The mercury-specific aptamer was synthesized by Shanghai Kehua Bio-Engineering Co. Ltd. (Shanghai, China). The sequence was HS-5′-(CH₂)₆-TTTT-3′. 6-Mercapto-1-hexanol, HAuCl₄, cysteine, hexaammineruthenium(III) chloride, potassium ferricyanide and tris(2-carboxyethyl)phosphine hydrochloride (TCEP) ware purchased from Aldrich (St. Louis, MO, USA) and used as received. Hg²⁺ stock solution (0.1 M) was prepared by dissolving HgCl₂ with 0.5% HNO₃ [12]. The Hg²⁺ stock solution was diluted to desired concentration with 0.05 M Tris-HCl buffer (pH 7.4) containing 0.1 M NaCl. All other reagents were of analytical grade and used without further purification. 0.05 M Tris-HCl buffer (pH 7.4) and 0.1 M phosphate buffered saline (PBS) (pH 7.4) were prepared according to standard procedures. All solutions were prepared with Milli-Q water (>18.2 MΩ·cm⁻¹).

The QCM was performed on a CHI 440A quartz crystal microbalance electrochemical workstation (Shanghai Chenhua Ltd., Shanghai, China). A 7.995 MHz AT-cut quartz crystal (13.7 mm diameter) covered with gold electrodes (Φ 5 mm) was mounted levelly in a homemade Taflon holder. This Taflon holder was linked with a six-channel BT100K peristaltic pump.

The scanning electron microscopy (SEM) images used to characterize the morphology of the QCM electrode surface were obtained on an S-4800 microscope (Hitachi Ltd., Tokyo, Japan). Electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (CV) were carried out with a CHI660c electrochemical workstation (Shanghai Chenhua Ltd., Shanghai, China). CV and EIS were performed in 5 mM [Fe(CN)₆]^3−/4− solution which was purged with highly purified nitrogen at least 5 min. All the measurements were carried out at room temperature.

The QCM gold electrodes were cleaned with phiranha solution [H₂O₂/H₂SO₄ 1:3 (v/v)] for 1 min, rinsed with copious amounts of purified water and ethanol, and dried with a gentle flow of nitrogen in order to remove grease and other pollutants from the electrode surface. Prior to the immobilization of the electrodes, aptamer stock solution (0.1 mM) was reduced in 10 mM TCEP for 2 h to cleave disulfide bonds. The resulting solution was then diluted with Tris-HCl buffer (pH 7.4) to the desired concentration (from 0.5 to 50 μM). Then, the QCM electrode was incubated in the aptamer solution for 16 h in the dark at 4 °C. Following incubation, the electrode was rinsed with water and then immersed in an aqueous solution of 1 mM 6-mercapto-1-hexanol for 1 h to passivate the electrode surface [31]. As a final step, the electrode was rinsed with water, dried with nitrogen, and stored at 4 °C prior to use.

Au-NPs were prepared according to a previously reported method with a slight modification [32,33]. In detail, solutions of HAuCl₄ and trisodium citrate were filtered through a 0.8 μm microporous membrane filter prior to use. 1% Trisodium citrate (2.5 mL) was added to boiling 0.25 mM HAuCl₄ solution (50 mL) and stirred for 10 min at the boiling point. The resulting solution was allowed to cool to room temperature. Concentration of the as-prepared Au-NPs was determined by UV-vis spectroscopy using Lambert-Beer's law (molar extinction coefficient of Au-NPs is 2.7 × 10⁸ M⁻¹·cm⁻¹ at λ₅₂₀) [34]. The as-prepared Au-NPs were further co-modified with the MSA probe. Briefly, Au-NPs (5 mL, 50 nM) was incubated with a MSA probe (100 μL, 100 μM) for 16 h at room temperature. The final mixture was slowly brought up to a final salt concentration of 0.1 M NaCl, 10 mM TCEP, 1 mM 6-mercapto-1-hexanol and 10 mM PBS (pH 7.4) and allowed to age for 40 h. Centrifugation was performed at 14,000 rpm for 40 min in order to remove excessive MSA. The precipitate was washed with 0.1 M NaCl, 10 mM PBS buffer (pH 7.4) solution, recentrifuged, and finally dispersed in 10 mM Tris-HCl buffer (pH 7.4) containing 0.3 M NaCl for the further use. The as-prepared MSA-(Au-NPs) was denoted as the linking probe 2. In control experiment, the original MSA was denoted as linking probe 1.

