Anal. Chem. 1986, 58,2103-2105
2103
Table 11. Selectivity Coefficients Ki, for the Coated Wire Nitrate Ion Selective Electrode CWE interferants
c1-
carbon paste'
PMMAb 0.0398
c10; CH8COOH2P04-
0.035 2.331 0.195 0.00064 0.347 1.628 0.026 0.028
useful pH range
2.5-10.3
3-8.5
1NO2-
so42c10,
0.158 0.00079 1.82
epoxy
PVC'
Orion-ISEb
Orion-ISEc
0.097 3.1 0.29 0.00075
0.063 7.7 0.15 0.012
0.00631
0.006 20 0.06 0.0006
2
>10
2.3
2.7-10.0
0.0631 0.00063
1000
2.5-10.7
2-12
2-12
'Present work. bData from ref 3. 'Data from ref 4. Reference solution, 5 X lo-, M KNO,; interfering solution, 0.09 M. Electrode response in series of pure test solution, 59.4 mV/lO-fold change in concentration. nitrate-CW in which that the plasticizer and graphite powder were not incorporated. Effect of Interfering Anion. The effect of foreign ions on the response of the electrodes was studied by making potentiometric measurements of the solution of 5 X M KNOBcontaining 0.09 M interfering anion of interest. Selectivity coefficients were calculated from the following Eisenman equation:
-200
> E
-2c
-100
c
AE = (slope)
L
0 a
c
v
?
L
-Y W
o
100
1
I 5
3
1
-Log[NO;]
Figure 2. Effect of thickness of polymer membrane on carbon nitrate-CWE: 1.0 mm (0),0.6 mm (O),0.4 mm (e), 0.2 mm (O),0.1
mm (e).
plasticizer was used,the resistivity was less. When epoxy resin and graphite powder were mixed a t ratios of 2:O.l-3, more graphite powder caused less resistivity. When the ratio of added graphite powder to epoxy resin was over 3 to 2, both the resistivity and the potential of the polymer membrane were reduced, because the surface of the polymer membrane was not uniform. As the result, the response of the carbon nitrate-CWE was considerably better compared to the epoxy
log
I
1
+ Ki-
1
where ai and z are the activity and the charge of the interfering anion, respectively. Selectivity coefficients are given in Table 11. Sulfated and chloride did not interfere even a t concentrations 1500 times and 28 times that of nitrate, respectively, and the selectivity of iodide and perchlorate for the carbon nitrate-CWE was improved much more than the selectivity of the other liquid membrane nitrate ion selective electrodes for the above ions. Registry No. Biphenol A, 80-05-7; graphite, 7782-42-5; nitrate, 14797-55-8.
LITERATURE CITED (1) Davies, J. E. W.; Moody, G. J.; Thomas, J . D. R. Analyst (London) 1972, 97. 87-94. (2) Coetzee, C. J.; Freiser, H. Anal. Chem. 1969, 4 1 , 1128-1130, (3) Kneebone, Barbara M.; Freiser, H. Anal. Chem. 1973, 45, 449-452. (4) Suzukl, K.; Wada, H.; Shirai, T.; Yanagisawa, S. Jpn. Anal. 1980, 29, 816-820. (5) Ansaidi. A.; Epstein, S. I. Anal. Chern. 1973, 4 5 , 595-596. (6) Mesaric S.; Dahmen, E. A. M. F. Anal. Cbim. Acta 1973, 6 4 , 431-438.
RECEIVED for review October 1,1985. Accepted April 10,1986. This work was supported by the Korea Research Foundation under grant in 1984.
Laser Two-Photon Excited FluoreScence Detector for Microbore Liquid Chromatography William D. Pfeffer and Edward S. Yeung* Ames Laboratory-USDOE and Department of Chemistry, Iowa State University, Arnes, Iowa 50011 The fluorometric detector in liquid chromatography (LC)
has the special features of high selectivity and high sensitivity compared to the absorption detector and the refractive index detector. With the addition of a laser as the excitation source, impressive results have been reported (1-4). In particular,
the laser allowed better stray light reduction (due to better beam quality), and better focusing into small volumes (due to better spatial coherence). Coupling with microbore LC, where detector volumes less than 1 pL are required, is then relatively straightforward, and mass detedability is enhanced.
