1. Introduction

Monitoring of pH value plays a significant role in a variety of applications, such as biomedical, analytical chemistry and environmental sciences [1–3], etc. Hence, the design of sensitive systems to trace pH values has been receiving much interest for many years. Among numerous techniques and methods reported previously, fiber-optic pH sensors have been of great interest due to their unique merits such as safety for in vivo pH detection, miniature in size, and immunity to the electromagnetic interference [4–6].

In general, most fiber-optic pH sensors have been exploited based on the spontaneous fluorescence emission principle. The fluorescence probe molecules are usually immobilized on the surface of the optical fiber and built upon an open detection system to monitor the fluorescence emission properties (e.g., the fluorescence emission intensity or wavelength shift) under different pH values [7,8]. These systems have provided valuable insights into the pH function in biological and biomedical processes. Nevertheless, a few drawbacks need to be addressed. (i) The open system is unsuitable for monitoring hazardous, toxic, and volatile solutions because of the potential risk to the human health and environment. (ii) The probe molecules need to be immobilized on the surface of the fiber in advance, which cannot be changed to suit different detection demands. Additionally, a washing step is required when the probe molecules are photobleached. Therefore, these analytical methods make the detection processes not only complicated but time-consuming and effort-consuming, which exhibits low efficiency and severely reduces the sample throughput. (iii) To date, fluorescence-based pH sensing which is based on the change of fluorescence property is the most exploited approach that is accompanied by decreasing (turn-off) or enhancing (turn-on) mechanism in the fluorescence intensity. Nevertheless, due to spontaneous emission properties of the fluorescence, the fluorescence signal is often too weak and even buried in the background noise. In addition, the sensor may suffer unwanted effects from omnipresent external noise sources. These factors increase the probability of pseudo signal and decrease both the signal-to-noise ratio and the sensitivity.

Compared to the fluorescence sensing, lasing has the potential to boost the sensitivity due to the stimulated signal amplification inherent to the lasing process, which results in the increased signal-to-noise ratio, the narrow linewidth, and the lasing threshold [9,10]. For example, Michele Gaio and coworkers recently demonstrated high sensitivity pH sensing based on random lasing (RL) [11]. However, the proposed RL sensing system still faces several limitations. First, the lack of an optical cavity gives the RL a high lasing threshold, because of the lack of lasing mode selections and a large optical loss of inefficient lasing modes. Second, the rhodamine 6G (Rh6G) aqueous solution is employed both as the gain medium and the pH probes, which also results in a high lasing threshold due to a low quantum yield induced by self-aggregation of Rh6G dye molecules in water solution [12]. Moreover, these pH-sensitive fluorophores typically exhibit poor photobleaching tolerance in the RL system, limiting their use in the experiments for an extended period of time. In contrast, fluorescent probes like disodium fluorescein (DSF) exhibit outstanding performance, including superior water solubility, sensitivity to pH values, and high quantum yield, which have been widely investigated in a variety of chemical and biological analyses [13,14]. However, the applications of DSF as lasing gain medium and probes for pH sensing have rarely been explored. Therefore, to achieve better sensitivity with a low lasing threshold and eliminate effects of photobleaching, it is still highly desirable to develop a new optical cavity as well as highly soluble probes that present distinct changes in lasing intensity when responding to different pH conditions.

