Anal. Chem. 1982,5 4 , 925-929
Laser Induced Fluorescence Detection for Conventional and Microcolumn Liquid Chromatography Staffan FoYestad, Lars Johnson, and Blorn Josefsson* Department of Analytical and Marine Chemistry, Chalmers University of Technology, S-4 12 96 Goteborg, Sweden
Bo Galle Department of Physics, Chalmers Universlv of Technology, and Swedish Water and Air Pollution Research Institute, Box 5207, S-402 24 Goteborg, Sweden
The cell voilume of detectors used In HP1.C should be kept small to mlnlmlze postcolumn band broadenlng. The offect of band dlsperslon and dllutlon in flow-through cells Is of extreme Importance In connectlon wlth mlnlaturlzed systems. The effluent from dlfferent kinds of columns was studied by a laser Induced fluorescence detectlon technique wlth respect to postcolumn band broadening, sensltlvlty, and utlllty. The effluent from conventional HPLC columns was arranged as a free-falling thin jet. Mlcrocolumn effluents were passed through a quartz caplllary to avold drop formatlon. The radlatlon from a krypton Ion laser in the UV mode for excltatlon was focused onto the lamlnar column effluent. The fluorescent light produced was (collected by a lens In a dlrectlon optlmlzed wlth respect to the influence of elastic scattering. By different optical adjustments, the deadi volume could be mlnlmlzed to practically zero (1 nL). The mlnlmum detection Ilmlt was 20 fg for fluoranthene. Reversed-phase HPLC was studied with dlfferent solvent comblnatlons and gradients.
Extracolumn band broadening in detectors has recieved attention since high-performance liquid chromatography (HPLC) was3 introduced. As columns and injectors have been improved, the requirement on detector performance has increased to minimize peak broadening by dispersion. In this respect the cell volume as well as connections play an important role. The whole detector arrangement may influence the flow prolperties, which in turn affect the band spreading. There is a growing interest in the use of miniaturized I-IPLC systems basted on packed rnicrobore columns or open tubular columns. These chromatlographic systems exhibit high resolving power together with a low flow rate of the mobile phase. The microbore packed column technique was introduced by Scott and Kucera (1,2)who pointed out the importance of reducing the detector cell volume to utilize the separation efficiency of the column. Ishii et al. (3, 4 ) constructed a rniniatiirized HPLC system with open tubular columns, which have essentially smaller inner diameters of about 50 pm. With this technique the detector cell requires still smaller volumes or in the range of 0.1-1 pL. Knox and Gilbert (5) calculated that the effective detection volumes should be in the order of 1-10 nL before any hope exists of operating capillary HPLC systems under optimal conditions. They also stated that the practical limitation on capillary HPLC arises from the dispersion caused by the detector. When different small volume detectors are considered, the fluorescence technique is promising because of its high sensitivity. HPLC fluorescence detectors utilizing laser excitation radiation have recently been introduced (6-8). Laser fluorimetry has some important characteristics, which may be advantageous with small volume detectors. The produced fluorescence radiation is directly proportional to the intense
laser excitation light. The good collimation of the laser beam gives a unique possibility to focus its high intensity onto a very small area. In addition the monochromaticity of the laser makes it easy to suppress scattered light from Rayleigh and Raman processes as well as reflections without sacrificing sensitivity. The amplitude stability of the laser also helps to1 minimize the background noise level. Diebold and Zare (6) presented a windowless flow cell where the HPLC effluents flow from a steel capillary tube down t o a rod 2 mm below forming a droplet bridge of 4 p L volume. Focusing the laser beam in a small spot inside the droplet facilitates the rejection of the elastically scattered light. With the aid of a pulsed HeCd laser and gated detection electronics, they achieved very high sensitivity when determining aflatoxins. Hershberger et al. (7) designed an HPLC cell based on a sheath flow principle, where the effluent is injected in the center of an ensheathing solvent stream under laminar flow conditions. The laser beam illuminated cell volume is very small or in the range of 6-150 nL. Since the windows are not in contact with the sample flow, the stray light from the windows is reduced. Sepaniak and Yeung (8)used a quartz capillary tube where the HPLC effluent moved upward. A laser beam was focused in the effluent underneath an optical fiber. The emission was collected through the fiber perpendicular to the laser beam. The construction leads to a minimal influence of scattered and fluorescent light from the capillary tube walls. The limiting factors of fluorescence detectability are to a great extent connected with the background fluorescence emanating from the cell, optical components, solvent, and sample contaminants a well as elastically scattered light. The reduction of the background emission is highly dependent on the detector cell design. The objective of this work was to design a laser-based detector ideal for use with conventional HPLC columns as well as microbore and capillary columns. Special attention was paid to the effective cell volume with regard to postcolumn effects by comparing the efficiency of the systems.
