Direct electrochemical detection of individual collisions between magnetic microbead/silver nanoparticle conjugates and a magnetized ultramicroelectrode† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c5sc02259b

Magnetic fields and silver nanoparticles increase the frequency and current signature of collisions between individual particles and electrode surfaces.


Introduction
In this paper, we report a new method for amplifying the current signature of collisions between single particles and electrode surfaces. 1 The specic approach we describe could evolve into a viable means for using single-particle collisions for low-level sensing applications: something that has not yet been achieved. The method involves direct electrochemical detection of silver nanoparticles (AgNPs) conjugated to conductive magnetic microbeads (cMmBs) via DNA hybridization. Detection limits as low as 20 aM are achieved due to two factors. First, the presence of multiple AgNPs on each cMmB, and, second, the ability of a magnetic ultramicroelectrode (UME) to increase the rate of mass transport of the cMmBs, relative to diffusion, to the UME surface. These results are signicant for the following three reasons. First, they demonstrate the feasibility of direct electrochemical detection of DNA-conjugated AgNPs, which can be used as labels for a variety of electrochemical assays. Second, we provide a detailed analysis of the parameters that control AgNP detection, which are relevant to future bioassays based on collisions. Third, we describe a simple method for preparing magnetic UMEs that will be useful for many different types of applications.
The experiment described in this article is set up as follows. First, as shown in Scheme 1a, the cMmBs are prepared by coating a magnetic microbead with Au to produce a conductive shell, and then this shell surface is modied with AgNPs using DNA hybridization to yield a conjugate of the form: cMmB-DNA-AgNP. Second, a magnetic UME is prepared by coating the surface of a Ni wire with a thin layer of Au, and then placing magnets around the wire to magnetize it (Scheme 1b). Third, as shown in Scheme 1c, an electrochemical cell is congured so that when the cMmB-DNA-AgNP composite is driven to the electrode surface by the magnetic eld, the associated AgNPs oxidize more or less simultaneously. This gives rise to an anodic current transient of the type shown in Scheme 1c.
This work was motivated principally by earlier reports from the groups of Lemay, 1 Compton, 2 and by one of our own prior studies. 3 Lemay and co-workers were the rst to describe electrochemical detection of collisions between individual nanoparticles and an electrode surface. Specically, they studied collisions between nonconductive latex beads, having diameters of 150 and 500 nm, and a 5 mm Au UME. 1 Upon striking the electrode surface, the latex particles were found to irreversibly adsorb to the surface of the UME, and this was signaled by a stepwise decrease in the faradaic current. The current is attenuated because each particle partially blocks the active surface area of the electrode, thereby hindering mass transport of a redox probe (ferrocenemethanol, FcMeOH). This work was extended by Bard and co-workers, who investigated the effect of low electrolyte concentration on collision frequency and amplitude of the current change via nite element simulations. 4 We further extended Lemay's ndings by correlating optical tracking and electrochemical measurements of collisions between insulated microbeads and a UME surface. 5 The collision trajectory was tracked using uorescence microbeads, and the highest current change was observed when the microbead struck or migrated to the edge of the UME. This nding is consistent with both theory and simulations that predict the highest current ux to be at the edges. 6 Additionally, Yoo et al. showed that it is possible to detect insulated magnetic microbeads (iMmB) in a microelectrochemical device at concentrations as low as 500 zM using a single, moveable magnet placed under the channel of the device. 3 Pre-enrichment steps collected the microbeads inside the channel inlet and then focused them at the working electrode.
The Compton 2,7-12 and Pumera 13 groups have described a different type of electrochemical collision experiment that is also highly relevant to the ndings reported here. In their work, individual or agglomerated metal nanoparticles, usually Ag, strike a UME surface resulting in a burst of anodic current. The important point about this type of approach is that the charge resulting from each collision can be directly correlated to the size of the colliding nanoparticle.

Results and discussion
Synthesis and characterization of conductive magnetic microbeads modied with AgNPs (cMmB-DNA-AgNP) As will be discussed later, it is not possible to carry out experiments like those reported here successfully using iMmBs. This is because the surface of the iMmB is not conductive, and therefore only the AgNPs within $1 nm of the electrode would be oxidized. As a result, only a tiny fraction of the total number of AgNPs dispersed on the surface of the iMmB would yield a signal, and assays built on the approach described here would be insufficiently sensitive to be worthwhile.
