Chromatography Column J

0 downloads 0 Views 669KB Size Report
among others, by Lue Yen-Bower I 6 and by Parker e/ al. I;. Unfortunately, room temperature phosphorescence in liquid solutions (R TPL) must be considered as ...
Chromatography Column Recent Developments in Luminescence Detectors for Continuous Flow Systems Including HPLC

Roland W Frei & John W Birks Introduction

Che~~e.cence

The ability to measure light fluxes at extremely low levels, down to- single photon events, makes· most analytical methods based on luminescence highly sensitive. Fluorescence, the most commonly used luminescence technique, has been used for many years for detection in continuous flow systems such as high performance liquid chromatography (HPLC) at the low picogram (10- 12 g) level. Fluorescence detection also is a highly selective technique in that relatively few molecules have high fluorescence quantum yields. For compounds that lack native fluorescence, a wide variety of derivatization procedure have been developed that tag target molecules with specific fluorophores. The selectivity of the derivatization reaction toward a particular class' of compound can be used to advantage when the analyte of interest is buried in a complex sample matrix. The reaction of amino acids with o-phthalaldehyde to form highly fluorescent products is a classic example. A recent review gives many examples of the rapidly growing use of post-column derivatization reactions in liquid chromatography. 1 Additional selectivity can be achieved in fluorescence detection by judicious choice of the excitation and emission wavelengths. Also, since each fluorescent compound has a characteristic lifetime of its first excited singlet state, time resolution of the fluorescent emission can be used to achieve additional selectivity and to aid in deconvolution of overlapping peaks in HPLC, although in practice this is seldom done for lack of commercial instruments having the capability of time resolution. Time-resolved fluorescence also has the potential of enhancing the signal-to-noise ratio (hence, improved limits of detection) by gating out scattered light from the excitation source. Weinberger recently has provided an excellent review of instrumentation for fluorescence detection. 2 In addition to this we would only emphasize the enormous potential for lasers in fluorescence detection. Aside from being expensive and rather difficult to operate, lasers are ideal light sources for fluorescence detection. The light produced is intense, highly collimated, monochromatic, and tunable lasers are available over the entire UV Iv is spectral region. Furthermore, pulsed lasers allow time resolution of the fluorescence. Taken together, these properties can result in extremely low limits of detection. For example, Diebold and Zare' quantified the carcinogenic aflatoxins B., B2 , G 1 and G 2 in HPLC using laser-induced fluorescence at levels as low as 0.75 pg. Certainly, lasers increasingly will be applied to the more difficult analytical problems as their prices are reduced and their reliability increased. In the past few years great strides have been made in the previously neglected areas of chemiluminescence and phosphorescence detection. We have chosen these two areas to highlight in this article.

Who has not been fascinated at one time or another by the blinking lantern of the firefly? Such a fascination has, no doubt, been an underlying stimulus to the many chemists who have investigated the fundamentals of the chemiluminescent (including bioluminescent) reactions involved. Not least among these have been the analytical chemists, fascinated by the prospect of an analyte that sends forth its own electromagnetic wave to signal its presence. In those instances where chemiluminescence has been applied to chemical analysis, the technique has proven to be both extremely sensitive and extremely selective. In fact, the selectivity of chemiluminescence is a two-edged sword; because so few reactions are accompanied by the emission of visible light, until recently the number of applications have been quite limited. The analyte can be any of the ingredients necessary to the chemiluminescent reaction or any species that accelerates the reaction, but these have often been exotic species, e.g. hydrogen peroxide or adenosine triphosphate (ATP). The scope of chemiluminescence detection was greatly expanded in 1976 when Seitz and co-workers· reported the first analytical application of peroxyoxalate chemiluminescence. This chemiluminescence reaction system, developed in the 1960s by Rauhu t, 5 is the most efficient man-made system (with quantum yields approaching 32%)6 and is highly versatile in that it may be used to detect a variety of fluorophores in addition to hydrogen peroxide. In this reaction an oxalate ester is reacted with H 2 0 2 in the presence of a fluorescent molecule such as a polycyclic aromatic hydrocarbon. The emitted light has a spectrum identical to the fluorescence spectrum of the fluorophor. The mechanism of the reaction is still not completely understood. All evidence to date suggests a double electron transfer in which the fluorophor is formed in an excited electronic state via a sequential oxidation and reduction, analogous to the established mechanisms for several other chemiluminescent reactions in solution 7 and to electrogenerated

ESN-EuropHn S,*",cc_ N..... 57 (1984)

g o

0-0

[ + ArH - -

0

0-0 }---( ArH 0 0

J

~

T[ArH~CO;l C02

[ArH~ CO~ fig."

