Analytical and Bioanalytical Chemistry Electronic Supplementary Material Observation of size-independent effects in nanoparticle retention behavior during asymmetric-flow field-flow fractionation Julien Gigault and Vincent A. Hackley 1. Experimental details Reagents used to formulate the mobile phase, calibrate the inductively coupled plasma mass spectrometer (ICP-MS) and synthesize Se nanoparticles (NPs) were purchased from VWR (Philadelphia, PA) as follows: ammonium nitrate (NH 4 NO 3, > 99 %), high purity nitric acid (HNO 3, > 99 %), Au, Se and Ag standards (1000 mg L -1 in 3 % HNO 3, Ricca ICP- MS NIST traceable standards), Na 2 SeO 3 (> 99 %), ascorbic acid and sodium dodecyl sulfate (SDS). Durapore 0.1 µ m filters (EMD Millipore, Billerica, MA) were used to remove particulates from the mobile phase. All solutions were prepared using ultrapure filtered deionized water obtained from an Aqua Solutions (Jasper, GA) Type I biological grade purification system. Table S1. Summary of instrument parameters and optimized performance conditions used to characterize nanoparticles by A4F-DAD-MALS/DLS-ICP-MS A4F DAD Light Scattering ICP-MS Injected volume: V inj = 100 µl Wavelength: (190 to 900) nm Wavelength: 658 nm Nebulizer flow: 0.16 ml min -1 Main flow: V out = 0.5 ml min -1 Sampling rate: 0.01 s Scattering volume: 0.07 µl Cones (sample and skimmer): Ni Injection flow: V I = 0.2 ml min - 1 Reference angle: 90 RF power = 1500 W Focus flow = 2.0 ml min -1 MALS Formalism used: Zimm first order fit Plasma gas flow rate: 15 L min -1 Cross flow: V c = 0.2 ml min -1 DLS angle: 107 The identification of any commercial product or trade name does not imply endorsement or recommendation by the National Institute of Standards and Technology. 1
2. Influence of the quantity injected The quantity of nanoparticles injected in the channel can have an influence on the retention time determination. If it is too low, detector sensitivity becomes a limiting factor. On the other hand, if the concentration is too high, sample overloading phenomena can occur and will significantly alter the retention behavior of analytes in the A4F channel. The overloading effect causes a decrease of retention time as illustrated in Figure S1, which represents the decrease of retention time with an increase particles injected for Au and PSL nanoparticles with a nominal size of d H =30 nm and d H =60 nm. This decrease results from an effective increase in the retention parameter λ (see Eq. 1 of main text) when overloading occurs. So for all studies we maintained the total particle concentration injected into the channel below 1 10 10. 60 nm 30 nm Fig. S1. Influence of the number of nanoparticles injected on the retention time for Au and PSL with having two nominal sizes. Conditions of analysis are: 0.8 ml min -1 of cross flow and spacer thickness of 350 µm, and mobile phase ionic strength of 0.5 mmol L -1 of ammonium nitrate We also tested the influence of the injected nanoparticle (NP) quantity on the measured retention time in mixtures. As an example, Figure S2(a) shows two superimposed MALS fractograms for suspensions containing the same mixture of NPs (i.e., PSL, Se, Ag and Au), but with one suspension containing 5x Ag (red trace) relative to the other containing 1x Ag (black trace). In the concentration range evaluated, we found no significant change in retention time corresponding to the extent of the MALS signal as exemplified in Figure S2(a). Furthermore, the ICP-MS trace for the 1x Ag suspension shown in Figure S2(b) and corresponding to m/z 107 (Ag), clearly shows that the retention time remains unchanged for the AgNPs regardless of their concentration. We therefore conclude that comparisons conducted using a constant NP injected mass (versus a constant number of injected NPs) are appropriate. 2
Fig. S2. (a) MALS fractograms of the mixed NP suspension (PSL,, SeNP, AgNP and AuNP) and of the same suspension doped 5x with AgNPs ; (b) ICP-MS trace at m/z 107 (Ag) corresponding to the original mixed NP suspension (the ICP-MS trace for the corresponding 5x Ag suspension is not presented due to saturation of the detector) 3
3. DLS derived volume-weighted particle size distributions for NP populations Batch mode DLS measurements of the stock suspensions for the four NP populations were performed and intensity-weighted size distributions calculated using a vendor-provided non-negatively constrained least squares inversion algorithm with the default setting for resolution (see Figure S3). Error bars represent one standard deviation for 3 to 5 replicate measurements made under repeatability conditions. Because of the strong size dependence for Rayleigh scattering, broader distributions will tend to overemphasize or shift toward larger particle size, and this is evidenced in Figure S3 for the Au and Ag NPs, relative to the more narrowly dispersed Se and PSL. Nominally, the distributions are in close alignment with the mean size near 100 nm. The density and the zeta potential of the nanoparticles are presented in Table S2 below. Fig. S3. Hydrodynamic diameter distributions obtained off-line by batch mode DLS for the four NP stock suspensions used in this study. Zeta potential values are reported along with the core density values used for comparative purposes Table S2. Density and zeta potential for the NPs used in this work PSL SeNPs AgNPs AuNPs Density (g cm -3 ) 1.05 4.3 10.49 19.3 Zeta Potential (mv) -40.6-39.1-37.1-39.1 4
