Anal. Chem. 1997, 69, 2109-2114
Influence of Imidazole and Bis(trichlorophenyl) Oxalate in the Oxalyldiimidazole Peroxyoxalate Chemiluminescence Reaction Malin Emteborg (b. Stigbrand), Einar Ponte´n, and Knut Irgum*
Department of Analytical Chemistry, Umea˚ University, S-901 87 Umea˚, Sweden
The complex role of imidazole when used as a catalyst in the bis(2,4,6-trichlorophenyl) oxalate (TCPO) peroxyoxalate chemiluminescence (PO-CL) reaction is explained by the transient formation and subsequent degradation of 1,1′-oxalyldiimidazole (ODI). When ODI was used directly as PO-CL reagent, the stability was improved by addition of TCPO as an “imidazole sponge”, since ODI is rapidly decomposed in the presence of imidazole. In this way, the imidazole-catalyzed degradation of ODI was hindered efficiently. The stability of ODI was also influenced by the storage vessel material. Polymeric bottles were found to be more suitable than glass containers. A comparison was made between the traditionally used reagent TCPO/imidazole (mixed on-line for formation of ODI) and the new reagent combination ODI-TCPO (premixed) with respect to sensitivity, noise, and background. There is a growing interest in chemiluminescence as detection principle in liquid chromatography (LC) as well as in flow injection analysis (FIA).1,2 This mainly stems from the potential for very low detection limits, due to light emission from an ideally totally dark background. Peroxyoxalate chemiluminescence (PO-CL) is one of the more frequently used reaction schemes.3 For applications in analytical chemistry, an ideal oxalate compound for use in PO-CL should have a fast reaction with a high quantum yield, show no quenching properties, give a low background signal, have a high solubility, and be completely stable. No reagent known today fulfills all these criteria. Bis(2,4-dinitrophenyl) oxalate (DNPO), for example, has a high reactivity, but this compound exhibits a limited stability.4 The most frequently employed reagent, bis(2,4,6-trichlorophenyl) oxalate (TCPO), is more stable than DNPO but less reactive, and a base catalyst is therefore needed. The combination of TCPO with imidazole is considered to be the most efficient reagent. However, the emission profile obtained when TCPO is used exhibits a complex dependence on the imidazole concentration.5 We have previously suggested that, when using imidazole as catalyst, one of the active intermediates is, in fact, 1,1′-oxalyldiimidazole (ODI),6 which is formed by a substitution reaction between TCPO and imidazole. By using ODI (1) Robards, K.; Worsfold, P. J. Anal. Chim. Acta 1992, 266, 147-173. (2) Ponte´n, E. Ph.D. Thesis, Umea˚ University, Sweden, 1996. (3) Kwakman, P. J. M.; Brinkman, U. A. Th. Anal. Chim. Acta 1992, 266, 175192. (4) Honda, K.; Miyaguchi, K.; Imai, K. Anal. Chim. Acta 1985, 177, 103-110. (5) Hanaoka, N.; Givens, R. S.; Schowen, R. L.; Kuwana, T. Anal. Chem. 1988, 60, 2193-2197. (6) Stigbrand, M.; Ponte´n, E.; Irgum, K. Anal. Chem. 1994, 66, 1766-1770. S0003-2700(96)01225-5 CCC: $14.00
© 1997 American Chemical Society
directly, instead of relying on its formation from TCPO and imidazole, it was found that light was emitted at a faster rate. Recently, a paper concerning the kinetics and mechanism of the reaction between TCPO and imidazole was published.7 The identity of ODI as an intermediate was confirmed by UV absorbance and 13C-NMR spectra. In flow systems, the chemiluminescence signal is generally affected by flow stream variations due to pump pulsations and incomplete mixing.8 Moreover, when TCPO and imidazole are to be mixed and introduced, a careful design of the flow system setup is necessary in order to avoid a high background. One way to cope with this problem is to reduce the number of different carrier solutions by means of immobilized fluorophores. This was first done by Gu¨bitz et al.,9 and this approach now constitutes an active research area in our laboratory.10,11 Different materials as well as different luminophores were recently evaluated in a FIA system for their suitability in the determination of hydrogen peroxide.12 The results showed that 3-aminofluoranthene immobilized onto a porous methacrylate resin was the most suitable combination among those