0003-2700/86/0358-2 103$01.50/0 0 1986 American Chemical Society
2104
ANALYTICAL CHEMISTRY, VOL. 58, NO. 9, AUGUST 1986
Table I. Relative Efficiences of Various Laser Systems for Two-Photon Excitation av power, W
type
f , 8-l
t, s
Ar ion
1
1
Cu vapor
3 0.6 1 0.06 0.045 1
3 x 10-8 1.2 x 10-10 7 x 10-11 a x 10-13 1.5 X 1 x 10-8 1 x 10-8
mode-locked Ar ion mode-locked YAG, doubled mode-locked dye cavity-dumped dye" Q-switched YAG, doubled excimer-pumped dye
1
1 5000 7.6 x 7.6 x 7.6 x 3.8 X 10 500
107 107 107
lo6
peak power, W 1 2 x 104 66
190 990
7.9 x 103 1 x 107 2 x 105
efficiency 1 6 X lo4 39 190 59 360 1 x 107 z x 105
Pumped by a mode-locked YAG laser. To further increase selectivity in fluorometry, one can use a two-photon excitation scheme ( 5 ) . The simultaneous absorption of two photons of light is guided by different selection rules than normal absorption, so that different electronic states are excited. This provides a separate dimension for selectivity. The process is dependent on the individual polarization conditions of the photons, so polarization selection is possible. The process is enhanced by electronic states matching the energy of one of the photons (6) to provide a unique type of wavelength dependence. Finally, fluorescence from twophoton states back to the ground state are actually forbidden by selection rules (1). There needs to be a third electronic state present, making this a more restrictive process than normal fluorescence. The additional selectivity has already been demonstrated in LC effluents (5). The key to successful application of two-photon excited fluorescence is sensitivity. The two-photon event is much less probable than the one-photon event in general. The probability however increases with photon density. The relationship between the absorbed optical power and the incident optical power is expressed as AP = P2CLA-'6 where AP and P are the absorbed power and the incident power, respectively, L is the path length, C is the concentration, A is the cross-sectional area of the optical beam, and 6 is the two-photon absorption strength. Values typical for 6 are 110-@cm4 s photon-' molecule-'. This is why lasers are essential to providing useful signals. Also, fluorescence rather than absorption is monitored to provide better detection. Unlike conventional fluorometry, where detectability is limited by stray light and Raman/Rayleigh scattering in the solvent, background in two-photon excited fluorometry is almost always negligible. This is because excitation is by visible photons and observation is via UV photons. Even simple filters can adequately reject background contributions. Furthermore, eq 1shows that the signal increases quadratically with incident power while any background signal will increase linearly with incident power. This favors lasers with high peak powers even though CW lasers provide adequate sensitivity (5). Naturally, another important consideration is the average power of the laser, since a low duty cycle implies poor efficiency in producing fluorescence photons. For a laser with average power P (W), pulse duration t (s), and a repetition rate off (s?), one can derive the peak power per pulse as Plft. The two-photon signal is then proportional to the square of this quantity. Over a 1-s period, the total number of two photon events will be proportional to
AP
a
P2/ft
Equation 3 is an indication of the efficiency in generating a signal if the fluorescence yield, path length, beam size, and concentration are identical. In Table I we list the efficiencies of several commercial laser systems relative to a CW laser used in ref 5. In interpreting Table I, one must also consider other
factors, such as beam quality (which allows focusing to small volumes), excessive power density (which leads to dielectric breakdown and other nonlinear processes), and intensity stability (which contributes to base line fluctuations). In this work, we describe a two-photon excitation scheme for fluorometric detection based on a copper vapor laser (LTPEF). The cell geometry allows the use of microbore columns. Substantial improvements in both concentration detectability and mass detectability are achieved.