4. Influence of A4F channel orientation Figure S4 demonstrates the influence of channel orientation on AuNPs retention. In the normal orientation, the cross-flow direction is downward and the channel (laminar) flow is horizontal. By inverting the channel such that the cross-flow direction is upward (opposite the gravitational force), the impact of particle density should be mitigated and the retention time thereby reduced. This experiment was repeated for several different orientations with similar results (data not shown). Fig. S4. A4F-UV/Vis fractograms for 100 nm AuNPs realized in the condition summarized in Table 1 of the main text, with channel position at 0 (black trace) and with the channel inverted 180 (red trace) 5. Proposed influence of Van der Waals forces on the retention time. The value of the effective Hamaker constant (A 123 ) depends on the polarizability of the two bodies (1 and 3) interacting through an intervening dielectric medium (2) according to the approximation 1 : A!"# = A!/!!/!!! A!! A!/!!/!!! A!! where A xx is the Hamaker constant for material x in a vacuum. In the present case, A 11 refers to the NP, A 33 to the membrane (accumulation wall) and A 22 is the Hamaker constant for the medium (water). The energy of interaction for a spherical particle at a planar surface is then: V!"# = A!"#r 6x where r is the radius of the sphere and x is the surface-to-surface separation distance. With the sphere size approximately the same for each NP population in the present study, and assuming similar approach distances to the membrane surface, the attractive force near the surface is then proportional to the effective Hamaker constant, and additive to the force arising from the applied cross-flow. 5
The value of A123 was calculated for each NP population assuming a spherical particle in water and using A11 values found in the literature.234 The same Hamaker constant was used for both the PSL NPs and for the membrane surface; a survey of published constants (both calculated and experimentally determined) for a range of polymeric substances (including polystyrene, but excluding fluorocarbons) indicated that A11 typically falls in the range from about 6.5 to 10 (x 10-20 J), so a mid-value of 8 was chosen. 6. Limitations for on-line size detection in A4F The obvious benefit of on-line particle size measurement is the avoidance of materialspecific elution phenomena that might interfere with the determination of diffusion coefficients based solely on calibrated retention times. However, on-line measurements are not always practical or possible. For particles with a size near to or greater than 100 nm (as used in the current study), the slower decay of the autocorrelation function combined with the fast transit through the scattering zone in the cell results in noisy data that are not conducive to accurate and reproducible analysis; DLS works well for smaller particles that have faster decays. Additionally, the use of elastic scattering (MALS derived Zimm plots, e.g.) to obtain size information is limited for metallic NPs like Au and Ag due to the strong absorptive component of the refractive index (e.g., due to surface plasmon resonance effects) and relatively low real component values. As a result, the Zimm plots for the 100 nm AuNPs and AgNPs present no clear angle dependency of the scattered light. The graphs in Figure S5 illustrate this point. On the other hand, Se and PSL have high real refractive indices and low absorption, yielding a strong angular dependence. By using MALS it is therefore possible to determine the geometric radius of both of these NPs, yielding 49.6 ± 0.2 and 50.4 ± 0.3 nm, for PSL and SeNPs, respectively. Fig. S5. Zimm plots of MALS data obtained on-line for PSL, Se, Au and Ag NPs 6
Alternatively, as shown in section S3 above, we can determine the size by off-line DLS under the same conditions as realized on-line in terms of dilution and aqueous media (i.e., diluted in the A4F mobile phase). As the coefficient of diffusion of SeNPs and PSL can be determined in this manner, we can calculate the effective spacer thickness of the channel according to the equation: ω eff = 6DtR V ln 1 + V c out This equation allowed us to predict the hydrodynamic diameter expected at the measured retention times for both AgNPs and AuNPs. The size expected at these retention times is 129 nm and 134 nm, for Ag and Au, respectively. Clearly, these values are not in agreement with the actual size known from off-line DLS analysis and information provided by the vendor. 1 J. N. Israelachvili, Intermolecular and Surface Forces, Academic Press: London, 2 nd Edition (1991). 2 P. C. Hiemenz and R. Rajagopalan, Principles of Colloid and Surface Chemistry, Marcel Dekker: NY, 3 rd Edition (1997). 3 A. T. Hubbard, Encyclopedia of Surface and Colloid Science Volume 4, Marcel Deker: NY (2002). 4 I. D. Morrison and S. Ross, Colloidal Dispersions: Suspensions, Emulsions, and Foams, Wiley Interscience: NY (2002). 7