Equilibration-Based Preconcentrating Minicolumn Sensors for Trace

Equilibration-Based Preconcentrating Minicolumn Sensors for Trace Level Monitoring of Radionuclides and Metal Ions in Water without Consumable Reagent...
4 downloads 0 Views 701KB Size
Download PDF
Anal. Chem. 2006, 78, 5480-5490

Equilibration-Based Preconcentrating Minicolumn Sensors for Trace Level Monitoring of Radionuclides and Metal Ions in Water without Consumable Reagents Oleg B. Egorov, Matthew J. O’Hara, and Jay W. Grate*

Pacific Northwest National Laboratory, P.O. Box 999, Richland, Washington 99352

A sensor technique is described that captures analyte species on a preconcentrating minicolumn containing a selective solid-phase sorbent. In this approach, the sample is pumped through the column until the sorbent phase is fully equilibrated with the sample concentration, and the exit concentration equals the inlet concentration. Oncolumn detection of the captured analytes using radiometric and spectroscopic methods is demonstrated. In trace level detection applications, this sensor provides a steady-state signal that is proportional to sample analyte concentration and is reversible. The method is demonstrated for the detection of Tc-99 using anion-exchange beads mixed with scintillating beads and light detection, Sr-90 using SuperLig 620 beads mixed with scintillating beads and light detection; and hexavalent chromium detection using anion-exchange beads with spectroscopic detection. Theory has been developed to describe the signal at equilibration and to describe analyte uptake as a function of volume and concentration, using parameters and concepts from frontal chromatography. It is shown that experimental sensor behavior closely matches theoretical predictions and that effective sensors can be prepared using low plate number columns. This sensor modality has many desirable characteristics for in situ sensors for trace level contaminant long-term monitoring where the use of consumable reagents for sensor regeneration would be undesirable. Initial experiments in groundwater matrixes demonstrated the detection of Tc99 at drinking water level standards (activity of 0.033 Bq/ mL) and detection of hexavalent chromium to levels below drinking water standards of 50 ppb. The development of in situ sensors for long-term environmental monitoring remains a challenge. Such sensors must be selective, be sufficiently sensitive to meet difficult detection limit requirements, and will ideally operate with no consumable reagents. To achieve the desired detection limit, some form of preconcentration may be necessary. These requirements are particularly true for monitoring radionuclides in groundwater, where the detection limits must be extremely low.1 * To whom correspondence should be addressed. E-mail: [email protected].

5480 Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

It is anticipated that many radiochemically contaminated sites will require long-term stewardship. These sites were contaminated during the production of nuclear weapons materials, resulting in large groundwater plumes that are very difficult to remediate.2 To be credible, long-term stewardship must include monitoring for radionuclides in groundwater. However, current procedures involving pumping, sampling, and laboratory analysis are quite costly. The development of in situ sensors could significantly reduce costs and enable more effective monitoring in long-term stewardship and other monitoring applications (e.g., monitoring barrier performance and remediation progress). We previously set out needs for radionuclide sensors for water monitoring and noted that radiometric detection methods are required in most cases in order to meet the trace detection limit requirements.1,3 Radiometric detection is significantly more sensitive than any conceivable chemical detection methods for many radionuclides, and it distinguishes between radioactive isotopes of interest and stable natural isotopes in the background. Direct radiometric detection of R- and β-emitters in solutions, however, is problematic because the radiation has short ranges in condensed media. Therefore, these radionuclides must be collected, concentrated, and localized in proximity to the detector in order to make the radiometric measurement. Since R- and β-radiation in solution does not provide distinctive energy spectra, the collection and concentration method must be selective for the radionuclide of interest over other potentially interfering radionuclides. We have previously described a preconcentrating minicolumn sensor for detection of radioactive pertechnetate in groundwater.1 The design of this sensor focused on using separation chemistry to selectively capture and localize the radionuclide of interest in a minicolumn for detection by radiometric means. In our original design, separation chemistry and scintillating fluors were combined in polymeric beads that were packed into the minicolumn, which was sandwiched between two photomultiplier tubes, as shown in Figure 1. These beads constituted a selective scintillating (1) Egorov, O. B.; Fiskum, S. K.; O’Hara, M. J.; Grate, J. W. Anal. Chem. 1999, 71, 5420-5429. (2) Hartman, M. J., Dresel, P. E., Eds. Hanford Site Groundwater Monitoring for Fiscal Year 1997; Pacific Northwest National Laboratory: Richland, WA, 1998; PNNL-11793 UC-402, pp 403, 702. (3) Grate, J. W.; Egorov, O. B.; O’Hara, M. J. ACS Symp. Ser. 2004, No. 904, 322-341. 10.1021/ac060355m CCC: $33.50