EXPERIMENTAL SECTION HPLC System. The LC system was assembled from the following: Pump: metering pump, ISCO, Lincoln, NE, pLC-500. Column: Alltech, Deerfield, IL, 5-pm microsphere C18packing, 25 cm long, 1 mm i.d. Injector: Rheodyne, Berkeley, CA, Model 7410 with 1.0-pL injection loop. Eluent: 90/ 10 acetonitrile/H,O. Solvent: The solvent for all solutions was 90/10 acetonitrile/ HzO. Compounds: 2-(4-biphenyl)-5-phenyl-1,2,3-oxadiazole, PBD (Eastman Kodak, Rochester, NY) and 2,5-diphenyl-1,3,4-oxadiazole, PPD (Pfaltz and Bauer, Waterbury, CT), were used. All separations were performed under normal room temperatures at a flow rate of 30 pL/min, and a pump pressure of 580 psi. LTPEF Detector. The laser radiation of a copper vapor laser (Plasma Kinetics, Pleasanton, CA, Model 151 with unstable resonator cavity) leaves the aperture and passes through one Corning 3-71 sharp cutoff filter and one Corning 4-96 wide band-pass fiiter mounted directly on the laser head. The purpose of the Corning 3-71 fiiter is to eliminate secondary laser radiation. The Corning 4-96 fiter is used to block out the orange fluorescence created by the 3-71. No attempt was made to separate the 510/578 nm lines from the laser. The radiation is then focused into the flow cell by a 8.5 cm focal length, 5 cm diameter lens. The flow cell consists of a 1 mm i.d. quartz tube directly attached to the outlet (bottom) of the column. A quartz rod is attached to the inside of the tube to limit the actual cell volume to approximately 0.8 p L (see Figure 1). The purpose of the direct column attachment and quartz rod is to lessen the effect of band spreading associated with HPLC detectors. The fluorescence leaving the flow cell passes through three UV band-pass fiiters: one Corning 7-54, one Corning 7-59, and one Corning 7-51. The fluorescence then enters a photomultiplier tube (Amperex, Hicksville, L.I., NY, Model 56-DVP) operated at 1360 V. Signal Processing. The signal from the photomultiplier tube was processed by a picoammeter (High Speed Picoammeter, Keithley Instruments, Inc., Cleveland, Ohio, Model 417). The A to measure the fluorescence. instrument was set at 1 X The signal was adjusted with the damp control until a reasonably smooth base line was obtained. This corresponds to a time constant of about 3 s. Chromatograms were obtained by sending the processed signal to a strip chart recorder (Fisher Recordall, Houston Instruments, Austin, TX, Series 5000). Detector Optimization. The base of the flow cell was mounted in a x,y,z translational stage that in turn was mounted to an optical bench. By viewing the focused laser radiation, the active region of the flow cell was aligned so that the radiation was focused inside the cell. The system was then fastened securely
ANALYTICAL CHEMISTRY, VOL. 58, NO. 9, AUGUST 1986
FTCS
FLB
nr
PBD
I
W
0
10
TIME Flgure 1. O.&fiL flow cell for two-photon excited fluorescence: CS,
microbore column outlet tube; FLB, focused region of copper vapor laser; QT, quartz tube (1 mm i.d.); QR, quartz rod (0.8 mm 0.d.). so no tension was placed on the microbore column. The entire system was then cloaked with a thick black fabric to eliminate background radiation.