EXPERIMENTAL SECTION Conventional HPLC System. An LDC chromatographic system consisting of two Constametric I11 pumps and a Gradient Master was used. Injections were made with a Valco injector with a 10-pL loop. Separations were performed on a 150 mm X 4.6 mm column packed with Nucleosil 7.5-pm silica and on a 200 mm X 4.6 mm Nucleosil 5-~rn RP18 column. The columns were mounted in a rack and pinion arrangement which allowed vertical adjustment. Hexane, methanol, and dichloromethane solvents from Rathburn Chemicals (HPLC grade) and double distilled water were used in different mobile phase compositions. For comparative studies a FS 970 LC fluorimeter from Schoeffel Instruments was used, which was equipped with a Corning 7-54 primary fdter and a 418nm cutoff secondary filter. The excitation wavelength was set to 360 nm. The time constant 0.5 s was used. Microbore HPLC System. The low flow rates were maintained with a Waters M-6000 A pump. The frequency generator
0003-2700/82/0354-0925$01.25/00 1982 American Chemical Soclety
ANALYTICAL CMMISTRY. VOL. 54. NO. 6. MAY 1982
npur 1. SdmmIk &gam
01 daecta cd arrangements: (a) hee laiiing !st. (b) mlcrobae column quartz capillary: (A) laser focusing
pdnt: (8)symgeneedeSGE.17mm.0.12mmi.d..0.5mmo.d.; (C) cyanoacrylate abslve: (0)SS tubing 'I,. in.. 15 mm. 0.2mm 1.d.; (E) fused silica capillary. 80 mm. 0.20m m 1.d.. 0.30mm 0.d.; (F) SS tubing 'I,@ in. 17 rnrn. 0.35 mm I d
of the pump was disconnected and replaced by a Tektronix FG 502 function generator. Samplea were injected with a Valco air actuated valve having a fixed sample volume of 0.5 #L. Columna in. o.d., 1 nun were constructed from 10 and 20 em lengtha of 'la i.d. SS tubing. Nucleoail7.5-pm silica particles were suspended in methanol and pumped into the columns a t a pressure of600 bar using a Haskel DST 150 air-driven fluid pump. Acetone, dichloromethane. and the appropriate eluent composition were used for conditioning purposes. Free Falling Jet Laser Deteetor Cell. The effluents from the conventional HPLC column were arranged as a falling thin jet. The jet stream was produced hy a small bore capillary which is shown in Figure 1. The capillary consist of a short piece of an SGE syringe needle. 0.5 mm 0.d. and 0.12 mm i.d., inserted 5 mm in a '/le in. SS tubing 0.2 mm i.d. and drilled out to fit the syringe needle. The needle is fixed with cyanoacrylate adhesive. The capillary was then connected to the column end. The construction gives a very fine jet of the column effluent a t flow r a m down to approximately 1 mL/min (Figure 2). In a similar arrangement a syringe needle 32 mm in length and 0.12 mm i.d. was used as a jet producing capillary. This tube was connected directly to the column end. Quartz Capillary Tube h r DetectorCell. An alternative detector design to the falling jet wan required a t flow ratea below 1 mL/min. Different quartz capillary tubes were tested for flow rates typical of miniaturized HPLC systems. It was found that fused silica capillaries 0.05,0.1, and 0.2 mm i.d. from HewlettPackard and SGE commonly used as GC/GC columns yielded the lowest fluorescenee background. The capillary was mounted to the column end inside a steel capillary tube as shown in Figure 1. Five millimeters of the polyimide coating on the fused silica capillary was carefully burnt off just below the end of the SS tubing using a gas flame. The uncovered area was then washed with dichloromethane. The laser beam was focused on the unmvered part of the capillary. The capillary tube allows any flow rate commonly used in conventional as well as microbore HPLC. Optical Arrangements. Figure 3 presents a schematic new of the optical system. As excitation source the monochromatic output from a Coherent Radiation CR 3000 K krypton ion CW laser operating in the UV-mode was used. This laser emits radiation a t 351 and 356 nm with a total power of 1-2 W. After the bluish fluorescence light emanating from the laser cavity was suppressed, by means of a band-pass color filter UG1, the laser beam was focused with a quartz lens onto the detector cell underneath the column. Lenses with a focal length of 15 and 60 em were used. In order to avoid the strong scattered light in the horizontal plane emanating from the cylindrical surface of the vertically mounted deteetor cell, detedion was achieved a t an angle of approximately 30" to this plane, After reflection in a front surface mirror, the fluorescence light ia collected and focused onto the entrance slit of a Jobin Yvon HL monochromator, using a quartz lens (f = 15 cm. 0 = 10 cm). Depending on the nature of the experiment, the entrance slit was varied between 1and 6
Flpm 2. closwp picture of the free-fallingjet call.
