- i -

Peroxyoxalate Chemiluminescence for Miniaturized Analytical

Flow Systems

by

Tobias Jonsson

AKADEMISK AVHANDLING

som med vederbörligt tillstånd av Rektorsämbetet vid Umeå Universitet för erhållandet av filosofie doktorsexamen framlägges till offentlig granskning vid

Kemiska Institutionen, Sal KB3B1 (Stora hörsalen) KBC-huset, Fredagen den 24 Januari 2003, Klockan 13:00.

Fakultetsopponent: Professor John W. Birks, Department of Chemistry and Biochemistry, and Cooperative Institute for Research in the Environmental Sciences (CIRES),

University of Colorado, Boulder, Colorado, USA.

- ii -

Title: Peroxyoxalate Chemiluminescence for Miniaturized Analytical Flow Systems

Author: Tobias Jonsson Department of Chemistry, Umeå University, S-901 87 Umeå, Sweden

Abstract: This thesis deals with the peroxyoxalate chemiluminescence (POCL) reaction and its application as a detection technique in flow systems for chemical analysis. Particularly, miniaturized flow systems aimed for separation of molecules. In such systems, a high light intensity and a rapid development of the emission are the desired reaction characteristics, for reasons discussed in this text. The work tries to develop an understanding of the chemical processes involved in POCL, with special emphasis to the species favoring or hindering a rapid light evolution. Hence, is the focus placed on the nature of catalysis and the desired properties of substances acting as catalysts in this reaction. Consequently, the scientific papers on which this work is founded includes both systematic stopped-flow studies of catalyst candidates and of the causes for diminished light emission. In addition, multivariate strategies for reaction optimization in practical analysis situations are treated, and the application of the POCL technique to detection of serum-extracted neuroactive steroids, derivatized with fluorescent moieties, is presented.

From the experiments in this thesis it is clear nucleophilic catalysts are the most efficient enhancing compounds, which means that they must possess a carefully balanced characteristics of nucleophilicity, leaving group ability, and basicity. The investigations also conclude that the feature of basicity efficiently can be delegated to a non-nucleophilic co-catalyst, which allow the use of nucleophilic catalysts that need to be deprotonated to be active. This thesis also shows the importance of minimizing the amount of competing nucleophiles at the site of reaction to maintain the emission. This implies that also solvents and buffer substances should be carefully chosen not to interfere with the emission process.

The most promising combination of catalysts found in this work was 4,5-dichloroimidazole together with 1,2,2,6,6-pentamethylpiperidne. This arrangement was capable of speeding the reaction more than tenfold while increasing the maximum emission intensity by about the same factor.

Keywords: peroxyoxalate, chemiluminescence, nucleophilic catalysis, non-nucleophilic base, imidazole, triazole, aminopyridine, miniaturized, flow system, capillary electrophoresis, multivariate

ISBN: 91-7305-377-5

© Copyright, Tobias Jonsson, 2002. All rights reserved.

Printed by Solfjädern Offset, Umeå.

- iii -

This thesis is partly based on the investigations presented in the following papers and manuscripts, that are hereafter referred to by their Roman numerals:

I. Determination of C-21 Ketosteroids in Serum Using Trifluoromethanesulfonic Acid Catalyzed Precolumn Dansylation and 1,1’-Oxalyldiimidazole Postcolumn Peroxyoxalate Chemiluminescence Detection

Patrik Appelblad, Tobias Jonsson, Torbjörn Bäckström, Knut Irgum

Analytical Chemistry, 70 (1998) 5002-5009

II. Heterocyclic Compounds as Catalysts in the Peroxyoxalate Chemiluminescence Reaction of bis(2,4,6-Trichlorophenyl)Oxalate

Tobias Jonsson, Malin Emteborg (b. Stigbrand), Knut Irgum

Analytica Chimica Acta, 361 (1998) 205-215

III. Very Fast Peroxyoxalate Chemiluminescence

Tobias Jonsson, Knut Irgum

Analytica Chimica Acta, 400 (1999) 257–264

IV. New Nucleophilic Catalysts for Bright and Fast Peroxyoxalate Chemiluminescence

Tobias Jonsson, Knut Irgum

Analytical Chemistry, 72 (2000) 1373-1380

V. Preliminary Kinetic Evaluation of Co-Catalyzed Acidic Azoles for Peroxyoxalate Chemiluminescence Detection in Miniaturized Flow Systems

Tobias Jonsson

manuscript

Papers I and IV are reprinted from Analytical Chemistry with permission from the American Chemical Society (copyright 1998 and 2000, respectively), whereas papers II and III are reprinted from Analytica Chimica Acta with permission from Elsevier Science (copyright 1998 and 1999, respectively).

- iv -

Also by the author but not included in the thesis:

A. Mechanism and Applications of Peroxyoxalate Chemiluminescence

Malin Stigbrand, Tobias Jonsson, Einar Pontén, Knut Irgum, Richard Bos

in A.M. García-Campaña, W.R.G. Baeyens (Eds), ‘Chemiluminescence in Analytical Chemistry’, Marcel Dekker, New York, 2001, Chapter 7, Pages 141-173.

B. Noise Characteristics and Analytical Precision of a Direct Injection High Efficiency and Micro Concentric Nebuliser for Sample Introduction in Inductively Coupled Plasma Mass Spectrometry

Erik Björn, Tobias Jonsson, Daniel Goitom

Journal of Analytical Atomic Spectrometry, 17 (2002) 1257-1263

C. The Origin of Peristaltic Pump Interference Noise Harmonics in Inductively Coupled Plasma Mass Spectrometry

Erik Björn, Tobias Jonsson, Daniel Goitom

Journal of Analytical Atomic Spectrometry, 17 (2002) 1390-1393

D. Characteristics of Vortexes Formed in Radial Entrance Fountain Detection Cells and their Impact on Chemiluminescence Analysis

Tobias Jonsson, Knut Irgum

manuscript in preparation for Analytical Chemistry

- v -

Table of Contents

1. Introduction................................................................................................................ 1

2. Peroxyoxalate Chemiluminescence........................................................................... 2

2.1 Emission Characteristics.......................................................................................... 3

2.2 Oxalic Reagents ........................................................................................................ 5

2.3 Light Generating Reaction....................................................................................... 7 2.3.1 The Overall Reaction ......................................................................................... 7 2.3.2 The High-Energy Intermediate.......................................................................... 7 2.3.3 The Fluorophore Excitation ............................................................................. 11

2.4 Catalysis .................................................................................................................. 13 2.4.1 General-Base Catalysis .................................................................................... 13 2.4.2 Nucleophilic Catalysis...................................................................................... 14 2.4.3 Other Enhancers .............................................................................................. 15

2.5 Side Reactions ........................................................................................................ 16 2.5.1 Hydrolysis ........................................................................................................ 16 2.5.2 Nucleophilic Deactivation................................................................................ 17 2.5.3 Base-Catalyzed Decomposition........................................................................ 17 2.5.4 Emissive Reactions and Miscellaneous............................................................. 18

2.6 Comprehensive Mechanism.................................................................................. 19

3. Chemiluminescence in Analytical Flow Systems.................................................... 21

3.1 Typical Applications .............................................................................................. 21

3.2 Conventional Systems ........................................................................................... 22

3.3 Miniaturization ...................................................................................................... 23 3.3.1 Reaction Kinetics.............................................................................................. 24 3.3.2 Mixing and Instrumentation ........................................................................... 25

3.4 Optimization.......................................................................................................... 27

4. Conclusions and Future Perspectives...................................................................... 28

5. Acknowledgement..................................................................................................... 30

6. References.................................................................................................................. 32

- vi -

- 1 -

Let there be light:

and there was light

The above words have been claimed to initiate the development of the universe1 and also the birth of rock’n’roll music,2 but more important in the context of this thesis, is that those words capture the very essence of chemiluminescence; to produce light at will.☺ Technically speaking, chemiluminescence is defined as the production of electromagnetic radiation, i.e., photons, with energies ranging from the ultra-violet to the infra-red region of the spectrum, when an electron in an atom or molecule, formed or excited as a result of a chemical reaction, returns to its natural, ground state energy level. Or to put it in the language of chemistry:

A + B → P* → P + hν (1)

There exists a multitude of chemiluminescent reactions and many of these have also been used for quantitative chemical analysis. A distinct advantage of chemiluminescence analysis in general is the absence of external light sources, which means that the analytically relevant emission can be measured against a completely dark background. One of the most efficient, and from an analytical chemical point of view, diverse synthetic chemiluminescence systems, is the peroxyoxalate chemiluminescence reaction, which relies on the oxidation of activated derivatives of oxalic acid by hydrogen peroxide in the presence of a fluorescent energy acceptor molecule acting as light emitter.

Currently, the strongest overall trend in analytical chemistry is miniaturization, a strategy that enables the assembly of analysis systems with several inherent advantages over larger conventional counterparts, such as the ability to produce reliable results from substantially smaller samples, a lower consumption of solvents and reagents and, more importantly, a higher potential for parallel processing of samples. Smaller dimensions also imply shorter transport and diffusion pathways and thus a potential for increased speed of most analyses. However, the miniaturization approach also imposes a variety of new problems, such as increased demands on instrument design, purity of chemicals and control of surfaces in contact with the sample, not to mention the need for adaptation of older reaction chemistries to meet new demands in terms of speed and sensitivity.

The aim of this thesis has been to develop conditions for generation of faster and more intense bursts of light from the peroxyoxalate chemiluminescence reaction because of the inherent advantages of such reaction characteristics in miniaturized analytical flow systems. This has included studies of the action and preferred nature of catalysts, the stability of reactants and approaches to optimization in practical analysis situations. This summary tries to illuminate the contemporary knowledge in the field of peroxyoxalate chemiluminescence, and discuss the possibilities and limitations of this technique in chemical analysis with a particular emphasis to the special requirements for miniaturized systems relying on fluid flow. So buckle up, put on your sunglasses and prepare for an enlightening explore of a bright research area…

☺ The obstinate reader might object that producing light at will is exactly what he does every morning merely by pressing the button on his bedside lamp, but to a true chemist this is light-years away from the charm and beauty of emission generated by a chemical reaction.

- 2 -

2. Peroxyoxalate Chemiluminescence As briefly outlined above, peroxyoxalate chemiluminescence,3-8 or simply POCL, is the

designation used for the reaction between hydrogen peroxide and activated derivatives of oxalic acid in the presence of a fluorescent substance, FL. The main overall reaction can be summarized as:4,8,9

FL R-CO-CO-R + H2O2 → 2 CO2 + 2 RH + hν (2)

This reaction is classified as a sensitized or indirect8,10 chemiluminescent reaction since the light is emitted from a sensitizer molecule, or fluorophore, FL, that gains its excitation energy from intermediates appearing along the reaction path, rather than directly from excited species formed in the reaction. The POCL reaction can, under certain conditions, produce light with an overall efficiency of up to 34%,11 which makes it the most efficient synthetic chemiluminescent reaction known. However, in typical analytical environments the efficiencies are much lower. The reaction has also been commercialized and made publicly available in the form of Cyalume® and Novaglow® portable ‘light-sticks’, capable of emitting light in a wide range of colors.

OO

O

OClCl

ClCl

Cl

Cl

NN

O

ON

N

ClCl

O

O

OO

O

OO

O

1 2 3 4 TCPO ODI

Figure 1. Structures of activated oxalic acid derivatives capable of producing chemiluminescence, tentatively according to the peroxyoxalate reaction scheme. Legend; 1, bis(2,4,6-trichlorophenyl) oxalate, TCPO; 2, 1,1’-oxalyldiimidazole, ODI; 3, oxalyl chloride; 4, diacetic oxalic anhydride.

A key feature of this reaction is the capability to excite a wide range of fluorophores, from the ultra-violet to the near infrared region, having excitation energies up to about 420 kcal mol-1,12,13 however, with the excitation efficiency being dependent on the type of fluorophore used.13-17 Although the original definition9 of POCL encompasses all the different types of oxalic acid derivatives displayed in Fig. 1 as possible reactants, it has not yet been explicitly clarified if all these reactions occur by the same mechanism. Due to this, and because the body of literature mainly concerns aryl oxalates8,9,18 substituted with electron-withdrawing groups (1) or secondary oxamides11,19-21 (2), this thesis will focus on these categories rather than the reactions of oxalyl chloride3,22 (3) and oxalic anhydrides23 (4). Still, it is clear that a relatively wide range of oxalic acid derivatives substituted with good leaving groups will promote the production of light if used as reactants for this reaction.4,8,9,18 The structural requirements on the peroxide reagent is conversely very stringent; only hydrogen peroxide is applicable.24 Whilst this is a feature related to the fundamental reaction mechanism as outlined below, it induces a unique selectivity from an analytical point of view, and hence, discriminating determination of hydrogen peroxide in the presence of organic peroxides can be performed.25 Another elementary attribute of the reaction of 1 or 2 is the requirement of a base catalyst to

- 3 -

achieve acceptable light intensities. However, it has recently been clarified that basicity is not the foremost characteristic of the most efficient catalysts.8,IV A fundamental limitation of the reaction is the sensitivity to water, primary alcohols, and other nucleophilic solvents that can attack and degrade the oxalic reagent and hence decrease the light intensity. All these aspects of the reaction will be further discussed below.

2.1 Emission Characteristics Before embarking on our journey through the world of POCL, we should first define some

common terms and expressions relevant for the subsequent discussion. For the sake of clarity let us begin by making a distinction between chemiluminescence and other luminescent phenomena. Fluorescence and phosphorescence are the designations normally used to describe the photoluminescent processes where externally applied light causes excitation of an electron in a compound, whereafter emission takes place from its first singlet and triplet state, respectively. Strictly speaking, the terms actually only refer to the emission processes26 themselves. Thus, the terminating step in a chemiluminescent reaction sequence is either the radiational transition between electronic states of the same multiplicity, i.e., fluorescence, or of different multiplicity, i.e., phosphorescence. Deserving to be mentioned here is also bioluminescence, i.e., chemiluminescent reactions occurring in living organisms, which usually, when occurring in their natural environments, show efficiencies unrivaled by synthetic chemiluminescence systems.

The efficiency of a chemiluminescence reaction is expressed as the chemiluminescence quantum yield,10 Φ

CL, which is defined as the fraction of reacting molecules that produce

photons, i.e., the number of light quanta emitted per reactant molecule. The chemiluminescence quantum yield is also called the overall quantum yield since it can be expressed as the product of the yield of the chemical reaction, Φ

CHEM, the yield of formation of

excited state, ΦEX

, and the efficiency of the fluorescence emission process, ΦFL

.10

ΦCL = ΦCHEM · ΦEX · ΦFL (3)

Chemiluminescent processes are inherently dynamic, i.e., the intensity of the produced light change as a function of time. Typically, this is manifested as a rise-decay curve as exemplified in Fig. 2, characterized by a rise rate constant, k

rise, a fall rate constant, k

fall, a maximum light

intensity, J, a time to reach the intensity maximum, tmax

, and the area under the curve, which corresponds to the total amount of light produced and, after appropriate calibration, also the quantum yield, Φ

CL.

