Detection of heavy metal ions using whispering gallery mode lasing in functionalized liquid crystal microdroplets

: We demonstrate a detection method for heavy metal (HM) ions based on whispering gallery mode (WGM) lasing in a liquid crystal (LC) microdroplet biosensor. By doping with stearic acid, nematic LC 4-cyano-4’-pentylbiphenyl (5CB) microdroplets are biochemically functionalized and used as both optical microresonators and sensing elements. Typical WGM lasing emission is observed in stearic acid-doped 5CB microdroplets under a pulse laser pump. Our results show that quantitative spectral shift of WGMs can serve as a real-time indicator of the adsorption of HM ions on the microdroplet surface. The detection limit of our sensor is as low as 40 pM for Cu(II) ions, six orders of magnitude better than the exposure threshold deﬁned by the World Health Organization. Furthermore, this sensing system has an ability to discriminate between heavy and light metal ions. We believe that this novel biosensor has great application potential for environmental monitoring and drinking water quality testing.


Introduction
Heavy metals (HMs), such as mercury (Hg), copper (Cu), cadmium (Cd) or lead (Pd), can be discharged into air, water and soil environments in a variety of ways [1]. Due to their non-degradability, HM ions can accumulate in the human body and can cause serious health problems. The toxicity and carcinogenicity of HM ions have been widely reported [2,3]. For example, Cu(II) ions, one of the HM ions discovered by human beings, excessive intake can lead to malignant diseases such as gastrointestinal bleeding, liver poisoning, and acute renal failure [4]. Therefore, many regulatory agencies, including the World Health Organization (WHO), have formulated limits for HM ions in order to protect human safety and health [5].
Many important techniques have been reported for HM ions detection, including atomic absorption spectroscopy [6], cold vapor atomic fluorescence spectroscopy [7], inductively coupled plasma mass spectrometry [8], and X-ray fluorescence spectrometry [9]. Although these methods can effectively monitor HM ions, there are inherent limitations in each of them such as low sensitivity, expensive instruments or complex sample preparation. In addition, the microfluidic / optofluidic platforms-based HM ions detection method has also been developed with the advantages of real-time and multiplexed detection [10][11][12][13]. The methods and techniques for HM ions detection are still under investigation. Recently, liquid crystals (LCs) have been considered as promising sensing materials because the orientation and long-range arrangement of optically anisotropic LC molecules are sensitive to the chemical composition of their surrounding environment and, thus, molecular events at the LC interface can be converted into visible optical signals. A variety of LC-based detection strategies have been established for biochemical molecules such as DNAs [14,15], proteins [16][17][18][19], lipids [20,21] and artificial polymers [22,23].
Previous studies have demonstrated micromolar detection of Cu(II) ions using urease immobilization techniques in LC-based sensors [24]. In addition, by target-induced DNA conformational change, a label-free LC biosensor for nanomolar Hg(II) ion detection has also been successfully designed [25]. These studies represent excellent pioneering works in the field of HMs detection using LC materials, but high detection limits or complex sample preparation restricts the practicality of these sensors. On the other hand, it has been demonstrated that the sensitivity of sensing elements based on LC microdroplets can be as high as 1 pg/mL (∼0.59 pM) for lipid A [26]. Such ultra-high sensitivity is attributed to the interaction between the target molecules and point topological defects in microdroplets. Though LC microdroplets provide a new strategy for ultra-sensitive biosensors, the existing readout methods for microdroplet sensors are still inaccurate. Conventional polarized pattern methods [27,28] rely on the naked eye for rough optical texture recognition, which readily leads to false-positive judgments. Therefore, a method to make the detection processes and results more quantitative and accurate while maintaining (or improving) the ultra-high sensitivity of LC microdroplets is clearly merited.
Herein, we recommend using spectral responses in functionalized LC microdroplets for more sensitive, quantitative and accurate detection. Light is trapped in an optical microcavity with refractive index higher than that of the surrounding medium due to total internal reflection to form WGMs [29]. WGMs are optical eigenmodes in microresonator with circular symmetry, such as microdroplets, microspheres and microtoroids [30][31][32]. The resonant frequency (wavelength) depends on the refractive index profile of the closed light path of the reflected light, so any adsorption behaviors on the microcavity are reflected in the WGM spectra in real-time. Highly sensitive detection of biomolecules such as DNAs [33,34], sugars [35], and even single molecule biological proteins (eg, interleukin-2, protein G and streptavidin) [36,37]. has been implemented by monitoring the resonant frequency shift of the WGM. In particular, the idea of using changes in the WGM spectra of LC droplets for chemical sensing applications was first introduced by Humar and Musevic in the context of surfactant sensing [30].
In this paper, stearic acid-doped 5CB microdroplets were used as both optical microresonators and sensing elements to detect HM ions. The active WGM cavities containing a fluorescent dye, which was pumped using an external source of pulsed laser light to achieve lasing. Lasing emission spectra were observed from micron-sized 5CB microdroplets and ascribed to WGMs after comprehensive analysis. We demonstrated that the detectable shifts in WGM spectra from the microdroplets were caused by interactions between metal ions and deprotonated stearic acid. The detection limit was as low as 40 pM for Cu(II) due to the double amplification effect of 5CB molecules and WGMs. Furthermore, stearic acid molecules have different adsorption affinities for various metal ions, which make the LC microcavity has an ability to discriminate between heavy and light metal ions.