© 2006 American Chemical Society Published on Web 06/27/2006

Figure 1. Schematic diagram of the radiometric sensor column configuration where the transparent column is placed between two photomultiplier tubes (PMT) for light collection.

material with dual functionality. Unretained radionuclides produced transient signals as they passed through the column in proximity to the scintillating fluors, while retained radionuclides accumulated in the column to provide a steady signal. This approach was based on quantitative capture of the radionuclide of interest and required either regeneration of the sensor material with reagents or renewal of the sensor material by manual or fluidic means. This sensing approach arose from recent developments in the automation of radiochemical analysis, where the radionuclides were collected and separated from the sample on a separation column, followed by downstream detection by scintillation counting.4-10 The sensor configuration combined the separation and detection steps in a single functional unit. The combination of a separation chemistry and scintillation in polymeric beads was first demonstrated several years ago using ion-exchange beads; however, collection and counting were done in separate manual steps.11 The concept has now been exploited in a number of reports for radionuclide sensors and analyzers.1,12-16 While the previous preconcentrating minicolumn sensor for pertechnetate1 represented a milestone in the development of radionuclide sensors for water monitoring, and succeeded at meeting stringent detection limit requirements, the use of reagents to regenerate the sensor column remains a drawback for in situ monitoring applications. In this paper, we describe a new sensing modality for the preconcentrating minicolumn sensors that we call “equilibration-based” sensing.3,10,17,18 A minicolumn sensor based on quantitative capture relies on avoiding breakthrough conditions, and the signal is proportional (4) Grate, J. W.; Strebin, R. S.; Janata, J.; Egorov, O.; Ruzicka, J. Anal. Chem. 1996, 68, 333-340. (5) Grate, J. W.; Fadeff, S. K.; Egorov, O. Analyst 1999, 124, 203-210. (6) Egorov, O.; O’Hara, M. J.; Grate, J. W.; Ruzicka, J. Anal. Chem. 1999, 71, 345-352. (7) Grate, J. W.; Egorov, O. B.; Fiskum, S. K. Analyst 1999, 124, 1143-1150. (8) Grate, J. W.; Egorov, O. B. Anal. Chem. 1998, 70, 779A-788A. (9) Egorov, O. B.; O’Hara, M. J.; Ruzicka, J.; Grate, J. W. Anal. Chem. 1998, 70, 977-984. (10) Grate, J. W.; Egorov, O. B. In Handbook of Radioactivity Analysis, 2nd ed.; L’Annunziata, M. F., Ed.; Academic Press: Boston, 2003; pp 1129-1164. (11) Li, M.; Schlenoff, J. B. Anal. Chem. 1994, 66, 824-829. (12) DeVol, T. A.; Roane, J. E.; Williamson, J. M.; Duffey, J. M.; Harvey, J. T. Radioact. Radiochem. 2000, 11, 34-46. (13) DeVol, T. A.; Egorov, O. B.; Roane, J. E.; Paulenova, A.; Grate, J. W. J. Radioanal. Nucl. Chem. 2001, 249, 181-189. (14) DeVol, T. A.; Duffey, J. M.; Paulenova, A. J. Radioanal. Nucl. Chem. 2001, 249, 295-301. (15) Roane, J. E.; DeVol, T. A. Anal. Chem. 2002, 74, 5629-5634. (16) Roane, J. E.; DeVol, T. A. J. Radioanal. Nucl. Chem. 2005, 263, 51-57.