O o " . r l l lens L
Flgwe 3. Schematic dlagam 01 the experlmenlal lnsrmnmnlatiOn
arrangement. mm. while the exit slit was fixed a t 17 mm, yielding a band-pass of 28 nm. For convenience, the monochromator was centered a t 450 nm during all the experiments. A compromise was made between the maximum fluorescence yield and rejection of the Raman peak of water and hexane as well as elastically scattered light. The elasticity scattered light was further suppressed with a cutoff filter WG 385 or GG 420 before reaching the photcmultiplier tube (EM 9558 QB). The output of the PMT was then,
ANALYTICAL CHEMISTRY. VOL. 54. NO. 6. MAY 1982
Flgun 1. The r e ! a h lluoresaxlceMSnSnyvs. Umdstance bmween Um lens and Um call m l i z e d to 1 a1 me focal length of me lens:
Fbwa 0. StabW of falling Jetwth hexane as Um mobb phase. Drop fonnatlon occus in Um unsheded area. The nOw rate is pbned against me distance between me column emuent exn and me laser
column. Nucleosll 7.5 rm: mobile phase. hexane mntainlng fluxan.
mene: fiow rate. 2 wmh: ( 0 )heming )et can. ( 0 )q
50 r m i.d.
Retention time (min) *re
5. Picture showing the falling jet horn me HPLC column l i b mhaled by a laser beam. The pcducs r e k t e d i5hI in Um hairomal plane is projected at the background as an intense sharp band.
after amplification, fed to a strip chart recorder. AU experiments were performed under darkroom conditions. Procedure. Investigationswere carried out to find the maximum signal response by varying the focusing point of the laser beam vis-a-vis the falling jet. Different distances between the lens and the jet were tested at which the foeus was locsted in front of, inside, and behind the jet. The results are shown in Figure 4. The experiment was facilitated by pumping a hexane solution aontaining a constant concentration of fluoranthene as a marker. The maximum response was found when the laser beam was focused ahead the falling jet. When the laser beam was scanned horizontally. the maximum response was achieved with the laser beam in the center of the jet. At the maximum response an intense disk of reflected l i h t appeared in the plane perpendicular to the jet stream; see Figure 5. The optical properties of the jet are dependent on, e.g., flow rates. solvent properties, and the diameter of the capillary. With a capillary diameter of 0.12 mm. different flow rates were run to establish the minimum flow without turbulence and droplet formation. As shown in Figure 6 hexane could be uaed as mobile phase down to a flow of 1.2 mL/min. For gradient runs (methanolwater 0 to 100% in 10 min). an unaffected base line was achieved at a flow rate of 1.5 mL/min. Droplet formation appeared at a flow rate of 1.0 mL/min when the water content was approximately 55%. An example of a gradient run is shown in Figure 7. A brief test of the dependence of the signal upon vmying the power of the laser resulted in no saturation being observed.
