2007
Anal. Chem. 1985, 57,2007-2009
ACs = (0.02009)(0.504) = 0.0101 cal K-’
Temperature (K) T,
270
250
260
RESULTS AND DISCUSSION
,
I
FREEZING
I I
lOmcal sec-1
FUSION
Figure 3 shows the endothermic and exothermic transitions as recorded by the DSC-2. The endothermic transition width was found to be 8.75 K. This temperature interval multiplied by AC, gives a correction of 0.088 cal, which is subtracted from the measured peak area of 1.691 cal to give the heat of fusion AHF = 1.691 - (8.75)(0.0101) = 1.603 cal ~
1
+
/
,- - - - - - - - - - - -
-3-
I
1 .
TF
+
The calculated exothermic transition energy is within 1% of ;\L the endothermic energy, the agreement being well within the
-‘AT,-;
270
The exothermic transition exhibited a peak heat flow rate of 15.6 mcal s-l, and a transition width of 1.5 K, followed by an exponential decay with a width of 2.2 K. The supercooling interval was 18.5 K, and the measured peak area was 1.413 cal. Thus the heat of crystallization is given by AHc = 1.413 (18.5 1.5)(0.0101) = 1.615 cal
290
280
Temperature (K)
Flgure 3. Fusion and freezing of distilled water.
terval AT,, is included in the correction. When the specific heat correction is necessary, the base line under the peak is constructed by extrapolating the pretransition and posttransition base lines and joining them by a perpendicular line through the point of maximum heat flow.
EXPERIMENTAL SECTION The analysis was performed with a Perkin-Elmer Model DSC-2, using a 0.25 in. diameter, 0.004 in. mica disk between the sample and the sample holder. The sample was 20.09 mg of distilled water, encapsulated in a gold pan. The temperature program rate was 1.25 K min-’ for both heating and cooling studies. Rowas found to have a value of 1340 K s cal-’, and the measured transition temperatures were corrected in the usual manner for thermal lag. The ordinate calibration of the instrument was established on the basis of the fusion of the water sample, corrected as described above. The heat of fusion and the associated change in specific heat were taken as 79.8 cal g-’ and 0.504 cal g-’ K-’, respectively. For this sample, the heat of fusion and the thermal capacity change are given by A& = (0.02009)(79.8) = 1.603 cal
normal experimental error for measurements of this kind. These results confirm that there is no difference between endothermic and exothermic measurements in power-compensated DSC, even when large temperature gradients are generated, provided that the instrument is operated within its linear range. The DSC-2 was chosen to demonstrate this fact; however, similar results would be obtained with the DSC-1B and other power-compensated DSC instrumentation. In conclusion, it is recommended that a DSC, like any other analytical instrument, should be operated within its specified linear range. Users should recognize that an off-scale deflection of a strip-chart recorder indicates saturation and that area measurements made under these conditions are inaccurate. The solution to overloading caused by supercooling of a sample material is either to reduce the sample size or to introduce enough additional thermal resistance to keep the indicated sample heat flow rate on scale.
LITERATURE CITED (1) (2) (3) (4)
Eckhoff, S. R.; Bagley, E. B. Anal. Cbern. 1984, 56, 2868-2870. O’Neill, M. J. Anal. Chem. 1984, 36, 1238-1245. O’Neill, M. J. Anal. Chem. 1875, 4 7 , 630-637. Gray, A. P. “Analytical Calorimetry”; Porter, R. S., Johnson, J. F., Eds.; Plenum Press: New York, 1968; pp 209-218.
M. J. O’Neill Perkin-Elmer Corporation Norwalk, Connecticut 06859-0093
RECEIVED for review March 6,1985. Accepted April 29,1985.