Preparation of stearic acid-doped 5CB microdroplet
A solution of 1% (v/v) 5CB in n-heptane was mixed with 0.01 wt.% stearic acid and 0.01 wt.% DCM. The mixed sample was then ultrasonically treated for 30 min to ensure adequate dissolution and uniformity. After evaporation of n-heptane, the 5CB sample was stored in a small reagent bottle for future use. Next, we fabricated a silica capillary tube into a tapered microtube with a diameter of about 7 µm using the flame tapering method, which is described in detail in our previous papers [38,39]. The other end of the tapered microtube was connected to a syringe pump, which served to dispense the 5CB sample. By adjusting the injection speed and the operating time of the pump, 5CB microdroplets with different sizes were produced in an aqueous solution ( Fig. 1(a, b)). The range of typical droplet sizes is from 40 µm to 60 µm used in the sensing experiments. The prepared microdroplets were suspended from the end of the microtube rather than connected to prevent fluctuations of the droplet size due to pressure changes in the microtube. To facilitate the excitation of WGM lasing, the LC microdroplets in all experiments were generated and controlled by a homemade microtube that can produce one microdroplet at a time. Detection of a specific HM ions concentration was conducted at least three times to avoid false-positive judgements.

Measurement of lasing emission spectra
Figure 1(a) shows the schematic diagram of the experimental setup. 5CB microdroplets were excited by a frequency-doubled pulsed laser (pulse width: 8 ns, repetition rate: 10 Hz, wavelength: 532 nm). A tapered tip drawn from standard single-mode fiber (SMF-28, Corning) was used to transmit the pump light and excite the microdroplets. Lasing emission spectra of dye-doped microdroplets were collected using a 20× (NA = 0.5) objective. The pump light was removed by a 532 nm filter. Filtered light was fed into the splitter and then delivered to a charge-coupled device (CCD, DP21, Olympus, Japan) and a spectrometer with a 0.05 nm resolution (PG2000, Ideaoptics Technology Ltd., China).