The procedure for preparing cMmBs is described in detail in the ESI † but is briey outlined here. 32 We start with commercially available iMmBs surface-functionalized with negatively charged carboxylate groups. These are mixed with AuNPs having positively charged 2-aminoethanethiol on their surface ( Fig. S1 and S2, ESI †). As shown by the micrographs in To ensure that most or all of the nanoparticles on the surface of the cMmB-DNA-AgNP conjugate are electrochemically addressable, we carried out the following experiment. The cMmB-DNA-AgNP beads were drop cast onto a glassy carbon electrode (GCE) congured in an electrochemical cell in a faceup orientation. Fig. 2 shows six consecutive ASVs obtained using a single electrode in a solution containing 100 mM NaCl and 10 mM phosphate buffer (pH 7, referred to hereaer as 100 mM PBCl). The black trace is the rst ASV, and it exhibits two Ag oxidation peaks: a large peak at $70 mV and a much smaller peak at $25 mV. It is worth noting, however, that the small peak at $25 mV is not present on every rst scan. The magnitude of the total charge under these two peaks is 3.37 mC, which corresponds to 6.77 Â 10 7 AgNPs. The shi in peak position for the second and subsequent scans will be discussed later.
To demonstrate the importance of the conductive Au shell in these studies, we carried out a control experiment using iMmBs (no conductive shell) functionalized with AgNPs. These materials were prepared by reacting streptavidin-coated MmB (sMmB, Fig. S4a †) with biotinylated DNA modied AgNPs. An SEM image of the resulting sMmB-DNA-AgNP conjugate is shown in Fig. S4b. † The important result is that when the experiment described in the previous paragraph is carried out using the sMmB-DNA-AgNP conjugate instead of that based on the conductive analog, no detectable ASV current is detected (Fig. S4c †). This conrms the necessity of rendering the iMmBs conductive prior to carrying out collision experiments.
To demonstrate that DNA hybridization is primarily responsible for attachment of cMmBs to AgNPs in the cMmB-DNA-AgNP conjugate, we carried out a control experiment in which noncomplementary DNA was used for the attachment link. ASVs for conjugates built using noncomplementary and complementary DNA are compared in Fig. 3a, and the integrated charge for the two resulting peaks is provided in Fig. 3b. The results clearly show that the average charge from the stripping peak is signicantly lower when noncomplementary DNA is used. In other words, there is only a very small amount of nonspecically adsorbed AgNPs on the cMmBs.
We return now to the second through sixth scans in Fig. 2. These result in just a single peak at $20 mV, which is the location of the rst small peak in the rst scan. For the following discussion we will refer to the peak at $20 mV as the rst peak and the one at $70 mV as the second peak. To understand the origin of these two peaks, the following experiments were performed using the procedure described earlier for the rst peak. First, when the potential of the GCE was held at 100 mV prior to returning it to the initial potential of À200 mV and then recording the ASV, the second peak did not appear. Second, when shorter DNA was used to link the cMmBs to the AgNPs, the second peak decreased in size and the rst peak increased. On the basis of these experiments, we believe that the origin of two peaks in the ASVs is related to the insulating DNA layer and the possibility that both AgCl and Ag + are  figure). At the conclusion of the individual scans, the electrode potential was stepped back to À0.20 V and held there for 3.0 s before the next scan was initiated. The electrolyte was 100 mM PBCl buffer. products of the electrooxidation of Ag. For the purposes of the present work this is not an especially important point, but it does direct us to hold the electrode potential more positive than the second ASV peak for the collision experiments. More information about the two peaks is provided in the ESI (Fig. S5-S7 †).
Fabrication of a magnetic Ni/Au UME The detection limit of electrochemical collision experiments is limited by the ux of particles to the electrode surface. 4,19,22 Normally, this ux is determined by diffusion, and because particles are large in comparison to molecules, their diffusion coefficients, the hence the limits of detection (LODs) of collision experiments, are generally not very low. To achieve lower LODs, diffusion must be supplemented by a second means of mass transfer like electrophoresis, 4 pressure-driven ow, 24,25 or, as in the present case, a magnetic force. [32][33][34][35][36][37] Note that attempts to use electrophoresis and pressure-driven ow to lower LODs have not been very successful.
