PDMS (Sylgard 184) was purchased from Dow Corning (Midland, MI, USA). Chloroauric acid trihydrate (HAuCl₄·3H₂O), HMIs and 11-mercaptoundecanoic acid (MUA, Figure 2) were obtained from Sigma (St. Louis, MO, USA). Deionized water (18.2 MΩ) produced by the Milli-Q system was used throughout the experiments.

Fifteen (15) nm Au nanoparticles (AuNPs) were prepared by citrate reduction of HAuCl₄ which is similar to Grabar's method [20]. 1% trisodium citrate solution (4 mL) was added to a boiling solution of HAuCl₄ (99 mL deionized water and 1 mL 1% HAuCl₄). The mixture was kept boiling and stirring for about 30 min until the color of the aqueous change from yellow to red. After that, the solution was cooled to room temperature while being stirred continuously and then the prepared AuNPs was stored at 4 °C.

The modification of AuNPs with 11-mercaptoundecanoc acid (MUA-AuNPs) was carried out mainly as reported [18] with some changes as follows: aqueous solution (500 μL) consisting of MUA (2.4 mM) and an equivalent amount of sodium hydroxide (2.4 mM) were added to 15 nm Au nanoparticle suspension (500 μL). The mixed solution was stirred with the speed of 450 rpm in 80 °C for 1 h on the Thermo mixer (Eppendorf, Hamburg, Germany). After cooling down to room temperature, the mixture was centrifuged twice (10,000 rpm, 10 mins, 4 °C) and the supernatant was replaced with deionized water.

Microfluidic chips with Y shape and zigzag microchannels were firstly fabricated according to standard photolithographic methods [21]. Then a negative master was prepared on the silicon wafer by SU-8 photoresist, and a plasma etcher was used to realize the passivation of the master. The prepared master was placed in a glass bottom dish. After that, PDMS prepolymer (10:1 v/v mixture) was degassed and cast onto the master. After heating for 2 h at 80 °C, the PDMS was removed from the substrate and the holes punched with corresponding metal pipes. We used a flat PDMS slab (3 mm thick) as the substrate that bonding with the prepared PDMS layer to form the channels for chemical reaction. The dimensions of the finished microchannels in the chip are 100 μm (width) × 30 μm (height), and the whole chip are 3.5 cm (width) × 5 cm (length).

After degassing at 10 kPa for 1 h, the microfluidic chip was immediately sealed with adhesive tape except for the inlet reservoirs to be ready for Pb²⁺ detection. The MUA modified AuNPs and different test samples were added into two inlet reservoirs using micropipette.

Our AuNP-based HMI sensor was based on the colour change caused by the surface plasmon resonance effect on AuNP aggregation. It is known that normal AuNPs with diameter between 10 nm to 50 nm appear red in colour in water [22,23], as we can see from the tube 1 in Figure 3, when Pb²⁺ was added, chelate interaction with the carboxylate groups of MUA-AuNPs occurred, which shortens the distance between the gold nanoparticles and formed aggregates. Owing to the inter-particle coupled plasmon excitons in the aggregated states, a red-to-purple color change can be observed in the solutions. As shown in Figure 3, the colorimetric response to Pb²⁺ resulted in an obvious color change in the number two and three tubes. We challenged the strategy with other metal ions, including Mn²⁺ and Zn²⁺, and the experimental results demonstrated that Pb²⁺ had a much stronger signal (data not shown). Yet to be improved, an obvious megascopic response can only be obtained with concentrations higher than 0.025 mM in solution.

PDMS has been widely used in microfluidic systems for their optical transparency, low toxicity and ease of fabrication properties [24,25]. More importantly, with its high gas solubility, PDMS was demonstrated to be a great potential material for power-free microchips [26]. The equilibrium concentration of gas dissolved in PDMS is directly proportional to the local gas pressure around the PDMS. Accordingly, when a vacuum degassed PDMS device returns to the atmosphere, it will absorb air to establish new state of equilibrium which will automatically cause a negative pressure in the microfluidics channels. This is the basic principle of a power-free PDMS device. Our PDMS chip was fabricated with two Y-shaped zigzag micro-channels, and the two inlets were at the equal position of the upper corners of Y shape [19]. Zigzag shaped channels were designed to favour solution mixing. The PDMS microfluidic chip was firstly degassed in an airtight vacuum desiccator for 1 h, then 3 μL of MUA-AuNPs solution and 3 μL of different concentrations of Pb²⁺ (or deionized water, unknown samples) were added into each inlet reservoir. Drawn by the negative pressure in the channels, the droplets from two inlets simultaneously flowed to the junction and mixed together in the channels. Since the passages have twists and turns these enhanced the degree of mixing, and MUA-AuNPs and Pb²⁺ reacted sufficiently. Carboxylate groups on the surface of MUA-AuNPs chelated Pb²⁺, and made the nanoparticles aggregate. Gradually, a dark line can be observed using microscope in the presence of Pb²⁺ whereas no line can be seen without Pb²⁺. As we can see from Figure 4, the Pb²⁺ solution with a concentration as low as 10 μM could form a very clear line, while the blank samples generated no visual signal (Figure 4A). The time for solution flowing from inlet to outlet reservoir was less than 15 min, which is a remarkable advantage for real use. It was demonstrated to be a simple, fast and feasible method which showed great superiority of the power-free microfluidic device.

With a high throughout microfluidic chip that include eight microchannels, different concentrations of HMIs can be detected simultaneously. The process of sample injection was the same with the operation of the single microchannel microfluidic chip. The same analysis effect was achieved even on a multi-microchannel microfluidic chip. As shown in Figure 5, the channels numbered from 5 to 8 that contained Pb²⁺ displayed obvious dark lines. In contrast, the control ions, such as Na⁺, Ca²⁺ and Mg²⁺ did not show any aggregation (lines 1–4 in Figure 5).

Interestingly, just with the help of a drop of water as a magnifier, the lines can be seen with the naked eye. As shown in Figure 6, the microchannels can be identified clearly under a normal camera. Therefore we can read out the results easily due to the visible aggregation lines, which is quite important for the field testing.