Results and discussion
3.1. Orientational behaviors of stearic acid-doped 5CB microdroplets induced by different pH solutions The orientational behaviors are referring to the orientation of molecular director within the 5CB microdroplets. In this study, we utilized stearic acid, which contains a long hydrophobic chain and a hydrophilic head group, to control the orientational behavior of 5CB molecules at the LC/aqueous solution interface. Previously, Liu et al. [40] demonstrated that stearic acid-doped hemispherical LC droplets produced on an octyltrichlorosilane (OTS)-functionalized silica substrate could be harnessed to detect enzymatic reaction between urea and urease. At the LC/ aqueous solution interface, pH increases promoted deprotonation of carboxylic acids, which then induced an orientation transition of 5CB molecules from planar to homeotropic anchoring at the droplet surface. Compared to hemispheres, intact spherical microdroplets have a larger surface area and do not require an OTS-functionalized silica substrate, so they may have higher sensitivity and different configurations under polarized optical microscope (POM). Figure 2 shows the POM images of stearic acid-doped 5CB microdroplets in PBS solution at different pH levels. Four experiments were conducted independently at each pH to ensure the reproducibility of the results. The optical appearance of the microdroplets exhibited a bipolar configuration at pH 6.5 ( Fig. 2(a)). Under these experimental conditions, the arrangement of 5CB molecules was parallel to the microdroplet surface [26]. When the pH value increased to 7.5 ( Fig. 2(b)), a disclination loop as typical transitional configuration appeared near the droplet equator. This appearance suggested that 5CB molecules gradually changed from planar to homeotropic anchoring [26]. At pH levels of 8.5 (Fig. 2(c)) or higher (data not shown), the radial configuration with a point topology defect at the center was observed, indicating that the 5CB molecules were anchored perpendicularly at the surface. In our case, the anchoring states of 5CB molecules were regulated by deprotonation of stearic acid at higher pH. Increasing the solution pH resulted in an increase of the deprotonated stearic acid concentration. The deprotonated stearic acid self-assembled at microdroplet surface due to amphiphilicity, ultimately inducing 5CB molecules to change from planar ( Fig. 3(a)) to homeotropic ( Fig. 3(b)) anchoring.

WGM lasing spectra of stearic acid-doped 5CB microdroplet
LC microdroplets have nearly perfect spherical structures and smooth surfaces, so they can be used as good WGM resonators [38,39,41,42]. The difference in refractive indexes between LC microdroplets and the immersion liquid prompts the fluorescence emitted by the dye to be trapped in microdroplets in the form of WGMs. There are size-dependent optical characteristics in excited WGM lasing. In principle, the free spectral range (FSR) of WGM decreases with the diameter (D) increase of the resonators. In spherical resonators, FSR can be calculated by FSR = λ 2 /(π n eff D), where λ is the resonant wavelength and n eff is the effective refractive index of the resonator [43]. Figure 4(a) shows a series of lasing emission spectra of stearic acid-doped 5CB microdroplets with diameters ranging from 40 to 60 µm. In this experiment, the pH of the immersion liquid is 6.5. The microdroplet's diameter is determined from the microscope image. The microdroplets were briefly excited (∼1 second) for spectral recording to avoid degradation of the dye molecules and thermal effects. Results clearly revealed a linear dependence of FSR on 1/D (Fig. 4(b)), supporting the WGM lasing mechanism. For bipolar 5CB microdroplets, only a set of transverse electric (TE) modes was observed. In a bipolar 5CB microdroplet resonator, the TE polarization is parallel to the microdroplet surface and senses an extraordinary refractive index of the LC molecules (n e =1.71), while the transverse magnetic (TM) polarization is perpendicular to surface and senses an ordinary refractive index (n o =1.54). Therefore, TE modes have a greater refractive index contrast relative to TM modes, which results in their higher Q-factors. The lasing modes number m of 5CB microdroplets can be calculated from the following equation [44] λ −1 (R, n 1 , n r , q, m Where λ is the resonant wavelength, R is the microcavity radius, q is the radial order and q = 1 in our calculations, n 1 is the refractive index of the microcavity, n r = n 1 / n 2 where n 2 is the refractive index of the surrounding medium, L = 1/q r for TM modes and L = q r for TE modes, and α(q) is the roots of the Airy function. For the TE modes in bipolar 5CB microdroplets, n 1 = 1.71 and n 2 = 1.33. The lasing peaks of stearic acid doped-5CB microdroplet with 60µm diameter were in good agreement with first-order TE modes from 539 to 549 as shown in the top half of Fig. 5(a). High pH-induced deprotonated stearic acid molecules can self-assemble at the LC/aqueous solution interface, inducing a change in anchoring behavior of 5CB molecules from planar to homeotropic. In this experiment, the droplets were generated in a PBS solution with pH = 8.5 and there was no change in the pH of the solution. Figure 5(b) shows that 5CB microdroplets change from a bipolar to a radial configuration within 45 s in PBS solution. Interestingly, the disclination loop configuration existed in a short time (approximately a few tenths of a second, data not shown) compared with the other two transition configurations (i.e., pre-radial configuration and escape-radial configuration). A blueshift of 2.45 nm was also observed correspondingly in spectra due to the decrease of refractive index perceived by TE modes during the evolution process ( Fig. 5(a)). Therefore, the WGM lasing can be used as an effective indicator of the structure evolution of stearic acid-doped 5CB microdroplets; it also provides a quantitative method for detection of foreign molecules.
It is worth noting that the contact between microdroplet and microtube has an effect on the symmetry and the smoothness of the resonator surface. The Q-factor of optical resonator is about 1000 ∼ 2200 in our work (calculated by Q = λ/∆λ, where λ is the central wavelength and ∆λ is linewidth of the lasing peak). Compared with the Q-factor of LC microdroplets suspended independently in aqueous solution or polymer, the Q-factor is about one order of magnitude lower at the same diameter in our work [30,42,45]. Therefore, some detailed phenomena are not been observed in our experiments, such as frequency splitting and spectral deterioration.