Fbum 7. Separaikm of a standard mlxhre of 11 (dlmemylamlno~
naphthalenesuifonyhyhaes: wlumn. 200 X 4.6 mm. NwtaoSII RP-18 5 tun: mobile phase. meUmr0l:watSr. linear gadient 50-100% in 12 min: Row rate. 1.3 mLlmin. 10 r L injected, 1 = (dimethylaminobaphmalenesu~oyhydrazine,2 = fwmaldehyde. 3 = acetaldehyde. 4 = acetone. 5 = propanal. 6 = butanal. 7 = pentanal. 8 = benzaldehyde. 9 = hexanal. 10 = hepianal. 11 = octanai. 12 =
nonanai. The laser beam was focused directly on the quartz capillary in the enme way 88 with the f a U i jet. The background i n c r d 1order of magnitude. However, the signal a h increaned resulting in signal to noise ratio in the w e range. The increased signal may be a result of a more favorable geometry for multiple reflections inside the cell. Thus, in practice, the cell volume is larger than 1 nL. The PMT viewing of the emission radiation area was screened off by an aperture to decrease the detection cell. RESULTS AND DISCUSSION The Minimum Detection Limit. Under the conditions used for the falling jet, the minimum detection limit was established. Figure 8 shows a chromatogram of an injected g in 10 sample containing 20 fg of fluoranthene (20 x r L of hexane). The capacity factor k'was 3.2 and the signal to noise ratio was 10. T h e linear velocity through the detector was about 300 cm/s. The laser excitation volume is about 1nL which means that very few molecules are present in the measured volume. Since the fluorescence lifetimes are trpicauy in the nanosecond range the molecules will be excited several times while passing the illuminated detector zone. Thereby different mobile phase flow rates will result in changed detector sensitivity. When
ANALYTICAL CHEMISTRY, VOL. 54, NO. 6, MAY 1982
Flgure 9. Plot of log h vs. log v for conventional HPLC with different detectors: column, 150 X 4.6 mm, Nucleosil 7.5 pm; mobile phase, hexane:dichloromethane (935); solute, anthracene; k’ = 1. The exB / v Cv by using perimental data were curve fitted to h = A Y~~~ a modified Marquardt method ( 7 7 (3) conventional detector, (A)laser quartz capillary cell, (0)laser falling jet.
Retention t i m e (min)
Flgure 8. The minimum detectable quantity of fluoranthene: column,
Table I. Band Broadening in the-Conventional HPLC System Using Different Detectorsa
150 X 4.6 mm, Nucleosll, 7.5 pm; mobile phase, hexane; flow rate, 2 mLlmln; injected volume, 10 pL; 1 = anthracene, 2 = fluoranthene (20 fgl.
lower flow rates typical for microcolumn systems are used, the limit of detection will be improved. Applications. There are a limited number of compounds which emit fluorescent light. Furthermore, they have to absorb a t the laser wavelengths available. Partly these restrictions may be overcome by using the fluorescence labeling technique (9, IO). A chromatogram of 11 (dimethylamino)naphthalenesulfonylhydrazone derivatives of carbonyl compounds is shown in Figure 7. The laser/free-fallingjet detector was used to illustrate a gradient elution application. Comparison of Detectors Coupled to a Conventional Column. A conventional fluorescence detector as well as the two different laser detectors were coupled to a conventional column for purposes of comparison. The column system and the separation conditions were the same in the three experiments. Thus the cell arrangements could be compared regarding band broadening on the detector side. The HETP curves for the three cases are shown in Figure 9. The respective detector dead volumes were Schoeffel detector approximately 5 pL, the falling jet 0.6 pL, and the quartz capillary 1.0 pL. The illuminated volumes with the two laser detectors were about 1 nL. Thus the differences in the HETP curves are mainly derived from the connections when the laser is used. In the case of the conventional detector, the cell performance yields a contribution to band broadening also. In addition, the detector filtering electronics (time constant 0.5 s) contributes to peak distortion. An attempt was made to estimate the dispersion in the capillary connections. Laminar flows were prevailing since the Reynolds numbers did not exceed 300 with the various laser detector cells used. As a rule of thumb from hydrodynamics, the parabolic velocity profile is developed in tubes with a length ( L ) to diameter ratio of 20-40 or greater and Poiseuille flow prevails. The band broadening as volume standard deviation may be calculated from the formula by Taylor (12)
where D, is the molecular diffusion coefficient, e.g., 3 X
conv detector laser quartz cell laser falling jet cell a
2674 1875 1597
difference 799 279
Column 200 x 4.6 mm, flow rate 1.5 mL/min, and
k ’ = 1.1.