I.

Mrrhanism

1- ArH* + CO 2

proposrd by McCapra' for chrmiluminncrnC't tXcitalion strp

thr

prroxyoxalate

chemiluminescence. 8 McCapra 9 suggested the mechanism of Fig. I which involves the formation of a charge transfer complex between the proposed high energy intermediate 15

dioxetanedione and the fluorophor. Positive identification of dioxetanedione has remained elusive. However, this mechanism is consistent with observations that fluorophores having low oxidation potentials are most efficiently excited in the peroxyoxalate reaction. I 0.1 I Of all ftuorophores yet examined, the amino-substituted polycyclic aromatic hydrocarbons (amino-PAH) are the most efficient, apparently as a result of their low oxidation potentials ana ability to form charge transfer complexes. I 0·1 I One of the principal advantages of chemiluminescence detection over fluorescence detection is the elimination of the excitation light source. Fluctuations in the source output and scattering of excitation light into the detector are the major sources of noise in fluorescence detection. In HPLC detectors, scattered light is reduced by proper choice of cell geometry and by use of a monochromator or optical filter to discriminate against the excitation wavelength. Noise from light source fluctuations can be reduced by ratioing the fluorescence signal to the response from a photodiode or photomultiplier tube (PMT) that continuously monitors the source output. Still, detection limits in conventional (non-laser) fluorescence detectors are limited by these sources of noise and not by the PMT dark current. The peroxyoxalate reaction system does produce a weak emission even in the absence of a fluorophor. I 0 This background is quite small, however, so that the P~IT can be operated at a much higher voltage

2

Chemiluminescence 3.47

Fluorescence

290/>389

20

10

30

40

50

60

70

Time(min) fig." 2. HPLC chromatogram. or shale oil with peroxyoxalate chemiluminescence and fluorescence det.ction. Fluorescence excitation at 290 nm with a 389 nm emission cutoff filter. (I) aminonaphthalenes. (2) C,-aminonaphthalenes. (3) aminophtnanthrenes. (4) aminoanthracenes, (7) aminopyrenes

Tab" /. Det.ction limits ror amino-PAH by optimiztd fluorescence detection and by peroxyoxalatt chemiluminescence detection

2

Detection limits (pg) Compound I-aminoanthracene 6-aminochrysene 3-aminofluoranthene 2-aminofluorene I-aminonaphthalene 9-aminophenanthrene I-aminopyrene

Fluorescence

Chemiluminescence

16.0

0.15 0.30 0.09 8.0 2.0 0.90 0.11

1.2 18.0 5.0 5.0

7.2 2.6

7

4 6

than in fluorescence detection. Under optimal conditions and for favourable fluorophores, limits of detection can be improved by more than two orders of magnitude by substituting chemical excitation for light excitation. For example, detection limits for seven amino-substituted polycyclic aromatic hydrocarbons by both fluorescence detection and chemiluminescence detection are given in Table I. I I These detection limits were obtained in HPLC under chromatographic conditions that resolve all seven compounds. The detection limit of 90 femtogram (10- I 5 g) or 400 attomole (10- 18 mole) for 3-aminofluoranthene is 200 times lower than for fluorescence detection using the same fluorometer. At the detection limit fewer than two million molecules are present in the detection cell when the chromatographic peak elutes. Advantage has been taken of the very high selectivity toward PAH-amines in the analysis of these mutagenic and carcinogenic compounds in synthetic fuels and their by-products. I t Figure 2 compares fluorescence and chemiluminescence chromatograms for a shale oil sample that has been pre-fractionated on both neutral alumina and silicic acid columns. All of the significant peaks in the chemiluminescence chromatogram were identified as amino-PAH. Despite extensive sample clean-up the fluorescence chromatogram still contains too many peaks to be resolved, none of which can be identified as amino-

5

10

15

20

25

30

35

Time (min) Fig." 3. H PLC chromatogram or coal oil with peroxyoxalate chemiluminescence detection. (I) aminonaphthalenes. (2) C,.aminonaphthalenes. (4) aminophenanthrenes, (6) aminofluoranthenes. (7) aminopyrenes