investigated. Fluorophores immobilized onto controlled pore glass (CPG) exhibited lower sensitivity, but the detection limits were similar to those attained using a porous methacrylate resin, due to a lower background level. One parameter which limits the attainable sensitivity and detection limits in practical work is the background emission. This problem has been addressed in several papers,13,14 and it has been suggested by Mann and Grayeski15 and Sigvardson and Birks16 that this background is caused by intermediates formed during the reactions between hydrogen peroxide and oxalate. Reagent impurities may also contribute.17,18 Moreover, Barnett et al.19 reported that the background emission spectrum was dramatically altered when imidazole was used instead of triethylamine (TEA) as catalyst in the absence of fluorophore. (7) Hadd, A. G.; Birks, J. W. J. Org. Chem. 1996, 61, 2657-2663. (8) Mann, B.; Grayeski, M. L. J. Chromatogr. 1987, 386, 149-158. (9) Gu ¨ bitz, G.; van Zoonen, P.; Gooijer, C.; Velthorst, N. H.; Frei, R. W. Anal. Chem. 1985, 57, 2071-2074. (10) Ponte´n, E.; Glad, B.; Stigbrand, M.; Sjo ¨gren, A.; Irgum, K. Anal. Chim. Acta 1996, 320, 87-97. (11) Ponte´n, E.; Viklund, C.; Irgum, K.; Bogen, S. T.; Nson-Lindgren, A° . Anal. Chem. 1996, 68, 4389-4396. (12) Ponte´n, E.; Stigbrand, M.; Irgum, K. Anal. Chem. 1995, 67, 4302-4308. (13) Milofsky, R. E.; Birks, J. W. J. Am. Chem. Soc. 1991, 113, 9715-9723. (14) Hanaoka, N.; Tanaka, H.; Nakamoto, A.; Takada, M. Anal. Chem. 1991, 63, 2680-2685. (15) Mann, B.; Grayeski, M. L. Anal. Chem. 1990, 62, 1532-1536. (16) Sigvardson, K. W.; Birks, J. W. Anal. Chem. 1983, 55, 432-435. (17) Mellbin, G. J. Liq. Chromatogr. 1983, 6, 1603-1616. (18) Kobayashi, S.; Sekino, J.; Honda, K.; Imai, K. Anal. Biochem. 1981, 112, 99-104. (19) Barnett, N. W.; Evans, R. N.; Rusell, R. A. Anal. Proc. 1994, 31, 241-244.
Analytical Chemistry, Vol. 69, No. 11, June 1, 1997 2109
The advent of ODI on the PO-CL reagent arena is important in a fundamental context, as it brings us one step closer to understanding both the complex reactions which result in the emission of light and the causes of the small but still troublesome background light emitted from the TCPO/imidazole CL reaction. In our previous investigation, the major problem when working with ODI was the limited stability. We have, therefore, sought ways to improve the stability of this otherwise attractive reagent. A clue to this could be found in the work of Imai et al.,20 where it was shown that the stability of DNPO could be improved if bis(4nitro-2-((3,6,9-trioxadecyloxy)carbonyl)phenyl) oxalate (TDPO) was added. We have, therefore, attempted addition of TCPO to ODI in order to suppress the reagent breakdown reactions. Here we report on some new findings which improve the stability of the reagent and explain the complex imidazole concentration dependence seen in the TCPO reaction. Further on, the stabilized ODI reagent is compared to the TCPO/imidazole system by use of two different flow system configurations. EXPERIMENTAL SECTION Reagents and Solutions. Bis(2,4,6-trichlorophenyl) oxalate (TCPO) was synthesized as described by Mohan and Turro21 and 1,1′-oxalyldiimidazole (ODI) according to the procedure outlined by Murata.22 The products were kept in darkness and stored in a desiccator at 5 °C. Oxalylchloride (98%), (trimethylsilyl)imidazole (TMSI; derivatization grade), trichlorophenol (98%), and 3-aminofluoranthene (97%; Warning: Suspected carcinogen!) were all obtained from Aldrich (Steinheim, Germany) and used as received. The hydrogen peroxide used was a 30% aqueous p.a. solution from Merck (Darmstadt, Germany), standardized by iodometric titration, and stored in the refrigerator. Acetonitrile was of “analyzed HPLC grade” (Baker, Deventer, Holland) and was further dried by adding 100 g/L of washed (water) and dried (120 °C for 72 h, followed by 380 °C for 24 h) 3 Å molecular sieves (KeboLab, Stockholm, Sweden). Withdrawal from the bottle was accomplished by a siphoning system comprising an in-line filter (Solvent IFD, Catalog No. 6725-5002; Whatman, Maidstone, England) to remove fine particles that may have been liberated from the molecular sieves. The content was protected from ambient water