The Search for Reactive Peroxides and Hyclroquinones in an Acrylic Copolymer Blend via Supercritical Fluid Extraction-Chromatography 2 0 0 3 , 57, 5 3 3 - 5 3 7
D. Johnston 1 / M.Ashraf Khorassani 2 / L. T. TayloP* 1 Esstech,Inc., Essington, PA 19029, USA 2 Department of Chemistry, Virginia Tech, Blacksburg,VA 2406]_, USA; E Mail:
[email protected]
In contrast to gas chromatography (GC) and liquid chromatography (LC), supercritical fluid chromatography (SFC) opens a broad range of tridimensional operating conditions by varying pressure in addition to composition (LC dimension) and temperature (GC dimension). With a unique array of physical, economical, and ecological properties, carbon dioxide (CO2) is the prominent component of SFC mobile phases. For separating highly polar materials, CO2 is mixed with a small
amount of common polar solvent (e.g. modifier) such as melhanol_ The modifier in turn may contain a secondary additive classically required in LC conditions such as isopropylamine. Berger [1] has noted that the advantages of SFC over LC are practical, not fundamental since there is no fundamental characteristic of supercrticial fluids which differentiates them from gases or liquids. The greatest difference is simply the need to hold the column outlet pressure above ambient to prevent expansion of the fluid. Berger, for example, has stated that "packed column SFC
can be thought of as an odd form of LC". One practical advantage of SFC that is seldom emphasized is that the mobile phase is inert to both oxidation and hydrolysis. Consequently, analytes which are prone to either oxidize, react with water, or are involafile prove to be excellent candidates for separation via SFC. As examples, peroxides, polyphenols, certain vitamins, and polymer additives immediately come to mind. We wish to describe, herein, a study where the inertness of the mobile phase proved vital for achieving a successful analysis. Specifically, the extraction, identification, and quantification of dibenzoyl peroxide, N,N-dimethyl-p-toluidine, hydroquinone, and the methyl ether of hydroquinone in an ethyl and methyl methacrylate copolymer blend cured at 35 ~ for two different time periods are considered. Our search of the open literature revealed no SFC studies of intact peroxides and hydroquinones; yet the components are widely used in making polymers. For example, both dibenzoyl peroxide and diclmlyl peroxide are chain reaction initiators for polymerization processes_ SFC ideally should allow reaction mixtures such as these to be assayed for decomposition products left after polymerization. Gere et al. [2] has reported in an application note the packed column SFC separation of the reaction products from two different GC stationary phases (e.g. SE-54 and SE-33) with two different polymerization initiators. Dicumyl peroxide was suggested to leave more benign reaction products than dibenzoyl peroxide as the chromatographic trace was considerably sim-
Original
Chromatographia 2003, 57, April (No. ?/8)
533
KeyWords Supercritical fluid chromatography Supercritical fluid extraction AcrylTc co polymer Additive residues
Summary The obTective of this work was to extract, identify, and quantify dTbenzoyl peroxide (DBPO), N,N-dimethyl-p-toluidine (DMPO, hydroquinone (HQ), and the methylether of hydroquinane (MEHQ) Tnan ethyl and methyl methacrylate copolymer blend cured at 35 ~ for iwo different time periods. The reactivity of BPO, HQ, and MEHQ made working with supercrifical fluids most advantageous. Exactly 0.85 gram of cured polymerwas placed in a 5 m/extraction yes sel. In orderto obtain effTdent collection of the analyie afier supercritTcal fluTd extractTon, a solTd phasetrap (Cls) was used.Various polymer samples were cured at different times and then extracted and analyzed. The level of DMPTdecreased by 50% (0.22% to O.12%) as the cure tTme increased from 30 minutes to 24 hours. The BPO level did not vary with increasing cure time (0.44% vs 0.43%). The MEHQ level could not be ascertained since it was below our detection ITmit for MEHQ.
Introduction
0009-5893/00/02
533- 05
$ 03.00/0
@ 2003 Friedr. Vieweg & Sohn Verlagsgesellschaft m b H
pier for the former peroxide_ The siluation regarding liquid chromatographic separation of peroxides is similarly sparse. Few sludies have appeared and conversion to the more stable acid decomposition product prior to analysis [3 5] appears to be the rule.