PAH. Figure 3 is a similar chromatogram of a coal oil fraction. Again, all of the peaks detected in this complex sample appear to be due to amino-PAH. The nitro-substituted polycyclic aromatic hydrocarbons constitute another important class of mutagens that may be detected with high sensitivity and selectivity using peroxyoxalate chemiluminescence. 12 These compounds are found associated with carbonaceous particles such as diesel exhaust particles and carbon blackl\ used as toners for photocopying machines. The nitro-PAH are not themselves chemilumophores, but may be reduced to the corresponding amino-PAH on zinc particles at near neutral pH. In HPLC a short (3-4 cm) column packed with a I : I mixture of glass beads and zinc particles may be placed before or after the analytical column to achieve quantitative reduction. In this way two characteristic 16

ESN-Euro_" SpectrMCopy News. 57 (1984)

TClcntlon times, one charactenstic of the nitro-PAH and one characteristic of the corresponding amino-PAH, are obtained for each component, as seen in the chromatograms of Fig. 4. Again, very low detection limits are obtained for these compounds. Because of its inherent selectivity, applications of peroxyoxalate chemiluminescence would be extremely limited were it not for the possibility of derivatization. Fortunately, two of the most common fluorescence derivatization.procedures, dansylation and reaction with fluorescamine, result in amine-containing fluorophores. In the first applications of peroxyoxalate chemiluminescence to HPLC, Imai and co-workers detected dansylamino acids '3 and fluorescamine derivatives of catecholamines 14 and achieved detection limits of 10 and 25 fmol, respectively. A major disadvantage of chemiluminescence detection is the requirement of post-column addition of the necessary reagents such as the oxalate esters, H 2 0 2 and an amine catalyst in the case of peroxyoxalate chemiluminescence detection of fluorophores. The post-column addition pumps must be as pulse free as possible, and the mixing tees should be designed so as to contribute as little as possible to band broadening in the case of chromatographic detection. A new approach to this simplification

sample plugs were injected. Light from a bed reactor packed with particles of the oxalate ester provided the signal. The limit of detection (LOD) was 6 x 10- 9 mole (0.2 ppb). It is unlikely that chemiluminescence detection will become widely used, except for a few analytes, until the required instrumentation is considerably simplified. Now that the advantages of chemiluminescence detection have been firmly established, we can expect additional improvements in methods of reagent addition and/or generation. Also, recent advances in our understanding of the electron exchange nature of the excitation step suggest the possibility of new approaches based on electrogenerated chemiluminescence with elimination of one or more of the reagents.

Phosphorescence Another area in the luminescence field which has intrigued many scientists in the past is phosphorescence. Again, it is one of the phenomena which can be obsen'ed in nature here and there since, contrary to fluorescence, its liftetimes are many orders of magnitude longer. The phosphorescence principle is best shown by using a Jablonski diagram as seen in Fig. 5. The major difference from fluorescence is the long triplet lifetime tp of the triplet state T I' and as a result phosphorescence has usually only been observed in rigid solutions at 77 K. The requirement of liquid N 2 conditions for the observation of phosphorescence appeared to be a serious hindrance for its use as a general analytical tool. However, phosphorescence of rigid glassy samples at 77 K is not applicable as a detection principle in dynamic flow systems. Hence, a real breakthrough was achieved when it was shown that phosphorescence can be observed at room temperature for compounds being absorbed to particular

Pre-column reduction

..... '--2 4

5 Post- column reduction

1 3

10

6

15

20

FiguTr 5. Simplifit'rl diagram showing tht- kint'tirs of molt-cular rxcittd stairs. hl'c. indical'" the frequency of the excitation li!(ht absorbed by the molerul .. : k" i, the ralt" constant of thr intt'mal convtnion prortss. draC'ti\"ating tht hi~h("r ("xci It'd singlrt statr n. k" k. r and Kia( in s - I art Iht ratt constants of thr fluorrscrncf procf'ss, intt'rnal convt'rsion procf'Ss and intr-rsystr-m C'rossin~ procrdure rr!'prcli\'eiy. thus dt'acti\'aling Iht' S, Slalr~ kp and k. p arr thr raIt' Constant!' in !' - I of thf' phosphor("S("rncr proeMS and intrnystrm crossing procf'Ss rf'Spt'ni\'rly. which deacti,"atr the T, Slate: finally ~IQJ is the effective rate ronstan' of 'h, bimolrcular qurn('hin~ rraC'tion of this triplrt Slalt. gi\'t'n by T, +Q ~ 50 + Q+ heal

25 30

Time (min) Fig." 4. HPLC chromatograms of a .ynlhrtir mixturr of six nitro-PAHs with rhemiluminesrrn