cmz/s for benzene in heptane from ref 13, and u is the mobile phase flow rate. The equation is valid under conditions of laminar flow and a residence time of t >> (1/3.8)2(r2/D,,J. However, the applied flow rates in the range of 1-3 mL/min results in a negligible diffusion contribution. Thus the dispersion is mainly caused by connector effects. From the HETP curves in Figure 9 it is possible to calculate the total dispersions caused by the three different detectors which are presented in Table I. The laser falling jet cell yielded the smallest band broadening from the connections. Comparison of Different Detectors Coupled to a Microbore Column. The quartz capillary detector cell was tested with a microbore column. A comparison under the identical conditions was made with a conventional HPLC fluorescence detector with a relatively small detector cell dead volume (5 pL). The HETP curve is shown in Figure 10. In the experiment the pump flow rates were in the range of 20-400 pL/min. At a mobile phase flow rate of 18.5 pL/min, typical for microbore columns, and an inside diameter of 0.1 mm of the quartz capillary, the volume variance:6 = 0.008 pL2 which should be compared with the total:6 = 32 pL2 a t k’= 2.2 for a 10-cmcolumn. In this case the formula by Taylor i s valid. In a recently published review Yeung and Sepaniak (14) discussed the potential of using laser fluorimetric detection in HPLC. The previously reported laser-based detectors are characterized by complicated designs, e.g., to suppress stray light. Furthermore the detectors are not designed to be used with miniaturized system or with a gradient elution mode. The only detector that may be used with gradient operation is the cell based on an optical fiber inside a quartz tube cell (20-pL cell volume). The free falling jet principle is a straightforward approach and it has been applied in turbidimeters to measure scattered light from particles. The jet is characterized by stability and a well-defined smooth surface also during gradient elution,
ANALYTICAL CHEMISTRY, VOL. 54, NO. 6, MAY 1982
D - 7
Quartz capillary column
Flgure 10. Plot of log h vs. log v for microbore column with different detectors: column, 400 X 1.0 mm, Nucleosll 7.5 pm; mobile phase, hexane:dichloromethane (965);solute, anthracene, k ' = 1.4; (M) conventlonal detector, (A)laser quartz caplllaiy cell.
with flow rates typical of conventional HPLC columns. The background is easy to suppress with the collecting optics out of the perpendicular plane to the jet stream. Actually a 30' angle results in a background signal decrease of 6 orders of magnitude. This arrangement yields the highest sensitivity or a detection capacity of the femtogram level. When laser excitation with its well-defined beam is used, it is far easier to manipulate the optics than to complicate the construction of the flow cell. The use of quartz capillaries as flow cells in fluorescence is very common. The mlost critical factor regarding quartz materials iei the background fluorescence. Besides, the optical properties connected with the evenness of the quartz surface play an important role. Compared to the falling jet cell, the quartz capillary flow cell has the very important advantage of allowing very low flow rates adequate for microbore or capillary column separations. As mentioned in the introduction, one main advantage of laser excitation is the possibility to irradiate small detector volumes. Therefore laser-based detectors inay be successfully used with miniaturized systems. Hershberger et al. (7) reported the possibility of using a detector volume down to 1 nL. This ie true for the laser irradiated volume in their detector, but by no means for the whole detector volume including the connections. Miniaturized FIPLC systems are characterized by narrow peaks. The peaks should1 not be dispersed anywhere in the detector system. Conventional detectors with dead volumes of 5 1 L cannot be used without an additional makeup liquid to sweep the detector ( I 3 , 1 5 ) . The decrease of sensitivity by dilution effects may be partly compensated by laser intensity; however, this does not coinstitute an adequate solution since the solvent background will also increase. Extracolumn dispersion has been discussed by Martin et al. (16), who proposed to use very short connection tubes and small cell detectors. Kirkland et al. (17)coupled the detector cell directly to the injector to study the cell variances. This instrumental setup gave very rapidly emerging peaks, which require a short detector response time. Reese and Scott (18) have tested extracolumn dispersion in commercial equipments without separation columns. The central point in achieving detectors with minimal or almost no dead volume is to take into account the connecting tubings. The results in this report show the advantage of transferring the solutes to the detection area in a short small bore capillary. Poiseuille flow in this connection will give a negligible dispersion. The smallest quartz capillary tested with an inside diameter of 0.05 mm corresponds to a total detector volume of 65 nL. When quartz capillary separation columns
Figure 11. Schematic view of laser induced fluorescence detectian directly on an HPLC quartz capillary column.