Experimental SupercriticalFluidExtraction All extractions were performed with an Isco Suprex AP-44 (Lincoln, NE, USA) supercritical fluid extractor. Exactly 0.85 gram of synthesied in-house polymer (or 0.5 gram of polymer powder as received) was placed in a 5 mL extraction vessel. Most of the dead volume in the vessel was then filled with sand. In order to obtain efficient collection of the analyte after supercritical fluid extraction (SFE), a solid phase silica trap derivatized with C18 units was used. After completion of the extraction, the solid phase trap was rinsed with 5 mL of 80:20 (v/v) methylene chloride/acetone directly into a collection vial. All extractions were performed with Air Products and Chemicals Inc. (Allentown, PA, USA) SFE-SFC-grade CO2 pressurized with 2000 psi helium which was modified in-line with acetone. The following parameters were used for all extractions: Extraction Fluid: 85/15 CO2/Acetone (vlv) CO2 Pressure: 450 arm Extraction Temperature: 70 ~ Extraction Time: Dynamic flow 30 rain CO2 Flow Rate: 0 mL min of liquid C02 Solid Trap Rinse Volume: 5 mL of 80/20 Methylene Chloride/Acetone Solid Trap Packing: octadecylsilica Solid Trap Temp. During Ext.: 25 ~ Solid Trap Rinse Temperature: 50 ~
SupercriticalFluid Chromatography A Berger Instrument (Newark, DE, USA) supercritical fluid chromatograph equipped with UV detector and high pressure flow cell was used. A Supelcosil cyano column (250 mm • 4.6 mm, 5 lxm dp) was employed for all separations. Air Products and Chemicals Inc. supplied the SFE/SFC grade CO2. The following chromatography conditions were used to obtain the separation of all compounds:
5 34
Column: Supelcosil CN (250 mm • 4.6mm OD) FlowRate: 1.8 mL rain 1 CO2 liquid Pressure: 150 aim CO2 Temperature: 55 ~ Modifier (MOD): Methanol + 0.5% Isopropylamine (v/v) Mobile Phase Gradient: 99.8/0.2 CO2/ MOD hold for 2 min, then to 99.5/0.5 CO2/MOD at 0.3% MOD min 2, then 75/25 CO2/MOD at 4% MOD min 2, hold for 5 rain. Injection Volume: 3.5 p~LAcetone solution Detector: UV, 267 nm
Results and Discussion ChromatographicAnalysis of Additives
The polymer powder containing approximately 2% BPO and liquid monomer containing HQ, MEHQ, and DMPT starting materials (provided to us for preparation of the polymer) were manually mixed in a 2 to 1 weight ratio of powder to liquid. The mixture was stirred thoroughly and quickly spread on a glass plate to prepare a film specimen of 0.5 to 1.0 mm thickness. Next, the resulting film of polymermonomer mixture was cured in a GC oven at 35 ~ for either 30 minutes or 24 hours. Each cured polymer was then cut into small pieces and placed in a high pressure vessel for SFE. Three samples were singly extracted for validation and statistical analysis.
The primary goal of our study was to extract, identify, and quantify residual dibenzoyl peroxide (initiator), N,N-dimethyl-p-toluidine, hydroquinone, (inhibitor), and the mono methyl ether of hydroquinone (inhibitor) in an ethylmethyl-methacrylate copolymer blend cured at 35 ~ as a function of two reaction times. Since the assignment of the various SFC peaks in the polymer (SFE) will become important because of the tendency for some of these materials to degrade in air and water (vide infra), a discussion of the separation behavior of these four additives prior to their inclusion into the copolymer is first provided. SFE and SFC are well suited for this study as CO2 which is inert and exhibits little affinity for water is both the extracting phase and the chromatographic mobile phase. Figure 1 shows the SFC separation of a mixture of the four components freshly dissolved in acetone. Surprisingly, more than four peaks were observed in addition to the solvent peak. This observation suggested to us that the additives were either impure as received or reactive under the sample preparation-chromatographic conditions employed in this work. For examples, samples in this particular experiment (e.g. antioxidants and initiators) were exposed to air/light, dissolved in a moderately polar solvent, and separated on a packed column with methanol containing isopropyl amine (0.5%) modified carbon dioxide. Thus, we believe that the multitude of peaks arose in part from the high reactivity of the analytes with their environment. The assignment of several of the peaks can be made based upon retention time comparison with single compound injected employing the same packed column and the identical elufion schedule as previously used with the mixture. Figure 2 shows individually the separation of BPO, DMPT, HQ, and MEHQ each dissolved in acetone. BPO yielded a single peak at approximately 2.9 minutes whether it was dissolved in acetone or methylene chloride. If the acetone contained a small amount of water, two peaks were observed separated by about 5 minutes. The later elufing component from the wet acetone solution can be attributed to benzoic acid, a breakdown product of
Chromatographia 2003, 57, April (No. 7/8)
Original
Preparation of Standard Solution Four stock solutions were prepared by individually dissolving a known amount (100 mg each) of BPO, DMPT, HQ, and MEHQ each in 10 mL of methylene chloride (for HQ the solvent was 80:20 methylene chloride/acetone). Next, 500, 200, 100, 50 and 25 lxL volumes of each stock solution were diluted to 5 mL with methylene chloride in order to prepare the various standard solutions for generation of the calibration plot. It is important to note here that a synthetic mixture of the combined four components was chemically unstable since the color of the solution changed to brown and additional peaks appeared in the SFC as time passed, thus the need for four different solutions. Methylene chloride was used as the solvent where possible instead of acetone because it was believed that moisture in the acetone may cause the degradation of certain additives such as BPO.