Inasmuch as the photons are products of the reaction, the emission intensity, I,‡ at every moment is a function of the rate of the production of photons,4,5,27 and of course also directly related to the quantum yield, Φ

CL.

I = ΦCL d(hν)/dt (4)

‡ The emission intensity I here refers to the total photon flux from the reacting molecules, which when measured by an analysis instrument, must be modified by a less-than-unity constant describing the system’s efficiency in collecting and detecting the photons.

- 4 -

tmax

J

krise

kfall

Time

Inte

nsit

y

Figure 2. Typical rise-decay curve from a chemiluminescent reaction with indications of its characteristic parameters. This theoretical curve was generated by plotting the integrated rate function (Eq. 6) of the pooled two-step consecutive irreversible reaction model (Eq. 5) and assigning the first-order rate constants k'1 and k'2 values of 1.5 and 0.5 time-1, respectively.

Thus, the rise-decay curve will not be a manifestation of the number of photons produced versus time, but rather a plot of the time-related rate of emission. For a direct chemiluminescence reaction the light intensity I will therefore be directly related to the rate of the chemical reaction,5 whereas for an indirect chemiluminescence reaction, the rate by which the acceptor molecules are excited and emit light is proportional to the concentration of the high-energy intermediate,5 i.e., the compound responsible for the energy transfer to the fluorophore. If neither the formation nor the disappearance of the high-energy intermediate is rate-determining for the overall reaction sequence, the rise-decay curve of such an indirect chemiluminescence reaction will picture how the concentration of the rate-limiting intermediate changes with time.

If the POCL reaction, or most sensitized chemiluminescence reactions for that matter, is carried out under pseudo first-order conditions with all but one reagent or catalyst present in sufficient excess, the rise and decay of the intensity can be described by two first-order rate constants, k

rise and k

fall (cf. Fig. 2), derived from a two-step, consecutive, first-order irreversible

reaction.5,28-30,II

k'1 k'2 A → B → C (5)

This pooled reaction model treats reactants, A, intermediates, B, and products, C, as groups rather than individually, and allows the reaction rate constants k'

1 and k'

2 to be directly

evaluated from the observed intensity rate constants krise

and kfall

, by fitting the experimental rise-decay curves to the integrated rate equation of the reaction model5,28 (see also Fig. 2).

I = M · k'1 · (e-k'

2t – e-k'

1t) / (k'1 – k'2) (6)

A word of caution is called for here when assigning the pseudo first-order reaction rate constants k'

1 and k'

2 to either the rise or decay of the chemiluminescence intensity, since the

integrated biexponential rate expression is symmetric with respect to exchanging the two rate constants and thus can equally well satisfy two solutions,29,30,II the only difference being the

- 5 -

maximum theoretical intensity, M.† There is thus a need for complementary spectroscopic analysis to determine the rate of concentration change of the limiting reactant or intermediate, in order to explicitly assign the reaction rate constants. It should also be emphasized that this reaction model can neither differentiate between chemiluminescent and non-chemiluminescent paths for the reaction of intermediates in pool B, nor take into account possible decomposition of the reagents in pool A.

An equivalent, although less illustrative, approach to describe the rise-decay curve is by the sum of two exponential functions.31-33

I = A1 e-k'

1t + A2 e

-k'2t (7)

Here, the second pre-exponential factor, A2, is bound to carry a negative sign to account for

the rise in light intensity and the two factors A1 and A

2 must be of similar absolute magnitude.

This interpretation of the rise-decay curve suffers from the same limitations as above and cannot be used to predict5 the maximum emission intensity.

Under other reaction conditions biphasic or “pulsed” emission profiles with two local intensity maxima have been observed,9,31,34-36,IV which clearly cannot be modeled by the simple rise-decay kinetics outlined above. To discuss the origin of this phenomenon, however, we must first develop an understanding of all the constructive and destructive reactions and the intermediates involved in POCL chemistry and of the environment where the reaction takes place. Hence, we will return to this topic later (see section 2.6).

2.2 Oxalic Reagents A plethora of different oxalic reagents have been synthesized and used in the POCL

reaction,18 see for example Figs. 1, 3 and 4. Initial research focused on the assembly of reagents with high quantum yields, which resulted in the development9,37 of efficient aryl oxalates like TCPO (1), DNPO (5) and PCPO (6). These investigations later culminated with the discovery11 of the trifluoromethylsulfonyl substituted oxamide 7, which provided an overall quantum yield of 0.34 and thus is the most efficient synthetic chemiluminescent compound known.

OO

O

O

NO2

NO2O2N

NO2

OO

O

OClCl

ClCl

Cl

Cl

Cl

Cl

Cl

Cl

NN

O

O

SO2

SO2

CF3

CF3

Cl

Cl

Cl

Cl

Cl

Cl

5 6 7 DNPO PCPO

Figure 3. Examples of activated oxalic acid derivatives designed for high quantum yield in the peroxyoxalate chemiluminescence reaction. Legend; 5, bis(2,4-dinitrophenyl) oxalate, DNPO; 6, bis(pentachlorophenyl) oxalate, PCPO; 7, N,N’-bis(trifluoromethylsulfony1)-N,N’-bis(2,4,5-trichlorophenyl)oxamide.

† If k'1 > k'2 the disappearance of one intermediate in the reaction sequence is the rate-limiting step and consequently there will be some accumulation of that intermediate and M will be large, but if k'1 < k'2 it is the rate of the reaction between the reactants that will adjust M and limit all intermediates to low concentrations. In any case, the decline of the intensity I will be determined by the slow reaction step.

- 6 -

Several studies of both aryl oxalates9,38 and oxamides19,11 have reported of a general trend of increasing quantum yield and light intensity with increasing electronegativity of the substituents, i.e., with better leaving group ability, although no direct correlation between quantum yield, Φ

CL, and maximum light intensity, J, was observed.9 It is, however, evident11,38

that steric effects also affect the quantum yield. This is particularly apparent by the negative deviation of aryl oxalates that is substituted on both ring position 2 and 6, like PCPO (6), from a plot of the quantum yield versus the sum of Hammet sigma values,38 describing the total electronegativity of the phenyl ring substituents.

Although delivering high quantum yields, a serious practical limitation of the aryl oxalates and oxamides above is the limited solubility in polar organic solvents and in water. Aryl oxalates carrying alkoxycarbonyl groups, like TDPO39 (8) and similar,34,39,40 together with oxamides such as METQ41 (9) and 10,42,43 are examples of successful approaches to increase the solubility up to one molar in such environments, compound 10 also with an improved stability in water.

OO

O

O

NO2

O2N

OO

OO

OO

OO

O

O

8 TDPO

NN

O

ON+

N+

O

OSO2

SO2

CF3

CF3

-OSO2CF3

-OSO2CF3

NN

O

O

SO2

SO2

CF3

CF3SO3H

HO3S

9 10 METQ

Figure 4. Examples of activated oxalic acid derivatives designed for good solubility in aqueous or polar organic solvents. Legend; 8, bis[2-(3,6,9-trioxadecylcarbonyl)-4-nitrophenyl] oxalate, TDPO; 9 4,4’-{oxalyl bis[(trifluoromethylsulfonyl)imino]-ethylene} bis(4-methyl morpholinium trifluoromethanesulfonate), METQ; 10, [Oxalylbis(trifluoromethanesulfonyl)imino]-[bis(bensyl)-4,4’-disulfonic acid.

When the POCL reaction developed into a detection principle in analytical flow systems, a rapid evolvement of the light intensity became an important design criterion for the oxalic acid derivative. This eventually led to the synthesis20,21,44 of ODI (2), but started off as an unrecognized8,45,46 act of nucleophilic catalysis (see section 2.4.2), which in practice forms the reactive reagent in-situ. Since the research line of controlled in-situ reagent generation only recently has started to be pursuedII,III,IV,V the desired properties of the initial oxalic reagent are still essentially unidentified, but it can be suspected that stability might be a more important attribute than reactivity.

- 7 -

2.3 Light Generating Reaction Even though numerous original research papers treating the kinetics and mechanism of the

POCL reaction have been published and reviewed5-8 since its discovery3 by Chandross in 1963, there is still debate regarding its detailed mechanism. For quite some time the nature of the catalytic mechanism of various bases was under debate,8 which also clouded the elucidation of the other reaction steps, however, this have recently come to a settlement8 (see section 2.4). Questions that still await their definitive answers include the identity and route of formation of the high-energy intermediate and the mechanism of its interaction with the fluorophore.

Below are the different existing theories and experimental evidence for each step in the reaction reviewed, twisted and turned, weighted against each other, and condensed to produce the most plausible comprehensive mechanism of the POCL reaction. In doing so the possibility of nucleophilic catalysis is initially intentionally disregarded, and only non-catalyzed and partly general-base catalyzed POCL is considered. Catalysis as a whole is instead dealt with under the apposite heading below (see section 2.4). The reactions of activated oxalic acid derivatives are similar regardless of if the substituent is an amine or a phenyl group, and in most situations the conclusions apply to both compound classes and the general denotations oxalate and oxamide can thus be used interchangeable in the discussion below, except where indicated and in the direct descriptions of previously published experiments.

2.3.1 The Overall Reaction

The overall POCL reaction (Eq. 2) with a 1:1 stoichiometry between the oxalic acid derivative and hydrogen peroxide has, however, been methodically established through titration experiments with oxalyl chloride,22 and the observation9 of a linear increase of the amount of emitted light with increasing hydrogen peroxide concentration up to a molar ratio close to 1:1 in tests with oxalates. The observed47 first order dependence of the pseudo-first-order fall rate constants of the light intensity on the concentration of both the oxalate and hydrogen peroxide do also confirm Eq. 2. In direct measurements of the gaseous reaction products of the POCL reaction, Rauhut et al. have identified carbon dioxide as the main final product, with some formation also of carbon monoxide.9 However, all oxalate carbon could not be accounted for, and oxygen was excluded as a reaction product.9 Quantitative yields of the phenolic leaving group from aryl oxalates have also been deduced,9 again corroborating Eq. 2 as an accurate description of the overall reaction.

2.3.2 The High-Energy Intermediate

One continuous and especially perplexing problem in POCL is the question of the number and nature of the high-energy intermediates responsible for exciting the fluorophore, and today two main conflicting theories exist. The original theory4,9,17,19,31-33,48 favors one single common high-energy intermediate for this reaction, whereas more recent theories support the thought of reagent specific13,29,30 or multiple30,35,36,49,50 high-energy intermediates with different excitation efficiencies.30 The discrepancy in structure of the proposed intermediates is illustrated in Fig. 5 which attempts to display the collection of different compounds that have been proposed as candidates for the role of high-energy intermediate in the POCL reaction. It should be accentuated that none of these molecules have been directly identified, despite several skilful efforts.9,51-54 To critically review these reaction theories, the experimental

- 8 -

evidence of both kinetic and structural characteristics of all intermediates, rather than just the conclusions, must be examined. Such a scrutiny is presented below, but readers less devoted to POCL might find it convenient to skip this section and head directly for the condensed reaction mechanism presented in Fig. 7.

O

O

RO

OH

O

OH

R

OH

O

R

O O

O

R

OH

R

H

O O

OOO

O

O

O O O

OO

O

O

O

HO

R

11 12 13 14 15

O

OOO

R

O

RO

O

OO

O

R

O

R

O

RH O O

OOH

R

O O

O O

OO O

O

16 17 18 19 20

Figure 5. Collection of structures proposed as high-energy intermediates in the peroxyoxalate chemiluminescence reaction. The substituent R does in most cases correspond to the leaving group of the reagent, however, in some structures it can relate to a temporarily attached nucleophilic catalyst. Legend; 11, monoperoxyoxalic acid (ref. 3,35,50); 12, (ref. 35); 13, (ref. 35,55); 14, (ref. 7,56); 15, (ref. 7); 16, (ref. 7,49,50,56); 17, (ref. 30); 18, (ref. 13,29,30,35); 19, 1,2-dioxetan-3,4-dione (ref. 4,9,22,31,32,35,37,57); 20, (ref. 37,58). A peroxyoxalic acid variant of structure 16 (rightmost R=OOH, ref. 7) has also been suggested, as has a six-membered modification of 14 (ref. 56) exclusively formed from photogenerated acyl radicals during laser illumination.

The initial reaction upon mixing hydrogen peroxide with an activated derivative of oxalic acid has, based on the first order dependence on both hydrogen peroxide and oxalate, been suggested9,29,30,47,56 to be a nucleophilic attack of hydrogen peroxide, or in basic solutions the hydroperoxide anion, on one of the electrophilic carbonyl carbons of the oxalate. The momentary, initially zwitterionic, tetrahedral intermediate thus tentatively formed quickly collapses and expels one of the original substituents to form the monoperoxyoxalic acid (11), as positively identified by several independent kinetic59 and synthetic48,60 studies. This step parallels the reaction between hydrogen peroxide and oxalic acid,61 and the rapid attack of hydrogen peroxide on 4-nitrophenyl acetate.62 It also most likely explains the reported63 generation of stable ‘active intermediates’ (presumably 11) during storage of mixtures of hydrogen peroxide, aryl oxalates and strong carboxylic acids at temperatures below 0 °C. The ability of structure 11 to produce light only32,48 in the presence of a strong enough base catalyst, excludes 11 as a candidate for the position of high-energy intermediate but firmly establishes its status as a precursor. This conclusion also effectively excludes structures 12 and 13 as possible high-energy intermediates in POCL since they either are precursors to 11 or incompatible with 11 as a reaction intermediate.

Under reaction conditions with an excess of oxalate or oxamide, kinetic studies of both photoinitiated7,56 and hydrogen peroxide triggered30 POCL have led to the proposed formation of a peroxydioxalate (21), see Fig. 6, to account for the first order dependence of both the rise and fall of chemiluminescence on oxalate,56 and the decay rate of intermediate 11 decreasing asymptotically while the quantum yield was increasing linearly with ODI concentration.30

A guess is that the supervisor, the opponent and members of the examination committee, automatically falls into the category of

devoted readers, whereas most others safely can consider themselves as less devoted, however, it is the subtle wish of the author that the community of devoted readers will increase as a result of this thesis, so please feel free to return to this section after consultation of Fig 7.

