A gold nanorod-based plasmonic platform for multi-logic operation and detection

Hydrogen tetrachloroaurate (III) trihydrate and cetyltrimethylammonium bromide (CTAB) were purchased from Sinopharm Chemical Reagent (Shanghai, China). L-ascorbic acid, sodium borohydride (NaBH₄) and silver nitrate (AgNO₃) were supplied by Alfa Aesar (Shanghai, China). Sodium iodide (NaI), mercury nitrate [Hg(NO₃)₂], hydrogen peroxide (H₂O₂, 30%) and copper chloride (CuCl₂) were products of the Beijing Chemical Plant (Beijing, China). 11-mercaptoundecanoic acid (MUA) was obtained from Energy Chemical (Shanghai, China), and (11-mercaptoundecyl)trimethylammonium (MTA) bromide was provided by Sigma (St. Louis, MO, USA). The reagents were of analytical grade and used as obtained without any further purification. All stock solutions were stored in a refrigerator at 4 °C. Working solutions were prepared by appropriate dilution of the stock solutions with triple-distilled water, and were filtered through 0.22 μM filters (Jiuding High Tech., Beijing, China) before use.

The AuNRs were prepared following an Ag-assisted seed-mediated method [25], with moderate modification. Scheme 1 shows the experimental flowchart. Briefly, 5 ml of 0.1 M CTAB was mixed with 25 μl 50 mM HAuCl₄ in a 25 ml glass bottle with gentle stirring for 5 min Then, 300 μl freshly-prepared, ice-cold NaBH₄ (10 mM) solution was quickly injected into the mixture, followed by vigorous stirring for 25 s. A dark yellow to brownish yellow change of the solution signified the formation of gold seeds. The seed solution was stored under 27 °C before use. To prepare the AuNR growth solution, 10 ml of 0.1 M CTAB was added with 200 μl 1 M hydrochloric acid and 100 μl 50 mM HAuCl₄ with gentle mixing, then was added with 120 μl 10 mM AgNO₃ and 100 μl 100 mM ascorbic acid. Under vigorous stirring, the growth solution was injected with 24 μl seed solution, and the mixture was left undisturbed at 30 °C for 24 h to ensure full growth of the AuNRs. Before use, the AuNRs were separated from the growth solution by centrifugation at 10 000 rpm for 15 min, washed by triple-distilled water two times and redispersed in 10 ml 0.1 M CTAB solution. The concentration of Au⁰ was estimated to be 0.50 mM.

Zoom In Zoom Out Reset image size

UV–vis spectra of the AuNPs were recorded on an UV-2450 UV–vis spectrophotometer (Shimadzu Corporation, Kyoto, Japan) using a 1 mm thickness quartz cuvette with 10 mm light pass. The morphology of the AuNRs was visualized on a TF20 (200 kV, FEI Tecnai, Oregon, USA) transmission electron microscope (TEM).

The etching solution (ES) was a freshly-prepared mixture of 200 mM H₂O₂, 0.4 mM NaI and 0.4 mM hydrochloric acid. The AuNRs were dispersed in 0.1 M CTAB. Unless otherwise stated, the base solution was 200 μl AuNR dispersion. The logic inputs included I⁻, H₂O₂, MTA, MUA, Hg²⁺ and Cu²⁺, which were added to the base solution to desired concentrations.

For logic operation, the base solution was transferred to a 0.5 ml centrifuge tube and placed in a water bath at 50 °C for 10 min Afterward, logic stimuli were introduced to the base solution, and the mixture was vigorously stirred for 1 min, then kept in the water bath to allow the logic input-regulated etching to take place. Ten minutes later, the mixture was cooled in an ice-water bath to terminate the etching process. The final solution was observed by naked-eye or measured by UV–vis spectroscopy.