Polymer FilmPreparation
BPO caused by reaction with either acetone or with water in acetone. The peak assigned to benzoic acid (BA) was later confirmed by first chromatographing BA dissolved in methylene chloride (tR = 7.7 rain) and then dissolved in dry acetone. This experiment was later followed-up with the baseline separation of a synthetic mixture of BPO and BA dissolved in methylene chloride (Figure 3). DMPT yielded a single sharp early eluting peak near 1.8 minute indicative of its high purity. HQ and MEHQ gave multiple tailing peaks. In the case of MEHQ, a pronounced shoulder (nearly resolved) accompanied the main peak (tR 5.9 rain) which could reflect a mixture of positional ring isomers or it could represent an oxidized product of MEHQ which in the copolymer serves as an antioxidant. The separation of our sample of HQ was even more complex than MEHQ in that three partially resolved peaks were observed near 10.5 min which could be due to positional aromatic ring isomers (i.e. orthometa-para) and/or oxidized products (i.e. quinones). 1,4-Benzoquinone, however, can be ruled out as one of the eluting components since an injection of this compound gave a single peak with tR retention time equal to 2.2 rain. With the single compound injection data discussed above, one can confidently make assignment of the chromatographic peaks arising from the separation of the synthetic mixture of the "four" components. Since our goal was to quantify the residual additives in a synthesized in-house copolymer, an on-column detection limit for each analyte at 267 nm was determined. Other wavelengths were investigated but 267 nm proved to be optimial. For HQ, the chromatographic peak areas of both peaks in the fresh solution were employed. Detection limits (on column, 3.5uL injection, 3 • S/N) were determined to be 50 ng (DMPT), 200 ng (BPO), 206 ng (HQ), and 240 ng (MEHQ).
DMIml9 Frah Solulioa
lIFO ~to ~o Io
o
t+Q
-Io
-3G -4o
2
10 TI~
12
(m:Lr~)
Figure 1. SFC separation of a mixture containing four standards freshly dissolved in acetone. See Experimental for chromatographic conditions.
it++ +;+; ...... + ..T 9
4
m
II.
10
I~
f
15
,!
Ms
.Io
z
+
6
8
tO
I
4
6
s
l0
I2
Ttme
(ml~)
Figure 2. Singlecompound injection of DMPT, BPO, MEHQ, and HQ.
Polymer samples for this investigation were prepared in-house. Not only were we interested in the amount of residual additive in the prepared polymer, but we wanted to know how this quantity varied with co-polymer synthesis time. Prior to our study, we were informed that the powder contained approximately 2% BPO (wl
w) and the liquid monomer contained MEHQ at 250 ppm and HQ at 25 ppm. A chromatogram of the liquid monomer injected (3.5 9L) onto the SFC column at two different concentrations (neat and diluted) was first obtained. HQ was not even detected with a neat injection. Upon reflection, at 25 ppm and 3.5 I~L injection, only 87.5 ng would have been placed on the column which is lower than our measured HQ on-column detection limit. Increasing the injection volume of liquid monomer from 3.5 pL to 10 p,L also did not show any trace of HQ. HQ obviously had oxidized already to 1,4-benzoquinone in the sample since a peak at retention time of 2.1 minutes due to the latter was
observed. MEHQ was observed but its elution was shifted to lower retention time (ta 3.8 min) via the neat injection because the packed column was overloaded. One therefore, concludes from these chromatographic results that HQ will not be observed in any polymer extract, and MEHQ will be at such a low level as to also not be detected. The powder sample which was to be mixed with the monomer liquid for polymerization was next examined for the content of BPO and any co-extractables. For this purpose, duplicate 0.5 gram samples of powder were subjected to SFE. BPO was expected to dissolve in SF CO2; while the polymer was anticipated to be immo-
Original
Chromatographia 2003, 57, April (No. 7/8)
535
Analysis of Residual Polymer Additives
mAU
BPO
BA
..........
; ........
a ......
; --'
:
'
;
. . . . . . . . . ;I, . . . . . .
,i. . . . . . . . . . . .
Time (ml~)
Figure 3. SFC separation of synthetic mixture o fbenzoic acid and benzoyl peroxide.