to the analyte concentration and the sampled volume. Signal enhancement can be achieved by delivery of a larger volume of sample to the sensor, so long as analyte breakthrough does not occur. The equilibration-based approach turns this idea on its head, seeking to deliberately achieve full breakthrough conditions where the analyte concentration exiting the column is the same as the analyte concentration entering the column. Under these conditions, the sensing material in the column has equilibrated with the analyte concentration in solution. As we shall describe below, the amount captured is proportional to the analyte concentration under trace detection conditions. Then the signal is proportional to concentration, and because it is based on dynamic equilibrium, it is also reversible. Although delivery of sufficient sample through the sensor to achieve equilibrium can be time-consuming, speed of response is not typically a requirement for an in situ sensor for long-term monitoring. A detailed theory for equilibration-based sensing, using concepts and mathematics from chromatography, will be described below. The sensing approach and its agreement with the theory is then demonstrated with 99Tc(VII) (pertechnetate), a pure β-emitter. This radionuclide is a significant contaminant at U.S. DOE sites; it has a long half-life and a high mobility in the environment. It is further shown that the method can be extended to other radionuclides through a change in the selective chemistry, using β-emitting 90Sr as an example. Although we demonstrate this new method primarily for radionuclide sensing, it is in principle extendable to other analytes where the amount of analyte collected on the column can be measured by other means compatible with on-column detection, such as absorbance or fluorescence. We briefly demonstrate UV-visible detection of hexavalent chromium using an equilibration-based preconcentrating minicolumn sensor as an example. THEORETICAL CONSIDERATIONS Equilibration-Based Sensing Using a Preconcentrating Minicolumn. In the equilibration-based approach, a sample solution with a given analyte concentration is delivered through the sensor column until complete chromatographic breakthrough has occurred. Once this breakthrough condition has been achieved, no further analyte preconcentration occurs and the analyte concentration in the sensor column effluent is equal to the analyte concentration in the sample solution upstream from the sensor column. As a result, dynamic equilibrium is established between the flowing sample solution and the entire volume of the stationary sorbent phase of the sensor column. The phase is equilibrated, but not necessarily saturated (see below). Because this is a dynamic equilibrium, the sensor is reversible. If a blank solution is run through the column, the sorbed analyte will be eluted from the column. The equilibrium analyte concentration in the sensor’s stationary phase, Cs, is determined by the analyte partition coefficient K:

K ) Cs/Ca

(1)

where Ca is the analyte concentration in the flowing sample (17) Egorov, O.; O’Hara, M. J.; Grate, J. W. Reno, NV 2002; American Nuclear Society, La Grange Park, IL; Spectrum 2002; pp 928-931. (18) Egorov, O. B.; O’Hara, M. J.; Addleman, R. S.; Grate, J. W. ACS Symp. Ser. 2004, No. 868, 246-270.

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

5481

solution. The equilibrium amount of the analyte, Meq, captured by the sensor column with the sorbent-phase volume, Vs, and mobile-phase volume, Vm, can be calculated as

Meq ) CsVs ) CaKVs

(2)

The Meq can be related to the analyte retention factor k ) KVs/ Vm and the analyte retention volume, Vr using the following equation:

Meq ) CaVmk ) Ca(Vr - Vm)

(3)

Meq = CaVr

(3a)

The retention factor describes the amount of analyte accumulating in the stationary phase relative to that of the mobile phase in the column and is obtained by multiplying the partition coefficient by the phase volume ratio. For Vm negligibly small relative to Vr, as is typically the case, eq 3 simplifies to eq 3a. In this manner, the amount of the analyte captured by the sensor column after equilibration is equivalent to the amount of analyte contained in the sample volume equal to Vr. In frontal chromatography, the retention volume is observed as the inflection point of the sigmoidal breakthrough profile.19-21 The effect of sample concentration on the amount of analyte captured on the column is dependent on the sorption isotherm in the concentration range of interest. At low concentrations relevant to trace level detection, the sorption isotherm is expected to be linear. For example, even if the overall isotherm saturates at high concentrations, there will be a linear region at trace concentrations where the quantity of analyte captured by the sensor column at equilibrium is much less than the sorption capacity of stationary sorbent phase. Then the partition coefficient is independent of the analyte concentration, and the amount of analyte captured by the sensor column after equilibration is directly proportional to the analyte concentration in the sample, as indicated in eq 2. Should the concentration in the monitored solution go down, the amount on the sensor column will go down in response as the new solution concentration is pumped through. Similarly, if the concentration of the monitored solution goes up, the amount on the sensor column will increase. Sensor Response Modeling. We are now interested in developing a model describing the amount of analyte retained on the sensor column as a function of the sample volume pumped into the column. Similar to frontal chromatographic analysis, the analyte input concentration profile in an equilibration sensing measurement can be represented by a general step function with prior and subsequent analyte concentrations equal to C0 and C1, respectively.19 (i) Analyte Load Step. Let us first consider an input step with prior concentration C0 ) 0 and subsequent analyte concentration C1. Analyte capture and migration through the sensor column result in the transformation of the initial step input function into a sigmoid-shaped concentration profile in the column effluent.19 (19) Reilley, C.; Hildebrand, G. P.; Ashley, J. W. Anal. Chem. 1962, 34, 11981213. (20) Wenzel, W. J. Chromatogr., A 2001, 928, 1-12. (21) Yang, G. L.; Hu, Z. React. Funct. Polym. 1996, 31, 25-29.