- 9 -

Peroxydioxalate (21) was speculated7,30,56 to be capable of cyclizing to highly efficient high-energy intermediate structures 14 and/or 15 directly, or 16 and/or 17 after an attack of a nuleophile acting as catalyst. The observations of imidazole-catalyzed decomposition of ODI21 and 1132 in dry organic solvents (see section 2.5.3), together with the large backgroundI chemiluminescence (see section 2.5.4) produced from reactions of only ODI and hydrogen peroxide at certain concentrations in post-column detection applications, and the presence of polar21,64 fluorescent64,§ impurities in the commercial ODI preparation used in these studies30 do, however, make the formation of 21 and succeeding intermediates a less straightforward interpretation. In addition, the same authors report29 a constant quantum yield with increasing oxalate concentration in the imidazole-catalyzed reaction of TCPO, i.e., under conditions which, apart from the released trichlorophenol, should be identical to those described above,30 since the decline of chemiluminescence intensity was reported29 to correspond to the disappearance of in-situ formed ODI. Also, other experiments report that the quantum yield is essentially constant9 in the range 1:5 to 5:1 of oxalate/hydrogen peroxide, and that the quantum yield is independent32 of the initial concentration of 11 in its imidazole-initiated intermolecular reaction. Against all large high-energy intermediates points also the decreasing probability that this many molecules meet with sufficient energy for reaction, together with the reported,3,9,22,64-66 albeit disputed,13 volatility of the high-energy intermediate.

RO

OR

O

O

O

O

OO

OH

O

O

HO

21 22

Figure 6. Compounds suggested as precursor or degradation intermediates in the peroxyoxalate chemiluminescence reaction. Legend; 21, peroxydioxalate; 22, peroxyoxalic acid.

With excess of hydrogen peroxide, the disappearance of an intermediate most probably identical to monoperoxyoxalic acid (11), has been observed29-32,47 to follow mainly first order kinetics over all, and zero order kinetics with respect to all starting reagents, witnessing of a unimolecular reaction of 11. A small first order component with respect to hydrogen peroxide in this decay,29-32,47 has been attributed to the formation of the non-chemiluminescent peroxyoxalic acid (22), see Fig. 6 and the discussion below (see section 2.5.2). The capability of 11 to produce chemiluminescence32,48 in company with bases and fluorophores, and the very poor ability of organic peroxides9 to produce chemiluminescence with aryl oxalates have led to the conclusion that the most probable unimolecular reaction is an intramolecular cyclization of intermediate 11. As would be expected do kinetic studies33,67 indicate that the rate of the cyclization of 11 is increased by electronegative substituents, i.e., with better leaving group abilities. Direct cyclization of 11 could theoretically give either structure 18, 19, or 20, depending on the target carbonyl for the cyclization, and whether the tetrahedral intermediate collapses and expels the second R substituent, or stabilizes by acquiring a proton. Rearrangements of 1,2-dioxetan-3,4-dione (19) have also been suggested58 to produce the isoelectronic isomer structure 20, yet the decomposition of 20 would probably58 produce

§ The excitation and emission maximum of this impurity in acetonitrile was determined to 396 and 440 nm, respectively (M. Stigbrand).

- 10 -

singlet molecular oxygen, carbon dioxide and carbon monoxide rather than the expected9 two molecules of carbon dioxide.

Early reports4,9,37 of POCL indicated that the fluorophore itself catalyzed the decomposition of the high-energy intermediate, which otherwise was claimed to be stable for several minutes at room temperature. Infra-red spectroscopic studies9 also hinted that this stable intermediate neither retained features of the oxalate ester nor of a peroxy acid or diacyl peroxide since the expected vibration peaks9,37 (1806 and 1700-1800 cm-1, respectively) were not observed. These results have, however, never been possible to reproduce13,66 and in retrospective it seems probable, as also previously suggested,66 that this stable intermediate, like others,32,48,63 was the monoperoxyoxalic acid (11), and that the delayed addition of fluorophores introduced trace amounts of base as well, which initiated32,48 the reaction. This conclusion is further supported by the observations37 that the reactions with delayed fluorophore addition proceeded slower than the normal reactions and provided higher quantum yields. A peculiar result,55 sustaining the thought of fluorophore-catalyzed reaction of the high-energy intermediate, is the reported linear relationship between the quantum yield and the amount of carbon dioxide produced at different fluorophore concentrations, but the apparent lack of degassing of the solutions in these experiments55 complicates the interpretation.

Current experimental data fail to explicitly differentiate between alternatives 18 and 19 but there exist experimental results that tend to give preferentiality to one of the molecules over the other. The earliest argument13 for proposing 18 as high-energy intermediate was the large differences in extrapolated initial intensity, and thereby quantum yield, between oxalates carrying substituents of different electronegativity, and the well established68 capability of the structurally similar dialkyldioxetanones to chemically excite fluorophores. Some have favored 18 because of the decrease in quantum yield when imidazole is employed as catalyst for TCPO POCL, and attributed the decrease to the base catalyzed degradation of 18 to generate 19, which thus was proposed to not being able to excite the fluorophore.29 Still others have preferred 19 because the cyclization of the monoperoxyoxalic acid (11) shows two measurable, quite similar, rate constants when catalyzed by a sterically hindered nitrogen base in aprotic solvent.32 In the interpretation of this, the two rate constants were attributed to the deprotonation of 11 by the base catalyst and the general-acid catalysis of the following cyclization by the corresponding acid of the base catalyst, respectively.32 As the rate constants were independent on activator concentration, the authors concluded that the release of the leaving group would precede the interaction of the high-energy intermediate with the fluorophore, thereby excluding 18 and favoring 19 for this function.32 The same paper32 did, on the other hand, also report of a base-correlated decrease in quantum yield that was less pronounced at higher fluorophore concentrations, which would possibly support structure 18. It has, however, been argued48 that the cyclization may occur in a concerted process where the catalyst must be able to act as both a general-base and a general-acid catalyst to minimize the development of a negatively charged tetrahedral intermediate. This because a negative charge near the peroxide would, according to the presumed electron transfer excitation mechanism outlined below (see section 2.3.3), cause an immediate cleavage of the peroxide ring.31 This reasoning would defend structure 19, as would again, the noted volatility3,9,22,64-66 of the high-energy intermediate. The results33,67 of a faster cyclization of 11 the better the leaving group attached to the carbonyl mentioned above could also be interpreted in favor of 19 as the high-energy intermediate. Theoretical thermodynamical calculations69 suggest that the

- 11 -

lifetime of 19, before it spontaneously decomposes to two molecules of carbon dioxide, is about half a second at room temperature. This is in good agreement with the ‘immediate decomposition’ noted for the structurally similar oxalic acid anhydride dimer.70 A successful attempt to condense a volatile high-energy intermediate from the POCL reaction at –75 °C has actually been reported,65 however, with very scarce direct results and poor information about the experimental procedure.

In summary, both 18 and 19 remains good candidates for the high-energy intermediate, however, with a tilt towards 19, whereas others (11-17,20) can be dismissed on the grounds outlined above. After nearly three decades of research, the early suggested,9 highly elusive, almost mythical molecule 1,2-dixoetan-3,4-dione (19) thus still remains the strongest candidate for the role of high-energy intermediate in POCL. Arriving at this conclusion, the conventional mechanism outlined in Fig. 7 can be used to describe the productive POCL reaction up to the formation of the high-energy intermediate.

- RH

O

O

RO

OH

- RH+ H2O2R

R

O

OO O

O O

Figure 7. The most likely reaction path for the formation of the high-energy intermediate, later responsible for an energy-transfer reaction with the fluorophore (cf. Fig. 9). The first step probably proceeds via a tetrahedral intermediate in an addition-elimination reaction, whereas the second step might involve a concerted replacement.

2.3.3 The Fluorophore Excitation

Whatever the definitive structure of the high-energy intermediate turns out to be, it certainly must be able to interact with a fluorophore upon decomposition and promote it into its first singlet excited state. The involvement of emission from a singlet state, rather than the triplet state, is unequivocally shown by the close match between the POCL emission spectrum and the fluorescence spectrum of the fluorophores.3,13,19,22,47 Several studies show that the excitation step is kinetically unobservable, i.e., much faster than the preceding reaction steps. This is supported by the observation that the fall rate of the light intensity in the peroxyoxalate reaction is independent of the concentration13,22,47,55,71 of the fluorophore and also of the type13 of fluorophore used, even though there exist some reports9,37 of fluorophore-induced alterations in the reaction rate. Although there is a linear55,72 relationship between the maximum light intensity, J, and the fluorophore concentration over at least two orders of magnitude, the quantum yield, Φ

CL, tends to increase with the fluorophore concentration only

up to a certain limit.13,47 This expresses13,17,22 itself as a linear relationship between the inverse of the quantum yield, 1/Φ

CL, and the inverse of the fluorophore concentration, 1/[F], which,

together with the observed stability22 of the fluorophore throughout the reaction, again is indicative of a transfer of energy from the high-energy intermediate to the fluorophore competing with a spontaneous decomposition of the high-energy intermediate (see section 2.5.3).

The quantum yield for the POCL reaction is markedly dependent on the structure and the properties of the fluorophore, and seems to slowly level off with increasing fluorophore excitation energy.12,13,15-17,73 This is in marked contrast to the chemiluminescence from dioxetanes, where the quantum yield is independent of the fluorophore excitation energy up to a sharp limit.12 More specifically, the quantum yield for the POCL system tend to decrease

- 12 -

asymptotically12,17 with increasing fluorophore singlet energy to reach a limit at about 400-440 kJ mol-1,12,13 which corresponds to an excitation wavelength of 270-300 nm. Several experiments have revealed13-17,73 a linear relationship between the chemiluminescence quantum yield, corrected for fluorescence quantum yield, and the half-wave oxidation potential of the fluorophore, placing easily oxidized amino-substituted polycyclic aromatic hydrocarbons14,16 (23-26), see Fig. 8, in the front line of fluorophores. It has moreover been claimed17 that the relative rate of interaction between the fluorophore and the high-energy intermediate is linearly correlated to this half-wave oxidation potential of the fluorophore, although this reaction is too fast to be measured directly. However, the remarkable efficiency of amino-substituted polycyclic aromatic hydrocarbons14,16 compared to their non-substituted source molecules cannot be explained only by their favorable oxidation potential, and it has thus been suggested16 that the solvation energy also affects the excitation efficiency.

NH2

NH2

NH2

NH2

23 24 25 26

Figure 8. Fluorophores with favorably low oxidation potentials, capable of efficient energy transfer in the peroxyoxalate chemiluminescence reaction. Legend; 23, 3-aminofluoranthene, 3-AFA; 24, 3-aminoperylene; 25, 1-aminopyrene; 26, 2-aminoanthracene.

Both the leveling off with singlet energy and the direct relation with ease of oxidation described above, strongly point towards a transfer of an electron or a charge as being directly responsible4,74 for the excitation. Although it has been argued75 that charge transfer via bond making and breaking is a better idea than that of a full electron transfer between two discrete radical ions, the reports9,22 of decreased POCL quantum yields but constant reaction rates with the addition of free radical chain inhibitors, seem to support the formation of a charge-transfer complex of radical ions. Similar observations from the chemiluminescence reactions of dioxetanes have led to the formulation of the general theory of chemically initiated electron exchange luminescence, CIEEL,76-78 which, although no clear-cut evidence for its existence yet have been presented, also seems to agree well with the experiments described above.

O O

O O

+ FL -CO2

..

O O

O O -

FL +

O

C

O -

FL +.

.

FL* FL + hv-CO2

Figure 9. Illustration of the chemically initiated electron exchange luminescence (CIEEL) mechanism, adapted to the peroxyoxalate chemiluminescence reaction supposing 1,2-dioxetan-3,4-dione as the high-energy intermediate and FL the representative for the fluorescent acceptor molecule.

According to the POCL-adopted CIEEL mechanism,74,79 see Fig. 9, the high-energy intermediate, here taken as 1,2-dioxetan-3,4-dione (19), is initially reduced by the donation of an electron from the fluorophore, whereafter the reduced intermediate rapidly decomposes to a radical anion of carbon dioxide while releasing one neutral molecule of carbon dioxide.

- 13 -

Still within the same solvent cage, the strongly reducing radical anion of carbon dioxide then transfers the electron back to the oxidized fluorophore radical cation. The annihilation of the two radicals releases enough energy in one step to promote the back-transferred electron to its first singlet excited state. The CIEEL sequence is terminated by the fluorescence emission from the fluorophore.

2.4 Catalysis Any action of catalysis in the PO-CL reaction affects the light emission profile by increasing

the maximum light intensity, J, and decreasing tmax

, the time to reach this maximum. The quantum yield, Φ

CL, should remain unaffected by catalysis unless, as unfortunately is the case,

competing side reactions also are accelerated (see section 2.5.3). As hinted above, and previously reviewed,8 the nature of catalysis in POCL has been under some debate over the years. Due to the extensive research efforts20,28,31,80,81 put into this area, and with the aid from previous work done on the catalysis of carboxylic ester hydrolysis,82,83 it is now established though, that the POCL reaction can be subject to both general-base catalysis and nucleophilic catalysis, independently and in parallel, as well as in combination.8 These different forms of catalysis are schematically illustrated in Fig. 10 and further characterized below together with other means of light intensity enhancement where the mechanism is more unclear.

+ RH + B + HA + B

O

A

O

R

O

R

O

Nu+ NuH + RH

+ RH + B+ NuH + B

O

Nu

O

R

+ NuH+ HA

O

A

O

Nu

+ NuH+ HA

O

A

O

Nu

general-base catalysis nucleophilic catalysis general-base catalyzed nucleophilic catalysis

Figure 10. Simplified schematic representation of the different mechanisms of catalysis encountered in the peroxyoxalate chemiluminescence reaction. For maximum simplicity only one carbonyl group is displayed in all structures and only initial proton transfers are shown. R denotes a carbonyl substituent group, HA an attacking nucleophilic acid, B a base catalyst and NuH a nucleophilic catalyst (some nucleophilic catalysts do, as indicated here, have an acidic proton attached, whereas others lack this feature).

2.4.1 General-Base Catalysis

Any adequately strong base can, as the name implies, accomplish general-base catalysis of the POCL reaction by subtracting a proton from acidic molecules such as hydrogen peroxide or the monoperoxyoxalic acid (11), and thereby shifting the acid-base equilibrium of these compounds toward their deprotonated, more reactive form. A range of different compounds such as salicylate, hydroxide, 2,6-lutidine and pyridine have historically been employed as general-base catalysts for the POCL reaction.8 The efficiency of general-base catalysts is expected to increase with basicity but concomitant with stronger basic properties is it quite likely that the nucleophilicity84 of these compounds increase as well. If such qualities are undesired, sterically hindered strong amines, see Fig. 11, like triethylamine (27), BDN (28) or even better PMP (29), should be chosen,IV whereas the claimed85 non-nucleophilic base DBU (30), is not suitable since it causes a rapid degradationIV of oxalates, possibly due to nucleophilic attack on the carbonyls as discussed below (see section 2.5.2).