There are a number of ways one might imagine integrating a magnet into an electrochemical cell for collision experiments. For example, a tiny permanent magnet could be located beneath a microfabricated UME, but this would be very difficult to implement. If the magnet was much larger than the UME, then MmBs would be trapped at locations other than the electrode surface. It is possible to fabricate very small electromagnets, but that is also experimentally challenging and in addition the heat resulting from the windings of the magnet introduces a new variable to the experiment. [38][39][40] To avoid these types of problems, we simply magnetized the UME itself using an external magnet that focuses the magnetic eld at the tip of the electrode.
The magnetic UME used here consists of a Ni wire with a thin layer of Au deposited on its tip using galvanic exchange. 41,42 Specically, a 50 mm Ni UME was prepared by sealing a Ni wire in a glass capillary. As shown in Scheme 2, the Ni UME was then sealed in an acrylic plate using epoxy glue. Next, the surface of the electrode was polished to remove excess epoxy. A thin layer of Au was added by submerging the electrode in a 10 mM HAuCl 4 solution for 10 s with gentle stirring. This results in spontaneous galvanic exchange between the Ni wire and Au 3+ in solution. At this point the electrode was washed with a copious amount of DI water and checked under an optical microscope (Fig. S8 †) to visually conrm Au deposition by a color change from gray to orange.
The Ni/Au UME resulting from this process was mounted face-up in an electrochemical cell (Scheme 2f). In some cases the electrode was magnetized using ring magnets placed around the UME as shown in the scheme and Fig. S9. † In other cases, the magnets were absent so that control experiments could be carried out. Using the Ni/Au UME (in the absence of the magnets) a cyclic voltammogram (CV) of the Au surface was recorded in 100 mM phosphate buffer solution (pH 7). The red trace in Fig. 4 reveals the characteristic peak potentials (E p ) associated with Au oxidation (E p ¼ $0.7 V) and oxide reduction (E p ¼ $0.3 V). This CV can be compared to the black trace in Fig. 4, which shows that these peaks are absent prior to Au galvanic exchange. One nal note: of course Ni is less noble than Au, and therefore one would expect peaks associated with Ni oxidation and reduction. Their absence is a consequence of the formation of an oxide of nickel resulting from air exposure. 43 Detection of collisions between cMmB-DNA-AgNP conjugates and a magnetic Ni/Au UME Fig. 5a shows representative chronoamperograms (i-t curves) for collisions between cMmB-DNA-AgNP conjugates and a Ni/ Au UME using a 100 mM PBCl buffer. This experiment was Scheme 2 Fig. 4 Cyclic voltammograms of a Ni UME before (black) and after (red) electroless deposition of Au. The scans started at À0.20 V, the scan rate was 100 mV s À1 , and the electrolyte was 100 mM phosphate buffer (pH 7). carried out by setting the potential of Ni/Au UME to 200 mV, which is positive of both the rst and second ASV peaks shown in Fig. 2. The black trace is a control experiment that was recorded in the absence of the cMmB-DNA-AgNP conjugate, and its shape is similar to that observed in previous collision experiments. 2 The decrease in the background current as a function of time may be due to blockage of the electrode resulting from progressive accumulation of trace contaminates from solution, or a reduction in activity of the Au UME due to electrodeposition of Ag. 3 The red and blue traces were recorded in solutions containing 100 aM of the cMmB-DNA-AgNP conjugate plus the PBCl buffer. The blue i-t trace was obtained in the presence of the magnetic eld, and it reveals numerous current transients associated with oxidation of AgNPs. When the magnets are removed from the electrode, the red i-t trace results. In this case, just a single, small current transient is observed. It is obvious that both the number and size of current transients are much larger when the magnetic eld is present. Fig. 5b and c are expanded views of the i-t data shown in Fig. 5a in the presence and absence of the magnetic eld, respectively. As mentioned earlier, the sharp current transients, which correspond to very fast oxidation of multiple AgNPs per cMmB, are much larger in the presence of the magnetic eld (Fig. 5b). Specically, the average charges for collisions in the presence and absence of the magnetic eld are 36.4 AE 33.7 pC and 8.5 AE 6.9 pC, respectively. Although one expects the collision frequency to increase in the presence of the eld (vide infra), it is not obvious that the magnitudes of the charges should differ so dramatically. We believe there are two possible explanations for this observation. First, the cMmB-DNA-AgNPs may aggregate in the presence of the magnetic eld, leading to larger current transients. 44 Second, it is possible that the cMmB-DNA-AgNPs are in better contact with the electrode or have a longer residence time on its surface in the presence of the eld.