5482

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

This breakthrough profile results from a distribution of the velocities of individual analyte molecules or ions through the column due to factors such as transport in the mobile phase through a packed bed, diffusion in the mobile phase, and a distribution in the amount of time analyte molecules spend being immobile in the stationary phase. In conventional chromatography, all the analyte molecules (or ions) are injected at the same time and the distribution in individual velocities results in a peak shape, which is usually Gaussian. Thus, in typical chromatographic theory, the overall distribution is represented by a Gaussian function, which is a valid solution for the governing differential equations and is consistent with typical symmetrical chromatographic peaks.22 In frontal chromatography, the analyte molecules or ions still traverse the column with a distribution of individual velocities, although they do not all start at the same time. The effluent concentration profile is then assumed to be represented by an integral of the Gaussian distribution function.19,23 The inflection point of the sigmoidal breakthrough profile corresponds to the maximum in the Gaussian distribution (and the retention volume). Let function f (V) represent the normalized analyte concentration profile in the column effluent (breakthrough curve) resulting from the unit step input with prior concentration equal to zero (C0 ) 0; C1 > 0). Then, the effluent concentration as a function of volume, Cef(V), resulting from a an input step with concentration C1 is

Cef(V) ) C1 f (V)

(4)

where f (V) goes from zero to one.19 The amount of the analyte that has passed through the sensor column after delivering sample volume V, Mel(V), is the integral of the effluent concentration over volume,

∫ f (V) dV

Mel(V) ) C1

V

0

(5)

The amount of analyte, Ms,C1(V), present on the sensor column after delivery of a sample volume, V, can be calculated as the amount delivered minus the amount that has passed through:

∫

Ms,C1(V) ) C1V - C1

V

0

f (V) dV ) C1V - C1F(V) (6)

where F(V) ) ∫Vo f (V) dV is a function corresponding to the integral of the normalized breakthrough profile (where the breakthrough profile is itself the integral of a distribution that is usually assumed to be Gaussian). (ii) Analyte Elution Step. Let us now consider a step input with the prior analyte concentration equal to C0 and subsequent analyte concentration C1 ) 0. This scenario corresponds to delivery of a blank sample to a column previously equilibrated with a sample solution of concentration C0 > 0. The analyte concentration in the sensor column effluent as a function of volume, Cef(V), can be described by the following equation: (22) Giddings, J. C. Unified Separation Science; John Wiley & Sons: New York, 1991. (23) Lovkvist, P.; Jonsson, J. A. Anal. Chem. 1987, 59, 818-821.

Cef(V) ) C0(1 - f (V)) ) C0 - C0 f (V)

(7)

The total amount of the analyte, Mel, eluted from the sensor column, as a function of volume, can be obtained by integrating the effluent concentration (eq 6) over volume:

Mel(V) )

∫ (C V

0

0

∫ f (V) dV )

- C0f(V)) dV ) C0V - C0

V

0

C0V - C0F(V) (8)

The initial amount of the analyte present on the sensor column equilibrated with sample solution C0 is given by eq 3 and is equal to C0Vr. Then, the amount of analyte, Ms,C0(V), remaining on the sensor column after delivery of V milliliters of the blank sample, can be calculated as the initial equilibrium amount minus the amount eluted:

Ms,C0(V) ) C0Vr - (C0V - C0F(V))

(9) f (V) )

(iii) General Input Step. An arbitrary step input with prior and subsequent analyte concentrations equal to C0 and C1 can be treated as a sum of two independent load and elution steps with initial and subsequent concentrations of C0 ) 0; C1 for load, and C0; C1 ) 0 for elution, respectively. The amount of analyte present on the sensor column during delivery of the sample with analyte concentration C1 can be then obtained by summing eq 6 and eq 9:

Ms,C0,C1(V) ) C0Vr + V(C1 - C0) - (C1 - C0)F(V)

(10)

(

1 1 + erf (τ - 1) 2 2

xN2 )

(11)

where erf(x) is the error function defined as erf(x) ≡ 2/(π)1/2 ∫x0e-t2 dt, τ ) V/Vr, and N is the number of theoretical plates. Integration of eq 11 over volume yields the following expression for F(V):

F(V) )

∫ f (V) dV ) V

0

[(

Vr 1 V+ Φ (τ - 1) 2 x2N

1 erfc 2

(x

)

