCMS RMN

GCMS RMN

0 1

time (min)

(1-αBP

(1-α BP

(a)

(b)

Fig 5 Fraction of non-reacted BPA (1a BPA ) in (a) epoxy/BPA (1/0.076) and (b) epoxy/DDS/BPA (1/1/0.076) blends both cured at

100 1C.

3.3.2 GC/MS analysis

GC/MS has been used to analyse phenolic analytes

either directly[12]or after silyl derivatization[13] By

incorporating a silyl group to a phenolic compound,

better reproducibility, sensitivity and resolution can be

achieved for the gas chromatographic analysis This

work is not dedicated to analysis of traces of polar

analytes or to complex mixtures of phenol compounds

with very similar chemical structure Thus, a direct

analysis of the LLE and SLE residues was performed

A typical chromatogram of the phenolic analyte

obtained after the LLE procedure is shown inFig 3

Only 7 min were necessary to complete the analysis

The mass spectrum obtained in scan mode for the

compound (2) ofFig 3is shown inFig 4and reveals

the molecular ion of BPA at m/z 228 and a base peak

at m/z 213 corresponding to loss of a methyl group

The GC-MS and1H NMR analysis were found to

give very similar results, as inferred from Fig 5

which compares the BPA content in epoxy/BPA and

epoxy/DDS/BPA blends determined from both

methods The fate of BPA in both blends will be

discussed in the next two sections

3.4 The fate of BPA in epoxy/BPA and epoxy/

amine/BPA blends

Fig 6 compares the fate of BPA in

stoichio-metric epoxy/amine blends from either aliphatic or

aromatic hardener

BPA is integrally recovered from the totally cured epoxy/jeffamine blend In such resins, BPA is only linked to the polymer network though physical bonds (hydrogen bonds) and is thus prone to be readily released in the environment Conversely, the higher temperature required to cure epoxy resins with aromatic hardener like DDS favours the etherification between BPA and epoxy DSC analy-sis (not shown) showed that BPA does not induce a change in Tg for epoxy/jeffamine blends (Tg onset measured during the second ramp near 32 1C) while

a 10 1C decrease of the Tg of epoxy/DDS/BPA (1/1/ 0.076) compared to the Tg of epoxy/DDS (1/1) was noted (respective Tg onset of 164 and 174 1C) Owing to the low amount of free BPA (o10% of the initial content from Fig 6) in the totally cured DDS based resins, the plasticization induced by BPA can undoubtedly be related to the chemical alteration of the polymer network (internal plasti-cization) rather than an external plasticization Low amounts of BPA (few wt%) incorporated in epoxy/ aliphatic amine blends only acts to catalyse the epoxy–amine reaction, as suggested by the mea-sured gel time of 161 min in epoxy/jeffamine (1/1) blend compared to 82 min in epoxy/jeffamine/BPA (1/1/0.076) blend, both cured at 60 1C

Fig 6also reveals a clear increase in the rate of etherification in the course of the reticulation at the three investigated temperatures This acceleration must be related to the accumulation of hydroxy

0 0.2 0.4 0.6 0.8 1

0 100 200 300 400 500 600 700 800

time (min)

(1-α BP

Fig 6 Fraction of non-reacted BPA (1a ) in epoxy/amine/BPA (1/1/0.076) blends isothermally cured.

and/or tertiary amine during the epoxy/amine

reaction This result suggests that trimolecular

complexes such as shown in Schemes 2 and 3 are

involved during the epoxy/BPA reaction It is

noteworthy that this is not strictly speaking an

autocatalytic effect Further experimental evidence

of such a catalytic effect is provided by comparing

the change in BPA content in epoxy/BPA (1/0.076)

and epoxy/DDS/BPA (1/1/0.076) mixtures cured

under the same conditions (Fig 7) Irrespective of

the cure temperature, the rate of the etherification reaction between epoxy and BPA is found to be higher when the amine hardener is incorporated into the epoxy/BPA blend

Since BPA/epoxy/BPA trimolecular complexes may be potentially involved as soon as the initial times of the reaction, it is believed that tertiary amines, despite their steric crowding, exert a more efficient catalytic effect than hydroxy groups

0 0.2 0.4 0.6 0.8 1

0 200 400 600 800 1000

time (min)

(1-αBP

0.2 0.4 0.6 0.8

(1-αBP

0.2 0.4 0.6 0.8

(1-α BP

without DDS with DDS

without DDS with DDS

without DDS with DDS

200 °C

0 1

0 500 1000 1500 2000 2500 3000

time (min)

150 °C

0 1

0 1000 2000 3000 4000 5000

time (min)

100 °C

Fig 7 Fraction of non-reacted BPA (1a ) in isothermally cured epoxy/BPA (1/0.076) and epoxy/DDS/BPA (1/1/0.076) blends.