The principle of the sensor is based on the mass change on the sensing interface. In order to obtain the sensing signal, the designed aptamer must be first immobilized on the sensing interface as a capturing probe. Then, in the presence of Hg²⁺, a sandwich structure between the capturing probe and the linking probe is formed through the Hg²⁺ mediated T-Hg²⁺-T base pairs. SEM was used to characterize the sandwich structure on formed the sensing interface. Figure 1 shows SEM images of different interfaces after hybridizing in 0.5 μM Hg²⁺ solution. Figure 1(A) is a bare electrode surface. Figure 1(B) is the sensing interface after hybridization between the capturing probe and linking probe 1. In contrast, Figure 1(C) is the interface after hybridization between the capturing probe and linking probe 2. Comparing Figure 1(C) with 1(B), it is obvious that a large number of Au-NPs were successfully linked on the sensing interface through linking probe 2, which would result in a larger mass change on the QCM-based sensor in the same Hg²⁺ solution. Therefore, the frequency response (Δf) should remarkably increase compared with using linking probe 1. Consequently, the sensitivity of the QCM-based Hg²⁺ sensor can be significantly improved by using Au-NPs as the signal amplifier. In addition, electrochemical techniques including CV and EIS were utilized to characterize the steps of modified electrode and sensing for Hg²⁺. The results were shown in Figure S1–S2 in supplement information. The electrochemical data also demonstrate that the sensor was successfully prepared.

All QCM experiments were performed in a QCM cell, in which the quartz crystal Au electrodes were modified with the capturing probe. The device for Hg²⁺ detection is presented in supplement information as Figure S3. The experimental process can be described as follows: 10 mM Tris-HCl buffer (pH 7.4) containing 0.1 M NaCl was injected firstly into the QCM cell with a flow rate of 20 μL/min and the crystal frequency was then measured as a background signal. After the signal change is less than 1 Hz of drift in 60 s, the experiment can be continued. 1 μM linking probe (1 or 2) solution was followed at a flow rate of 20 μL·min⁻¹ and the signal (ΔF) of the QCM-based sensor was recorded. The response of the sensor to Hg²⁺ was shown in Figure 2. As can be seen in the a-b sections of Figure 2, this step was stopped after 600 s, Then, Hg²⁺ was injected into the QCM cell by using micro-injector and the solution in QCM cell was remained to quiet until a stable signal (ΔF) was obtained (typically, about 900 s). In this process, Hg²⁺ was captured with a sandwich structure, T-Hg²⁺-T, formed through the capturing probe, Hg²⁺ and linking probe (1 or 2). This step provided a readout signal and frequency change for quantitative detection of mercury ions, as shown in the b-c sections of Figure 2. Subsequently, 10 mM Tris-HCl buffer (pH 7.4) containing 50 mM cysteine was followed at a flow rate of 25 μL·min⁻¹. In this step, the sandwich structure T-Hg²⁺-T was dissociated and Hg²⁺ was released from the sensing interface. Such that, the signal of the sensor recovered gradually to the level before Hg²⁺ was injected (c-d sections of Figure 2), indicating that the sensing interface of sensor has been renewed. The regenerated sensor could be reused again and the above process could be repeated continuously. This strategy provides a facility for the continuous determination of Hg²⁺.

As a highly sensitive mass sensor, the QCM signal depends on the mass change on its sensing interface. For the same amount of target, if the mass change on the sensing interface can be increased greatly, the sensitivity of detection should be improved remarkably. In this work, Hg²⁺ was detected through a specific sandwich structure formed by the capturing probe, Hg²⁺ and the linking probe, which caused the mass change on the sensing interface [35]. Au-NP was employed to link to the linking probe to get much more mass change on the sensing interface.

Figure 3 shows the response curves of the QCM-based sensor to different concentrations of Hg²⁺. As shown in Figure 3(A), without linking probe, no frequency change (Δf) can be observed. In Figure 3(B), an obvious Δf can be detected due to the use of linking probe 1. When linking probe 2 was used instead of linking probe 1, the response signal, Δf, increases greatly, about 15.2-fold under the same Hg²⁺ concentration, as shown in Figure 3(C). Since a large mass of Au-NP was immobilized on the linking probe, an amount of Au-NPs were loaded on the sensing interface through the formation of the sandwich structure, which led to more mass change on the sensing interface, as shown in Figure 1(C). Thus, the detected signal was greatly amplified by the Au-NPs.

The sensitivity of the QCM-based sensor is dependent on the amount of sandwich structures formed under the same Hg²⁺ concentration, which depends in turn on the surface density of capturing probe molecules on the sensor surface. The dependence of the sensor signal on the surface density of capturing probe in same Hg²⁺ solution is depicted in Figure 4, which shows that a low sensor sensitivity was obtained when a low density of capturing probe was used. The sensitivity of the sensor increases with the increase of density of the capturing probe. When the density reached about 3.25 nmol·cm⁻², the response signal reaches the maximum, which is ascribed to the saturation of immobilized capturing probe on the sensor surface (the surface density was calculated on the basis of the Δf measured during QCM experiments) [36,37]. The surface density of capturing probe molecules on the sensor surface was also calculated to be 3.75 × 10¹⁴ molecules·cm⁻² by Tarlov's electrochemical method [31]. This value is less than 3.25 nmol·cm⁻² (1.96 × 10¹⁵ molecules·cm⁻²) by the QCM method. The surface density of probe calculated by QCM may also contain contributions from the thimbleful of adsorbed water on the sensor surface. Compared with other reports employing hairpin aptamer ((6.57 ± 0.55) × 10¹³ molecules·cm⁻²) [38], the surface density of our sensor is larger, which is likely attributable to the use of short apatmers (only containing four thymines). The high surface density should greatly improve the sensitivity of the sensor and significantly reduce nonspecific adsorption and contaminant trapping.