- 14 -

N

N N

N

N

N

27 28 29 30 BDN PMP DBU

Figure 11. Bases successfully (27-29) and unsuccessfully (30) employed as non-nucleophilic general-base catalysts in the peroxyoxalate chemiluminescence reaction. Legend; 27, triethylamine (ref. IV); 28, 1,8-bis(dimethylamino)naphthalene, BDN (ref. 32); 29, 1,2,2,6,6-pentamethylpiperidine, PMP (ref. IV); 30, 1,8-diazabicyclo[5.4.0]undec-7-ene, DBU (ref. III,IV).

As the organic solvents employed for POCL generally have poor hydrogen donating and accepting properties and thus a low degree of autoprotolysis,86-88 the solvent buffering capacity is exceedingly low over a wide pH range. In addition, the low dielectric constant and poor hydrogen bonding ability in most of these environments makes ionization so energetically unfavorable that ions formed by protonation or deprotonation tend to associate themselves with one or a few of its neutral precursors in so called homoconjugates87,89 or with other neutral polar species in heteroconjugates. From this, it is understandable that both absolute and relative acid-base properties89,90 of most molecules are different in organic solvents compared to in water. The amphiprotic water molecule is, in for example acetonitrile, a considerably stronger87,90,91 base than in water as it readily accepts protons by forming hydrates. Since the most stable hydrate contains four water molecules,92 the effective basicity of water increases with concentration.87 Provided that this relation holds for other similar organic solvents as well, it can account for the reported POCL experiments9,93,IV that, at low water concentrations, show an initial increase in reaction rate and light intensity with increasing water content. In summary, it can therefore be expected that the effect of even minute amounts of bases, added to or formed in the POCL reaction mixture, on the actual proton activity is quite large in these non-aqueous environments when no buffer is present in ample excess. There is thus a risk of changes in pH and thereby reaction rate during the course of the reaction in these situations. This can be expected to influence the appearance of the light emission profile, as discussed below (see section 2.6). It is also quite likely that differences among various organic solvents in these characteristics, i.e., autoprotolysis, hydrogen bonding ability and charge separation efficacy, largely can explain the observed94 variance in the light emission profile induced by change of solvents. In contrast, the larger autoprotolysis, higher dielectric constant, and better hydrogen bonding ability in mixed aqueous solvents should restrict any such effects, provided that the potion of water is sufficiently high.

2.4.2 Nucleophilic Catalysis

On the other hand, nucleophilic catalysis83 works by providing an alternative reaction pathway by replacing the substituents on the oxalate with the catalyst itself, thereby forming a more reactive intermediate reagent. Hence, with imidazole as catalyst both the two original substituents are replaced and ODI (2) is formed as a transient intermediate.20,31,80,81 Further experimentsII,IV with different catalysts for the POCL reaction have indicated that the corresponding intermediates are formed with several other nucleophilic catalysts as well, and it has been concludedII that this is a requirement for efficient catalysis. Multiple catalysis95 in the form of general-base catalyzed nucleophilic catalysis has also been indicated31,80,81 in the POCL reaction, which does imply96 that an additional catalyst molecule is assisting the

- 15 -

nucleophilic catalyst in the attack on (cf. Fig. 10), or release from (not shown), the electrophilic oxalic carbon, by subtraction of a proton. Another possibility, which may at first appear similar but which actually is not a true multiple catalysis,95 is when a general-base catalyst is used to adjust the basicity of the solution to deprotonate the nucleophilic catalyst in a pre-equilibrium before the nucleophilic attack.IV This is the presumedIV mechanism for less basic azoles such as triazoles and imidazoles substituted with electron-withdrawing groups.

An efficient nucleophilic catalyst must thus have both the properties of a good nucleophile and those of a good leaving group,83,II and if used alone, also sufficient basicity.II,V Indeed, it has been observedII that the more potent catalyst imidazole reacts with TCPO (1) nearly as fast as its more basic relative 4-methylimidazole (32), while forming a much more reactive intermediate, i.e., ODI (2). Also in line with this, is a pronounced decrease in catalytic efficiency and synchronous decrease in ability to form any reactive intermediate, that has been observedII when the acidic proton of imidazole is replaced by a methyl group as in the isobasic, but evidently less nucleophilic derivative, 1-methylimidazole. In the POCL reaction, imidazoles31,45,46,80,81,II,III,IV (31-33), triazolesII,III,IV,V (34) and 4-aminopyridines72,III,IV (35-37), see Fig. 12, have proven to possess the characteristics required for nucleophilic catalysts. Among these, imidazole (31) and the 4-aminopyridines are strong enough bases themselves to be used alone,IV,V whereas triazoles and more acidic derivatives of imidazole bearing electron-withdrawing substituents such as in 4,5-dichloroimidazole (33), requireIV,V the use of an ancillary, preferably non-nucleophilic,IV base to deprotonate the catalyst to its active ionic form.

N

N

H

N

N

H

CH3

N

N

H

Cl

Cl

N

NN

H N

NH2

N

N

N

N

31 32 33 34 35 36 37

Figure 12. Examples of compounds successfully employed as nucleophilic catalysts in the peroxyoxalate chemiluminescence reaction, singly (31-32, 35-37) or together (33-34) with strong non-nucleophilic general-base catalysts (cf. Fig. 11). Legend; 31, imidazole (ref. 45,46,II,III,IV,V); 32, 4-methylimidazole (ref. II); 33, 4,5-dichloroimidazole (ref. II,V); 34, 1,2,4-triazole (ref. II,III,IV,V); 35, 4-aminopyridine (ref. III,IV); 36, 4-(dimethylamino)pyridine (ref. 72,III,IV); 37, 4-pyrrolidinopyridine (ref. IV).

The above statement that the act of catalysis does not change the quantum yield, ΦCL

, is, however, not expected to be completely true in the case of nucleophilic catalysis, because this mode actually involves the formation of a new reagent in-situ (cf. Fig. 10). Hence, it could be expected that the trend of increasing quantum yield with increasing leaving group ability9,11,19,38 (see section 2.2) will be obeyed here as well. In support of this are the experimental resultsV in the presence of a non-nucleophilic co-catalyst showing an increase in quantum yield with the catalysts imidazole (31) < 1,2,4-triazole (34) < 4,5-dichloroimidazole (33), i.e., with decreasing base strength97 and thus increasing leaving group ability.

2.4.3 Other Enhancers

Some transition metal ions, e.g., copper(II), lead(II), and Zn(II) ions, have been claimed to enhance the quantum yield, Φ

CL, by reducing the rise and fall rates in semi-aqueous POCL

reactions catalyzed by imidazole.98,99 However, a more recent investigation100 conclude that this

- 16 -

is mainly an effect of a decrease in solution pH by the formation of complexes between the metal ions and imidazole.

Micelles101,102, reversed micelles,103-105 cyclodextrins,106 microemulsions,107,108 solid silica particles,109 filter surfaces,110 and other systems that can provide local microenvironments with properties different from the bulk, and are distributed throughout the reaction medium, have been shown to affect the quantum yield, Φ

CL, and the maximum in light intensity, J, of the

POCL reaction. In principle, there are several possible explanations that could account for these observations. Examples of observed or anticipated mechanisms include i) locally higher reagent concentrations caused by ionic or hydrophobic interactions,102,103,106,107 ii) severance of light-generating reagents and side reaction initiators or emission quenchers,102,103,107 iii) high-viscosity regions favoring stable charge transfer complexes and relaxation by fluorescence rather than by collision,107 and iv) catalysis of both light-generating and dark reactions by functional groups in the added aggregate.110 However, since several effects seem to prevail in every particular case, it is difficult to pin down the grounds for the changes in kinetics and efficiency independently.

2.5 Side Reactions As if the POCL reaction was not complicated enough already, there are in addition to the

productive reaction pathways described above, several opportunities for non-light producing decomposition of the reagents and intermediates involved in this reaction. The visible effects of such routes are a reduction of the quantum yield, Φ

CL, and most often also of the maximum

in light intensity, J. Although these processes frequently are referred to as ‘quenching’, it should be stressed that the negative effects are mostly due to the diversion of the reacting molecules to alternative dark reaction paths rather than, as this designation implies,26 an effect of non-emissive interactions of the excited specie. In addition to these processes, there are other, not directly competitive but still undesired side reactions that can take place in the POCL reaction cocktail.

2.5.1 Hydrolysis

Hydrolysis of the activated oxalic acid derivatives is probably the most troublesome and the most well studied decomposition reaction. In most situations of semi-aqueous environments this is what actually limits the maximum in light intensity, J, and the quantum yield, Φ

CL.

Quite unsurprisingly, the initial steps of this reaction is very similar to the chemiluminescent reaction and it has actually been used to study the features of nucleophilic catalysis in POCL.81 The monosubstituted oxalic acid derivative initially formed after the nucleophilic attack of a water molecule, or hydroxide ion, on one of the carbonyl carbons of the disubstituted oxalic acid derivative, suffers different fates depending on the nature of the substituents. With strongly electron-withdrawing substituents, i.e., good leaving groups such as 2,4-dinitrophenol, the main decomposition products are carbon monoxide and carbon dioxide.111-

113 These gases are produced by successive decarboxylation and decarbonylation113 of the monosubstituted oxalic acid derivative, or alternatively by a cyclic concerted mechanism similar to the suggested path for the action of ODI (2) in the dehydration of aldehyde oximes114 and carboxylic acids.115 Conversely, less activated oxalates, are hydrolyzed by two molecules of water in a stepwise process,113 generating as the final product oxalic acid,112 which is not reactive enough24 to proceed down the POCL path. In concord with the light-producing

- 17 -

reaction with hydrogen peroxide, the hydrolysis process is also subject to both nucleophilic80 and general-base catalysis.80 The action of nucleophilic catalysis in hydrolysis is evident by the difference in decomposition rates of ODI with different buffer substances at constant pH.I In general, the reagent decay ratesI with inorganic buffers like ammonium acetate and phosphate are much faster than with the less nucleophilic, so called Good-buffers116 and their analogues, i.e., zwitterionic amino acid-like substances comprising secondary or tertiary aliphatic amines and sulfonic acid groups.

2.5.2 Nucleophilic Deactivation

Another process, which in effect is similar to the hydrolysis generating oxalic acid described above, is the reaction between the oxalic reagentsIV or intermediates32 and nucleophiles that are bad leaving groups. This nucleophilic deactivation produces less reactive reagents carrying leaving groups too poor9,24 to allow the oxalic compound to be engaged in the POCL reaction. This has been observed as a decrease in both maximum light intensity, J, and quantum yield, Φ

CL, in solutions containing solvents like the lower alcohols methanol and ethanol,9 and with

nucleophilic base catalysts like morpholine and piperazine,IV whereas non-nucleophilic bases akin to the sterically hindered PMP (29) cannot initiate this kind of reaction.IV

At too high concentrations of hydrogen peroxide, another form of deactivation can occur, which is evident by the decrease in quantum yield and the small first order component in hydrogen peroxide observed in the disappearance of the monoperoxyoxalic acid (11).29-32,47 In this side reaction, 11 is attacked by another molecule of hydrogen peroxide, leading to the formation of peroxyoxalic acid (22), which cannot participate further in the POCL reaction.

2.5.3 Base-Catalyzed Decomposition

Numerous studies of the POCL reaction have noted a decline in quantum yield, ΦCL

, with increasing concentration of added base catalyst.9,21,28,29,47,IV In semi-aqueous environments this could naturally be attributed to reagent hydrolysis (see section 2.5.1) accelerated by general-base catalysis or nucleophilic catalysis,80 but results from water-free environments21,IV show that this not is the whole story.

During inspectionIV of different bases it was noted that TCPO (1) and ODI (2) decompose even when only non-nucleophilic bases are added to simple aprotic systems as dry acetonitrile. Reagents with less good leaving groups, i.e., TCPO (1), are much less affected by this admixture of bases.IV The decline in quantum yield with increasing concentration of non-nucleophilic base in the 1,2,4-triazole (34) catalyzed POCL reactionIV of 1 also point towards a base induced, or at least a base catalyzed, decomposition reaction. The decrease in maximum intensity and quantum yield, but increase in reaction rate with increasing basicity of the 4-aminopyridine catalysts, that was observedIV with ODI (2) as reagent in pure acetonitrile, also indicates that the decomposition is related to the strength of the added base. Whereas the mechanism of this decomposition remains concealed, it is evident that any attempt to speed up the POCL reaction using general base catalysts will inevitably reduce the quantum yield, Φ

CL, and, at high enough base concentration, also the maximum in light intensity, J.IV

It has also been observed that an accelerated,20,21 concentration dependent21 decomposition of the oxalic reagent ODI (2), is induced by the addition of different amounts of this reagent’s protonated leaving group imidazole (31) to dry acetonitrile solutions. This process produces carbon monoxide20 and can be halted21 by the addition of TCPO (1) in sufficient amounts

- 18 -

because it depletes the solution of imidazole by forming ODI (2). Other experimentsIV in dry basic acetonitrile environments with 1,2,4-triazole (34) as nucleophilic catalysts also show a decrease in quantum yield with increasing concentration of 34 when other conditions remain constant. This decay cannot be an effect of increased basicity since the concentration of basic species in the solution in both these cases would be expected to remain constant or decrease by the addition of more protonated azole. Whether the mechanism behind this decomposition, here exemplified by ODI (2) and imidazole (31), includes successive nucleophilic attack of imidazole followed by loss of the very unstable neighboring N-carbonylimidazole group117 and immediate dissipation of the latter into carbon monoxide and imidazolium ion, or general acid catalysis due to the amphiprotic properties of the azoles, or homolytic cleavage of the bond between the two carbonyls followed by decarbonylation and recombination, or still another path, is unknown.

The detrimental effect of base catalysts is, however, not limited to the oxalic starting reagents, as manifested by the decline in quantum yield in the base-induced POCL reaction of the monoperoxyoxalic acid (11).32 As briefly discussed above (see section 2.3.2), a probable mechanism for this decline is the spontaneous decomposition of the anion of the high-energy intermediate precursor 18 due to the destabilizing force of the negative charge near the peroxide ring.32 However, the less pronounced decrease in quantum yield observed at higher concentrations of the fluorphore,31,32 indicate a competition between the fluorophore and the base catalysts for the high-energy intermediate by another, still unidentified mechanism.

In conclusion, it is clear that bases, whether nucleophilic or not, cause decomposition of both reagents and intermediates. Any attempt to accelerate the POCL reaction using base catalysts will therefore unavoidably lower the quantum yield, Φ

CL, thereby setting a practical

upper limit for the concentration range of nucleophilic and base catalysts in every application.

