Increasing the Sensitivity of Asymmetrical Flow Field-Flow Fractionation

Aug 15, 2006 - sensitivity of the asymmetrical flow field-flow fractionation, with and without ICPMS online coupling for elemental detection. For the ...
0 downloads 0 Views 181KB Size
Anal. Chem. 2006, 78, 6664-6669

Increasing the Sensitivity of Asymmetrical Flow Field-Flow Fractionation: Slot Outlet Technique Harald Prestel,*,† Reinhard Niessner,† and Ulrich Panne‡

Institute of Hydrochemistry, Technical University of Munich, Marchioninistrasse 17, D-81377 Munich, Germany, and Federal Institute for Materials Research and Testing (BAM), Richard-Willstaetterstrasse 11, D-12489 Berlin, Germany

A new enrichment approach is described to improve the sensitivity of the asymmetrical flow field-flow fractionation, with and without ICPMS online coupling for elemental detection. For the slot outlet technique, a part of the laminar carrier stream is removed through an additional pump. This allowed an enrichment of the colloidal particles in the separation channel up to a factor of 14. Additional improvements of the separation efficiency permitted us to separate colloids with differences in their molecular masses of only 2 kDa. Different polymer standards, proteins, and synthesized tracer colloids, as well as real samples (wastewater, liquid manure, serums) were used to assess the performance of the new technique. Field-flow fractionation (FFF) is a separation technique that has been broadly applied to biological macromolecules, polymers, and colloidal particles over a considerable mass or size range.1-4 Of the different FFF techniques, asymmetrical flow field-flow fractionation (AF4) is the most versatile and widely used, because displacement of the sample components by a cross-flow acting as the force field is universally applicable to colloid systems with characteristic dimensions between 0.001 (∼1 kDa molecular mass) and 50 µm.5 However, the sample dilution in the separation channel can result in a significant decrease in the signal-to-noise-ratio for online detection6 via absorbance, fluorescence, or ICPMS. Not surprisingly different prechannel and in-channel enrichment methods have been described earlier. All prechannel methods, e.g., filtration or centrifugation, are time-consuming, can change the sample composition through coagulation, and can introduce artifacts.7,8 * To whom correspondence should be addressed. E-mail: harald.prestel@ ch.tum.de. Phone: +49 2180 78240. Fax: +49 2180 78255. † Technical University of Munich. ‡ Federal Institute for Materials Research and Testing (BAM). (1) Giddings, J. C.; Thomson, G. H.; Myers, M. N. Sep. Sci. 1967, 2, 797800. (2) Giddings, J. C. Science 1993, 260, 1456-1465. (3) Reschiglian, P.; Zattoni, A.; Roda, B.; Cinque, L.; Parisi, D.; Roda, A.; Dal Piaz, F.; Moon, M. H.; Min, B. R. Anal. Chem. 2005, 77, 47-56. (4) Schimpf, M.; Caldwell, K.; Giddings, J. C. In Field-flow Fractionation Handbook; Wiley: New York, 2000. (5) Gimbert, L. J., Andrew, K. N., Haygarth, P. M. and Worsfold, P. J. TrAC, Trends Anal. Chem. 2003, 22, 615-633. (6) Prestel, H.; Schott, L.; Niessner, R.; Panne, U. Water Res. 2005, 39, 35413552. (7) Buffle, J.; Perret, D.; Newman, M. Environ. Part. 1992, 1, 171-230. (8) Horowitz, A. J.; Elrick, K. A.; Colberg, M. R. Water Res. 1992, 26, 753763.

6664 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

Figure 1. Principle of in-channel sample enrichment using the SO technique.