N (1 - τ) + 2τ

(x

1 exp(2N) erfc 2

xN2 ) - Φ(xN2 )] (12)

)

N (1 + τ) (13) 2τ

where the erfc(x) is a complimentary error function erfc ) 1 erf(x).25 Equation 13 can be analytically integrated over volume to give the following expression for F(V):23

F(V) )

∫ f (V) dV ) 2τV V

0

[

(x (x

(τ - 1) erfc

(τ + 1) exp(2N) erfc

where V is the volume of the sample solution with concentration C1. (iv) Breakthrough Curve Equations. Analyte species enter and exit the column with a distribution of individual velocities through the column or, equivalently, with a distribution of volumes that eluted them. This distribution is typically assumed to be Gaussian. Equation 10, which describes the amount of analyte present on the sensor column as a function of sample volume and analyte concentration, uses a function f (V) as the normalized breakthrough profile. In conventional linear frontal chromatography, the shape of function f (V) is usually assumed to be represented by an integral of a Gaussian distribution:19,23,24

f (V) )

where Φ(x) is the integral of the erf(x) of the form Φ(x) ) x erf(x) + exp(-x2)/(π)1/2. The integrated Gaussian breakthrough profile assumption is strictly valid for columns with relatively high numbers of theoretical plates.23 However, preconcentrating sensor columns have small dimensions, use sorbent materials with relatively big particle sizes, and are typically operated at high linear flow velocities. Moreover, in the radionuclide sensors to be described, the packed columns consist of a mixture of sorbent and scintillator particles. As the result, these sensor columns are not expected to have high chromatographic efficiencies or large numbers of theoretical plates. Therefore, the integral of a Gaussian distribution may not be applicable as a breakthrough profile when describing these column systems. Lo¨vkvist and Jonsson have reviewed several previously reported breakthrough profile equations and proposed that the following function can be used to describe the breakthrough profile in frontal chromatographic systems with low plate numbers:23,24

) )]

N (1 - τ) + 2τ N (1 + τ) 2τ

(14)

We will refer to this as the low plate number model, while the models based on the integration of Gaussian distributions will simply be called the Gaussian model. (v) Radiometric Sensor Responses. The amount of analyte collected on the sensor column could be detected in a variety of ways, depending on its properties. In this paper, we will demonstrate radiometric detection in detail with a pertechnetate sensor and optical detection in a proof-of-principle example with hexavalent chromium. In radiometric detection, the number of radioactive decay events per second is directly proportional to the number of analyte atoms. The fraction of the total decay events being detected or counted is determined by the absolute detection efficiency, Ed. Therefore, equations describing the total amount of analyte present on the sensor column can be converted to radiometric sensor responses using the following equations:

Rc/s,eq ) EdAaVr

(15)

Rc/s,C0,C1 ) Ed[A0Vr + V(A1 - A0) - (A1 - A0)F(V)] (16) The first of these equations gives the radiometric count rate, Req, in counts/second (c/s) of a sensor column that is fully equilibrated (24) Lovkvist, P.; Jonsson, J. A. J. Chromatogr. 1986, 356, 1-8. (25) mathworld.wolfram.com/Erf.html.