We have shown an optofluidic ring resonator (OFRR) laser based on evanescent wave excitation in our previous works [15,16], which can solve various issues toward the current fluorescence and RL system for pH sensing. In the present work, we attempt to adopt this type of OFRRs as a novel strategy for pH sensing, in which DSF aqueous solution is employed as both the gain medium and fluorescent probes. As shown in Fig. 1(a), an optical fiber (refractive index (RI) = n₁ = 1.458), serving as both the optical cavity and the sensing reactor, is integrated into a polydimethylsiloxane (PDMS, RI = 1.405) microchannel. The DSF aqueous solution (n₂ = 1.333) flows through the enclosed microchannel and serves as the cladding of the fiber. The circular cross section of the fiber forms a ring resonator that supports whispering gallery modes (WGMs) of high Q factors (>10⁶) [17,18]. The pump light propagates along the fiber axis due to the total internal reflection (TIR) at the fiber/cladding interface. The evanescent field of the pump light extends out of the fiber surface and efficiently interacts with the DSF molecules residing in the evanescent field region of the WGMs to produce a lasing emission. We choose this OFRR system for pH sensing for several reasons. First, the gain medium solution used for probes is enclosed in a microfluidic channel, which provides an ideal platform for the hazardous, toxic, and volatile solutions detection. Second, the OFRRs have successfully removed unintended fluorescence background from its optical signal, as all the gain molecules in close proximity to the fiber/cladding interface contribute to the laser phenomena, which is favorable for low lasing threshold and highly sensitive sensing. Third, the probe molecules are dissolved in water and flow through the microfluidic channel rather than being immobilized on the surface of the fiber. Therefore, the OFRR system not only provides a powerful platform for label-free sensing but significantly increases the sample detection efficiency and throughput but eliminating the effects of photobleaching. More importantly, by monitoring the dynamic changes of the lasing properties due to different pH conditions at the same position, real-time and in situ sensing can be achieved.

 

Fig. 1. Schematic diagram of the chip design and the experimental setup (a) The schematic design of the PDMS chip. The dimension of the microfluidic channel is 5mm × 300 μm × 300 μm (length × width × height). The inset on the right shows the cross section of the OFRR, which is used to illustrate the principle of the OFRR lasing generation and sensing. (b) Schematic of the experimental setup. Optical parametric oscillator (OPO) laser: pulse width: 5 ns, repetition rate: 20 Hz, output wavelength: 488 nm. The focal length of the lens is 75 mm.

Download Full Size | PDF

In this paper, we first demonstrate the lasing emission characteristics (i.e., linewidth, threshold, and polarization) of the OFRRs with DSF aqueous solution as the gain medium. Then, we experimentally and theoretically characterize the lasing threshold and emission intensities under different pH conditions to evaluate the sensing performance. Finally, we showcase the highly sensitive pH sensing by using the proposed OFRRs laser.

2. Experimental section

Figure 1(a) shows the schematic design of the chip, in which an optical fiber (diameter D =200 µm) is integrated into a PDMS microfluidic channel. The inset of Fig. 1(a) shows the cross-section of the fiber in the microfluidic channel, which is used to illustrate the principle of the lasing generation and sensing. The microfluidic chip is fabricated by using the standard PDMS soft lithography procedures. The details of the fabrication process are described in Ref. [19]. The size of the chip is 20 mm × 10 mm × 2 mm (l × w × h), and the dimension of the microfluidic channel is 5mm × 300 μm × 300 μm. The fluidic outlet is connected to a syringe pump (PHD 2000, Harvard Apparatus) for fluidic delivery. The DSF dye dissolved in deionized (DI) water is employed as the gain medium as well as the cladding of the fiber and flowed through the microfluidic channel at a flow rate 1 µL/min. Figure 1 (b) shows the schematic of the experimental setup. The fiber is longitudinally pumped by an optical parametric oscillator (OPO) laser at 488nm (5 ns pulse width and 20 Hz repetition rate). As the RI of the optical fiber is larger than that of the surrounding cladding solution, the beam inside the fiber would propagate along the fiber axis through TIR at the fiber/cladding interface. The evanescent field of the pump light extends out of the fiber surface and efficiently excites the dye molecules residing in the evanescent field region of the WGMs to produce a lasing emission. The emission light is collected along the X-axis through a multimode fiber and subsequently transmitted to a spectrometer (Spectrapro 500i) mounted with an ICCD detector (PI-Max 1024RB). In the experiment, the concentration of DSF dye is fixed at 0.5 mM. The pH value of the aqueous solution is adjusted by adding NaOH or HCl solution to DI water. A calibrated pH meter (FE28, Mettler Toledo) is utilized to measure the pH values. The detection range of the pH meter is from 0 to 14 with a pH-relative accuracy of ±0.01. All measurements in the experiment are conducted at room temperature (25 °C).