RESULTS AND DISCUSSION An inspection of Table I shows that three laser systems stand out as having substantially hgher conversion efficiencies than the others. A Q-switched Nd:YAG laser system (Quantel, Santa Clrira, CA, Model VG480) was tested with a steady flow of PBD solution (7).Because of the poor mode structure, the beam waist a t the focused region is quite large, and the two-photon efficiency in Table I was not achieved. Furthermore, the existence of hot spots in the laser beam caused damage to the capillary tube after even brief exposures. The slow repetition rate of the YAG laser and the pulse-to-pulse intensity fluctuations ( i 5% ) introduced additional fluctuations in the base line. Improvements in detectability over CW laser excitation ( 5 ) is estimated to be only about a factor of 10. On comparison of the copper vapor laser and the excimer-pumped dye laser, detectabilities are comparable but the former is easier to operate (no alignment or degradation of dye or gas) and has better mode structure if an unstable resonator cavity is used. So, the copper vapor laser is used for subsequent studies. A chromatogram of the test mixture is shown in Figure 2. The injected quantities are 45 pg of PPD and 3.0 pg of PBD. The peaks are well separated and the manufacturer's specification for the column of 100000 plates/m is achieved, showing negligible extra-column band-broadening. We find however that if the quartz rod in Figure 1 is removed, substantial band-broadening occurs, presumably due to turbulent mixing after the optical region. The quartz rod can be replaced by an optical fiber (I) so that the visible fluorescence can be monitored to achieve two-dimensional detection in complex samples. Despite the filters, the background is not at exact zero. So, laser intensity fluctuations (fl%) still contribute to base line noise. The slow drift in base line also corresponds to a change in average laser power during observation. Spikes in the base line correspond to an occasional dust particle crossing the laser beam and increasing the stray light. The PBD peak is higher because of a better match in the absorption wavelength for this laser system. The detectability (SIN = 3) of PBD is estimated from Figure 2 to be 250 fg injected. Over the range of 1 X lo-* M to 1 X lo+?M PBD injected, the signal is found to be linear after corrections for laser power variations. The detectabilities here are more than 4 orders of magnitude better than those in ref 5. Since the excitation wave-
2105
20
30
(MIN)
Flgure 2. Two-photon excited fluorescence chromatogram of test mixture: [PPD] = 2 X lo-' M; [PBD] = 1 X lo-' M; injection loop, 1 pL; flow rate, 30 pL/min; eluent, 90/10 acetonitrilelH,O; column, 5-pM C,', 1 mm i.d., 25 cm long.
lengths are almost identical, the differences reflect real improvements. The use of microbore columns immediately provides a factor of 25 (ratio of inner diameter squared) in mass detectability. The chromatographic efficiency here is substantially better, probably enhancing detection by a factor of 4. So, a gain of 300 can be attributed to the lasers alone. The unstable resonator cavity of the copper vapor laser still does not provide a TEMWmode like the Ar ion laser. A spot size roughly 5 times the diffraction limit can be expected, so the peak powers in Table I are not realized. Thus, only an enhancement of 2.4 X lo3 is theoretically possible. With a less stable base line because of larger intensity fluctuations in the laser here, there is further degradation in SIN. So, the observed improvement of 300 is consistent with the characteristics of the laser systems. With the improved detectabilities here, two-photon excited fluorescence is competitive with most other LC detectors in sensitivity. Normal fluorescence will still be slightly better in ideal cases where emission is far away from solvent Raman bands. The fact that there is so little background from the solvent here implies that solvent purity requirements can be relaxed. The transverse excitation geometry and small optical volume make this detector suitable for even on-column measurements in capillary LC, supercritical fluid chromatography, or capillary zone electrophoresis. The 510-nm and 578nm laser lines effectively excite bands around 255 nm and 289 nm, respectively, allowing coupling with a large group of chromophores. Future work in resonance enhancement and polarization selection can make this an even more powerful technique. Registry No. PBD,852-38-0;Cu, 7440-50-8.
LITERATURE CITED (1) Sepaniak, M. J.; Yeung, E. S. Anal. Chern. 1980, 5 2 , 1465A-1481A. (2) Diebold, G. J.; Zare, R. N. Science 1977, 196, 1439-1441. (3) Folestad, S.; Johnson, L.; Josefsson, B.; Galle, B. Anal. Chem. 1982, 5 4 , 925-929. (4) Hershberger, L. W.; Callls, J. B.; Christian, G. D. Anal. Chem. 1979, 5 1 , 1444-1446. (5) Sepanhk, M. J.; Yeung, E. S. Anal. Chem. 1977, 49, 1554-1556. (6) Huff, P. B.; Tromberg, B. J.; Sepaniak, M. J. Anal. Chern. 1982, 5 4 , 946-950. (7) Hule, C. W.; Yeung, E. S., unpublished results, Iowa State University, 1981.
RECEIVED for review March 10,1986. Accepted April 14,1986. The Ames Laboratory is operated for the U.S. Department of Energy by Iowa State University under Contract No. W-7405-eng-82. This work was supported by the Office of Basic Energy Sciences.