during withdrawal by a drying tube filled with silica gel (KeboLab). The water content of the acetonitrile on direct withdrawal from the bottle was determined to 5 µg/mL, or 0.27 mM, by coulometric Karl Fischer titration.23 Super-Q water (Millipore, Bedford, MA) was treated as described previously24 to reduce the hydrogen peroxide content and used for preparation of 0.01-2 µM hydrogen peroxide standards. All other chemicals were of reagent grade and used as received. The PO-CL reagent carrier solutions were prepared in volumetric glass flasks by dissolving ODI and/or TCPO in dried acetonitrile. Immediately after their preparation, the solutions were transferred to polypropylene bottles (Nalgene, Nalge Co., Rochester, NY) equipped with a drying tube. Elemental Analysis. The purity of the ODI was verified by elemental analysis: ODI purchased from Aldrich, 32.1% N, 49.6% (20) Imai, K.; Nishitani, A.; Tsukamoto, Y.; Wang, W.-H.; Kanda, S.; Hayakawa, K.; Miyazaki, M. Biomed. Chromatogr. 1990, 4 (3), 100-104. (21) Mohan, A. G.; Turro N. J. J. Chem. Educ. 1974, 51, 528-529. (22) Murata, S. Chem. Lett. 1983, 1819-1820. (23) Cedergren, A. Anal. Chem. 1996, 68, 784-790. (24) Stigbrand, M.; Karlsson, A.; Irgum, I. Anal. Chem. 1996, 68, 3945-3950.
2110
Analytical Chemistry, Vol. 69, No. 11, June 1, 1997
C, 3.6% H; TMSI-synthesized ODI, 29.4% N, 47.6% C, 3.1% H. Calc for C8H6N4O2: 29.5% N, 50.5% C, 3.2% H. NMR Analysis. Samples of ODI were dissolved in acetonitrile-d3 from ampules (99.5% Dr. Glaser AG, Basel, Switzerland) or DMSO-d6 (99.5% Dr. Glaser AG) and the NMR tubes were sealed by melting. The acetonitrile preparations took place in a glovebox under argon atmosphere for protection against moisture, while DMSO solutions were prepared at ambient conditions. 1HNMR spectra were recorded at ambient temperature with a Bruker AC-P 250 NMR spectrometer. The shifts were measured relative to acetonitrile-d2 at 1.95 ppm or DMSO-d5 at 2.6 ppm. The chemical shifts were identical in both solvents: for ODI, 7.2, 7.8, and 8.4 ppm, and for imidazole, 7.0, 7.6, and 11.7 ppm. Flow Injection System. A dual-reagent manifold system described elsewhere12 was used for the TCPO experiments, while a single-line system was used for the experiments with ODI as reagent. The latter system consisted of a CMA 250 inert HPLC pump (CMA Microdialysis, Stockholm, Sweden) operated at a flow rate of 1 mL/min. The hydrogen peroxide was injected by means of a pneumatically operated poly(ether ether ketone) (PEEK) sixport injection valve with a Tefzel rotor seal (Rheodyne Model 9010, Cotati, CA), fitted with a 10 µL PEEK loop. The injection interval was controlled by a Chron-Trol timer (Lindburg Enterprises, San Diego, CA), and an Alitea (Ventur Alitea AB, Stockholm, Sweden) peristaltic pump was used to fill the injection loop with hydrogen peroxide solution. All interconnections were made with PEEK capillaries of 0.17 mm i.d. (Jour Research AB, Onsala, Sweden). The chemiluminescence detector was the same as described previously,6 operated at a photomultiplicator tube (PMT) voltage of 600 V. The detector cell was filled with 3-aminofluoranthene (3-AFA) immobilized on 125-177 µm diameter controlled pore glass (CPG), prepared as described previously.12 In some experiments, the luminophore 3-AFA immobilized on an in situ polymerized support, similar to PGT 9-12 in Ponte´n et al.,11 was used. Stopped-Flow System and Experimental Conditions. A stopped-flow spectrophotometer (SF-4 series; HiTech Scientific Ltd., Salisbury, U.K.) was used with the light source blocked. Equal volumes of two reagent solutions were introduced into the 40 µL cell by two pneumatically operated syringes. An external mixing chamber was employed, and useful data could be collected 10 ms after mixing was initiated. The voltage applied to the PMT was 700 V, and the temperature was 20 ( 0.1 °C. The output was transferred to a 3396A integrator (Hewlett-Packard, Palo Alto, CA) for recording and integration. One of the syringes contained a 1 mM solution of ODI in dried acetonitrile (solution A). The other syringe contained 3-AFA (50 µM) and hydrogen peroxide (1 mM) (solution B). For some experiments, combinations of imidazole (5 mM) and/or water (40%) were added to solution B, as denoted in Table 2. Caution: Trichlorophenol, the degradation product of TCPO, is a known carcinogen! RESULTS AND DISCUSSION Effect of Imidazole as an Impurity in ODI Preparations. In a previous paper dealing with ODI as a reagent for peroxyoxalate chemiluminescence,6 we reported briefly on a few important differences between a commercially available preparation (Aldrich) and ODI synthesized via the TMSI route of Murata,22 both with respect to solubility and stability in dry acetonitrile. The commercially available ODI was less stable over time and was more difficult to dissolve, always leaving a hazy brownish residue,