3. Results and discussion

To verify the possibility of lasing generation with DSF aqueous solution (pH = 6.51) as gain cladding, we first explore the lasing emission intensity and the full width at half maximum (FWHM) varied with the pump energy density (PED) by utilizing a low-resolution grating (grating density = 150 g/mm). As shown in Fig. 2(a), the turn-on process of the lasing emission can be observed by recording the emission spectrum while increasing the PED. A sudden increase in the emitted light intensity is observed when the PED reaches 2.62 μJ/ mm² (i.e., lasing threshold, marked by a green dotted line). After the threshold, stimulated emission becomes the dominant emission process and WGM lasing turns on. The change of emission regime is presented by both an FWHM narrowing (blue full spheres) and an increase of lasing emission intensity (black full squares). Figure 2(b) shows the transition spectra below the threshold (1.0 μJ/mm²), and above the threshold (4.0 μJ/mm²) with a 150 g/mm grating. When the pump power is below the threshold, only a broadband spectrum and a weak fluorescence emission (inset) are observed around the fiber in the microfluidic channel.However, the fluorescence spectrum changes to a sharp peak and a strong green light (inset) emitting from the rim of the fiber is observed in the vertical direction of the fiber axis. Additionally, because the lasing wavelength depends on the interplay between loss and gain, a small wavelength shift (from 520.7 to 522.8 nm) of the emission spectra upon increasing PED is observed, which is typical for WGM lasing. Figure 2(c) shows the high-resolution lasing spectrum (I_p= 4.0 μJ/mm²) with a 2400 g/mm grating. The mean spacing between the adjacent sharp peaks is about 0.29 nm, corresponding to the theoretical spacing ∼λ²/(2πn₁a) of WGM lasing, where λ ( = 521.6 nm), n₁ ( = 1.458), and a ( = 100 μm) are the lasing central wavelength, the RI of the fiber, and the fiber radius, respectively. In addition, polarization analysis (Fig. 2 (d)) shows that the lasing spectrum is a typical transverse electric (TE) wave lasing emission [16,20], i.e., the electric field vector of the lasing emission is perpendicular to the fiber axis, as shown in the inset of Fig. 2(d). All these are powerful evidence of WGM lasing emission with DSF aqueous solution as the gain medium.

To assess the feasibility of DSF as the sensing probes, we further explore the lasing emission intensity and the FWHM varied with the pH value of the DSF aqueous solution under a constant PED (∼3.0 μJ/mm²). As illustrated in Fig. 3(a), when the pH value is below the pKa (∼6.4) of DSF aqueous solution [21], the lasing emission turns off and only a broadband fluorescence spectrum is collected. However, a sharp peak exhibits distinct pH-dependent intensity changes when the pH value is higher than the pKa. Simultaneously, the FWHM of the spectra shows a nonlinear transition at around pH = 8, from ≈ 48 nm FWHM at pH 6.18 to 8-10 nm FWHM at pH = 12.30, as shown in Fig. 3(b). The inset of Fig. 3(b) shows that the DSF aqueous solution undergoes a color transition from dark to bright with the increasing pHs.

To better understand the lasing emission properties caused by the pHs, the high-resolution lasing spectra are collected when the pH is higher than the pKa. As shown in Fig. 3(c), the lasing wavelengths are restricted in the range of 517-528 nm with increasing lasing emission intensity but no significant emission spectral shift is observed. Figure 3(d) exhibits the integrated lasing intensity over the entire lasing spectrum (517-528 nm) as a function of the pH value. As shown in Fig. 3(d), when the pH value is lower than 8.13, the relationship between the lasing intensity and the pH value is linear. However, the lasing intensity increases slowly and gradually approaches a saturated value beyond pH = 8.13, and the relationship between the lasing emission intensity and the pH value is nonlinear in this region. Furthermore, to study the reversibility and repeatability of the pH sensing, a solution of pH 12.30 is prepared first and gradually added to a pH 6.51 solution to obtain a series of pHs between 6.51 and 12.30. Afterwards, to decrease the pH, the pH 6.51 solution is gradually added to the pH 12.30 solution to obtain the desired pH value of 12.30 - 6.51. Figure 3(d) shows that the lasing intensity variation under different pHs from 6.51 to 12.30 after 3 times of testing with ascending and descending order are repeatable and reversible.