2.5.4 Emissive Reactions and Miscellaneous

Side-reactions that do not lower the quantum yield and maximum light intensity by competing for the available reagent molecules can still be very troublesome, especially if the side-reactions produce light and the POCL scheme is to be implemented in a chemical analysis system. Phenoxyl radicals formed in the oxidation of phenols by hydrogen peroxide are, with strong indirect evidence,118 assumed to cause faint chemiluminescent emission centered at about 440-480 nm when diaryl oxalates are used as reagents for POCL in alkaline environments. No such chemiluminescent side-reactions could be detected when oxamides similar to METQ (9) were used as reagents,118 despite the fluorescent properties of the leaving groups.18 However, when ODI and hydrogen peroxide are mixed18,I in the absence of fluorphores there is a time-dependentI background emission around 570 nm,18 more pronounced at higher ODI/peroxide ratios,18,I indeed indicating the formation of some luminescent specie. Thus, unlike earlier suggestions,119 the vague fluorescence of buffers like imidazole is not a significant source for the POCL background emission, nor is the presence of trace impurities in the solvents.120 Kinetic studies121 of emission in the absence of fluorophore suggested that this luminescence was fast relative to the light producing reaction with fluorophores and hence could be discriminated, but this will most certainly be dependent on the reaction conditions.

It has also been shown122 that the addition of halogen ions has a negative influence on the light intensity, but since the fluorescence emission under the same conditions remained

- 19 -

constant and the chemiluminescence intensity was only determined at a fixed delay after reaction initiation, this could simply be due to a change in reaction kinetics. Some studies9,123,124 have also noted a decrease in quantum yield due to the degradation of certain fluorophores, however, without being able to determine the mechanisms for these destructive processes.

2.6 Comprehensive Mechanism With the above survey of the constructive and destructive reactions involved in POCL, it is

more than evident that the reaction altogether is very complex. The scheme in Fig. 13 tries to summarize the main features of the whole reaction sequence, including nucleophilic catalysis (step 1, see also section 2.4.2), formation of the high-energy intermediate (step 2-3, see also section 2.3.2), excitation of the fluorophore (step 4, see also section 2.3.3), fluorescence emission (step 5, see also section 2.3.3), and dark breakdown reactions (step 2b-4b, see also sections 2.5.1-2.5.3). Note that most steps are subjected to general-base catalysis (denoted B, see also section 2.4.1) and that step 1 with nucleophilic catalysis is not necessary for the reaction. It should also be mentioned that even in the imidazole-catalyzed POCL reaction with TCPO (1) a small contribution from a direct formation of a monoperoxyoxalic derivative (path not shown in Fig. 13) have been suggested33 to account for the kinetic results obtained at high concentrations of hydrogen peroxide.

O

O

R'R'

O

O

RR

B B

O

O

RO

OH

B O O

O O

B H2O2 B

decomp. decomp. decomp.

FL FL*

CO2

hv

1 2 3 4

5

2b 3b 4b

Figure 13. Comprehensive schematic illustration of peroxyoxalate chemiluminescence, including nucleophilic catalysis (step 1), formation of the high-energy intermediate (step 2-3), excitation of the fluorophore (step 4), fluorescence emission (step 5), and dark breakdown reactions (step 2b, 4b). B denotes an influence of general-bases catalysis, R’ and R are the substituents in the original and reacting oxalic reagent, respectively, and FL indicates the fluorophore. The scheme focuses on the transformations of the oxalate moiety, thereby only including intermediates with an appreciable stability or a significant importance in the production of light and excluding some reagents and products (cf. Fig. 7, 9 and 10).

When discussing the complexity of the POCL reaction it becomes unavoidable to comment on the phenomenon of biphasic or “pulsed” emission profiles with two local intensity maxima that have been observed9,31,34-36,49,50,IV under some reaction conditions. Although this has been interpreted35,36,49,50,125 as evidence for the existence of multiple high-energy intermediates, an alternative explanation, as also previously pithily suggested,8,31,IV is that the pulsed emission profiles are caused by changes in the reaction environment during the course of the reaction. This explanation can be justified by the fact that the biphasic emission profiles exclusively have been recorded at low concentrations of, or a claimed complete absence of, general-base catalysts in aprotic organic solvents, i.e., in an environment where the effective proton activity is expected to be unusually unstable towards small changes in concentrations of acids and bases (see section 2.4.1). Also, the pulsation pattern has been shown to change with changing ratios between the oxalic reagent and the base catalyst,35,36 which further supports this thought.

A simulation, see Fig. 14a, of the comprehensive POCL reaction scheme in Fig. 13, while including a rapid decline in base concentration, further strengthens the indications that a

- 20 -

fluctuating pH is responsible for generating the pulsed appearance of the light when only general-base catalysis is prevailing. This crude simulation tracks the concentration of the high-energy intermediate, which, for an indirect chemiluminescence reaction, is proportional to the light intensity, I (see section 2.1). The model assumes a non-catalyzed and a general-base catalyzed path in parallel for both the attack of hydrogen peroxide (k

1, k

3) and for the

cyclization (k2, k

4), a spontaneous dissociation of the high-energy intermediate (k

5), a base-

catalyzed decomposition of the starting reagent (k6) and the high-energy intermediate (k

7),

and a decline in base concentration governed by the release of the leaving groups (k8), see Fig.

14b. In agreement with the experimental reports35,36 of the pulsation phenomenon, the integrated chemiluminescence intensity decrease with increasing base catalyst also in this simulation. Although this simple scheme cannot explain all aspects of the experimentally observed biphasic emission profiles, it clearly shows that the pulsation phenomenon can be accounted for by the familiar comprehensive mechanism outlined in Fig. 13, without introducing multiple speculative high-energy intermediates. Possible practical explanations for the decline in effective basicity assumed above could include loss of trace amount of water or simply that the leaving group from the reagent is a stronger base than the catalyst.

0

2

4

6

8

10

0 5 10 15 20 25 30

Time (a.u.)

[Z]

(a.u

.)

A

B

C

C

A

a

X

- AHB

B- AH

Y

Z

k5

k2

k1

k4

k3

k7BQ

Q

QB k6

Qk8B + AH

b

Figure 14. Simulation of the peroxyoxalate chemiluminescence reaction under changing solution basicity at different initial base concentrations (unpublished results, T. Jonsson). Part (a) shows the time-dependent concentration of the high-energy intermediate, which is proportional to the light intensity, and (b) exhibits the exact simulation model used to produce the curves. The plot (a) show the results for base-to-oxalate ratios of 0.8, 0.4 and 0.2 as trace A, B and C, respectively. In the simulation model (b), X symbolizes the oxalic reagent, Y the monoperoxyoxalate, Z the high-energy intermediate, B a base catalyst, AH the protonated leaving group from the oxalic reagent, and Q other reaction products. Values used for the rate constants were k1=k2=0.1, k3=k4=k6=1, k5=k7=k8=100. The simulations were performed by the Chemical Kinetics Simulator (CKS) v1.1 (IBM Corporation, Almaden, CA) computer software, which used a stochastic algorithm to propagate the reaction of 107 molecules in an unspecified constant volume under constant pressure.

- 21 -

However, in the vast majority of situations where POCL is employed in chemical analysis systems the pulsation phenomenon does not appear. This is simply because the reaction conditions for generating the light pulses is not appropriate for producing light of high intensity, which is demanded in analytical situations.

3. Chemiluminescence in Analytical Flow Systems Having covered the mechanisms of both light producing and dark reactions possible within

the POCL domain, we can now focus the spotlight on the appliance of this chemistry to detection of various substances in analytical flow systems.

A major feature of chemiluminescent schemes in general, and of the POCL reaction in particular, is the selectivity due to the strict structural demands on the participating reagents. Therefore, chemiluminescence can never be a general detection principle or find its way to the big audiences, yet this technique can be invaluable to specific and demanding applications in many areas. To be completely honest, however, the interest in POCL for detection applications in chemical analysis has decreased during the last couple of years, as judged from the relative number of publications in this area. Ironically this decline in interest coincides with a more rapidly expanding knowledge of the underlying reaction mechanisms, as outlined in the preceding sections. Possibly this betokens for a renewed interest in this detection technique, as it now should be possible to more rationally design reaction conditions for specialized tasks.

Generation and detection of chemiluminescence for chemical analysis in flow systems is conceptually and instrumentally quite simple and well suited to be joined to flow injection systems5,126,127 or liquid based separation techniques like liquid chromatography,5,27,126,128,129 capillary electrophoresis130-137 and electrochromatography. In principle, this arrangement only requires reagents to be constantly supplied to the sample flow line and allowed to mix before or after entering a light-tight box containing a transparent detection cell that is continuously monitored with a light-sensing device. The dynamic condition under which this micro-synthesis of light takes place does, however, considerably complicate for example the optimization processes. The following sections provides an overview of the principles and approaches for employing POCL for detection applications in analytical flow systems, taking off at the conventionally sized systems, but focusing on the needs of tomorrows miniaturized arrangements.

3.1 Typical Applications This section aims to provide a glimpse into the typical ways of employing POCL in

analytical flow systems, without claiming to make a comprehensive survey of the available applications. Consequently, this section is mainly a summary of previously published reviews,8,44,128,138 which together provides excellent in-depth coverage of the area.

The flexibility of the POCL reaction allows for quite a few modes of operation, and depending on the limiting reagent the scheme can in principle be targeted for determination of fluorophores, hydrogen peroxide or oxalates, and in addition compounds that enhance, quench or change the kinetics of the light emission. The high light intensities produced by the POCL system under optimized conditions in pure or mixed organic solvents, together with the low levels of background light, constitutes a powerful combination to determine small

- 22 -

amounts of analyte. In fact, what confines the detection limit is in many cases not the chemiluminescence reaction itself, but rather the efficiency and purity of derivatization reactions8,44,128,138,139 or, as for the determination of hydrogen peroxide, the difficulty of obtaining blanks with sufficiently low content of the analyte.8,44,140

The wide range of fluorophores excitable by the POCL reaction has more than anything established its usefulness in analytical chemistry,8,128 largely due to the vast number of fluorescent tags that have been developed8,128,138 for different functional groups. For example have amino acids, proteins, alkaloids, steroids and polycyclic aromatic hydrocarbons (PAH) been determined by attaching fluorescent labels to functional groups like carboxylic acids, ketones, aldehydes, amines and phenols, or by the direct detection of natively fluorescent moieties.8 This mode of operation is usually combined with a liquid chromatographic,128,129 or lately capillary electrophoretic,136 separation step since the detection itself is rather indiscriminating. Yet, the selectivity in this mode has proven high enough to differentiate between fluorophores with large differences in oxidation potentials, and hence amino-PAH:s have been determined in the presence of other PAH:s.14

The extreme selectivity for hydrogen peroxide over organic hydroperoxides24,25,141 and a range of other oxidizing species141 forms the foundation for the second large area of applications for the POCL reaction.8 Naturally this approach is most often combined with a simple flow injection system, although it sometimes also includes a sophisticated automated sampling system,8 e.g., for air measurements.25 This mode can under favorable conditions give linear ranges140 extending over more than seven orders of magnitude, and reach detection limits of about one nanomolar.21,140 As hydrogen peroxide is a by-product of many enzymatic reduction reactions, serial connection to the peroxyoxalate reaction in the form of immobilized enzyme reactors, IMER:s, is a repeatedly practiced approach for discriminating determination of several analytes of biological interest, including for example glucose, cholesterol and choline.8,128

Other compounds, such as certain amines142,143 and metal ions,98,99 that possess the ability to change the kinetics of the peroxyoxalate reaction, can cause either a concentration related increase or decrease in signal intensity, depending on whether the peak of emission is brought further in to, or out of, the detection cell, and consequently either enhanced143 or ‘quenched’142 chemiluminescence could be applied for their determination. In addition have reaction, or emission, quenchers been determined via indirect detection.8 Innovative derivatization reactions forming the oxalic reagent also allow for determination of oxalate or phenols with the POCL reaction chemistry.8

3.2 Conventional Systems Typical POCL systems for flow injection analysis of hydrogen peroxide (system A-B) or

detection of chromatographically separated fluorophores (system C-D) are shown in Fig. 15. In the former case, the sample is injected into a stream of oxalate (system A) or into an on-line prepared mixture of oxalate and catalyst (system B), while the fluorophore preferably is immobilized44,144,145 on a suitable solid support145,146 inside the detection cell. System A is thus only applicable in situations where the reactivity of the reagent is sufficient. During detection of fluorophores, the required hydrogen peroxide and oxalate can either be supplied as a reagent blend via only one input line (system C), or from two separate sources, which in turn can be either pre-mixed on-line and introduced at one point (system D), or less commonly,

- 23 -

successively merged via two separate inputs (not shown). Any catalysts are in these cases typically contained in the eluent or blended into the hydrogen peroxide solution. In applications with IMER:s, system designs C and D prevails since the reagents need to be supplied after the enzyme reactor column.

P I D W

P I D W

P

C

A C

P

I D W

P

P I D W

P

C

P

B D

Figure 15. Typical instrumental set-ups employed for peroxyoxalate chemiluminescence detection in flow injection analysis (A-B) and liquid chromatography (C-D). Systems C-D also depict set-ups used with immobilized enzyme reactors. Legend; P, pump; I, injector; C, separation column or enzyme reactor or both; D, detector; W, waste.

Simplified system set-ups has been attempted by using constant dissolution of oxalate from a solid, granular in-line source147 and by generating hydrogen peroxide in photochemical in-line processes,56,148 but the most functional approach to system simplification is the clever immobilization44,144,145 of fluorophores in the detection cell noted above. Similar benefits could be envisioned by also attaching a nucleophilic catalyst to a support material in the detection cell, since the light emission in that case could be confined to the region of the detection cell only.

3.3 Miniaturization Miniaturization has remained a fashion word in analytical chemistry for more than a decade

and the incentive for this is not hard to understand as reduced size of analysis systems inherently should mean reduced consumption of samples and reagents, shorter diffusion pathways, larger surface-to-volume ratios and a practical possibility of parallel sample processing. Although the term ‘miniaturization’ often tend to limit a discussion to chip-based149,150 analysis systems, this text will mainly consider capillary-based flow systems, largely because POCL detection on chips151-153 still remains in its very infancy and the problematic issues are similar. Detection after a separation process is inherently more difficult than in flow injection analysis because of the conflicting issues of sensitivity and conservation of separation efficiency, and hence focus will here be placed on post-column reaction detection.

While there is a multitude of reviews available covering chemiluminescence detection in capillary electrophoresis,130-137 essentially only original research papers are offered in the area of capillary liquid chromatography.154,155 The majority of all these studies tend to deal only with the design of suitable interfaces and detection cells, while largely ignoring the effect of the chemiluminescence reaction kinetics131,154,IV on the overall performance. The sections below

Here the author must confess his failure to obtain reproducible data with solid-phase nucleophilic catalysts, i.e., ‘Carthage still stands’.

- 24 -

aims to discuss the chemical and instrumental requirements on POCL detection in miniaturized separation systems and review the currently available interface designs.