The logic values of the inputs 0 and 1 were defined as the absence and the presence of the inputs, respectively. For outputs observed by naked-eye, the wine red color of the AuNP colloids before etching was defined as 0, and the color of the etched AuNR solution displaying light blue was defined as 1. Because the degree of AuNR color change reflected the extent of etching, which could be quantitatively characterized by the wavelength shifting of the L-LSPR peak, a parameter named relative L-LSPR shift R_λ was adopted as the criterion:

$$\begin{eqnarray}&&{{R}}_{\lambda }=\displaystyle \frac{{\rm{\Delta }}\lambda }{{\lambda }_{{\rm{\max }}}}=\,\displaystyle \frac{{\lambda }_{{\rm{\max }}}-\lambda \,}{{\lambda }_{{\rm{\max }}}},\end{eqnarray} \tag{ 1 }$$

where λ_max and λ corresponded to the wavelengths of the L-LSPR peaks of the AuNRs before and after etching, respectively. Herein, the outputs 0 and 1 in UV–vis measurements were experimentally established as below R_λ = 0.005 and above R_λ = 0.24, respectively, so that the output signals from all logic gates are coherent.

The AuNRs prepared by the seed-mediated approach displayed a color of wine red. The UV–vis spectra depicted a strong L-LSPR band and a weak T-LSPR band near 840 and 520 nm, respectively (figure 1). The TEM image indicated that the AuNPs were about 40 nm in length and 10 nm in diameter (figure S1 is available online at stacks.iop.org/NANO/30/055503/mmedia), with an average aspect ratio of ca. 4. The tips of the nanorods were less-covered, because during the seed-mediated growth of AuNRs in the presence of CTAB as the shape-directing surfactant, CTAB preferentially bound to the {100} Au faces along the length of the rods, compared with the end {111} Au faces, owing to the size of the CTAB headgroup [26]. The exposed tips were eroded in the presence of the ES consisting H₂O₂ and I⁻ [23]

$$\begin{eqnarray}&&2\,\mathrm{Au}+{{\rm{I}}}_{3}^{-}+{{\rm{I}}}^{-}\to 2{{\rm{AuI}}}_{2}^{-}.\end{eqnarray} \tag{ 2 }$$

Furthermore, it was proposed that presence of cetyltrimethylammonium ion (CTA⁺) favored the corrosion of AuNRs by the following reaction [27]

$$\begin{eqnarray}&&2{\rm{Au}}+{{\rm{I}}}_{2}+2{{\rm{I}}}^{-}+4{{\rm{CTA}}}^{+}\to 2{{\rm{AuI}}}_{2}^{-}-{({{\rm{CTA}}}^{+})}_{2}.\end{eqnarray} \tag{ 3 }$$

As a consequence, the AuNRs would gradually shrink along the longitudinal direction (figure 1, Panels (a), (b) and (c)), which was accompanied by a blue-shift in the L-LSPR band (figure 1(d)).

Zoom In Zoom Out Reset image size

However, the etching is affected by some factors. For example, the exposed tips can be covered by mercapto compounds via chemisorption and, hence, the etching process is hindered [28, 29]. Moreover, as found by our experiment, the etching process can also be impeded by the formation of Au-Hg amalgam. On the other hand, the AuNR erosion process can be promoted by addition of some species, such as Cu²⁺, as a moderate oxidant. Therefore, by employing mercapto compounds (MTA and MUA), Hg²⁺, Cu²⁺, I⁻ and H₂O₂ as inputs, numerous logic gates could be built.

To make the color change of AuNRs occur smoothly and repeatably, the conditions of the logic platform were optimized with regard to the dispersion medium of the AuNRs, the volume of the base solution, and the acidity and volume of the ES. The dispersion medium for the AuNR colloids played an important role in the color transition. The AuNRs washed by triple-distilled water two times and redispersed in 0.1 M CTAB offered the maximum blue-shift of the L-LSPR absorption (figure S2). The reason might be that CTA⁺ ion facilitated the etching process [27]. With regard to the volume of the base solution (AuNR solution), it was found that long etching time (>15 min) was required with 3 ml AuNR solution placed in a 5 ml centrifuge tube, probably due to the relatively poor heat transfer efficiency during etching. However, when the base solution volume decreased to 100 μl, it was difficult for the naked-eye discrimination of the color changes. A volume of 200 μl could fulfill the requirements for naked-eye observation and instrumental measurement, and save the precious AuNRs.