whereas the TMSI-synthesized reagent dissolved completely to form an essentially clear solution. The manufacturer of the commercial preparation is hesitant to reveal how their ODI is synthesized, and this certainly makes it difficult to figure out which impurities might be present. However, in the product documentation, the assay is listed as 90%, with the major impurity being imidazole. We must, therefore, assume that up to 10% free imidazole could be present in the preparation. It would be desirable to purify the reagent, but because of the limited stability we have not been able to devise a method for this. Based on the amount of nitrogen found in the elemental analysis, the commercial preparation was estimated to contain approximately 80% ODI, while the value for TMSI-synthesized ODI closely matched the calculated value. This observation was further strengthened by NMR measurements, which showed that the commercial preparation contained substantially more free imidazole as an impurity compared to TMSI-synthesized ODI. We will refrain from attempts to quantify the amount of imidazole in each preparation because some imidazole will inevitably form due to hydrolysis by residual water present in the deuterated acetonitrile. These experiments were hampered by the limited solubility of ODI in acetonitrile, which prevented us from preparing samples of high concentration, which would have masked the hydrolysis effect. However, on a relative scale, we feel secure in concluding that the amount of imidazole is considerably higher in the commercial preparation. Nevertheless, from an analytical point of view, the most critical characteristic is how these reagents behave in a flow system designed to make use of the rapid kinetics of ODI as a PO-CL reagent for the determination of hydrogen peroxide. There is only a minor difference in sensitivity for the commercially available and TMSI-synthesized ODI. When carrier solutions prepared from Aldrich and TMSI-synthesized ODI were compared in the flow system, the sensitivity did not differ significantly. On a relative scale, the sensitivity, expressed as the calibration curve slope, was 87% (RSD 4.8%) for Aldrich ODI compared to TMSIsynthesized ODI (RSD 8.3%; three determinations for each preparation). During these tests, we noted, however, that the in situ polymerized sorbent with the immobilized luminophore got a brownish tint when the commercial ODI preparation was used, whereas no discoloration was evident with the TMSI-synthesized ODI. This discoloration could be washed off with water after the experiments were concluded. We interpret this as retention of a polar impurity or a reaction product; therefore, the commercial preparation is unfit for use with immobilized luminophores. Factors Affecting the Stability of the ODI Reagent. In our previous work with ODI, we made an attempt to stabilize ODI through addition of imidazole. Somewhat to our surprise, we observed very fast breakdown (90% decrease after 20 min) instead of the increase in stability we expected. We hypothesized that the reason for this breakdown was that the reaction was so strongly shifted toward the reaction products of a conceived hydrolysis that the addition of imidazole had no effect. We also attributed the lower signal to a mechanism similar to the decrease in sensitivity that is seen when imidazole is added above the optimal concentration in TCPO PO-CL.5 We are now certain that this was caused by a catalytic breakdown, which was triggered by the initial addition of imidazole and accelerated by the imidazole liberated in the subsequent breakdown reaction.