 

Fig. 2. (a) Lasing peak intensity (full black squares) and linewidth (full blue spheres) varied with pump energy density. The pink rectangle illustrates the lasing region above lasing threshold (2.62 μJ/mm²). (b) Comparison of spectra above and below the threshold collected by a low-resolution grating (150 g/mm). Inset: weak fluorescence emission below threshold, and strong green lasing emission above threshold around the rim of the fiber, respectively. (c) High-resolution lasing spectrum (I_p= 4.0 μJ/mm²) collected by a 2400 g/mm grating. (d) Dependence of lasing output intensity on the polarization angle (θ). The red spheres are measured data, and the solid black curve is drawn by cos²θ. Inset: Scheme of the polarization measurements. The red arrows label the angle between the polarizer and the fiber axis, which is zero when the polarization direction of the polarizer is along the fiber axis. Error bars are obtained with five measurements.

Download Full Size | PDF

 

Fig. 3. (a) Emission spectra under different pH values of the cladding solution. (b) The FWHM of the emission spectra vs the pH values of the cladding solution. Inset: DSF aqueous solution with different pH conditions. (c) High-resolution lasing spectra for five representative pHs of 6.51, 7.40, 8.13, 10.02, and 12.30, respectively. (d) The reversibility of lasing emission intensities varied with the pHs. Error bars are obtained with five measurements.

Download Full Size | PDF

We consider that the pH dependence shown in Fig. 3 is mainly attributed to a change in the gain medium, as induced by deprotonation or protonation processes rather than a structural modification [22], i.e., when the pH value is close to or surpasses the pKa, which induces the deprotonation process drastically increasing, resulting in a fast turn-on of lasing emission in the pH range 6.51-8.13. While the deprotonation process is nearly completed when the pH further increases up to 12.30, then the lasing emission intensity approaches a saturated value. On the contrary, when the pH value decreases, the increasing H⁺ ion will induce the protonation process, which will lead the lasing emission intensity to shift back and gradually turn off.

The variation of lasing emission intensity under different pHs is directly reflected by the difference of the lasing threshold. Figure 4(a) plots the integrated lasing intensity under different pHs as a function of PED, which indicates that the lasing threshold decreases with the increase of the pH. Therefore, to further study the lasing capability and performance, Fig. 4(b) quantitatively presents the evolution of the lasing threshold with various pHs, which shows that the lasing threshold is drastically reduced about 2.7-fold in the pH range 6.51- 8.13, resulting in a large slope in the lasing threshold due to the rapidly increasing deprotonation of acid groups. However, the deprotonation of the acidic groups is completed when the pH values further increase up to pH 12.30, thereby the lasing threshold decreases slowly and gradually approach a saturated value. Accordingly, the slope change of the lasing threshold decreases.

 

Fig. 4. (a) The integrated lasing emission intensities under different pH conditions as a function of PED. The solid line denotes a linear fit above the threshold. The lasing threshold is about 2.68 μJ/mm² (pH 6.51), 2.07 μJ/mm² (pH 7.40), 0.98 μJ/mm² (pH 8.13), 0.86 μJ/mm² (pH 9.17), 0.83 μJ/mm² (pH 10.02), 0.76 μJ/mm² (pH 11.19), and 0.79 μJ/mm² (pH 12.30), respectively. (b) Lasing threshold varied with the pHs. Error bars are obtained with five measurements.