Hydrochloric acid was added to the ES to facilitate corrosion, because an acidic environment would improve the yield of iodine and coordination between Cl⁻ and Au⁺ would reduce the electric potential of Au⁺/Au [30]. Under typical conditions, the ES contained 0.4 mM hydrochloric acid. Finally, the volume of ES was optimized. Under a fixed etching time of 10 min, mixing different volumes of ES with 200 μl AuNRs gave rise to varying final colors (figure 1). The final solution became brown upon the addition of 10 μl of ES. However, with addition of 20 μl ES, a light-blue solution was obtained, which was readily distinguishable from the original wine red. Therefore, the volume of ES was set to 20 μl. The optimal etching conditions were adopted throughout the work for the good performance.

Presence of iodine is a prerequisite for the AuNR corrosion. We found that neither 1 mM I⁻ (0/1) nor 1 M H₂O₂ (1/0) as the input could induce a color change, implying no etching took place. Only when the two inputs were simultaneously applied (1/1) did the color of the resultant solution change from wine red to light blue. Based on this phenomenon, an 'AND' logic gate was developed (figure 2(a)).

Zoom In Zoom Out Reset image size

Two mercapto compounds, MTA and MUA, were employed as inputs to build a 'NOR' logic gate (figure 2(b)) by utilizing their anti-etching effects. The tips of CTAB-stabilized AuNRs were more exposed compared with the sides, and were more vulnerable to etching reactions, leading to an anisotropic reshaping. However, the mercapto compounds introduced could attach to the tips via formation of Au-S bonds. The chemisorbed long-chain mercapto compounds formed a new layer on the AuNR tips, serving as a barrier hindering the etching process. The concentrations of MTA and MUA in the final solution were optimized to be 29 and 33 μM, respectively, by means of relative L-LSPR shifts (figures S3 and S4). Individual input (0/1 or 1/0) could prevent the corrosion of AuNRs. Upon simultaneous stimulation (1/1), however, a similar level of protection was obtained, rather than a higher one. The phenomena were ascribed to the fact that in the base solution, the number of binding sites provided by the AuNRs was less than the number of MTA or MUA molecules inputted. The binding sites were completely occupied upon introducing either of the mercapto compounds. The minimum input value was lower for MTA. Although the two mercapto compounds have same chain length, they are differentiated by the terminal groups. MTA is terminated with a quaternary ammonium group, which is water-soluble and hence more accessible to AuNRs in aqueous solution [31, 32]. However, the relatively poor water solubility of MUA restrained its chemisorption on the AuNRs [33].

It was reported that Hg²⁺ could deposit on the tips of the AuNRs by forming Au-Hg amalgams in the presence of a reducing agent [34]. In our experiment, ascorbic acid was employed in synthesizing the AuNRs, and the residual ascorbic acid, which physically adsorbed on the tips of AuNRs, might initiate the in situ reduction of Hg²⁺. We found that with increasing amount of Hg²⁺ added, the intensity of the L-LSPR peak decreased gradually, concomitant with a very slight, almost negligible, hypochromic shift (figure S5), suggesting the formation of Au-Hg amalgams on the AuNR surfaces [35]. Furthermore, in the presence of the Hg²⁺, the wavelength of the L-LSPR band virtually did not change with the introduction of ES (figure S6), implying that the Au-Hg amalgams on the AuNR tips hindered etching. According to the above results, an 'XNOR' gate was constructed using MTA and Hg²⁺ as stimuli (figure 2(c)). Introducing either MTA (29 μM) or Hg²⁺ (0.5 μM) prevented AuNR etching and, accordingly, the solution maintained its original wine red color. However, the AuNR colloids turned into light blue upon application of both inputs (1/1). At input (1/0), MTA chemisorbed on the AuNR tips via Au-S bond. In addition, because it had a linear structure, which was relatively flexible, MTA could bind to the rod side via ligand exchange [36]. Further, this linear mercapto compound might even bind all over the nanorod [26], making the positive charge density on the rod side higher than that on the ends. On the other hand, at input (0/1), Hg²⁺ would preferentially attach on the AuNR tip owing to the lower Coulombic repulsion at this region relative to the longitudinal side. The above mechanisms explain why much higher concentration of MTA (nearly 60 folds as Hg²⁺) was needed for the protection. Similarly, at input (1/1), Hg²⁺ selectively attached to the AuNR tips due to the relatively weaker electrostatic repulsion. Because the mercapto group has stronger affinity with Hg⁰ than with Au⁰ [37], the simultaneous detachment of anti-etching reagents MTA and Hg²⁺ from the AuNR tips was responsible for the morphological change.