are used, the laser beam could ideally be located directly on the column, as illustrated in Figure 11. In an experiment a quartz capillary, 0.2 mm i.d., was illuminated at two different spots spaced 4 cm apart. The observed dispersion was about the same regardless of the distance from the separation column outlet. However, since the flow rates were high, 1-2 mL/min, the dispersion effects were difficult to estimate. The total dispersion at the detector side (see Table I) is probably mainly created by end effects of the connecting tubings. These effects are very difficult to estimate by callculation. In the case of the falling jet detector, the laminar flow is maintained after leaving the capillary. The pressure drop and the change of shear forces create a jet with a lower flow rate, resulting in an expanded diameter from 0.12 to 0.3 mm. The parabolic flow profile from the capillary is maintained but it is probably deformed by the expansion of the stream. Turbulence, caused by, e.g., air diffusion, will not occur until after some distance (see Figure 2).
ACKNOWLEDGMENT The authors are grateful to Sune Svanberg and Morgain Gustavsson for kind support of this work and to R. P. W. Scott for stimulating discussions. Thanks also to Derek Biddle for comments on the manuscript. LITERATURE CITED (1) Scott, R. P. W.; Kucera, P. J . Chromatogr. 1976, 125, 251-263. (2) Scott, R. P. W.; Kucera, P. J . Chromatogr. 1979, 169, 51-72. (3) Ishli, D.; Asai, K.; Hibl, K.; Jonokuchl, T.; Nagaya, M. J . Chromatogr . 1977, 144, 157-168. (4) Ishii, D.; Takeuchl, T. J . Chromatogr. Sci. 1980, 18, 462-472. (5) Knox, J. H.; Gilbert, M. T. J. Chromatogr. 1979, 186, 405-409. (6) Diebold, G. J.; Zare, R. N. Science 1977, 196, 1439-1441. (7) Hershberger, 1.. W.; Caiiis, J. B.; Christian, G. D. Anal. Chem. 19791, 51, 1444-1446. (8) Sepaniak, M. ,I.; Yeung, E. S. J . Chromatogr. 1980, 190, 377-3831. (9) Johnson, L.; Josefsson, B.; Marstorp, P.; Ekiund, G. I n t . J. fnviron. Anal. Chem. '1980, 9 , 7-26. (10) Blau, K.; King, G. "Handbook of Derivatives for Chromatography' ; Heyden and Smith: London, 1977. (1 1) Nash, J. E. "Compact Numerical Methods for Computers: Linear Aigebra and Function Minimlsation"; Adam Hiiger: Bristoi, 1979. (12) Taylor, G. Proc. R. SOC.London, Ser. A 1953, 219A, 166-203. (13) Tsuda, T.; Novotny, M. Anal. Chem. 1978, 50, 271-275. (14) Yeung, E. S.;Sepaniak, M. J. Anal. Chem. 1980, 52, 1465 A-1481 A. (15) Krejci, M.; Tesarik, T.; PaJurek,J. J. Chromatogr. 1980, 191, 17-24. (16) Martin. M.; Eon, C.; Gulochon, G. J. Chromatogr. 1975, 108, 229-241. (17) Kirkland, J. J.; Yau, W. W.; Stoklosa, H. J.; Dilks, C. H., Jr. J. Chromarogr. Sci. 1977, 15, 303-316. (18) Reese, C. E.; Scott, R . P. W. J . Chromatogr. Sci. 1980, 18, 479-486.
RECEIVED for review November 18,1981. Accepted January 25, 1982. This paper was presented in part in the Forth International Symposium on Capillary Chromatography, Hindelang/Allgau, FRG, May 1981. The work was supported by the National Science Research Council. We are also indebted to Waters for providing a pump.