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

5483

with a sample containing an analyte activity, Aa (expressed in Bq/ mL). The latter equation describes the sensor count rate, Rc/s,C0,C1, as a function of sample volume for an arbitrary activity step with initial and subsequent activities A0 and A1, respectively. EXPERIMENTAL SECTION Materials. The scintillating beads in the radiometric sensor columns were Bicron BC-400, with 100-250-µm particle size distribution (Saint-Gobain Crystals & Detectors, Newbury, OH). The sorbent used for uptake of 99Tc(VII) or hexavalent chromium was AG 4-X4 100-200 mesh weakly basic anion-exchange material (Bio-Rad Laboratories, Hercules, CA). Strontium-selective solidphase extraction material, SuperLig 620 (IBC Advanced Technologies, Inc., American Fork, UT) was used for 90Sr uptake. Reagents and Standards. All chemicals were of analytical grade or higher. Deionized water (Barnstead E-pure (18.3 MΩ‚ cm), Dubuque, IA) was used as a carrier solution without degassing. Nitric acid solutions were prepared using appropriate dilutions of a TraceMetal grade concentrated nitric acid solution (Fisher Scientific, Pittsburgh, PA). Hexavalent chromium solution was prepared from dilution of a potassium dichromate volumetric standard, 0.1043 N solution in water (Aldrich Chemical CO., Milwaukee, WI). Radioactive solutions were prepared from highpurity, in-house 99Tc(VII) and 90Sr standards. Ultima Gold (PerkinElmer Life & Analytical Sciences, Inc., Boston, MA) liquid scintillation cocktail was used for verification of radioactive solution activities. For experiments in groundwater matrixes, uncontaminated, unpreserved groundwater from well 699-19-88 on the Hanford Site was filtered using a 47-mm-diameter, 0.45-µm type HA filter (Millipore, Billerica, MA). Hexavalent chromium, 99Tc, and 90Sr spikes were added postfiltration. In standard addition experiments on contaminated groundwater samples, a lower concentration 99Tc standard was prepared in a groundwater matrix. Strontium-90 groundwater standard was prepared to 0.01 M HNO3 concentration by appropriate volume addition of nitric acid. Radiometric Equilibration Sensor. A minicolumn flow cell was designed to be inserted between the two PMTs of the Packard 500TR Series flow-through radioactivity detector (see below). The cell was fabricated in-house from a block of BC-800 (Saint-Gobain Crystals), which is cast from ultraviolet-transparent methyl methacrylate monomer material. Outer dimensions of the flow cell were 22 mm W × 9 mm D × 51 mm H. The sensor column internal dimensions were 4 mm i.d. × 29 mm (bed volume, 0.364 cm3). Frit material from the Quick-Snap Column system (Isolab, Inc., Akron, OH) was used to hold the packed material in place. Fluid inlet and outlet lines were constructed from black 0.8-mm-i.d. Teflon FEP tubing and were attached using 1/4-28 flangeless nuts of black Delrin (Upchurch Scientific, Oak Harbor, WA). A homogeneous mixed bed of sorbent and scintillator material was prepared using the following method: Sorbent and scintillator particles were weighed and then combined into a single vial. Contents were stirred thoroughly. For the 99Tc sensor, the AG 4-X4/BC-400 (dry weight) ratio was prepared at ∼1:4. For the 90Sr sensor, the SuperLig 620/BC-400 weight ratio was prepared at 1:1. The homogeneous mixtures were then dry packed into the sensor cell. For the low plate number 99Tc sensor, a shorter bed of 6 mm was packed within the same flow cell, using frits above 5484

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

and below the bed to define its length. The remaining column void space was filled with inert glass beads. Radioactivity Measurements. The activity of all radioactive standards, solutions, and manually collected fractions was verified off-line using a Tri Carb 2550 TR/AB liquid scintillation spectrometer (Packard Instrument Co., Meriden, CT). Radiometric measurements of the equilibration sensor were performed using a Flow Scintillation Analyzer 500TR Series (Packard) flow-through radioactivity detector. The detector was operated by Flo-One software (version 3.55, Packard) running on a laptop PC connected to the detector via serial line. The detection cycle was initiated externally by a signal from a laptop PC controlling the reagent delivery system. The detector integration time was typically 15 s. For experiments requiring flow rates below 0.5 mL/min, the detector integration time was extended to 1 min. Fluid Handling System for Radiometric Sensor Testing. An automated fluid handling system was configured using a Kloehn 48 000-step digital syringe pump utilizing a zero-deadvolume syringe (2.5 mL volume) (Kloehn Co., Las Vegas NV). The pump was configured with a six-port distribution valve, which enabled delivery of four separate solutions (with ports dedicated to waste and delivery to the detection instrumentation). The fluid handling system was controlled by LabWindows/CVI version 5.0.1 software (National Instruments, Austin, TX) running on a laptop PC and a four-port Mini Smart Switch (B&B Electronics, Ottawa, IL) utilizing RS-232 connections. Unless otherwise specified, programmed syringe pump flow rates were 2 mL/min; the pump was refilled periodically during the large-volume experiments, introducing intermittent short periods of no flow during the experiments. Including these pauses, the net average flow rate was 1.63 mL/min. Radiometric Sensor Detection Efficiency Determination. Standard injection experiments to determine detection efficiency of 99Tc sensor cell consisted of the following: (1) wash cell with 0.01 M HNO3, (2) 0.5-mL injection of 167 Bq/mL 99Tc standard in 0.01 M HNO3, and (3) 10-mL wash of 0.01 M HNO3. The load/ wash fraction was collected and counted off-line using the liquid scintillation spectrometer to ensure that 100% of the 99Tc was captured on the sensor cell. Absolute detection efficiency was net count rate/activity loaded on cell. Absorbance Equilibration Sensor. The equilibrium sensing of Cr(VI) from groundwater was perfomed by adapting a SMA-Z cell from FIAlab Instruments, Inc., with a nominally 1-mm optical path length. In this flow cell, fiber-optic cables with SMA connectors are set against quartz windows, which are held in place with Teflon O-rings. Due to the small size of the cell, the true configuration of the fluid path is not a “Z”, but rather a “cross”, and the true length of the optical path was closer to 2 mm. Set perpendicularly to the optical connectors are the fluid connectors, which are machined to secure two 1/4-28 fittings. The volume of the fluid chamber within the optical path was ∼6 µL. Frit material (IsoLab, Inc.) was placed in the downstream position of the cell in order to hold sorbent beads in place. A packed bed of AG 4-X4 was formed by injecting a dilute slurry of AG 4-X4 particles in water with a syringe. The packed bed of sorbent filled the fluid chamber and extended just above the upper threshold of the optical path. Because of the cross geometry, portions of the