In-channel methods include large-volume injection, use of a potential barrier, concentration via flow focusing, and flow-splitting methods.9-13 In general, increasing the injected sample volume to a few milliliters at low flow rate (0.1 mL/min) is time-consuming and leads to band broadening and subsequent loss of sample.14,15 A potential barrier based on a high ionic strength carrier and a low ionic strength carrier for elution can lead to aggregation of the sample.16,17 Splitting of the flow stream at the end of the channel can be achieved by inserting different flow splitters: split outlet, concentric tube stream, or frit outlet (FO).10 The FO technique permits a gentle separation and is easier to implement than the other enrichment methods mentioned above. However, the pore inhomogeneity of the employed frit and clogging of the frit for higher sample loads can be a considerable disadvantage. In this work, the slot outlet (SO) technique is introduced, which utilizes a slot outlet at the top of the channel. Through an additional piston pump a part of the carrier stream is removed via this slot. Figure 1 shows the principle of sample enrichment: the carrier is split into a sample free part and a substream with increased sample concentration. Distinct advantages of this technique are a homogeneous flow, no clogging of frits, compatibly with a wide range of samples and carriers, no foreign elements, and contamination. (9) Li, P.; Hansen, M.; Giddings, J. C. J. Microcolumn Sep. 1998, 10, 7-18. (10) Giddings, J. C.; Lin, H. C.; Caldwell, K. D.; Myers, M. N. Sep. Sci. Technol. 1983, 18, 293-306. (11) Giddings, J. C. Anal. Chem. 1985, 57, 945-947. (12) Wahlund, K. G.; Winegarner, H. S.; Caldwell, K. D.; Giddings, J. C. Anal. Chem. 1986, 58, 573-578. (13) Lee, H.; Williams, S. K.; Giddings, J. C. Anal. Chem. 1998, 70, 2495-503. (14) Beckett, R.; Nicholson, G.; Hart, B. T.; Hansen, M.; Giddings, J. C. Water Res. 1988, 22, 1535-1545. (15) Koliadima, A.; Karaiskakis, G. J. Liq. Chromatogr. 1988, 11, 2863-2883. (16) Koliadima, A.; Karaiskakis, G. Chromatographia 1994, 39, 74-78. (17) Koliadima, A.; Karaiskakis, G. J. Chromatogr. 1990, 517, 345-359. 10.1021/ac060259l CCC: $33.50

© 2006 American Chemical Society Published on Web 08/15/2006

THEORETICAL BASIS The principle of different FFF subtechniques, including sedimentation, flow, gravitational, electrical, and thermal FFF is based on the balance between mass transport due to the applied field and that due to diffusion that governs the retention and selectivity. Using flow FFF, differences in the diffusion coefficients of the colloids in the sample cause their separation. The separation mechanisms of the different techniques are described elsewhere in detail.4 Only a short description of the AF4 basics, which apply to this work, will be given here. The separation takes place in a thin, ribbonlike separation channel in which a laminar carrier flow with a parabolic flow velocity distribution is subjected to a perpendicular cross-flow field. The colloids are driven to the bottom of the channel, while diffusion causes them to be distributed at characteristic heights above the channel wall. Thus, the diffusion coefficient D establishes the average height l of a colloidal fraction and determines the average retention time tr directly. Due to their higher diffusion coefficient, smaller particles diffuse faster back into the channel and elute the channel earlier as opposed to larger particles. Under the action of a cross-flow field, the individual colloidal species are present at a certain average distance l from the accumulation wall:

l ) D/u

at high split ratios, introducing errors in the size and molar mass determined from the retention times of the sample colloids. An alternative approach is to measure the elution time tr of a series of molecular weight standards. Through determination of the diffusion coefficients for the individual standards using eq 3, a calibration curve of log D versus log Mr can be generated:

log D ) log R - β log Mr

From the axis intercept (log R) and the slope (-β) of the resulting straight lines, the parameters R and β can be obtained. With eq 5, the appropriate molecular weights Mr can be calculated from the corresponding retention times. For calculation of the peak resolution Rs of the fractograms, the following equation was used:

Rs )

r)

λ ) l/w

(2)

where w corresponds to the channel width or thickness (cm). In principle, AF4 allows the determination of the diffusion coefficient D and via the Stokes equation the hydrodynamic colloid diameter dp without further calibration. Due to the laminar flow in the AF4 channel, the individual colloidal species are separated as they experience different flow velocities. Using some approximations, which are described elsewhere in detail,4 from the experimentally determined retention time tr, the associated diffusion coefficient D can be calculated with defined laminar and transverse flow rates through an appropriate calibration:

D)

λw2Vc w2Vc ) V0 6 trVout

(3)

where Vc is the volumetric cross-flow rate (cm3 s-1), V0 the void volume (cm3), and Vout the flow rate at the laminar outlet (cm3 s-1). The knowledge of D allows the linkage to the retention time

dp )

2kTtRV0 πηVcw2t0

)

2kT 6πηD

(4)

where k is the Boltzmann constant, T absolute temperature (K), η the dynamic viscosity (g cm-1 s-1), and t0 the void time (s). It has to be mentioned that perturbations in cross-flow at the membrane wall are likely to cause departures from eq 4, especially