The data in Fig. 5d were obtained by carrying out experiments like those described for Fig. 5a, but using concentrations of the cMmB-DNA-AgNP conjugate ranging from 20 to 200 aM. This plot of collision frequency vs. the conjugate concentration very clearly demonstrates that the magnetic eld enhances the rate of mass transfer of the MmBs to the electrode surface. Although it is difficult to draw a meaningful line through the data points obtained in the absence of the magnet, the ratio of the slopes of the best linear ts through the two sets of data is 4, suggesting that the magnet is responsible for a four-fold increase in signal.
The charge resulting from each collision in the presence of the magnetic eld was analyzed by measuring the area under the individual current transients as a function of the conjugate concentration (Fig. 6). Regardless of concentration, the majority of the charges range from 20 to 70 pC, with an overall average of 36.4 AE 33.7 pC. The latter value corresponds to oxidation of $732 AgNPs. 2 By measuring the concentration of the AgNPs before and aer incubation with the cMmBs (using the Nano-Sight particle counter), and taking into account the average diameter of the AgNPs (23.3 nm), we nd the average number of AgNPs per cMmB to be 3054 AE 260. This value is $4 times higher than the average measured from the collision data ($732 AgNPs). Although the agreement is actually pretty good, the differential is probably due to the different methods used for measurement. The value of 3054 AgNPs/cMmB should include all AgNPs, while the number measured using the collision data only takes into account those that are electrochemically accessible. Electrochemical inaccessibility could arise from AgNPs immobilized on patches of Au that are not in electrical contact with the electrode at the time of collision. Similarly, the presence of the DNA linkers could render some AgNPs too far from the surface of the cMmBs to be oxidized.
To determine if oxidation of the AgNPs is dependent on the potential applied to the Ni/Au UME, collision experiments were performed at three different potentials (Fig. 7). Fig. 7a shows representative i-t traces for these experiments. The blue, red, and black traces correspond to electrode potentials of À100, 0, and 200 mV, respectively. These three potentials were chosen because À100 mV is more negative than the Ag oxidation potential, 0 mV is at the onset of Ag oxidation, and 200 mV is well into the Ag oxidation potential. The black ASV in Fig. 7b was obtained by drop casting AgNPs onto a Au macroelectrode, and it shows the location of the Ag oxidation peak relative to these three potentials. It reveals a sharp anodic peak at $20 mV having an onset potential at $0 mV. Fig. 7b shows that the frequency of anodic current transients is signicantly higher at 200 mV compared to À100 and 0 mV (red data points). Additionally, the average charge for Ag oxidation is signicantly higher at 200 mV. These results are consistent with previous reports of naked AgNP collision experiments, which showed that the collision frequency and charge increase dramatically aer the onset potential for silver oxidation. 2

Summary and conclusions
In summary, we have described direct electrochemical detection of AgNPs linked to individual cMmBs. Importantly, the use of a magnetized Ni/Au UME increases the ux of this conjugate to the electrode surface, relative to diffusion, and therefore the collision frequency is higher. Moreover, for reasons we can only speculate on, the magnitude of the collisions is also greater in the presence of the eld. This has allowed detection of cMmB-DNA-AgNP conjugates down to a concentration of 20 aM, which corresponds to $61 fM AgNPs (recall there are $3000 AgNPs/ cMmB).
In addition to improving the limit of detection for collision experiments through the use of a magnetic eld, the other important aspect of this work is that the AgNP labels are linked to the cMmBs via DNA. That opens up the possibility of using collision experiments for DNA detection, which we are currently exploring as a possibility.   . 7 (a) i-t curves obtained for cMmB-DNA-AgNP conjugate collisions in the presence of a magnetic field. The concentration of cMmB-DNA-AgNP was 100 aM, and the potentials applied to the Ni/Au UME for each experiment are indicated in the legend. (b) Plot of frequency and charge vs. the applied potential. The black curve is a representative Ag stripping voltammogram obtained for AgNPs dropcast onto a Au macroelectrode (2 mm). The scan started at À0.2 V and ended at 0.3 V. The scan rate was 50 mV s À1 and the electrolyte was 100 mM PBCl buffer. The error bars were determined from three independent experiments at each potential.