sorbent bed within the optical path are not in line with the fluid path. The spectrometer was a USB 2000 miniature fiber-optic spectrometer, which was optimized to 200-575 nm using a no. 1 grating and controlled by OOIBase32 version 2.0.1.3 software (Ocean Optics, Inc., Dunedin, FL) on a laptop PC. The light source was a HL-2000 tungsten halogen lamp (Ocean Optics) optimized for the visible-NIR (360-2000 nm). The optical fiber connecting the lamp to the sensor cell was a 1-m, 300-µm-diameter solarization resistant fiber, and the optical fiber connecting the s cell to the spectrometer was a 1-m, 600-µm-diameter solarization resistant fiber (Ocean Optics, Inc.). Automated fluid handling of hexavalent chromium solutions was performed using a FIAlab-3000 (FIAlab Instruments, Inc., Bellevue, WA) sequential injection system with 5-mL syringe (with two-way distribution valve) and a separate 10port distribution valve for selection of aqueous standards. Automated computer control was accomplished with FiaLab for Windows software running on a laptop PC. Unless otherwise specified, flow rates were 1 mL/min. RESULTS AND DISCUSSION Dual Functionality Composite Bed Sensor Columns for Radiometric Sensing. The key features of the equilibration-based preconcentrating minicolumn sensors are illustrated in this paper with radiometric detection of pertechnetate, and the results are compared with theory. The sensor columns were prepared using a uniform mixture of separate solid-phase extraction and nonporous plastic scintillating beads as column packing.3,10,18 These two materials provide chemical selectivity and scintillation properties, respectively. Because the range of the 99Tc β-particles (∼750 µm in water) is greater than the diameter of the analytical grade sorbent bead (typically 10. However, with decreasing numbers of theoretical plates, the model predictions become increasingly different, especially for N < 5. Figure 4 compares predicted response profiles for the Gaussian and low plate number models (LPM) at various values of N.

Table 2. Results of the 99Tc(VII) Sensor Response Model Fit for a 6-mm-long Columna

source

model used

detection efficiency, Ed, %

Figure 5A fit to models

Gaussian low plate number

85.6 38.6

a

retention volume, Vr, mL

number of plates, N

1.5 19.0

0.01 2.6

Matrix is 0.01 M nitric acid.

Figure 4. Predicted responses for systems with increasing number of theoretical plates (N) using low plate (LPM) and Gaussian models. Both models provide good response profiles for systems with N > 10. For systems with N < 5, model predictions become highly dissimilar.

Figure 6. Ratio of the breakthrough volume at 99% response to the retention volume as a function of column plate number.

Figure 5. Experimental traces showing uptake and equilibration with 99Tc(VII) (A) and 99Tc(VII) breakthrough as determined in collected fractions (B) for a 6-mm column bed equilibrated with 120 mL of a 3.2 Bq/mL 99Tc solution in 0.01 M HNO3. The data in each plot were fitted with Gaussian (dashed) and low plate number (solid line) models.