2(tB - tA) WA + W B

(6)

where tA and tB are the retention times of the peaks and WA and WB the peak widths at half-maximum, respectively. The theoretical enrichment factor r using the SO technique is

(1)

where D is the colloid diffusion coefficient and u the induced crossflow field. Instead of l, the dimensionless retention parameter λ can be introduced via

(5)

V˙ slot (out) + V˙ s(out) V˙ s(out)

(7)

where V˙ slot (out) is the removed carrier flow and V˙ s (out) the carrier flow with an increased sample concentration, which is led to an appropriate detector. EXPERIMENTAL SECTION Asymmetrical Flow Field-Flow Fractionation System with Slot Outlet Technique (SO-AF4). An AF4 system (Postnova Analytics, Landsberg, Germany)18 was upgraded to a SO-AF4. The cylindrical-shaped slot outlet port (diameter, 0.5 mm) is placed at the top of the channel (channel width w ) 350 or 500 µm, depending upon the type of hydrocolloids) in front of the laminar outlet port (distance, 13 mm; see Figure 1). The port has to be manufactured with high precision to ensure minimal flow perturbations, especially at high split ratios, and must be placed absolutely in the center of the channel positioned on one straight line from the channel tip to the channel end. The cross-flow and the SO flow are provided by double-piston precision pumps. A frit at the bottom of the channel (277-mm effective length, 20mm width) is covered with a membrane. Different membranes from Postnova Analytics (Landsberg, Germany) were tested for each application: regenerated cellulose with molecular weight cutoffs (MWCO) of 10 and 1 kDa; polyethersulfone with molecular weight cutoffs (MWCO) of 2 and 0.3 kDa. For each application, the membrane, which gave the best results, was used throughout all experiments. Depending on the type of sample, the channel flow is fixed at 0.5 or 1 mL min-1, respectively. SO flow rates of 0-0.8 mL min-1 (0-80% of the total flow rate) were used. Crossflow gradients were optimized for each application. Sample volumes of 5-100 µL are injected automatically using an autosam(18) Jiang, Y.; Miller, M. E.; Myers, M. N.; Kummerow, A. M.; Tadjiki, S.; Hansen, M. E.; Klein, T. Postnova Analytics, Inc. U.S. Patent No. 2004000519, 2004.

Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

6665

pler. Higher sample volumes are injected directly into an injector valve (Rheodyne) with a fixed loop (V ) 1000 µL). From the channel, the effluent is directed through an UV absorbance detector (Lambda 1000, Bischoff Analysentechnik, Leonberg, Germany). The resulting plots of UV absorbance versus retention time or elution volume are termed fractograms. The following parameters have to be optimized for each application: cross-flow ratio, laminar flow, SO flow, eluent composition, membrane type, injection volume, and focusing time. Table S-1 in the Supporting Information displays the optimized parameters for each application. Online AF4-ICPMS Coupling. An ICPMS (Sciex/Elan 6100, Perkin-Elmer Instruments, Shelton, CT) was used to determine total heavy metal concentrations in the samples, as well as an elemental detector behind the UV absorbance detector of the AF4 system. Owing to the similarity of the AF4 channel-flow rates and ICPMS sample-flow rates typically used for analysis, an ICPMS cross-flow nebulizer was connected directly to the UV or the fluorescence detector outlet with a polyetheretherketone tubing (PEEK, 0.2-mm i.d.) of 1000-mm length. The experimental operating parameters are summarized in Table S-2 in the Supporting Information. Reagents. All reagents used were of analytical grade dissolved in Milli-Q water ((Millipore, Schwalbach, Germany). The molecular weight standards were polystyrenesulfonates (PSS, Polymer Standards Service, Mainz, Germany and Postnova Analytics, Landsberg, Germany) ranging from 1.3 to 1330 kDa. Metal standards were prepared by dilution of 1000 mg L-1 stock solutions (VWR International, Ismaning, Germany). The AF4 carrier solution was degassed (ERC-3215, ERC INC., Kawaguchi, Japan) before use. The following carrier liquids were tested: A 10-3 mol L-1 NaCl solution was prepared by dissolving ∼0.058 g of NaCl (VWR International, Darmstadt, Germany) in 1000 mL of deionized water. A PBS solution was prepared by dissolving 1.36 g of KH2PO4 (10-2 mol L-1), 12.20 g of K2HPO4 (0.07 mol L-1), and 8.5 g of NaCl (0.145 mol L-1) (VWR International) in 1000 mL of deionized water. A 3 × 10-2 mol L-1 Tris buffer (VWR International) was prepared by dissolving ∼3.6 g of tris(hydroxymethyl)aminomethane, (Aldrich Chemical Co., Inc., Milwaukee, WI) in 1000 mL of deionized water and adjusted to pH 7 with concentrated nitric acid. A solution of 0.001% Tween 20 (polyethoxic sorbitan laurate, VWR International), at an ionic strength of 10-4 mol L-1 (NaClO4) buffered to pH 9 using a 5 × 10-3 mol L-1 Tris buffer was prepared to a final volume of 1000 mL with deionized water. A 0.1% sodium dodecyl sulfate (SDS) solution was prepared by dissolving 1 g of SDS in 1000 mL of deionized water. Fresh carrier liquids were prepared daily and were agitated in an ultrasonic bath for ∼20 min before usage. Colloid Samples. Sewage plant samples were taken from the municipal sewage plants of Garching, Germany, Munich 1, and Munich 2 and from an industrial sewage plant of a papermill (UPM Kymmene, Schongau, Germany). A detailed description of sampling and sample preparation is given elsewhere.6 Magnetite particles, coated with dextran and mitoxantrone for targeted chemotherapy with magnetic nanoparticles, were provided from the group of F. G. Parak. Preparation, characterization, and applica6666 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