and
High-Sensitivity Laser Fluorometer Sir: It is well-known that laser-induced fluorescence is a very sensitive technique for ultratrace analysis. Various combinations of laser sources and detection systems have been used in the past to achieve increased sensitivity (1-4). Hirschfeld et al. reported on the detection of one molecule of polyethyleneimine bound to y-globulin, tagged with 80-100 fluorescein isothiocyanate molecules (I). This was accomplished by illuminating the sample a t light intensities high enough to produce photochemical bleaching during the observation period, thus producing a short fluorescence pulse which contained the maximum signal the fluorescent tags could produce. More recently, Dovichi et al. achieved a de0003-2700/85/0357-2007$01.50/0
tection limit of 8.9 X M for Rhodamine 6G (R6G) in the liquid phase using a flow cytometer system with a probe volume of 11pL (2). This is the equivalent of 22000 molecules of R6G flowing through the probe volume during a 1-s integration time. In this work, we describe a relatively simple and highly sensitive laser fluorometric system which has been used to achieve a detection limit of approximately 8000 molecules of R6G. The R6G was adsorbed onto the surface of small (10 wm diameter) silica spheres which were viewed individually with a fluorescence microscope. Advantages of this technique are (i) there is no solvent fluorescence or Raman scatter and 0 1985 American Chemlcal Society
2008
ANALYTICAL CHEMISTRY, VOL. 57. NO. 9, AUGUST 1985
1 SPECTRAL FILTERS -SPATIAL
FILTERS
-LENS
OPTICAL FIBER
r
-OBJECTIVE
Flgure 2. SEM of a monolayer of silica spheres, mounted with double-sided tape.
1 -SLIDE
1. schematic diagram of laser excited mkroscwlc fluaaneter.
(ii) the particles are viewed from a stationary position.
EXPERIMENTAL SECTION A Spectra-Physics Model 171-18 CW Ar-ion laser (SpectraPhysics, Mountain View, CA) was used as the excitation source with all-line visible output in the light regulated mode. A laboratory-constructed fiber optic coupler (-70% efficient)was used to direct the output to an optical fiber. The coupler included a solenoid shutter to block the beam from the microscope when the sample was not heing observed. This minimized fading of the fluorescence. The output end of the fiber was positioned to illuminate the area under the obiective of a Nikkon Laboohot microscone (Nippon Kogaku K.K., Tokyo, Japan). This configuration h'Bs advantages over the method of incident illumination where the laser is d u d through the rem of the miamcope, reflected down through the objective by a dichroic mirror, and focused onto the sample. In this method, autofluorescence of and scatter by the optical components can be a problem. Also. the high power density achieved by focusing the laser through the objective may he high enough to vaporize the sample OF interest. In this work, the laser beam does not pass through the microscope and hence these problems are minimized. The laser waq operated to give 126 mW output at the fiber end. At higher output levels. photndemmpositiun and fading of the fluorescence became obvious. The fluorescence was collened by a Nikkon CF 40X objective and directed upward to a phototube attachment which contained a lens to image the fluorrscence onto a l.mm aperture (Figure 1). The microscope focus was adjusted so that the image of a silica sphere would fill the apenure, giving the maximum rexpome at the photodetector. Spectral filtering of the fluorescence was accomplished with a 550 nm, 40 nm fwhm band.paas filter. Corion Corp., and a long wavelength passing S O nm filter, type LC-SO0 Corion Corp. An IP28 photomultiplier (Hamamatqu Corp., Middlesex, NJ) operated at 700 V was used as the detecwr. The signal was amplified hy a Keithley Model 610B electrometer and then fdtered hy a laborawry-ronsuuned MFlO switched capacitnr filter (second order) with a cutoff frequency set at 1 Hz. The output of the filter was fed tn a stripchan recorder (Fisher Model I)5117-5AO). The shca spheres were prepared as 'standards" in the following way. Weighed amounts of the silica spheres (Spherisorb. Phase Separatiuns, Hauuppauge. NY)were placed in test tuhes. To each tube was added a 0.1-mL aliquot of a different R6G solution in ethanol, the solutions ranging in concentration from 10.' M to M. The solvent was allowed U) evaporate, leaving the Fl6G adsorhed w the silica spheres. The test tubes and glassware used in these steps were silanized with dichlorodimethyl8ilane to minimize loss of R6G by adsorption onto the glass surfaces. Rinsing the rest tube walls with small portions of ethanol combined with treatment in an ultrasonic bath was repeated several
-
times to minimize loss of R6G and to ensure uniform coating of the spheres. To estimate the concentration of theae standards in terms of numbers of molecules of R6G per silica head, a 2-mL portion of the silica beads was packed into a test tube, with constant tapping to aid settling of the spheres. Two milliliters of the spheres weighed 1.38 g. The number of spheres id 1.00 g can be estimated by assuming a certain packing arrangement in the 2-mL tube, for instance, cubic packing. The percent free space in a cubic packing arrangement is 47.6% (5); hence 52.4% is volume occupied by the packing material. From the above and the volume of a 10-rm diameter sphere, one can estimate there are 1.5 X los silica spheres/g in a cubic packing arrangement. From the known weights of the silica in each tube, and the known concentration and volumes of R6G added, concentrations were calculated in terms of molecules of RGG/sphere. It should be noted that the assumption of cubic packing is a 'worst case" possibility, Le., this is a very inefficient packing arrangement. It might he more reasonable to assume an arrangement like hexagonal closest packing, in which one sphere would be surrounded by six spheres, all of which lie in the same place and touch each other. By assuming cubic packing, we are probably underestimating the number of spheres that are present in a gram of the material, and hence overestimating the amount of RGG/sphere. In theory, thiswould mean a detection limit for R6G is even lower than we have found here. To prepare a calibration curve, the spheres from a given standard were spread on a quartz slide. The fluorescence emission was measured for each of ten individual spheres to obtain a mean value and a percent relative standard deviation. A halogen lamp in the base of the Nikkon microscope was used to illuminate the field while positioning a sphere under the pinhole to minimize fading of the fluorescence by laser irradiation.
