Fractionation and Characterization of High Aspect Ratio Gold Nanorods Using Asymmetric-flow Field Flow Fractionation and Single Particle Inductively Coupled Plasma Mass Spectrometry

Gold nanorods (GNRs) are of particular interest for biomedical applications due to their unique size-dependent longitudinal surface plasmon resonance band in the visible to near-infrared. Purified GNRs are essential for the advancement of technologies based on these materials. Used in concert, asymmetric-flow field flow fractionation (A4F) and single particle inductively coupled mass spectrometry (spICP-MS) provide unique advantages for fractionating and analyzing the typically complex mixtures produced by common synthetic procedures. A4F fractions collected at specific elution times were analyzed off-line by spICP-MS. The individual particle masses were obtained by conversion of the ICP-MS pulse intensity for each detected particle event, using a defined calibration procedure. Size distributions were then derived by transforming particle mass to length assuming a fixed diameter. The resulting particle lengths correlated closely with ex situ transmission electron microscopy. In contrast to our previously reported observations on the fractionation of low-aspect ratio (AR) GNRs (AR < 4), under optimal A4F separation conditions the results for high-AR GNRs of fixed diameter (≈20 nm) suggest normal, rather than steric, mode elution (i.e., shorter rods with lower AR generally elute first). The relatively narrow populations in late eluting fractions suggest the method can be used to collect and analyze specific length fractions; it is feasible that A4F could be appropriately modified for industrial scale purification of GNRs.


Supplemental Materials
. Additional TEM images of as received commercially sourced GNRs. To prepare the sample for analysis, 6 µL of suspension was drop-cast on a formvar-coated copper gird. The sample was then air dried at room temperature. Table S1. To estimate the length distribution from TEM imaging, 237 imaged particles were analyzed. Only single rods were counted (i.e., no aggregates or over lapping rods). The TEM scale bar was the only reference of measurement; therefore intrinsic errors in the length measurements may exist.

Determination of A4F Recovery
Sample recovery during A4F fractionation was estimated by replicating the experiment under identical conditions without applying cross-flow during elution. The integrated intensity of the UV-Vis absorbance, for example, can be used as a concentration metric. Recovery is estimated by difference between the integrated signals from the two measurements, assuming 100% recovery in the absence of cross-flow and a Beer-Lambert law relationship between Au mass and absorbance at an SPR peak. Given the presence of different morphologies, lengths and aspect ratios, the mass-absorbance relationship yields a rough estimate of recover in the present system. This is a commonly employed procedure in A4F.

Single Particle ICP-MS Data Acquisition and Processing
Single particle ICP-MS measurements were conducted on a Thermo Fisher X series II quadrupole ICP-MS (The identification of any commercial product or trade name does not imply endorsement or recommendation by the National Institute of Standards and Technology). The 197 Au intensity was recorded in time-resolved analysis (TRA) mode using Thermo Fisher PlasmaLab software. A typical sample series consists of blanks (deionized water and 0.15 mmol L −1 CTAB), a 27 ng L −1 (1.5 × 10 4 particle mL −1 ) AuNP standard (NIST RM 8013, nominal diameter 60 nm), ionic gold standards in the concentration range of (0.5 to 28) µg L −1 , and the diluted GNR sample suspensions in 0.15 mmol L −1 CTAB. GNR samples were diluted in triplicate and measured for 600 s at a dwell time of 10 ms.
Each measurement of a sample results in 60,000 readings consisting of background signals (ions of the same element, particles smaller than the detection limit (10 nm for sphere AuNPs) and instrument noise) and pulse signals on top of the background produced by particles. Data in the unit of counts per second (cps) were converted to counts per event and then processed in Microsoft Excel to calculate GNR length and length distribution. The particle signals were first differentiated from the background S3 using a five times standard deviation (5σ) criterion [29,30]. To eliminate the bias of mean particle intensity by false positive signals (high background signals that are erroneously recognized as particle events), data points with intensity less than 5 counts per 10 ms were further removed [29]. After split pulse correction [29], the average particle intensity of 60 nm AuNP standard was used to calculate the transport efficiency by comparing with the intensity of soluble gold mass flow [35].
The GNRs used in this study have a relatively uniform diameter of approximately 20 nm. Assuming cylindrical geometry and fixed diameter of 20 nm, the lengths of GNRs were calculated as follows: = 4 × 10 15 × 2 ⁄ where L = length of GNRs (nm), d = diameter of cross section (20 nm), and ρ = density of Au (19.3 g cm −3 ).