Figure 2. Fractograms of a PSS standard mixture under different separation conditions: carrier 1 mmol L-1 NaCl; membrane 800 Da MWCO PES; laminar-flow rate 0.5 mL min-1; cross-flow rate 5-0 mL min-1). Variation of (A) injection volume (SO flow, 0); (B) SO flow rate.

tions of these nanoparticles are described by Alexiou et al.19,20 For investigations of interactions of colloids with biofilms and colloidal tracer experiments, thulium phosphate particles (dp ) 211 ( 26 nm) were synthesized by heating solutions of phosphoric acid, Tm(NO3)3, urea, and SDS at 80° C for 3 h, resulting in a slow increase of pH and forming of spherical TmPO4 colloids (dh, 211 ( 26 nm). Optimization of preparation conditions, characterization, and applications of these particles as tracer colloids will be described elsewhere. Diclofenac antiserums were provided inhouse; the immunization of adult random-bred rabbits with diclofenac-BSA conjugate is described elsewhere in detail.21 Albumin and γ-globulin (Aldrich Chemical Co.), were used to for optimization and identification of the protein signals in the fractograms. Data Treatment. Raw fractograms were translated into size distribution profiles by applying an Excel (Microsoft Excel 2002, Redmond, WA) spreadsheet. Origin 7.0 (OriginLab Corp., Northampton, MA) was used to evaluate peaks, adjust baselines, and plot cumulative areas. RESULTS AND DISCUSSION After upgrading the AF4 system6 to the SO variant, all existing fractionation methods had to be adapted to the new flow conditions. During this optimization, it became clear that in some cases not only the signal intensity but also the separation performance could be increased, especially for small colloids such as humiclike substances and proteins. While a signal enhancement through an increase of the injected sample volume leads to a reduction of the retention time, i.e., loss of separation power, for a given colloid (19) Alexiou, C.; Jurgons, R., Parak, F. G.; Weyh, T.; Wolf, B.; Iro, H. IEEENANO 2004, Fourth IEEE Conference on Nanotechnology, Munich, Germany, 2004; pp 319-321. (20) Alexiou, C.; Jurgons, R.; Schmid, R.; Hilpert, A.; Bergemann, C.; Parak, F.; Iro, H. J. Magn. Magn. Mater. 2005, 293, 389-393. (21) Deng, A.; Himmelsbach, M.; Zhu, Q.-Z.; Frey, S.; Sengl, M.; Buchberger, W.; Niessner, R.; Knopp, D. Environ. Sci. Technol. 2003, 37, 3422-3429.