Figure 1. Decomposition rate for different concentrations of imidazole. The absorbance at 240 nm as a function of time after adding imidazole to a solution of ODI in dried acetonitrile. The concentration of ODI was 0.5 mM in all experiments, and the concentrations of added imidazole were none added (4), 0.5 mM (0), 5 mM (2), and 10 mM (9).
However, before we treat our experimental evidence for this, let us first review the findings of Neuvonen,25 who recently reported on the neutral and imidazole-catalyzed decomposition of 4-nitrophenyl oxalate (4-NPO). In these experiments, it was found that the decomposition of the intermediate generated in a mixture of oxalate and imidazole (suggested by Neuvonen to be ODI) is dependent on the imidazole concentration, i.e., that excess imidazole catalyzes the breakdown of ODI. All of Neuvonen’s experiments were carried out in acetonitrile with addition of water (440 mM), and since we are working with practically water-free (0.27 mM H2O) systems, we wanted to check if ODI would also decompose in the presence of imidazole under these conditions as well. The difference in UV spectra for imidazole and ODI in acetonitrile makes it possible to observe this breakdown by monitoring the absorbance at 240 nm as a function of time. This study showed that Neuvonen’s observation was valid also for ODI in acetonitrile without intentionally added water and that ODI decomposed at a faster rate in the presence of imidazole than in a solution without imidazole added. The decomposition rate for different concentrations of imidazole can be seen in Figure 1. A similar observation was also made during the NMR measurements, where the peaks attributable to imidazole were increasing as function of time, an increase that was most pronounced for commercial ODI. This catalytic breakdown clarifies why the commercial ODI preparation is less stable than TMSI-synthesized ODI, since the major impurity found in the commercial preparation is imidazole. Another important implication of the imidazole-catalyzed breakdown of ODI is that it is now, for the first time, possible to explain the decrease in sensitivity seen when TCPO is used with increasing concentrations of imidazole as catalyst. Depending on the system design, an optimum is often seen around 10 mM imidazole, above which the sensitivity starts to decrease. It is, however, still unclear what/which compound(s) form(s) during (25) Neuvonen, H. J. Chem. Soc., Perkin Trans. 2 1995, 951-954.
Analytical Chemistry, Vol. 69, No. 11, June 1, 1997
2111
Figure 2. Stability of 1 mM ODI from different sources and the combination of ODI with TCPO in dried acetonitrile. The reagents were stored in polypropene vessels. TMSI synthesized with addition of 0.25 mM TCPO (2), TMSI-synthesized ODI alone (9), and commercial ODI ([).