We conducted additional experiments using sensor columns of reduced length in order to test the two models at lower plate numbers. We observed that both response models continued to provide satisfactory empirical fits to the experimental data for sensor columns with plate numbers ranging from 1 to 5. However, the sensor parameters obtained from the fits told a different story. In the case of N < 5, the Gaussian model (eq 12) failed to provide realistic prediction of the sensor column parameters. In contrast, sensor parameters predicted by the low plate number model (eq 14) remained in satisfactory agreement with the independent measurements. These observations are exemplified in Figure 5, which shows the sensor response profile (graph A) and concurrent 99Tc(VII) breakthrough profile (graph B) for a 6-mm sensor column achieving equilibrium in a 0.01 M HNO3 solution with 3.2 Bq/ mL 99Tc. This column length compares with a 29-mm length in the prior studies. Model fit results in Figure 5A show that both

models provide good empirical fits to the experimentally observed response profile. Sensor parameters obtained from the fits are given in Table 2. The low plate number model (eq 14) predicts Ed ) 39%, which is consistent with the result of the standard injection experiment (Ed ) 41%). Moreover, predicted values of Vr ) 19 mL and N ) 3 for the 6-mm column are much smaller than the values of Vr ) 82 mL and N ) 13 for the 29-mm column (Table 1), as would be expected for a shorter column (∼4.8-fold reduction in the sensor column length). In contrast, the Gaussian model yielded quite unrealistic values of Ed ) 86%, Vr ) 1.5, N ) 6 × 10-3. Furthermore, results in Figure 5B clearly indicate that the Gaussian model (eq 12) fails to provide an adequate fit to the observed breakthrough data for the sensor system with low N. The results show that practical sensors can be implemented using low-efficiency sensor columns (N e 5), and their performance can be described by models provided that an appropriate integrated breakthrough profile model, such as the low plate number model in eq 14, is used. In the case of more efficient column sensor systems (N > 5), sensor responses can be adequately described by a model based on the integrated Gaussian breakthrough profile function given by eq 12. (iv) Equilibration Volume. The volume of the sample solution, Veq, that must be delivered to the sensor column in order to achieve equilibrium steady-state signal is determined by the analyte retention characteristics, size of the sensor column, and its chromatographic efficiency. For practical sensing applications, the Veq, can be defined as a volume of the sample solution necessary to achieve >99% of the steady-state signal level. For a sensor system with known Vr and N, Veq can be calculated numerically from an inverted form of eq 16.23 Figure 6 shows a plot of the calculated relative equilibration volume, Veq/Vr, as a function of the number of plates.23,29 For high plate numbers, the equilibration volume is similar to the retention volume, but for (29) Senum, G. I. Environ. Sci. Technol. 1981, 15, 1073-1075.

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

5487

lower plate numbers, increasing sample volume in excess of the retention volume is required to quantitatively equilibrate the column. The Veq/Vr value increases sharply with decreasing chromatographic efficiency for N values of less than 1. This observation indicates that chromatographic efficiency of at least several plates is desirable for practical sensing applications in order to avoid excessively large sample volumes necessary to achieve sensor equilibration. (v) 99Tc(VII) Quantification in Groundwater. The feasibility of reagentless equilibration-based 99Tc(VII) sensing in Hanford site groundwater, as opposed to 0.01 M nitric acid, was evaluated using a baseline (radiologically uncontaminated) well water spiked with 99Tc(VII). A series of 99Tc(VII) calibration standards were prepared in unacidified groundwater with successively increasing 99Tc(VII) activity levels for a calibration experiment identical to that described above for the 0.01 M nitric acid samples, using the 29-mm column sensor. The lowest level standard in this series, with a 99Tc(VII) activity of 0.033 Bq/mL, corresponded to the drinking water limit. The observed sensor responses (results not shown) confirmed sensor reversibility and linearity of the activity response in the actual groundwater matrix, and a response was visible at the drinking water limit. These results have provided an initial demonstration that pertechnetate can be sensed in a groundwater matrix. It was observed that the VrEd ) 17 counts/s (Bq/mL) in groundwater, as compared to VrEd ) 31 cps/(Bq/mL) in a matrix of 0.01 M HNO3. Nevertheless, the analyte preconcentration in the groundwater matrix was still sufficient to observe a signal above the background for the lowest level (drinking water limit) standard. At the selected flow rate, sensing required ∼90 min of pumping to achieve equilibration and a few minutes of counting to achieve satisfactory counting statistics. Four actual 99Tc-contaminated groundwater samples, each collected at different locations at the Hanford site, were also analyzed with the sensor. Standard addition methodology was used to correct for possible matrix effects on the analyte uptake due to variations in chemical composition of the groundwater collected at different locations. Sensor equilibration with the groundwater sample was followed by equilibration with the same sample that was spiked with a known activity of 99Tc(VII). For each of the four samples, analytical results obtained by the equilibration sensing technique were in excellent agreement with the results of the baseline ICPMS measurements (relative percent difference