3.3.1 Reaction Kinetics

A consequence of the rise-decay emission behavior of chemiluminescent reactions is that when the climb of light intensity is slow it might be necessary to introduce extra flow elements in the analytical flow system to allow for a delay before entering the detection cell. In turn, this will increase the tendency for dispersion127 in the flow line, which will easily broaden analyte zones in any analytical flow system and thus effortlessly cause an unacceptable degradation of the separation efficiency achieved by modern, miniaturized, highly efficient, separation techniques. Similarly will the observation time allowed in these systems be limited156 to one tenth of the width of the eluted peak in order to conserve the bandwidth in the detection system, demanding a brighter chemiluminescence light emission to preserve the sensitivity. With the dimensions157,158 of today’s systems for capillary electrophoresis and the expected157 volume of the emerging analyte peaks, this would imply a observation time of less than a second and a detection cell with a physical size of less than a millimeter. However, if the decay rate of chemiluminescence is sufficiently fast, the effective detection cell volume will be independent of the physical cell volume and only rely on the light decay kinetics.154 In system configurations where the physical detection volume cannot be reduced this can be used as a means of reducing the detection cell related band-broadening,154,159 a phenomenon sometimes referred to as ‘chemical band-narrowing’.154,159 For other chemiluminescence reactions, the rapid decline in intensity have been traced to competing dark reactions,160 hinting that their presence might be important for this effect also in POCL. It can be feared, however, that this ‘band-narrowing’ can introduce a skewing of the analyte peaks under certain conditions because of the spatial non-uniformity within the detection cell, as indicated by the results from simple calculations presented in Fig. 16. These computations are based on a stepwise repositioning of a normal distributed peak passing through a hypothetical detector cell experiencing a decline in sensitivity from its entrance to its exit, while neglecting the possibility of dispersion inside the detector cell. These simulations hint that peak skewing will be most severe for reactions with a fast second order decay in intensity when the volume of the detector cell is considerably larger than the analyte peak volume.

A B C

Figure 16. Simulated distortion of a normal distributed peak during passage through a large detector cell experiencing a decline in sensitivity from its entrance to its exit (unpublished results, T. Jonsson). Note that trace A is attenuated twice relative to traces B and C. The normal distributed peak (trace A) had a standard deviation (σ) of 1 (peak width is then about 4 σ) and the detector sensitivity declined with a first order (trace B) and a second order (trace C) decay with initial half-lifes of 2 σ and 0.5 σ, respectively. The length of the hypothetical detector cell was 8 σ. Simulations were performed in an iterative custom-made BASIC computer program, which calculated the output intensity for every iteration by integrating a 0.05 σ slice of the normal distributed peak, multiplying this with the integral of the corresponding 0.05 σ slice of the decay curve, and finally adding together all the corresponding slices inside the detector. The normal distributed peak was then moved one step (0.05 σ) relative to the detector cell and the calculation process repeated until the entire peak had passed the cell.

- 25 -

In summary, the ideal, and in reality unattainable, reaction characteristics for chemiluminescence reactions employed for detection applications in miniaturized flow systems are an instantaneous climb of light intensity to a high and constant level, followed by an equally rapid decay to complete darkness (cf. Fig. 17, trace A). With the POCL reaction, emission profiles approaching the desired shape can be obtained (cf. Fig. 17, trace B), by either using a reagent20,21 that itself is sufficiently electrophilic (see section 2.2) to be easily attacked by the peroxide in both the initial replacement and the following presumed cyclization (cf. Fig. 7), or by employing a nucleophilic catalyst to transiently form such a reagent21,IV from a less reactive precursor (see section 2.4.2). The advantages with the former are fewer chemicals and thus easier optimization and most probably a simpler system set-up,21 whereas nucleophilic catalysis offers the possibility to use a less reactive and thereby stable oxalic reagent and still achieve high light intensities within short times.IV

A B C

Figure 17. Schematic illustration of the desired (A), obtainable (B) and redundant (C) light-emission profiles for chemiluminescent reactions aimed for detection applications in miniaturized flow systems. Trace B was captured from the peroxyoxalate chemiluminescence reaction of bis(2,4,6-trichlorophenyl) oxalate (TCPO) with 4,5-dichloroimidazole and 1,2,2,6,6-pentamethylpiperidine as catalysts,V whereas trace C is from TCPO catalyzed by 1-methylimidazole.II The relative maximum intensity of trace B is about 70 times that of trace C and the displayed time for trace B is about 1 s compared to 80 s for trace C.II,V

An extra dimension of flexibility and a possibility of independently varying the basicity and catalyst concentration, is obtained by the combined use of an acidic nucleophilic catalyst and a non-nucleophilic general-base catalyst,161,162,III,IV although this also complicates the optimization of the detection system (see section 3.4) with one more variable. Nevertheless, several experimentsIII,IV,V show that this approach is unmistakably the way to reach the fastest POCL reactions and highest light intensities, and that the employment161,162 of this chemistry in miniaturized flow system detection applications results in improved sensitivity. Currently, the joint use of 4,5-dichloroimidazole (32) and the sterically hindered non-nucleophilic base PMP (29), which together produce chemiluminescence light intensities nearly ten times higher than imidazole-catalyzed systems,IV,V seem to be the primary choice. In any case, it is evident the traditional combination of a stable reagent and general-base catalysts is not capable of producing the desired emission shapesII,III,IV with light intensities sufficient for small-scale systems (cf. Fig. 17, trace C).

3.3.2 Mixing and Instrumentation

Naturally, fast reaction kinetics requires reagents to be merged and efficiently mixed in immediate proximity to the detection cell, which is not effortlessly achieved within analytical flow systems typically operating exclusively within the laminar flow region. Because diffusion is the only natural force for mixing under laminar flow conditions, the key to increased mixing efficiency is to minimize the diffusion pathways between the flows to be mixed or to introduce physically enforced secondary flow paths across the boundary between the bulk

- 26 -

flows. A complicating fact in the context of separation flow systems is that mixing is only sought in the radial direction perpendicular to the bulk flow since any mixing in the axial direction would ruin the separation efficiency.

R

W

hv

S

S

W

hv

R

W

R

hv

S

R hv

S

hv

A B C D

Figure 18. Drawings of a coaxial mixer reactor (A), a sheath flow reactor (B), a mixing tee reactor (C) and an end-column reactor (D) used for interfacing capillary-based separation processes with chemiluminescence detection. Legend; S, Sample separation column; R, reagent inlet or position; W, waste; hv and arrow, approximate position of emission.

Several designs130-137 of the reactor interface between the capillary-based separation system and the chemiluminescence detection instrument have been presented for continuous monitoring of the column effluent. Although there exist some discrepancy in the naming of the different reactor types, the designations used here are the coaxial mixer reactor (A), the sheath flow reactor (B), the mixing tee reactor (C) and the end-column reactor (D), see Fig. 18. The first two reactor categories are instrumentally very similar, as they both comprise coaxial positioning of an ending separation capillary within a reaction capillary of larger diameter delivering the reagents parallel to the separation flow. Nevertheless, their principles of operation differ fundamentally. Whereas the achievements of the coaxial mixer reactor154,155,163 (A) primarily is the good mixing induced by the higher velocity of the outer flow, however, at the cost of a marked band-broadening, the sheath flow reactor164 (B) is primarily constructed for excellent bandwidth conservation by delicately matching the linear velocities of the two flows resulting in slow mixing by diffusion only. Both these configurations require the separation capillary to be carefully centered within the reaction capillary for reproducible performance. A simpler way of merging the outlet from a separation capillary with a reagent solution is via a micro sized mixing tee135 (C). While the dead volumes inside the mixing tee and in the initial connector-covered part of the detection capillary, do, on the one hand, provide enough time for thorough mixing as manifested in the reported135 high sensitivity, they on the other hand also create a pronounced band-broadening. The simplest and most rugged configuration is the end-column reactor165 (D), which consist only of a separation capillary ending in a buffer solution vial containing the reagents for chemiluminescence. Modifications of this design include continuous reagent renewal166 via a reagent inlet in the vicinity of the separation capillary and a continuous withdrawal from the top of the reservoir. Apart from maintaining a constant reaction environment by removing reaction products, this also prevents the stepwise increase in POCL background level that otherwise easily occur with the elution of every fluorescent peak167,168 in the end-column approach. Improved sensitivity, at the cost of some increase in bandwidth, have been noted in both the sheath flow169 (B) and end-column166 (D) reactors when the

- 27 -

capillary has been placed less than one millimeter away from a flow disturbing object like the tip of an optical fiber or the bottom of the outlet reservoir mounted flush against a light sensing photomultiplier tube.

Chemiluminescence detection systems for capillary electrophoresis are commonly further classified as on-column (or in-column), if the electrode is situated downstream of the detection site, or off-column, if it is placed upstream. This distinction is important for reagents or products that tend to be destroyed by high electric fields or migrate backwards into the separation capillary, as the detection system is in the latter case is isolated from the high-voltage region.161 Typically, hydrogen peroxide is prepared as a component of the running buffer as are any catalysts, whereas the oxalic reagent is supplied to, or contained in, a reservoir after the column.161 Reagent delivery can be accomplished by using either hydrostatic pressure or electrokinetic forces135 to pump the solutions, alternatively by the use of an immobilized in-line reagent170 as shown for other chemiluminescence systems. Naturally, the reagent supply system must deliver the solutions a constant rate with a minimum of pulsations, to minimize the contribution to noise.7

It is evident from the above survey that the conflicting issues of sensitivity and separation efficiency have not been fully solved with the currently available interface designs, which of course hamper the development of ultra-fast POCL detection in miniaturized systems. The recent developments of passive mixing in microchannels by incorporating bas-relief structures171 or surface heterogeneities172 with oppositely charged patches173 do, however, give some hope for the future, but these innovations remain to be tested for their effect on axial dispersion and peak broadening.

3.4 Optimization The aim of an optimization of the POCL reaction for analytical purposes is to maximize the

signal-to–noise ratio for the analyte in the concentration range of interest, which in trace analysis usually means at or near the limit of detection. This procedure actually begins with the selection of instrumental components and the subsequent assembly of the analysis system (3.3.2), and the selection of reagents and catalysts to be employed (see section 3.3.1). The selected reagents should of course, in addition to providing the desired light emission profile (see section 3.3.1), experience adequate stability and no solubility problems (see section 2.2), and give minimum background contribution (see section 2.5.4). Moreover should, if possible, solvents and buffer substances be chosen to minimize the influence of dark reagent breakdown reactions (see sections 2.5.1-2.5.3).

After the selection of suitable chemicals and instrumental hardware, the procedure of optimizing POCL for a detection application involves fine-tuning of the reaction conditions by adjusting reagent concentrations and flow rates to maximize the signal-to–noise ratio. For maximum sensitivity in flow-based chemiluminescence detection schemes, the whole rise-decay light emission curve (cf. Fig. 2) should appear within the detection cell. In practice this is not often that convenient due to a slowly rising and decaying signal and instead this becomes a matter of positioning the peak of the rise-decay curve within the detector cell, a principle commonly referred to as the ‘time-window concept’.5,126

In theory, optimization of the time-window might be an easy task, yet in practice, the parameters to adjust are numerous since all factors that either change the light emission profile or the transport speed to and inside the detector, will affect the recorded amount of

- 28 -

light and thereby the sensitivity, background level and noise of the system. For example will changes in pH and concentrations of reagents, catalysts and solvents affect both the light-producing reactions and the dark breakdown reactions in a complex and interdependent manner as outlined above (see section 2). Moreover will changes in flow rate affect a range of different parameters, including reaction related variables like concentration of all reagents and solvents, time related variables like delay time before and residence time inside the detector and fluid dynamic related variables like speed of mixing and analyte peak dispersion.174,175

Although attemptsV have been made to utilize the pooled intermediate model5 (Eq. 5-6) to provide mathematical guidelines for a general approach to reaction optimization for general-base catalyzed POCL, the non-linearity in the relation between concentrations and light intensity in nucleophilic catalysis45,IV makes this methodology inapplicable. The wide variety of non-linear, mutually dependent variables calls for a multivariate optimization approach, which in principle is unique for every analysis system. Despite that such a method is practically essential for finding the optimum in this large multivariate space, it is only too seldom practiced,I possibly because with the vast number of variables available this rapidly can become an overwhelming assignment within the flow system. A feasible approach could then be to first study some reaction parameters in a static stopped-flow system under conditions mimicking the anticipated environment in the detection cell.45,I By integrating only the time-window part of the light emission profile,45,I a maximum in light intensity at a time suitable for the detection system in question could be predicted. After such a pre-optimization of the reaction conditions, concentrations of reagents and solvents, and flow rates to the analysis system are chosen to produce the desired environment inside the detection cell. In the end, however, the optimization of the signal-to–noise ratio still comes down to at least an adjustment of the flow rates of the reagents fed to the analysis system. This is also preferentially performed in the form of an experimental design followed by a multivariate evaluation of the results.I

4. Conclusions and Future Perspectives Although this thesis demonstrates the necessity of fast and light-intense chemiluminescence

reactions for miniaturized flow-based chemical analysis systems, it must be emphasized that this work by no means is able to deliver the definitive solution to this problem. However, the chain of investigationsI,II,III,IV,V presented here strongly indicate that the route to such results for POCL proceeds through the appliance of nucleophilic catalysts in an environment of carefully controlled basicity and minimized amount of competing nucleophiles. This research also suggests that the desired characteristics of the employed nucleophilic catalyst probably coincide with the properties of an imidazole derivative substituted with non-bulky electronegative functional groups.

Several beneficial directions for expanding the knowledge and applicability of POCL detection in small-scale systems can, however, be identified from this work. One such area, previously succinctly encountered in this thesis is heterogeneous nucleophilic catalysis, which would confine the light emission to the site of the solid phase and possibly also assist in the mixing of merging reagent streams. Improved speed of mixing could also probably be achieved by merging solutions in the vapor or aerosol phase rather than as condensed fluids, but such an approach is deemed to encounter a range of other problems.

- 29 -

and, at last, there was light

- 30 -

5. Acknowledgement There is a range of people who in different ways have assisted in the production of this

thesis and to whom the author therefore owe great thanks. But before acknowledging their help I would first like to express my gratitude to professor John Birks for taking his time examining this thesis and to the financial sponsors of parts of this project, i.e., Stiftelsen Bengt Lundqvist Minne and Stiftelsen J C Kempes Minnes Stipendiefond. Since most of the persons recognized below are Swedes, or at least have (or should have) an appreciable understanding of the authors’ mother tongue, this part is written in Swedish.