Figure 3. Fractograms and molecular weight and size distributions of sewage plant colloids with different SO flow rates: carrier 1 mmol L-1 NaCl; membrane 800 Da MWCO PES; laminar-flow rate 0.5 mL min-1; cross-flow rate 5-0 mL min-1).

size,22 an increased SO flow is not paired with a shorter retention time. Figure 2 illustrates this effect for a PSS standard mixture under different separation conditions. While in the upper graph the injected sample volume was increased, in the lower graph the SO flow was increased. In both cases, an increase of the injection volume or SO flow results in an increase of signal intensity. Additionally, increased injection volume is concurrent with a decrease of retention time, whereas enhanced SO flow leads to an increase in retention time. So the resolution with increasing SO flow is improved, whereas increasing the injection volume degrades the resolution. The changes of resolution of the signals related to the PSS standard fractionations in Figure 2 are shown in the Supporting Information (Table S-3). Resolution could be increased up to 32%, while enlarging the injection volume reduced the resolution up to 60%. The related peak heights and retention times for the different PSS standards for the lowest and highest injection volumes and SO flow rates used are listed in Table S-4 in the Supporting Information. Figure S-1 shows the related differences in peak heights and retention times for the different PSS standards for all tested injection volumes and slot outlet flow rates. In all cases, the signal increase is higher than theoretically expected (see eq 7). The decreasing retention time with increasing concentration or injection volume is well known.22,23 Hence, an increase of tr with increasing SO flow rate is also expected, as the flow rate of the main sample flow decreases with increasing SO flow. However, it has to be noted that changes in retention time by varying the SO flow depend on other parameters too, e.g., the size and the kind of colloids. Most, but not all, of the colloids investigated in this work displayed an increase of retention time with increasing SO flow rate. In Figure S-2 (Supporting Information), a PSS calibration using optimized separation conditions and the calculated molecular weight and size distributions is given. Standard deviations (SD) of the retention times are in the range (22) Martin, M.; Feuillebois, F. J. Sep. Sci. 2003, 26, 471-479. (23) Melucci, D.; Zattoni, A.; Casolari, S.; Reggiani, M.; Sanz, R.; Reschiglian, P.; Torsi, G. Ann. Chim.-Rome 2004, 94, 197-206.

of 0.25-1.32%, the SD of the peak maximums in the range of 0.892.18%. The related calibration data are listed in Table S-5 in the Supporting Information. For comparison, a fractogram of a PSS mixture using the AF4 system without SO technology is shown in Figure S-2, too.6 Such calibrations were carried out for calculating molecular weight and size distributions of real samples such as sewage and liquid manure. It is usually assumed that PSS standards are appropriate for calibration of humic substances (HS) and humic-like substances,4,24,25 whereas for other colloids, e.g., blood proteins, PSS standards are unsuitable. Sewage Plant Colloids. Fractograms of real samples also showed an improvement of resolution and sensitivity.6 Figure 3 displays optimized fractograms as well as molecular weight and size distributions of samples from an industrial sewage plant with and without SO flow. Calibration was carried out using a PSS standard mixture as described above. Two colloidal fractions with retention times of 2.2 and 24.5 min are observed. The first signal derives from HS, the other basically from inorganic particles such as iron and alumina hydroxides.6 The signal enrichment factor using a SO flow rate of 60% is 7.5 for HS and 1.8 for the inorganic colloids. Detailed results on the characterization of sewage plant colloids using an AF4 system without enrichment technique have been described elsewhere.6 Compared to our earlier results, the SO technique enables the characterization of samples with lower colloid concentrations. The resolution for smaller colloids is also increased, e.g., for the void peak and the HS signal from 0.62 without SO to 0.80 using 80% SO flow. Liquid Manure. Preliminary results from analysis of liquid manure are given in Figure 4. Similar to sewage fractograms, a HS fraction (tr ) 4 min) can be observed as well as three other fractions with higher retention times (tr ) 25-40 min). Scanning electron micrograph measurements revealed that these fractions (24) Yohannes, G.; Wiedmer, S. K.; Jussila, M.; Riekkola, M. L. Chromatographia 2005, 61, 359-364. (25) Ngo Manh, H., Geckeis, J. I., Kim, H. P. and Beck, C. S. Colloid Surf., A 2001, 181, 55-64.

Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

6667

Figure 4. Fractograms and molecular weight and size distributions of liquid manure with different SO-flow rates: carrier 1 mmol L-1 NaCl; membrane 800 Da MWCO PES; laminar-flow rate 0.5 mL min-1; cross-flow rate 5-0 mL min-1).