Download Full Size | PDF

To further understand the lasing threshold at different pH conditions, a theoretical calculation is performed in this section. According to the laser theory, the OFRR lasing threshold, I_th, is expressed as [23,24]

A change in any of the parameters in Eq. (2) will affect the lasing threshold. Hence, to further understand how the deprotonation processes affect the lasing threshold behaviors, we carried out the measurements of absorption (σ_a), fluorescence lifetime (τ), and fluorescence quantum yield (Φ) at different pHs starting from 6.51 to 12.30 (see details in the SI). As seen from Fig. 5, in the pH range 6.51-8.13, σ_a, τ, and Φ increase concomitantly and are linearly proportional to the pH value due to deprotonation process. However, the increase rate of σ_a and τ on the pH value is far less than that of Φ. In Figs. 5(a) and (b), σ_aincreases from 5.31×10⁻¹⁸ (Q_abs = 3×10⁶) to 6.72×10⁻¹⁸ cm² (Q_abs = 2.4×10⁶) and τ increases from 4.92 to 5.23 ns (see details in the SI), which are ∼1.2-fold and 1.1-fold, respectively. In contrast, the quantum yield Φ (Fig. 5(c)) exhibits very distinct pH dependence and increases from 0.29 to 0.73, which is as high as ∼ 2.5-fold and results in a significant increase of σ_e(λ) from 3.83×10⁻¹⁷ to 1.08×10⁻¹⁶ cm² at the lasing wavelength 523 nm. Accordingly, the lasing threshold is drastically reduced in this pH range. Beyond pH 8.13, σ_a, τ, and Φ simultaneously undergo a small increase and eventually level off, then the lasing threshold gradually approaches a saturated value. As demonstrated in Fig. 6, the normalized experimental and theoretical threshold behaviors show very good qualitative agreement. From these measurements, we conclude that it is the quantum yield Φ that is mainly responsible for the lasing threshold observed in Fig. 4, i.e., when the pH value is in the range 6.51-8.13, the deprotonation process facilitates the rapid increase of Φ, resulting in a sharp decrease of the lasing threshold. Beyond pH 8.13, the deprotonation process is gradually completed, then the quantum yield Φ increases slowly and show certain degree of saturation. Accordingly, the lasing threshold tapers off and further approaches its saturated value.

Based on the analysis above, we fully convince that the DSF dye possess the potential to be the pH probes for high sensitive sensing by employing the proposed OFRR lasing. The integrated lasing spectral intensity (I) at different pH conditions is recorded, which is used to determine the evolution of the light intensity as a function of the pH value. As shown by red spheres in Fig. 7, the experimental result is apparently divided into two parts. When the pH increases from 6.51 to 8.13, the relationship between the light intensity and the pH value is linear (R² = 0.97), with a remarkable increase intensity ∼ 40-fold. However, when the pH value is higher than 8.13, the light intensity increases slowly and gradually levels off and the relationship between the light intensity and the pH value is nonlinear in this region.

The sensitivity of the OFRR lasing can be defined as [11]

To obtain the sensitivity, we perform a linear fitting to the evolution of light intensity in the pH range 6.51-8.13. Based on Eq. (3), we acquire that the sensitivity (S_pH) is as high as 178 at pH = 6.51. Meanwhile, the limit of detection (LOD) is ∼ 0.01pH, defined as the ratio of 3 times of the standard deviation of the lasing emission intensity to that of the slope of the fitting line.

 

Fig. 5. DSF properties as a function of pH value. (a) Absorption (σ_a), (b) fluorescence lifetimes (τ), and (c) quantum efficiency (Φ) as a function of pH.

Download Full Size | PDF

 

Fig. 6. Comparison of the normalized lasing threshold of experiment and calculation. The black squares are the experimental results and the red solid line is the calculation results, respectively. Error bars are obtained by five repeated measurements.

Download Full Size | PDF

 

Fig. 7. Normalized intensity for both lasing and fluorescence at various pH values. The inset figure shows the linear relationship between the light intensity and pHs. Error bars are obtained by five repeated measurements.

Download Full Size | PDF

As a comparison, we also collect the fluorescence intensity under different pH conditions, in which a continuous-wave (CW) laser at 488 nm is chosen as the exciting source. As shown by black squares in Fig. 7, the obtained results very similar to the previous one in the lasing experiment, i.e., when the pH increases from 6.51 to 8.13, the relationship between the light intensity and the pH value is also linear (R² = 0.94), but the increased intensity is just about 1.2-fold. Also, beyond pH 8.13, the light intensity increases slowly and gradually saturated. The calculated sensitivity (S_pH) is ∼ 0.8. It is obvious that the sensitivity of the lasing based sensing is remarkably improved by nearly two orders of magnitude to that of fluorescence.