RESULTS AND DISCUSSION For the five standard groups measured, the percent relative standard deviations varied from 18 to 48%. Figure 2 shows an SEM micrograph taken of a monolayer of the spheres clearly showing a large variation in sphere size (manufacturer's literature states the mean diameter as 10 r m &ZOO/,). Generally, spheres which deviated from the mean diameter by more than -15% were not selected for measurement. The mean fluorescence intensities were background corrected (the background was taken as the mean signal from 16 blank spheres). The background consisted of contributions from the solvent-coated blank silica spheres and the quartz slide. The fluorescence intensity was proportional to concentration for the 10' to lo' molecules of RGG/sphere standards used, with a linear correlation ooeffcient (r) of 0.9994,The detection limit (3a) was -8 X lo3 molecules of RGG/sphere. Further, there are several reasons why the detection limit should be lower than what we have found here. As previously mentioned, a worst case estimate was made of the number of spheres/g of the material, giving a high estimate for the
2009
Anal. Chem. 1985, 57,2009-2011
number of R6G molecules/sphere. Also, it was assumed that during evaporation of the solvent from the test tubes containing the silica spheres, all of the R6G was adsorbed to the spheres. Loss of R6G by adsorption to the walls of the glassware would mean there is actually less rhodamine per sphere than is believed, and hence a lower limit of detection, LOD. Despite the difficulties presented in making a more accurate calculation of the LOD, we have shown through conservative estimates that the detection power of this technique is quite good. This particular instrumental setup may have future application for the ultratrace measurement of polycyclic aromatic hydrocarbons on air particulates.
ACKNOWLEDGMENT The authors express their appreciation to Michael Kosinski for the SEM analysis.
LITERATURE CITED (1) Mlyaishi, K.; Kunlake, M.; Imasaka, T.; Ogawa, T.; Ishibashi, N. Anal. Chlm. Acta 1981, 125, 161-164. (2) Bradley, A,; Zare, R. J. Am. Chem. SOC. 1976, 98, 620. (3) Hirshfeld, T. Appl. Opt. 1976, 15, 2965-2966. (4) Dovlchl, N.; Martin, J. C.; Jett, J. H.; Trkula, M.; Keller, R. A. Anal. Chem 1984, 5 6 , 346-354. (5) Masterton, W. L.;Slowinski, E. J. "Chemical Prlnciples"; W. B. Saunders Co.: Philadelphia PA, 1973;p 262.
.
Barbara Kirsch Edward Voigtman James D. Winefordner* Department of Chemistry University of Florida Gainesville, Florida 32611 RECEIVED for review February 21,1985. Accepted May 6,1985. This research was supported by DOE-DE-AS05-780R06022.