=AU t A 20 1
I
I 2
[
Ex..
eq, 4
8
a
t0
12
min
Extfuci of polymer semplo ~e~ uplkB
~n
Time (min)
Figure 4. SFC separation of polymer extract (A) with and (B) without a spike of HQ and MEHQ. Polymer cure time = 24 hours. Table I. Raw area, Wt. (mg), % Wt., and %RSD of DMPT, BPO, and MEHQ extracted and quantified in 0.85 gram of polymer cured at 35 ~ (1 part powder and 2 parts liquid). DMPT
BPO
MEHQ
HQ
Sample cured in 30 minutes Raw Area Wt. (mg) % Wt. %RSD
965 1.88 0.22 15.2
326 3.72 0.43 4.2
4.3 0.032 0.004 18.1
ND*
Sample cured in 24 h Raw Area Wt. (mg) %Wt. %RSD
562 1.05 0.12 9.5
319 3.63 0.43 6.9
ND
ND
* Not Detected.
5 36
C h r o m a t o g r a p h i a 2003, 57, April (No. 7/8)
bilized. The same S F E conditions t h a t were to be used o n the " p r e p a r e d " cured co-polymer samples were employed here. Inert s a n d was introduced to minimize the vessel d e a d volume. The c h r o m a t o g r a n l o f the p o w d e r extract solution showed the BPO peak a n d two other peaks. T h e peak which ~ at a retention time of 9.1 minutes is due to benzoic acid which is the hydrolysis p r o d u c t of BPO. The second peak which ~ at retention time o f 10.4 minutes can n o t be assigned. Although, the elution time of this peak is very close to the retention time of HQ, this peak c a n n o t be attributed to H Q because the p o w d e r did n o t contain HQ. Next, various polymer samples were cured at two different times a n d then extracted a n d analyzed using the S F E a n d S F C conditions stated in the Experimental section. Table I shows the raw area, actual weight of analyte, calculated weightpercent of analyte in the co-polymer, a n d relative s t a n d a r d deviation for three replicate extractions of each polymer sample cured at 30 minutes a n d 24 hours. The level of D M P T decreased by 50% (0.22% to 0.12%) as the cure time increased from 30 minutes to 24 hours. The BPO level did n o t vary with increasing cure time (0.44% vs. 0.43%). The M E H Q level could n o t be ascertained since it was below our detection limit for M E H Q . While a peak was observed near the retention time of HQ, it n o d o u b t is due to the "mystery" peak t h a t arises from extraction o f just the powder/ BPO mixture a n d alluded to earlier. Figure 4 shows the separation of a polymer extract (cure time 24 h) with a n d without a spike of M E H Q a n d HQ. Polymer extract w i t h o u t a spike of H Q p r o v i d e d a single small peak at 10.3 minutes, while the polymer extract with a spike showed several peaks with higher intensity near 10.3 min retention time. Thus, the single peak observed c a n n o t be ascribed to HQ. It is i m p o r t a n t to note here t h a t S F C o f the polymer extract also showed a peak which eluted at exactly the same retention time as M E H Q . It is i m p o r t a n t that ones does n o t assign this peak to M E H Q either. Usually the M E H Q peak elutes with some tailing, however the u n k n o w n c o m p o n e n t eluted with n o tailing a n d good symmetry. Figure 4 shows the c h r o m a t o g r a m o f the same polymer extract w i t h o u t a spike o f M E H Q a n d with a spike of M E H Q . It can be seen t h a t the extract which was spiked at a low level concentration of M E H Q yielded a peak with distinctive tailing while the polymer extract w i t h o u t a
Original
MEHQ spike gave a peak which eluted with no tailing. In conclusion, SFC has been shown to be an effective medium to monitor the presence of residual peroxide and its decomposition products as well as N,N-dimethyl-p-toluidine in polymeric formulations. Although the SFC/UV detection limits for the pair of hydroquinones were higher than the expected residual trace quantities, SFC of the raw materials afforded information concerning the integrity of the sample. With a more selective detector, no doubt residual hydroqui-
nones could be monitored. This study also demonstrates the danger in making peak assignments based upon retention time comparisons.
Original
Chromatographia 2003, 57, April (No. 7/8)
References [1] Berger,T.A. Packed Column SFE, The Royal Society of Chemistry, Cambridge, UK, 1995.
[2] Gere, D.R.; Stark, T.3.; Tweeten, T.N. Se-
phy, Hewlett Packard Application Note AN
800-4. [3] Salem, I.; Mena, P.; Gallardo, V.; Ruiz, M. STP Pharma Nci. 19955, 238. [4] Su, S.C.; Chu, H.; Chem, C.M.; Lee, S.C.; Chou, S.S. ,L Food Drug Anal. 1996,4, 223. [5] Chert, Q.C.; Mou, S.F.; Hou, X.P.; Ni, Z.M. Z Liq. Chrornatogr. Re[. Tech. 1998, 21, 705. Received:Ju129, 2002 Revisedmanuscript received: Nov 22, 2002 Accepted: Dec 11, 2002
paration of Aromatic Peroxides and Their Reaction Products with Methyl Vinyl Silicones by Supercritical Fluid Chrornatogra-
5 37