Finally, to assess the effect of the dye parameters (σ_a, τ, Φ) on the pH sensitivity. Let n_ss be the number of the output photons and r = I_p/I_th. Then, the intensity of the output light In_ss. According to Eq. S10 in the SI, the intensity of the output light I can be expressed as

Through Eqs. (2)–(4), the sensitivity below and above the threshold can be calculated by

According to Eq. (5), one can calculate the sensitivity for different parameters (σ_a, τ, and Φ) and insight its role in lasing process under different pHs. It is worth noting that these parameters are typically related to each other in real cases, but in our calculation, we consider them independently to isolate their role. The linear response sensitivity for the fluorescence regime can be defined as S_α = 1. Figure 8 plots S_α for the various parameters at different pump energy densities. The color map highlights areas with a linear response (blue) and a highly nonlinear response (red, S_α >1). The lasing threshold is below the yellow to red areas. The area between white dashed lines indicates the measured range of variation of the corresponding parameter, as reported in Fig. 5. For all parameters, there are regions with increased sensitivity (S_α >1) when compared to fluorescence. According to Fgure 8, the calculated highest sensitivities are appeared in the range of fluorescence to lasing transition, with S_σa = 76, S_τ = 52, and S_Φ= 198, respectively. As expected, a pH change affects all of them. However, the quantum yield Φ is the most remarkable.

 

Fig. 8. Sensitivity analysis. The relative sensitivity defined as S_α= (dI/I)/(dα/α) is calculated for the same system parameters, when varying the value of α, for α = σ_a (a), τ (b), Φ (c), and for different pump intensities. The areas between white dashed lines indicate the measured range of variation of the corresponding parameter, as reported in Fig. 5.

Download Full Size | PDF

Finally, we also carry out a control experiment to evaluate if the sensing response of the proposed pH sensor would change when expose sequentially to different salt concentration. Figure 9 presents the fluorescence intensity and quantum yield Φ varied with the concentration of NaCl under a constant pH value (pH = 6.51), which indicates that there is no significant change in the fluorescence intensity and Φ in the salt concentration range of 0 to 4 mM. Therefore, we consider that the influence of salt concentration on the sensing ability of the proposed pH sensor can be neglected.

 

Fig. 9. Fluorescence intensity and quantum yield of DSF varied with the concentration of NaCl.

Download Full Size | PDF

4. Conclusion

In summary, we develop a chip-scale pH sensor with high sensitivity based on an evanescent wave pumped OFRR lasing platform, in which DSF dye is employed as the gain medium and the sensing probes. The exponentially decaying evanescent wave interacts with DSF dye dissolved in the cladding DI water at the fiber-cladding interface, which provides a uniform excitation scheme and a significant reduction of the lasing probe volume, maintaining a low lasing threshold and high sensitive sensing. We identify the proposed OFRR lasing sensing platform, with a 2-order-of-magnitude sensitivity enhancement to fluorescence in the pH range 6.51-8.13. We further conclude that the quantum yield Φ is mainly responsible for the lasing threshold and the high sensitivity due to the deprotonation process with the increasing pH value. Furthermore, the fiber used as a sensing reactor is simply integrated into the microfluidic channel, and thus the probe molecules are free of any complex labeling process prior to analysis. Therefore, as compared with other label-free pH sensors (such as ion-sensitive field-effect transistors (ISFETs) [25], and GaInAsP photonic crystal nanolaser [26,27]), the proposed OFRR lasing sensors have great potential in the medicine, and hazardous/toxic/volatile solutions, which require low sample consumption, real-time, and in situ dynamic detection. Additionally, it should be pointed out that the PDMS chip used is not suitable for the detection of highly corrosive chemicals (such as hydrofluoric acid) as they may cause the PDMS to deform.