Spectral response of stearic acid-doped 5CB microdroplets induced by Cu(II) ions
LCs have been successfully harnessed as sensing materials to detect various biological macromolecules with high sensitivity. However, few LC sensors have been proposed for HM ions detection [24,25]. The main obstacle is that the low concentration of metal ions cannot effectively interfere with the orientation of LC due to the small interaction force. Similar problems exist in WGM-based sensors. Despite the complicated surface preconditioning of microresonators, detection limits have been limited to the nanomolar scale [46,47]. On the other hand, using a thermodynamic-and-electrochemical model, Hyun et al. showed that a stearic acid monolayer has different affinities for different metal ions [48]. This makes it possible to employ stearic acid doped-LC materials to detect HM ions with sensitivity and selectivity. However, the conventional polarized pattern method relies on the naked eye for rough pattern recognition, and artificial false-positive judgments are easy to be made. Real-time monitoring of the reaction process by WGM lasing is a feasible way to make the results more quantitative and more accurate.
We hypothesized that the electrostatic interaction between HM ions and negatively charged carboxylate head groups of stearic acid can interfere with the anchoring orientation of 5CB molecules (Fig. 3(c)). To prove this, Cu(II) ions were used as the HMs representative for a feasibility test. In the sensing application of WGMs, the sensitivity of the sensor strongly depends on the Q-factor of the resonator. The larger the Q-factor of the resonator, the higher the sensitivity of the sensor accordingly. Normally, big cavities have larger Q-factor than small ones [42,44]. Due to the droplets are suspended at the end of the microtube, the large droplets (more than 60 µm in diameter) are easy to fall from the microtube. Therefore, we choose 60 µm-diameter LC microdroplets as both optical microresonators and sensing elements, which not only ensure the stability but also the sensitivity of the sensor. Copper(II) chloride (200 pM) and stearic acid-doped 5CB microdroplets were incubated in a PBS solution with pH = 8.5 and the evolution states are shown in Fig. 6(a). We found an obvious disclination loop appeared on the surface of the microdroplets and remained for a relatively long period of time in the slow evolution. The slower transition process provided more detailed information of the configuration change. Disclination loop configuration indicated that the 5CB molecules underwent a transition from homeotropic to slant anchoring. Next, we studied the spectral response of stearic acid-doped 5CB microdroplets during Cu(II) adsorption. A 60-µm-diameter microdroplet was produced in PBS solution (pH = 8.5) containing 200 pM of copper(II) chloride. Lasing spectra were repeatedly recorded within a specific time interval as shown in Fig. 6(b). A redshift of 1.31 nm was observed between 0 s and 200 s. It should be noted that no significant differences were recorded in polarized optical microscope patterns between 160 s and 200 s ( Fig. 6(a)); however, a spectral shift of ∼0.12 nm was observed clearly during this time period. This indicated that the 5CB molecules underwent a slight deflection that cannot be recognized by the naked eye, but this change was captured sensitively by WGM lasing. Therefore, the adsorption of metal ions can be quantitatively characterized by wavelength shift of WGM lasing, which makes the detection process more reliable.