We found that presence of Cu²⁺ significantly promoted the etching. Therefore, a 'YES' logic switch was rationally developed (figure 2(d)), which was a 1-input 1-output gate performing logic affirmation. The base solution (mixture of 200 μl AuNRs and 20 μl 11 mM NaI) was wine red initially; it became light blue upon the addition of 1 mM Cu²⁺. The standard potential of the Cu²⁺/Cu⁺ redox couple [${{\varphi }_{{\rm{A}}}}^{\theta }$ (Cu²⁺/Cu⁺) = 0.17 V] is lower than that of I₂/I⁻ [${{\varphi }_{{\rm{A}}}}^{\theta }$ (I₂/I⁻) = 0.536 V] [38], suggesting that the reaction of equation (4) could not take place spontaneously.

$$\begin{eqnarray}&&{{\rm{Cu}}}^{2+}+{{\rm{I}}}^{-}\to {{\rm{Cu}}}^{+}+\displaystyle \frac{1}{2}{{\rm{I}}}_{2},\end{eqnarray} \tag{ 4 }$$

$$\begin{eqnarray}&&2{{\rm{Cu}}}^{2+}+4{{\rm{I}}}^{-}\to 2{\rm{CuI}}\downarrow +{{\rm{I}}}_{2}.\end{eqnarray} \tag{ 5 }$$

However, formation of CuI precipitates will promote the reaction, such that oxidation of iodide by Cu²⁺ through equation (5) can proceed smoothly (refer to the supporting information is for detail), yielding iodine to facilitate AuNR etching.

Because Hg²⁺ and Cu²⁺ functioned oppositely on AuNR etching, we adopted the mechanisms and built an 'IMPLY' logic gate (figure 2(e)). The base solution consisted of 200 μl AuNRs and 20 μl ES. When the two inputs were 0 (0/0), the AuNRs were etched, displaying a color of light blue. Presence of 0.05 μM Hg²⁺ (1/0) inhibited the corrosion, whereas addition of 50 μM Cu²⁺ (0/1) accelerated the process. Notably, application of both inputs (1/1) led to a color change of the AuNR colloids from wine red to light blue. The in situ reduction of Hg²⁺ by the residual ascorbic acid produced a thin layer of Au-Hg amalgam covering the AuNR tips. The anti-etching effect of the amalgam layer was effective against the mild ES consisting H₂O₂ and I⁻, but was overwhelmed by the iodine-mediated etching synergetically boosted by H₂O₂ and Cu²⁺.

Since the output of one logic gate can be the input signal of another, combination of different logic gates through cascade reactions can lead to more intricate logic units. We utilized the mixture of 200 μl AuNRs and 20 μl ES as base solution. By carefully balancing the protection effect of MTA, the anti-etching effect of the Au-Hg amalgams and the oxidation effect of Cu²⁺, two basic binary logic gates XNOR and IMPLY were successfully integrated into a ternary combinational logic gate (figure 3), which worked with the stimuli of Cu²⁺, Hg²⁺ and MTA at concentrations of 50 μM, 0.5 μM and 29 μM, respectively. The different results from inputs (1/0/1) and (0/1/1) revealed the varying protection efficiencies of MTA and Hg²⁺. The anti-etching mechanism by MTA was based on formation of the Au-S bond, while the protection by Hg²⁺ relied on the formation of Au-Hg amalgams. With the input of (1/1/1), an output of 1 was obtained because the simultaneous presence of MTA and Hg²⁺ resulted in the detachment of the anti-etching agents by forming stable Hg²⁺-MTA complexes, whereas introducing Cu²⁺ accelerated the etching process. The combinational logic gate had eight input states and eight corresponding output signals, providing more complicated information related to the etching/protection reactions. Moreover, the three-input logic gate had the potential to function as a multi-task sensor, thereby fulfilling the colorimetric detection of multiple analytes.