Först ett stort tack till Knut Irgum, min handledare och guru i kemi, för alla idéer och all hjälp, inte minst med finansieringen. Men främst vill jag tacka för den otroliga friheten i val av mina forskningsprojekt. Även om detta troligen har betytt att jag varit mindre effektiv i avhandlingsarbetet och att fast-fas katalysatorerna inte fick någon riktig chans, så har jag lärt mig väldigt mycket - speciellt av misstagen. Sedan vill jag också tacka Malin Stigbrand, min mentor i kemiluminescens, min Livingstone i djungeln, mitt stöd i bokhyllan. Stort tack för guidningen i kemiluminescensvärlden, för svar på alla dumma frågor och för de experimentella erfarenheterna. Ett tack även till Einar Pontén, experten på kemiska analyser, för oräkneliga kloka förslag, insiktsfulla slutsatser och belönande sidouppdrag. Kort sagt; Knut, Malin och Einar, det är från era axlar jag kunnat betrakta peroxyoxalat reaktionen.

Varmt tack till alla medlemmar i ”plastgruppen” för hjälpande händer, avslappnad atmosfär och smakrika sammankomster. Speciellt till de gamla rävarna Patrik Appelblad, Camilla Viklund, Anna Sjögren, Wen Jiang och klipporna ovan, men också till de färskare förmågorna Erika Wikberg, Petrus Hemström och Anna Nordborg. Ett extra tack till Patrik som agerat personlig samtalsterapeut under avhandlingsförfattandet och stabilt bollplank tidigare.

Hjärtligt tack också till alla nuvarande och tidigare doktorander på analytisk kemi för trevligt långdragna fikaraster, fredagsöl, fester, vinprovningar, innebandy och allmänt stödjande stämning. Jag vill dessutom vidarebefodra en från Malin tidigare erhållen44 lyckospark till samtliga doktorander. Extra hårt hade jag velat sparka på Erik Björn för ett mycket spännande, utmanande och tidskrävande samarbete utanför mitt avhandlingsområde, men eftersom han hann disputera177 före mig och då jag hört att man inte ska sparka uppåt så får jag låta bli. En beundrande bugning också till Svante Jonsson för skicklig konstruktion och tillverkning av flertaliga mätceller och instrumentdetaljer till mina projekt. Likaledes ljuvliga lovord till Lars Lundmark för hjälp med allt från spänningsaggregat till oscilloskåp och till övrig teknisk-administrativ personal för den stabilitet och kontinuitet ni medför. Uppskattande uppmuntringsord också till alla seniora forskare på analytisk kemi för omåttlig kunskap, kritiska frågor och tålmodigt lyssnande på mina upprepande seminarier. En tacksam tanke även till de drygt hundratal studenter som passerat på examensarbeten, projekt och kurser, för er entusiasm och de mängder av frågor som tvingat mig att tänka efter ett varv till.

En skara som på intet sätt bidragit direkt till den här produktionen men ändå varit oumbärlig för dess färdigställande är min familj och mina vänner. Tack för att ni finns! En extra tanke till föredetta kurskamrater tillsammans med vilka jag haft många oförglömliga exkursioner, diskussioner och fester genom åren. En stor gruppkram till Mor och Far för ändlöst stöd i alla tänkbara former och till lillebror Andreas för att han finns och dyker upp när man minst anar det. Slutligen en, jag menar fyra, kramar till mina söner Måns, Isak, Noah och det nya tillskottet Love, och förlåt för minskad uppmärksamhet och att ni inte fått

- 31 -

använda datorn på ett par månader. Allra sist, men ack så viktigt, en lång, dock icke-reproduktiv, omfamning till min fru Nina.

Avslutningsvis hade jag tänkt komma med ett visdomsord, nu när man ändå har chansen. Problemet var bara att det fanns så mycket att välja på. Först ratade jag det något barnsliga och resignerande ”None tam lata sum quam bos?” och sedan det mer raka, men tråkiga ”Quod scripsi, scripsi”. Valet föll till slut på det internationellt gångbara, sant realistiska och för avhandlingen väl sammanfattande:

a giant leap for a man, but a small step for mankind

- 32 -

6. References 1. Genisis 1:3, in The Holy Bible.

2. M. Young, A. Young, B. Scott, ‘AC/DC’, ‘Let there be rock’, Albert Studios, Sydney, 1977.

3. E.A. Chandross, Tetrahedron Lett., 12 (1963) 761-765.

4. M.M. Rauhut, Acc. Chem. Res., 2 (1969) 80-87.

5. R.S. Givens, R.L. Schowen, in J.W. Birks (Ed), ‘Chemiluminescence and Photochemical Reaction Detection in Chromatography’, VCH Puplishers, New York, 1989, ch 5, pp. 125-147.

6. P.J.M. Kwakman, G.J. de Jong, U.A.Th. Brinkman, Trends Anal. Chem., 11 (1992) 232-237.

7. A.G. Hadd, J.W. Birks, in R.E. Sievers (Ed.), ‘Selective Detectors’, Chemical Analysis Series, Vol. 131, John Wiley & Sons, New York, 1995, ch. 8, pp. 209-241.

8. M. Stigbrand, T. Jonsson, E. Pontén, K. Irgum, R. Bos, in A.M. García-Campaña, W.R.G. Baeyens (Eds), ‘Chemiluminescence in Analytical Chemistry’, Marcel Dekker, New York, 2001, ch. 7, pp. 141-173.

9. M.M. Rauhut, L.J. Bollyky, B.G. Roberts, M. Loy, R.H. Whitman, A.V. Iannotta, A.M. Semsel, R.A. Clarke J. Am. Chem. Soc., 89 (1967) 6515-6522.

10. A.K. Campbell, ‘Chemiluminescence: Principles and Applications in Biology and Medicine’, Ellis Horwood, Chichester, 1988, ch. 1, pp. 40-42 and 121-124.

11. S.-S. Tseng, A.G. Mohan, L.G. Haines, L.S. Vizcarra, M.M. Rauhut, J. Org. Chem., 44 (1979) 4113-4116.

12. P. Lechtken, N.J. Turro, Mol. Photochem., 6 (1974) 95-99.

13. C.L.R. Catherall, T.F. Palmer, R.B. Cundall, J. Chem. Soc. Faraday Trans. 2, 80 (1984) 837-849.

14. K.W. Sigvardson, J.M. Kennish, J.W. Birks, Anal. Chem., 56 (1984) 1096-1102.

15. K. Honda, K. Miyaguchi, K. Imai, Anal. Chim. Acta, 177 (1985) 111-120.

16. S.C. Kang, K.-J. Kim, Bull. Korea Chem. Soc., 11 (1990) 224-227.

17. C.V. Stevani, S.M. Silva, W.J. Baader, Eur. J. Org. Chem., (2000) 4037-4046.

18. R. Bos, Ph.D. Thesis, ‘Water Soluble Reagents for Peroxyoxalate Chemiluminescence’, Deakin University, Geelong, Australia, 2001.

19. D.R. Maulding, R.A. Clarke, B.G. Roberts, M.M. Rauhut, J. Org. Chem., 33 (1968) 250-254.

20. M. Stigbrand, E. Pontén, K. Irgum, Anal. Chem., 66 (1994) 1766-1770.

21. M. Emteborg (b. Stigbrand), E. Pontén, K. Irgum, Anal. Chem., 69 (1997) 2109-2114.

22. M.M. Rauhut, B.G. Roberts, A.M. Semsel, J. Am. Chem. Soc., 88 (1966) 3604-3617.

23. L.J. Bollyky, R.H. Whitman, B.G. Robets, M.M. Rauhut, J. Am. Chem. Soc., 89 (1967) 6523-6526.

24. M.M. Rauhut, D. Sheehan, R.A. Clarke, A.M. Semsel, Photochem. Photobiol., 4 (1965) 1097-1110.

25. M. Stigbrand, A. Karlsson, K. Irgum, Anal. Chem., 68 (1996) 3945-3950.

26. J.D. Ingle Jr., S.R. Crouch, ‘Spectrochemical analysis’, Prentice-Hall, London, 1988, ch. 12, pp. 325-351.

27. T.A. Nieman, in J.W. Birks (Ed), ‘Chemiluminescence and Photochemical Reaction Detection in Chromatography’, VCH Publishers, New York, 1989, ch. 4, pp. 99-123.

28. M. Orlovic, R.L. Schowen, R.S. Givens, F. Alvarez, B. Matuzewski, N. Parekh, J. Org. Chem., 54 (1989) 3606-3610.

29. A.G. Hadd, A.L. Robinson, K.L. Rowlen, J.W. Birks, J. Org. Chem., 63 (1998) 3023-3031.

30. A.G. Hadd, A. Seeber, J.W. Birks, J. Org. Chem., 65 (2000) 2675-2683.

31. C.V. Stevani, D.F. Lima, V.G. Toscano, W.J. Baader, J. Chem. Soc. Perkin Trans 2, (1996) 989-995.

- 33 -

32. C.V. Stevani, W.J. Baader, J. Phys. Org. Chem., 10 (1997) 593-599.

33. S.M. Silva, F. Casallanovo Jr., K.H. Oyamaguchi, L.F.L.M. Ciscato, C.V. Stevani, W.J. Baader, Luminescence, 17 (2002) 313-320.

34. C.D. Dowd, D.B. Paul, Aust. J. Chem., 37 (1984) 73-86.

35. F.J. Alvarez, N.J. Parekh, B. Matuszewski, R.S. Givens, T. Higuchi, R.L. Schowen, J. Am. Chem. Soc., 108 (1986) 6435-6437.

36. J.H. Lee, S.Y. Lee, K.-J. Kim, Anal. Chim. Acta, 329 (1996) 117-126.

37. M.M. Rauhut, A.M. Semsel, Final report AD-775882 ‘A study of oxalic ester chemiluminescent reactions’, to Office of Naval Research, Arlington, USA, from American Cyanamid Company, Chemical Research Division, Bound Brook, NJ, USA, 1974.

38. A.G. Mohan, R.G. Dulina, A.A. Doering, Final Report FHWA-RD-77-79 ‘Investigation of water-soluble chemiluminescent materials’, to Federal Highway Administration, Office of Research & Development, Washington, USA, from American Cyanamid Company, Chemical Research Division, Bound Brook, NJ, USA, 1976.

39. K. Imai, H. Nawa, M. Tanaka, H. Ogata, Analyst, 111 (1986) 209-211.

40. K. Nakashima, K. Maki, S. Akiyama, W.H. Wang, Y. Tsukamoto, K. Imai, Analyst, 114 (1989) 1413-1416.

41. M.L. Grayeski, E.J. Woolf, P.J. Helly, Anal. Chim. Acta, 183 (1986) 207-215.

42. N.W. Barnett, R. Bos, S.W. Lewis, R.A. Russell, Analyst, 123 (1998) 1239-1245.

43. N.W. Barnett, R. Bos, R.N. Evans, R.A. Russell, Anal. Chim. Acta, 403 (2000) 145–154.

44. M. Emteborg (b. Stigbrand), Ph.D. Thesis, Umeå University, Umeå, Sweden, 1997.

45. N. Hanaoka, R.S. Givens, R.L. Schowen, T. Kuwana, Anal. Chem., 60 (1988) 2193-2197.

46. K. Imai, A. Nishitani, Y. Tsukamoto, W.-H. Wang, S. Kanda, K. Hayakawa, M. Miyazaki, Biomed. Chromatogr., 4 (1990) 100-104.

47. C.L.R. Catherall, T.F. Palmer, R.B. Cundall, J. Chem. Soc. Faraday Trans. 2, 80 (1984) 823-836.

48. C.V. Stevani, I.P. de Arruda Campos, W.J. Baader, J. Chem. Soc. Perkin Trans 2, (1996) 1645-1648.

49. J.H. Lee, J.C. Rock, S.B. Park, M.A. Schlautman, E.R. Carraway, J. Chem. Soc. Perkin Trans 2, (2002) 802-808.

50. J.H. Lee, J.C. Rock, S.B. Park, M.A. Schlautman, E.R. Carraway, J. Chem. Soc. Perkin Trans 2, (2002) 1593-1597.

51. H.F. Cordes, H.P. Richter, C.A. Heller, J. Am. Chem. Soc., 91 (1969), 7209-7209.

52. J.J. DeCorpo, A. Baronavski, M.V. McDowell, F.E. Saalfeld, J. Am. Chem. Soc., 94 (1972) 2879-2880.

53. M.-M. Chang, T. Saji, A.J. Bard, J. Am. Chem. Soc., 99 (1977) 5399-5403.

54. C.V. Stevani, W.J. Baader, ChemWeb preprint, (2001), available at http://preprint.chemweb.com/orgchem/0110004.

55. G. Orosz, Tetrahedron, 45 (1989) 3493-3506.

56. R.E. Milofsky, J.W. Birks, J. Am. Chem. Soc., 113 (1991) 9715-9723.

57. H. Neuvonen, J. Biolumin. Chemilumin., 12 (1997) 241-248.

58. M.D.F. Steinfatt, Bull. Soc. Chem. Belg., 94 (1985) 85-86.

59. H.P. Chokshi, M. Barbush, R.G. Carlson, R.S. Givens, T. Kuwana, R.L. Schowen, Biomed. Chromatogr., 4 (1990) 96-99.

60. J.R. Hohman, R.S. Givens, R.G. Carlson, G. Orosz, Tetrahedron Lett., 37 (1996) 8273-8276.

61. R.M. Milburn, H. Taube, J. Am. Chem. Soc., 81 (1959) 3515-3517.

- 34 -

62. W.P. Jencks, J. Am. Chem. Soc., 80 (1958) 4585-4588.

63. P. Prados, T. Santa, H. Homma, K. Imai, Anal. Sci., 11 (1995) 575-579.

64. M. Stigbrand, unpublished results.

65. J. Stauff, W. Jaeschke, G. Schlögl, Z. Naturforsch., 27 b (1972) 1434-1435.

66. E.H. White, P.D. Wildes, J. Wiecko, H. Doshan, C.C. Wei, J. Am. Chem. Soc., 95 (1973) 7050-7058.

67. W.J. Baader, S.M. Silva, K.H. Oyamaguchi, L.F.L.M. Ciscato, C.V. Stevani, ’Proc. 2nd Dresden Chemilumin. Days’, 2000.

68. S.P. Schmidt, G.B. Schuster, J. Am. Chem. Soc., 102 (1980) 306-314.

69. W.H. Richardson, H.E. O’Neal, J. Am. Chem. Soc., 94 (1972) 8665-8668.

70. P. Strazzolini, A. Gambi, A.G. Giumanini, H. Vancik, J. Chem. Soc. Perkin Trans. 1, (1998) 2553-2557.

71. G. Orosz, R.S. Givens, R.L. Schowen, Anal. Chim. Acta, 266 (1992) 219-223.

72. G. Orosz, K. Torkos, J. Borossay, Acta Chim. Hung., 128 (1991) 911-917.

74. F. McCapra, Prog. Org. Chem., 8 (1973) 231.

75. F. McCapra, in J.W. Hastings, L.J. Kricka, P.E. Stanley (Eds), ‘Proc. 9th Int. Symp. Biolumin. Chemilumin.’, John Wiley & Sons, Chichester, 1997, pp. 7-15.