Sensitivity for Cu(II) ions detection
In order to study the sensitivity of WGM spectral shifting to the concentration of Cu(II) ions, stearic acid doped-5CB microdroplets were produced in PBS solution (pH = 8.5) containing different concentrations of Cu(II) ions, ranging from 0 pM to 400 pM. In this step, three independent experiments were conducted at each specific concentration. In each experiment, a new microdroplet with the same diameter (60 µm) was generated. Temporal dependence of WGM wavelength shift with different Cu(II) concentrations was measured as summarized in Fig. 7(a), where the error bars were calculated according to the standard deviation of three tests for each concentration. In the presence of 400 pM Cu(II) ions, the sensing system had a rapid initial response and the wavelength shift approached saturation at about 300 s. As the concentration of Cu(II) ions decreased, the initial response rate of the system became gradually slower because fewer Cu(II) ions were adsorbed in a given time window. The initial binding kinetics of the reaction at different concentrations of Cu(II) was determined from the slopes of linear fits of data points of Fig. 7(a) recorded within the first two minutes. The results shown in Fig. 7(b) indicated that the use of different HM ions concentrations allowed for simple quantitative control of the reaction rate. The detection limit of our sensing system was about 40 pM for Cu(II) with a final wavelength shift of 0.42 nm. On the other hand, the WHO has defined 31.5 µM for Cu(II) to be a safe threshold in drinking water [5]. Obviously, the sensitivity of our sensor is much lower than that set out by WHO. We also investigated lower concentrations of Cu(II) ions such as 5 pM and 20 pM. However, we have not been able to observe stable wavelength shift signals at these reaction concentrations due to the limitation of sensor performance (data not shown). We believe that the lower Q-factor is the main factor limiting the performance of our sensor.

Spectral response to different HM ions
Previous studies have shown that stearic acid has a different affinity for different metal ions [48]. To evaluate the detection ability of our system for different HM ions, several other common HMs (Hg(II), Cd(II) and Zn(II)) and alkaline earth metals (Mg(II) and Ca(II)) were tested under the same assay conditions at concentrations of 400 pM. The test was performed in PBS solution with pH = 8.5. All experiments were carried out in triplicate and the reproducibility is shown in Fig. 8. Among the four detected HM ions, Hg(II) had the greatest system response; the remaining ions were ordered (in descending fashion) Cu(II), Cd(II) and Zn(II). For alkaline earth metals, however, the final system responses were determined at much smaller values. The maximum wavelength shift caused by Mg(II) ions at 400 pM was ∼0.21 nm, only about one-ninth of Cu(II) under this condition. We hypothesize that the difference of system response was due to different adsorption affinities of stearic acid toward metal ions. Furthermore, Na(I) and K(I) were present in the PBS solution used during these tests, and we did not observe a significant wavelength shift when the reaction environment was pure PBS solution without metal ion doping (black line in Fig. 7(a)). These results indicated that the interference of Na(I) and K(I) in PBS can almost be ignored. Thus, the sensing system has an ability to accurately discriminate between heavy and light metal ions.

Conclusion
In this study, a biosensor for real-time detection of HM ions based on WGM lasing from LC droplets doped with stearic acid has been demonstrated. Compared with the conventional polarized pattern observation of LC sensor, WGM lasing can monitor the adsorption process of HM ions more quantitatively. Due to the double amplification effect of 5CB molecules and WGMs, the detection limit was down to 40 pM for Cu(II) ions, six orders of magnitude lower than the threshold defined by WHO (31.5 µM). Moreover, on account of the different affinity of stearic acid to various metal ions, HM ions at the identical concentrations produce diverse resonance shift behaviors in WGMs. Repeated experiments have effectively avoided false-positive judgments in the detection and also demonstrated the reproducibility of the results. This WGM