As shown in Figure 3(C), with the increase of Hg²⁺ concentration, the response signal of the sensor (Δf) obviously increases. As a result, the relationship between the response signal and the Hg²⁺ concentration can be established. Under the optimal conditions in which linking probe 2 was used, a linear relationship between the response signal Δf and Hg²⁺ concentrations was obtained (shown in an inset of Figure 5). The Δf increased linearly with Hg²⁺ concentrations in the tested range from 0.5 nM to 100 nM. According to the experimental data, a linear equation could be calculated: Δf = 1.82C + 1.60 (C is the concentration of the target Hg²⁺, 10⁻⁹ M; Δf is the signal of sensor, Hz) with a relationship coefficient of 0.9986. The detection limit was evaluated to be 0.24 ± 0.06 nM. Under the same conditions expect for the using of Au-NPs as amplifier, in which linking probe 1 was used, the detection limit was about 0.3 μM. The detection limit for Hg²⁺ detection was increased remarkably, about three orders of magnitude, by using Au-NPs as amplifier. Compared with data reported in the literature for other aptamer-based Hg²⁺ sensors, the detection limit in this work was lower. Table 1 summarizes the sensitivity reported for several aptamer-based Hg²⁺ sensors. The present Au-NPs amplification strategy reveals a superior sensitivity as compared to protocols of aptamer-based Hg²⁺ sensors. It is clear that the detection limit is dependent on the sensitivity of assay and the stability of detecting signal. The lower detection limit suggests that both sensitivity and stability of the sensor have improved when Au-NP was used as the amplifier.

In the present work, a short ss-DNA containing only four thymine (T) bases was designed as an aptamer to selectively capture Hg²⁺, which can selectively and strongly bind Hg²⁺ and form a stable ds-DNA through metal-mediated T-Hg²⁺-T structure. To investigate the specificity of the designed QCM-based sensor, we also challenged this sensor with nine interference ions: Ca²⁺ and Mg²⁺ (0.5 mM, each), Cu²⁺, Cd²⁺, Pb²⁺, Zn²⁺, Ni²⁺ and Ag⁺(0.5 μM, each), Au³⁺(0.1 μM). Figure 6 depicts the signal response of the sensor to the above mentioned ions. Compared with the significant frequency change (12 Hz) of 5 nM Hg²⁺, the 100-fold or higher concentrations other interference ions, except for Au³⁺, only produced a slight frequency change (<2 Hz), suggesting that the present sensor has highly selectivity owing to the specific coordination between T bases and Hg²⁺ ions. However, the interference of Au³⁺ was noted to be relatively great than that of other common metal ions.

Compared to antibody-based affinity biosensors, aptamer-based sensors provide an advantage of chemical stability [11]. The chemical stability of aptamer ensures that the sensor can be regenerated under proper conditions. In this work, the sensor was regenerated with 50 mM cysteine solution, because the sandwich structure of T-Hg²⁺-T can be disrupted by cysteine. The sensor was firstly challenged with 5 nM Hg²⁺ to obtain a response signal. Then, the sensor was washed with a flowing cysteine solution and the signal of blank solution was determined again. The above procedure was repeated continuously for 10 times and the results were presented in Figure 7.

The data suggested that the response signal to the same concentration Hg²⁺ only attenuates about 5.7% after 10 cycles and the blank signal almost remains unchanged, implying that the sensor can be regenerated and reused at least 10 times. In addition, when the sensor was stored in pH 7.4 Tris-HCl buffer at 4 °C, 97% and 91% response signal are retained after 7 days and 15 days, respectively. Linking probe 2 is also stable within 15 days when it is stored in pH 7.4 Tris-HCl buffer at −10 °C.

To validate the application of our sensor in real water samples, it was used to determine the levels of Hg²⁺ in waste and tap water samples. Prior to water sample analysis, the waste water samples were filtered through a 0.45 μm membrane. As described in the earlier section, 50 μL of waste water or tap water spiked with different concentration of Hg²⁺ was injected into the QCM cell. The concentration of Hg²⁺ was determined by applying the standard curve method and the results were summarized in Table 2. It can be seen that mean recoveries range between 96.8% and 101.6% with relative standard deviations under 4.2%. The results indicate that the QCM-based sensor was highly accurate, precise, and reproducible, and can be used for the direct analysis of traces of mercury in real samples.