Multiple-Use Polymer-Modified Electrodes for Electroanalysis of Metal Ions in Solution Sir: We recently demonstrated (1) the analytical utility of electrodes modified with functionalized polymer films for performing electroanalysis of metal ions in solution. The approach is based on the use of polymer films that bear both an electroactive group as well as a ligand. The former serves to immobilize (or deposit) the polymer onto the electrode surface via electroprecipitation. This is advantageous since it allows for the direct control and determination of the coverage of the polymer on the electrode surface by controlling the deposition conditions and by monitoring the electrochemical response of this electroactive center after deposition, respectively. It also serves in the determination of saturation (I). The ligand itself is chosen so as t o have a high affinity for the metal ion of interest and in addition show high selectivity. The ligand can be incorporated by being part of the polymer backbone or, alternatively, by ion exchange to a polycationic polymer film. We have demonstrated the effectiveness of this approach to the determination of low levels of iron and copper with high sensitivity and selectivity ( I ) . A drawback of this approach is that in many instances the electrodes can only be used for a single determination and that they need to be modified prior to each use. This procedure not only is tedious but also requires careful normalization of the data (with respect to coverage on the electrode surface) prior to comparison. Clearly the development of reagents suitable for multiple determinations is highly desirable. One way to accomplish this would be through the use of metal ligand complexes whose stability constant i s a very strong function of the oxidation state of the metal and whose electrochemical response is metal localized so that the ligand remains intact. This strategy would allow for the use of a single modified electrode in multiple analytical determinations. In essence, since the redox process will be metal localized, after the redox transformation (oxidation or reduction) the metal/ligand complex dissociates, but the immobilized ligand is left unaltered, so that the electrode is ready to be reused. We wish to demonstrate this with the use of 2,g-dimethyl sulfonated bathophenanthroline (also known as sulfonated bathocuproine) for the determination of copper. This reagent has a very high affinity for Cu(1) (2) but the complex dissociates when oxidized to Cu(I1) leaving the ligand intact. This is due to the fact that as Cu(1) the d10 metal center strongly favors a tetrahedral geometry. Upon oxidation to Cu(II),
however, a square planar geometry is preferred by the now d9 metal center. However, due to the steric constraints imposed by the 2,g-dimethyl substituents, such a geometry cannot be accommodated and the complex dissociates ( 2 , 3 ) .
EXPERIMENTAL SECTION 1. Reagents. The synthesis of the quaternized vinylpyridine-vinylferrocene copolymer ( I ) and the electropolymeri2f (v-bpy is 4-vinyl-4'-methyl-2,2'-bipyridine) zation of [R~(v-bpy)~] ( 4 ) were as previously described. Homogeneous polymerization of [ R u ( ~ - b p y ) ~was ] ~ effected + in acetonitrile by free radical initiation using AIBN (azobis(isobutyronitri1e)). Sulfonated bathocuproine was obtained from G.F. Smith and was used as received. Tetra-n-butylammonium perchlorate (TBAP) (G. F. Smith) was recrystallized three times from ethyl acetate, dried in vacuo at 90 " C for 72 h and stored in a desiccator. Sodium perchlorate (G.F. Smith) was used as received. Acetonitrile, dimethylformamide (DMJ?),and methylene chloride (Burdick and Jackson "distilled in glass") were dried over 3-a molecular sieves. Water was purified by passing through a Hydro Systems unit. All other reagents were of reagent grade quality and were used without further purification. Electrochemical experiments were performed with a Princeton Applied Research Model 173 potentiostat with a Princeton Applied Research Model 175 universal programmer or an IBM EC-225 voltammetric analyzer. Data were recorded on a Soltec Model 6423 or a Hewlett-Packard Model 7045-B X-Y recorder. Platinum disk electrodes (of area ranging from 0.01 to 0.03 cm2) were used and these were polished with 1-pm diamond paste (Buehler) prior to use. Electrochemical cells were of conventional design. Modification of the electrodes with the quaternized vinylpyridine vinylferrocene copolymer (q-vp/v-fc) was performed by electroprecipitationfrom methylene chloride/TBAP solutions as previously described ( I ) . The coverage of polymer on the surface was controlled so as to be in the (1-5) X 10-lo mol/cm2 range. Modification with poly[Ru(~-bpy)~]~+ was effected by immersion of a polished electrode into an acetonitrile solution of the polymer (5 mg/25 mL) for 1 min after which the electrode was air-dried and rinsed with acetone and water. This procedure gives a coverage of adsorbed polymer on the order of (0.3-3) X mol/cm2 as determined by integration of the cyclic voltammetric wave for the oxidation of the [ R u ( ~ - b p y ) ~centers ] ~ + in the polymer. The sulfonated bathocuproine was incorporated by ion exchange by contacting the electrode modified with either the (qvp/v-fc) or p~ly[Ru(v-bpy)~]'+polymers (both of which are
0003-2700/85/0357-2009$01.50/00 1985 American Chemical Society