Zoom In Zoom Out Reset image size

All the outputs in terms of the relative L-LSPR shifts [(λ_max − λ)/λ_max] were successfully labeled as logic 0 or 1 since they were not significantly different from the reference values 0.005 and 0.24, respectively (p > 0.05, table S1). Meanwhile, the values for logic 0 were significantly different and easily distinguishable from the values for logic 1 (p < 0.05, table S2; Cohen's d > 21.5, table S3).

Theoretically, a logic device is a platform for analysis. We selected the 'XNOR' logic gate as an example to study its applicability in detecting Hg²⁺. Hg²⁺ in the water body has drawn on-going concerns because the long-term accumulation of mercury in human body can cause immunological disorders, kidney damage and neurological diseases. We found that without resorting to sophisticated instruments, the limit of detection (LOD) for Hg²⁺ by naked-eye was 50 nM. Spectroscopic detection possessed higher sensitivity, offering LOD of 4.9 nM, which could be applied to detecting Hg²⁺ in drinking water that satisfied the US EPA guidelines (∼10 nM for Hg²⁺) [39]. Moreover, as altering the AuNR aspect ratio gave rise to optical signals in a wide range of wavelengths [24], this AuNR morphology change-based method provided wider linear ranges (Hg²⁺: 0.01 ∼ 0.5 μM, R_λ = 0.455× + 0.00438 (μM), r² = 0.9930, see figure 4 and table 1) than the AuNP aggregation-based counterparts [40].

Zoom In Zoom Out Reset image size

As suggested by a reviewer, we studied the influence of AuNR aspect ratio on R_λ and LOD of Hg²⁺. To this end, we synthesized additional AuNRs with aspect ratios of ca. 3 and 5, with rod length/diameter (in nm) of 60/20 and 40/8, respectively (please refer to the supporting information for synthesis procedures and figures S7 and S8 for TEM images), and applied them in the detection of Hg²⁺. The LODs obtained with all the AuNRs satisfied the US EPA guidelines, but the AuNRs of aspect ratio 4 offered the most sensitive detection (table 1). The AuNR aspect ratio determined the wavelength of L-LSPR peak, which were 783, 840 and 891 nm for AuNRs with aspect ratios of ca. 3, 4, and 5, respectively (figures 4 and S9 and S10). In addition, according to equation (1), aspect ratio affected the R_λ value. This is also reflected by the experimental results; for example, in the presence of 0.4 μM Hg²⁺, the R_λs for the above AuNRs were 0.2181, 0.1909 and 0.1818, respectively. Furthermore, table 1 indicates that AuNRs with higher aspect ratio provided wider linear range, because their L-LSPR wavelength could be changed in a broader range during etching.

Selectivity of the logic detection was investigated by evaluating the relative L-LSPR shifts generated by the potential interfering ions. Our results (figure 5) indicates that presence of 0.5 μM Hg²⁺ generated a relative L-LSPR shift of 0.24, but the coexistence of other metal ions at a 100-fold excess could not give rise to a noticeable wavelength shift or color transition of the AuNRs. The logK_f value of Hg(SCN)_n is ca. 21.8, whereas the corresponding values for the SCN-complexes with Zn²⁺, Cd²⁺, Co²⁺, Ni²⁺, Mn²⁺, Pb²⁺, Fe²⁺, Cr³⁺, Cu²⁺, and Au⁺ are ca. 2.0, 2.8, 1.72, 1.76, 1.23, 1.48, 1.31, 3.08, 10.4, and 16.98, respectively [41]. The largest logK_f value of Hg(SCN)_n, which is nearly 20 orders higher than most of the interfering ions, endowed the 'XNOR' logic gate with high selectivity.

Zoom In Zoom Out Reset image size