Figure 4. Stability of TMSI-synthesized ODI (1 mM) on storage in PTFE (2), polypropene (9), and borosilicate glass ([) vessels, presented as the change in peak area response relative to that immediately after preparation. Table 1. Effect of Imidazole and Water Content on the Sensitivity and Reaction Rate of ODI in the Reaction with Hydrogen Peroxide Investigated in Stopped-Flow Experiments expt no. 1 2 3 4
Figure 3. Stability for different amounts of TCPO added to a TMSIsynthesized ODI solution. The ODI concentration was 1 mM in all experiments, and the TCPO concentrations were none added ([), 0.1 mM (9), and 0.25 mM (2).
the degradation. Orosz and Dudar26 found that, when activated phenyl oxalates are hydrolyzed, gases are formed together with the phenols. The exact mechanisms for these reactions are yet to be determined. We suggest that some ODI will be hydrolyzed immediately when a solution is freshly prepared, because of the residual water present in the dried acetonitrile. Released imidazole, together with the imidazole present as an impurity, will then catalyze further decomposition of ODI. The residual water content of the acetonitrile is, therefore, critical, and the acetonitrile used to prepare the ODI reagent should be dried carefully to maximize the stability of the reagent. Using TCPO To Enhance the Stability of the ODI Reagent. After having confirmed the connection between the presence of (26) Orosz, G.; Dudar, E. Anal. Chim. Acta 1991, 247, 141-147.
2112 Analytical Chemistry, Vol. 69, No. 11, June 1, 1997
water, % (v/v)
imidazole, mM 2.5
20 20
2.5
peak height mV % RSD 1.65 15.5 7.82 12.4
1.2 2.7 1.3 3.2
s 47.4 9.0 19.8 10.2
t1/2 % RSD 1.9 9.8 5.5 4.8
n 3 3 4 5
imidazole and the breakdown of ODI in dried acetonitrile, we were intrigued by the thought of finding a reagent that could capture the imidazole formed in the initial water-induced breakdown, thereby stabilizing the ODI reagent. The answer proved to be very simple, namely to add a small amount of TCPO to the ODI solution. The difference in stability between the commercial and TMSI-synthesized ODI preparations is shown in Figure 2. It is evident that the commercial preparation is less stable than the TMSI-synthesized ODI and that the stability of the latter can be further enhanced by addition of small amounts of TCPO. When a combined reagent consisting of 1 mM TMSI-synthesized ODI and 0.25 mM TCPO is used, there is no significant decrease in the response for at least 8 h. The mechanism for this stabilization can be explained by capture of released imidazole by TCPO under formation of new ODI, or possibly the mixed ester amide (2,4,6trichlorophenyl) oxalylimidazole, accompanied by release of 2,4,6trichlorophenol (TCP). The function of TCPO will thus be that of an “imidazole sponge”, which prevents imidazole-catalyzed degradation of ODI. Consequently, the concentration of free imidazole will remain low, as long as TCPO is still present in the solution. When this initial discovery was made, different amounts of TCPO were combined with ODI, and the stability of each combination was studied in the flow system. The results can be seen in Figure 3 and reveal that at least 0.25 mM TCPO should be combined with 1.0 mM ODI to obtain a stable reagent. This concentration of TCPO is sufficient for eliminating the imidazole formed.
Table 2. Comparison of the Reagent Combinations TCPO/Imidazole (Mixed On-Line) and ODI/TCPO (Premixed) in the Detection of Hydrogen Peroxidea reagent combination
relative height sensitivity,b %
relative area sensitivity,b %
background level, mV
noise, µVp-p
S/N,c µM-1
limit of detection,d nM
1 mM TCPO + 5 mM imidazole 1 mM TCPO + 10 mM imidazole 1 mM TCPO + 20 mM imidazole 1 mM ODI + 0.25 mM TCPO
100 ( 0.8 89 ( 0.2 63 ( 0.4 98 ( 0.6
100 ( 1.1 96 ( 1.6 64 ( 0.2 104 ( 0.3
0.46 0.19 0.02 0.10
33 17 19 17
494 855 544 938
13.0 5.4 4.1 3.8
a Data were collected using 3-aminofluoranthene immobilized on CPG as solid luminophore. b The relative sensitivity values are based on calibration curve height and area slopes in the range up to 20 µM hydrogen peroxide. The slope and the 95% confidence interval are based on the mean response of each standard. c S/N represents the peak height sensitivity-to-noise ratio. d LOD was experimentally determined as 3 times the standard deviation of the blank or a low-concentration standard (