76. J.-Y. Koo, G.B. Schuster, J. Am. Chem. Soc., 99 (1977) 6107-6109.

77. G.B. Schuster, Acc. Chem. Res., 12 (1979) 366-373.

78. G.B. Schuster, S.P. Schmidt, Adv. Phys. Org. Chem., 18 (1982) 187-238.

79. F. McCapra, K. Perring R.J. Hart, R.A. Hann, Tetrahedron Lett., 22 (1981) 5087.

80. H. Neuvonen, J. Chem. Soc. Perkin Trans 2, (1995) 945-949.

81. A.G. Hadd, J.W. Birks, J. Org. Chem., 61 (1996) 2657-2663.

82. M.L. Bender, Chem. Rev., 60 (1960) 53-113.

83. M.L. Bender, ’Mechanism of Homogeneous Catalysis from Protons to Proteins’, John Wiley & Sons, New York, 1971, ch. 6, pp. 147-193.

84. W.P. Jencks, J. Carriuolo, J. Am. Chem. Soc., 82 (1960) 1778-1786.

85. J. March, ‘Advanced Organic Chemistry: Reactions, Mechanisms and Structure’, 4th ed., John Wiley & Sons, New York, 1992, p 1023.

86. H.A. Laitinen, ‘Chemical Analysis. An advanced text and reference’, McGraw-Hill, New York, 1960, ch. 4, pp. 57-83.

87. J.F. Coetzee, in A. Streitwieser, R.W. Taft (Ed’s), Progr. Phys. Org. Chem., Vol. 4, Interscience, New York, 1967, pp. 45-92.

88. I.M. Kolthoff, M.K. Chantooni, Jr., J. Phys. Chem., 72 (1968) 2270-2272.

89. J.F. Coetzee, G .R. Padmanabhan, J. Am. Chem. Soc., 87 (1965) 5005-5010.

90. I.M. Kolthoff, M.K. Chantooni Jr., J. Am. Chem. Soc., 95 (1973) 8539-8546.

91. I.M. Kolthoff, M.K. Chantooni Jr., J. Am. Chem. Soc., 90 (1968) 3320-3326.

92. M.K. Chantooni, Jr., I.M. Kolthoff, J. Am. Chem. Soc., 92 (1970) 2236-2239.

93. M.M. Nakamura, S.A. Saraiva, N. Coichev, Anal. Lett., 32 (1999) 2471-2487.

94. R. Weinberger, J. Chromatogr., 314 (1984) 155-165.

95. M.L. Bender, ’Mechanism of Homogeneous Catalysis from Protons to Proteins’, John Whiley & Sons, New York, 1971, ch. 10, pp. 321-396.

- 35 -

96. J.F. Kirsch, W.P. Jencks, J. Am. Chem. Soc., 86 (1964) 833-837.

97. M.R. Grimmet, in A.R. Katritzky and C.W. Rees (Ed’s), ‘Comprehensive Heterocyclic Chemistry’, Vol. 5, part 4A, Pergamon Press, Oxford, 1984, ch. 4.07, p. 384.

98. O.M. Steijger, P.H.M. Rodenburg, H. Lingeman, U.A.Th. Brinkman, J.J.M. Holthius, Anal. Chim. Acta, 266 (1992) 233-241.

99. O.M. Steiger, H.C.M. den Nieuwenboer, H. Lingeman, U.A.Th. Brinkman, J.J.M. Holthuis, A.K. Smilde, Anal. Chim. Acta, 320 (1996) 99-105.

100. M.M. Nakamura, S.A. Saraiva, N. Coichev, Anal. Lett., 33 (2000) 391-404.

101. M.S. Abdel-Latif, G.G. Guilbault, Anal. Chem., 60 (1988) 2671-2674.

102. N. Dan, M.L. Lau, M.L. Grayeski, Anal. Chem., 63 (1991) 1766-1771.

103. N. Dan, M.L. Grayeski, Langmuir, 10 (1994) 447-453.

104. Z. Zhi, X. Yang, L. Lu, X. Wang, Spectrosc. Lett., 32 (1999) 273-279.

105. O.M. Steijger, H.M. Mastbergen, J.J.M. Holthuis, Anal. Chim. Acta, 217 (1989) 229-237.

106. E.J. Woolf, M.L. Grayeski, J. Lumin., 39 (1987) 19-27.

107. R.B. Thompson, S.E. Shaw-McBee, Langmuir, 4 (1988) 106-110.

108. S. Cohen, S. Magdassi, Langmuir, 12 (1996) 3759-3762.

109. W.J. Horvath, N. Wu, Anal. Lett., 27 (1994) 2107-2116.

110. T. Yoshinaga, M. Morita, K.-I. Wada, T. Ichimura, H. Hiratsuka, Bull. Chem. Soc. Jpn., 66 (1993) 1-8.

111. R.N. Jennings, A.C. Capomacchia, Anal. Chim. Acta, 205 (1988) 207-213.

112. G. Orosz, E. Dudar, Anal. Chim. Acta, 247 (1991) 141-147.

113. H. Neuvonen, J. Chem. Soc. Perkin Trans 2, (1994) 89-95.

114. T. Kitagawa, H. Sasaki, N. Ono, Chem. Pharm. Bull., 33 (1985) 4014-4016.

115. T. Kitagawa, H. Kuroda, H. Sasaki, Chem. Pharm. Bull., 35 (1987) 1262-1265. “

116. N.E. Good, G.D. Winget, W. Winter, T.N. Connolly, S. Izawa, R.M.M. Singh, Biochemistry, 5 (1966) 467-477.

117. P.A. Kongshaug, K.R. Haugen, R.G. Miller, J. Am. Chem. Soc., 104 (1982), 627-629.

118. N.W. Barnett, R. Bos, S.W. Lewis, R.A. Russell, Anal. Commun., 34 (1997) 17-20.

119. N.W. Barnett, R.N. Evans, , R.A. Russell, Anal. Proc., 31 (1994) 241-244.

120. K.W. Sigvardson, J.W. Birks, Anal. Chem., 55 (1983) 432-435.

121. N. Hanaoka, H. Tanaka, A. Nakamoto, M. Takada, Anal. Chem., 63 (1993) 2680-2685.

122. K. Honda, J. Sekino, K. Imai, Anal. Chem., 55 (1983) 940-943,

123. Y. Qiu, F. Zhang, F. Zhao, Y. Tang, X. Song, J. Photochem. Photobiol., 85 (1995) 281-284.

124. Y. Qiu, F. Zhang, F. Zhao, X. Song, J. Photochem. Photobiol., 87 (1995) 231-233.

125. J.H. Lee, Ph.D. Thesis, Texas A&M University, Collage Station, Texas, USA, 2000.

126. T.A. Nieman, in W.R.G. Baeyens, D. de Keukeleire, K. Korkidis (Eds), ’Luminescence Techniques in Chemical and Biochemical Analysis’, Practical Spectroscopy Series, vol. 12, Marcel Dekker, New York, 1991, ch. 17, pp. 523-565.

127. A.C. Calokerinos, L.P. Palilis, in A.M. García-Campaña, W.R.G. Baeyens (Eds), ‘Chemiluminescence in Analytical Chemistry’, Marcel Dekker, New York, 2001, ch. 12, pp. 321-348.

128. P. Kwakman, U.A.Th. Brinkman, Anal. Chim. Acta, 266 (1992) 175-192.

- 36 -

129. N. Kuroda, M. Kai, K. Nakashima, in A.M. García-Campaña, W.R.G. Baeyens (Eds), ‘Chemiluminescence in Analytical Chemistry’, Marcel Dekker, New York, 2001, ch. 14, pp. 393-425.

130. W.R.G. Baeyens, B.L. Ling, K. Imai, A.C. Calokerinos, S.G. Schulman, J. Microcol. Sep., 6 (1994) 195-206.

131. T.D. Staller, M.J. Sepaniak, Electrophoresis, 18 (1997) 2291-2296.

132. A.M. García-Campaña, W.R.G. Baeyens, Y. Zhao, Anal. Chem., 69 (1997) 83A-88A.

133. R. Zhu, W.T. Kok, J. Pharm. Biomed. Anal., 17 (1998) 985-999.

134. X.-J. Huang, Z-L. Fang, Anal. Chim. Acta, 414 (2000) 1-14.

135. C. Kuyper, R. Milofsky, Trends Anal. Chem., 20 (2001) 232-240.

136. A.M. García-Campaña, W.R.G. Baeyens, N.A. Guzman, in A.M. García-Campaña, W.R.G. Baeyens (Eds), ‘Chemiluminescence in Analytical Chemistry’, Marcel Dekker, New York, 2001, ch. 15, pp. 427-472.

137. Y.-M. Liu, J.-K. Cheng, J. Chromatogr. A, 959 (2002) 1–13.

138. P. Kwakman, Ph.D. Thesis, ‘Derivatization for Column Liquid Chromatography with Peroxyoxalate Chemiluminescence Detection’, Vrije University, Amsterdam, The Netherlands, 1991.

139. P. Appelblad, Ph.D. Thesis, ‘Separation and Detection of Neuroactive Steroids’, Umeå University, Umeå, Sweden, 2001.

140. E. Pontén, P. Appelblad, M. Stigbrand, K. Irgum, K. Nakashima, Fresenius J. Anal. Chem., 356 (1996) 84-89.

141. D. Klockow, P. Jacob, ’Chemistry of Multiphase Atmospheric Systems’, Springer, Berlin, 1986, pp. 117-130.

142. J.K. DeVasto, M.L. Grayeski, Analyst, 116 (1991) 443-447.

143. M. Katayama, H. Takeuchi, H. Taniguchi, Anal. Chim. Acta, 281 (1993) 111-118.

144. G. Gübitz, P. van Zoonen, C. Gooijer, N.H. Velthorst, R.W. Frei, Anal. Chem., 57 (1985) 2071-2074.

145. E. Pontén, Ph.D. Thesis, ‘Solid Phase Detection Reactors for Analytical Flow Systems’, Umeå University, Umeå, Sweden, 1996.

146. C. Viklund, Ph.D. Thesis, ‘Monolithic Columns for Analytical Flow Applications’, Umeå University, Umeå, Sweden, 2001.

147. P. van Zoonen, D.A. Kamminga, C. Gooijer, N.H. Velthorst, R.W. Frei, Anal. Chim. Acta, 168 (1985) 249-256.

148. J.R. Poulsen, J.W. Birks, P. van Zoonen, C. Gooijer, N.H. Velthorst, R.W. Frei, J. Chromatogr., 360 (1986) 371-383.

149. D.R. Reyes, D. Iossifidis, P-A. Auroux, A. Manz, Anal. Chem., 74 (2002) 2623-2636.

150. P-A. Auroux, D. Iossifidis, D.R. Reyes, A. Manz, Anal. Chem., 74 (2002) 2637-2652.

151. L.J. Kricka, X. Ji, O. Nozaki, P. Wilding, J. Biolumin. Chemilumin., 9 (1994) 135-138.

152. X.-Z. Wu, M. Suzuki, T. Sawada, T. Kitamori, Anal. Sci., 16 (2000) 321-323.

153. M. Hashimoto, K. Tsukagoshi, R. Nakajima, K. Kondo, A. Arai, J. Chromatogr., 867 (2000) 271-279.

154. G.J. de Jong, N. Lammers, F.J. Spruit, C. Dewaele, M. Verzele, Anal. Chem., 59 (1987) 1458-1461.

156. R.P.W. Scott, Liquid Chromatography Detectors, 2nd ed., Journal of Chromatography Library, Vol. 33, Elsevier, Amsterdam, 1986, p 27.

157. L. Cruz, S.A. Shippy, J.V. Sweedler, in ‘High Performance Capillary Electrophoresis. Chemical Analysis’, M.G. Khaledi (Ed), John Wiley & Sons, New York, 1998, Vol. 146, pp 303-354.

158. A. Sjögren, Ph.D. Thesis, ‘New Approaches in Ion Chromatography and Capillary Electrophoresis’, Umeå University, Umeå, Sweden, 2001.

159. G.J. de Jong, N. Lammers, F.J. Spruit, U.A.Th. Brinkman, R.W. Frei, Chromatographia, 18 (1984) 129-133.

160. L. Zhang, J.T. Maloy, Anal. Chem., 70 (1998) 1100-1107.

- 37 -

161. C. Kuyper, K. Denham, J. Dickson, J. Murray, R. Milofsky, Chromatographia, 53 (2001) 173-178.

162. A. Carr, J. Dickson, M. Dickson, R. Milofsky, Chromatographia, 55 (2002) 687-692.

163. R. Dadoo, L.A. Colón, R.N. Zare, J. High Resolut. Chromatogr., 15 (1992) 133-135.

164. J.-Y. Zhao, J. Labbe, N.J. Novichi, J. Microcol. Sep., 5 (1993) 331-339.

165. R. Dadoo, A.G. Seto, L.A. Colón, R.N. Zare, Anal. Chem., 66 (1994) 303-306.

166. N.W. Barnett, B.J. Hindson, S.W. Lewis, Analyst, 125 (2000) 91.

167. K. Tsukagoshi , M. Otsuka, M. Hashimoto, R. Nakajima, Anal. Sci., 15 (1999) 1257-1260.

168. K. Tsukagoshi , M. Otsuka, Y. Shikata, R. Nakajima, J. Chromatogr., 930 (2001) 165–169.

169. M. Hashimoto, T. Nakamura, K. Tsukagoshi, R. Nakajima, K. Kondo, Bull. Chem. Soc. Jpn., 72 (1999) 2673.

170. J. Lin, H. Goto, M. Yamada, J. Chromatogr. A., 844 (1999) 341.

171. A.D. Stroock, S.K.W. Dertinger, A. Ajdari, I. Mezic, H.A. Stone, G.M. Whitesides, Science, 295 (2002) 647-651.

172. O. Kuksenok, J.M. Yeomans, A.C. Balazs, Langmuir, 17 (2001) 7186-7190.

173. D. Erickson, D. Li, Langmuir, 18 (2002) 1883-1892.

174. G. Lillig, H. Engelhardt, in I.S. Krull (Ed), ’Reaction Detection in Liquid Chromatography’, Chromatographic Science Series, Vol. 34, Marcel Dekker, New York, 1986.

175. J. Ruzicka, E.H. Hansen, ‘Flow Injection Analysis’, Chemical Analysis Series, 2nd ed., John Wiley & Sons, Chichester, 1988.

176. G. Orosz, R.S. Givens, R.L. Schowen, Trends Anal. Chem., 26 (1996) 1–27.

177. E. Björn, Ph.D. Thesis, ‘Fundamental Studies of ICPQMS with Alternative Sample Introduction for Improved Reliability of Measurements’, Umeå University, Umeå, Sweden, 2002.