Figure 5. Fractograms of artificial inorganic particles with different SO flow rates. Left: dextran and mitoxantrone-coated magnetite particles. Right: TmPO4 tracer colloids (carrier 0.1% SDS (magnetite particles), 1 mmol L-1 NaCl (TmPO4 particles); membrane 10 kDa MWCO cellulose; laminar-flow rate 0.5 mL min-1; cross-flow rate 0.5...0 mL min-1).

consist of inorganic particles, pollen, and different bacteria (data not shown). Detailed results of these investigations will be published elsewhere. Compared to fractionations without SO flow enrichment, factors of 4.7 for the HS fraction and 2.9 for the large particles could be achieved. Additionally, it can be seen that the only colloidal signals are increased by the SO flow, whereas the void peak remains constant, as expected. As a result, the overlap of the void peak with the HS fraction is reduced. Inorganic Particles. In Figure 5, fractograms with different SO flow rates of different artificial inorganic particles, are shown. Magnetite particles, coated with dextran and mitoxantrone for magnetic drug targeting, were described by Alexiou et al.19,20 An enrichment factor of 3.6 could be obtained (left). Figure 5 (right) 6668 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

shows fractograms of spherical TmPO4 tracer colloids (dh, 211 ( 26 nm) in a biofilm reactor matrix using online AF4-ICPMS coupling. No evaluable AF4-ICPMS fractograms of those samples could be obtained before upgrading the AF4 system. A signal increase of 6.1 was obtained. Serum Proteins. To separate the immunoglobulin (Ig) fraction in sera from other proteins, especially albumin, a separation method for rabbit antisera was developed.26 The separation of the Ig from albumin and other serum proteins is usually done via size exclusion chromatography (SEC). Using SO-AF4 is more efficient as the separation can be carried out in ∼15 min. Further investigations relating to the activity of the separated fraction in (26) Matschulat, D.; Prestel, H.; Haider, F.; Niessner, R.; Knopp, D. J. Immunol. Methods 2006, 310, 159-170.

Table 1. Enrichment Factors for SO-AF4 in Different Applications

application

maximum enrichment factor(s)

13.7 (Mr ) 1.3 kDa)a 3.7 (dh ) 100 nm) 7.5 (HS) 1.8(inorg particles) liquid manure 4.7 (HS) 2.9(inorg particles) coated magnetite 3.6 (dh ) 100 nm) particles TmPO4 tracer 6.1 (dh ) 211 nm) colloids proteins/serums 2.5 (albumin) 2.4 (Ig) PSS standards latex standards sewage

a

Figure 6. Fractograms of a serum with different SO flow rates and γ-globulin solution (without SO flow (carrier 1 mmol L-1 Tris buffer; membrane 1 kDa MWCO cellulose; laminar-flow rate 0.5 mL min-1; cross-flow rate 4-0 mL min-1)).

SO-AF4

ELISA tests will reveal whether yields better results than SEC. In Figure 6, fractograms of a serum21 with different SO flow rates are shown. The maximum enrichment factor is 2.5 for the albumin (Mr ≈ 68 kDa) and 2.4 for immunoglobulin (Mr ≈ 150 kDa) fraction, which agrees with the expected value of 2.5. In contrast to the other applications described here, the retention time decreased in this case with increasing SO flow rate (see Figure S-3 in the Supporting Information). Further work has to reveal in which way other factors (e.g., the kind of colloids fractionated or the ionic strength of the carrier) influence the retention times. Especially, the composition of the carrier solution may affect the charge of the membrane surface and the corresponding interactions with the colloids. Table 1 summarizes the obtained and theoretical enrichment factors for all samples investigated in this work. In most cases, the enrichment factors are higher than theoretically calculated; i.e., the enrichment is not completely described

theoretical enrichment related SO factor flow rate (%) 5 2.5 2.5

80 60 60

5

80

5

80

2

50

2.5

60

Other sizes; see Table S-4 in the Supporting Information.

via eq 7. Further investigations are necessary to reveal other parameters influencing the signal intensities and retention times. ACKNOWLEDGMENT This work was financial supported by the Deutsche Forschungsgemeinschaft (DFG). The authors gratefully thank Mrs. Christine Sternkopf for assistance with the ICPMS measurements. Special thanks go to Mr. Heiko Hilbert for allowance to take samples from the sewage plant of the UPM Kymmene papermill (Schongau, Germany), and Mr. Walter Bertl for assistance with the sampling. Last but not least sincere thanks are given to Prof. Dietmar Knopp for providing diclofenac antisera, as well as to Prof. Fritz G. Parak and Mrs. Olga Stroh for providing dextran and mitoxantrone-coated magnetite particles. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review February 9, 2006. Accepted June 22, 2006. AC060259L

Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

6669