Label-Free Detection of HeLa Cells Activity Excited by Blue LED

Highlights

• Label-free detection of HeLa cells’ intrinsic chemiluminescence.
• Characteristic or resonance frequency of HeLa cell light emission.
• a-Si:H p-i-n photodiodes are posted below and laterally of the analyte as a transducer.
• a-Si:H DOS responsible for weak light biological signal detection.
• a-Si:H photodiode as a transducer for label-free biosensors.

Abstract

This paper investigates a novel optical method that uses a high-responsivity a-Si:H photodiode for label-free detection of luminescence from HeLa cervical cancer cells excited by a blue LED. We examine the energy distribution of the energy-gap density of states (DOS) from the photodiode’s long-time transient current, which shows exponential decay kinetics in the HeLa cell reaction. We analysed the transient response of a-Si:H p-i-n photodiode upon the illumination of the analyte with a pulsed blue LED light to better understand the HeLa cells activity and the fundamental defect kinetics processes in the a-Si:H material. Results suggest that the characteristic very low-level, time-varying light response of HeLa cells is due to chemiluminescence within cells, resulting from the reaction between nitric oxide (NO) and hydrogen peroxide (H₂O₂). Given the low signal intensity and noise, we applied a Savitzky–Golay (SG) filter to post-process the data. By reducing noise without attenuating chemiluminescent peaks, the Savitzky–Golay filter enabled accurate, reproducible quantification of the photocurrent response, reflecting the kinetics of cellular reactions. Further studies and more precise measurement instruments are needed for this real-time, label-free, non-destructive method, which applies SG-filtered signal processing to microfluidic optical biosensors.

1. Introduction

In recent years, efforts have focused on developing microarrays and miniaturised analytical devices integrated into lab-on-a-chip (LOC) systems [1]. These devices can be implemented in several pharmaceutical [2], clinical, diagnostic [3], immunotherapy [4], and environmental applications [5]. In particular, optical techniques for detecting bioluminescence and chemiluminescence, which detect photons emitted by chemical reactions, are used in life sciences research to monitor and measure biological processes and outcomes.

In recent years, K. Manekar and M. A. Hasamnis [6] presented a microfluidic-flexible test strip integrated with a 3D-printed chemiluminescence (CL) detection platform for on-chip micromixing and glucose detection. In this system, the enzymatic reaction generates hydrogen peroxide (${\mathrm{H}}_2{\mathrm{O}}_2$), which undergoes a CL reaction for glucose quantification. Despite the rapid development of microfluidic technology, there remains a need for a method that is low in background noise, highly sensitive, fast, and simple.

The potential photodetector application of thin-film amorphous silicon as an optical transducer in a point-of-care testing (PoCT) molecular biomarker detection platform [7,8] based on microfluidics technology is often characterised by the magnitude of the mobility-lifetime ($\mathsf{\mu }\mathsf{\tau }$) product of photo-generated excess carriers [9].

The a-Si:H p-i-n photodiode has already been used for DNA fluorescence detection [10,11,12,13] and chemiluminescence detection [1,6,14,15], and as a transducer in the most promising label-free biosensors [16]. The LOC system, with a micron-sized thin-film hydrogenated amorphous silicon photodiode microfabricated on a glass substrate, exhibits high sensitivity and portability in microfluidic technology. At last, a fully integrated laser-induced [13] and more promising LED-induced [17] fluorescence detection device has also been developed for fluorescence and chemiluminescence measurements.

The a-Si:H photodiode has high quantum efficiency in the range of the visible spectrum, low dark current in comparison with the photocurrent, low temperature dependence in the range of ambient temperature [18,19], and a small contribution of peripheral current to the total reverse current at low applied voltages less than 1 V, which is used in the above-mentioned LoC systems.

Despite the relatively high defect density, band-tail defects play an important role in detecting weak light signals. Finally, advanced microfabrication technologies on glass and polymers, compatible with a-Si:H thin films, enable integration into miniaturised bioassays that rely on optical labels for signal detection [14].

Light-emitting diodes (LEDs) are increasingly used as excitation light sources in fluorescence detection systems [17,20] due to their compatibility with integrated microfluidic systems, low cost, ease of handling, and low-voltage power consumption.

There are pros and cons of blue LED light on the human body. Because it has a detrimental effect on tumour cells, blue LED light is phototoxic to some cells. To date, it has been reported that exposure to blue light from LEDs causes retinal toxicity and disrupts the circadian clock [21]. However, antitumour effects have been reported against tumour cells, including HeLa cells, malignant melanoma, colon cancer, malignant lymphoma, pancreatic cells, and leukaemia [22,23]. Blue LED light forms reactive oxygen species (ROS), including singlet oxygen, superoxide anions, hydrogen peroxide, and the hydroxyl radical [21]. It has been reported that nitric oxide synthase levels are elevated in HeLa cells [24]. Blue light affects nitrosinated proteins by breaking the bond between nitric oxide and the proteins [22].

This research aims to develop a label-free method to capture cellular responses in real time using a hydrogenated amorphous silicon (a-Si:H) photodiode, in which light serves as a convenient, non-destructive reactant.

In this study, we first examined the kinetics of the influence of light-induced defects on long-time photocurrent degradation under constant blue light and secondary light from HeLa cells. Based on the observed current amplification, the analysis of the measured photovoltage transient responses of an a-Si:H p-i-n photodiode to blue LED light pulses superimposed on the light bias from an analyte is performed. Our approach examines how the free carriers created by visible light from the analyte, via trapping–detrapping and recombination processes in the energy gap states, affect the device photocurrent or photovoltage. Based on the examined switch-on and switch-off responses by fitting the measured transient photocurrent to two exponential functions, we suggest possible mechanisms of the kinetic performance in a-Si:H as a consequence of the analyte light emission. We conclude that the amplified photocurrent/photovoltage response in our case occurs only at the frequency we identified, which corresponds to the HeLa cells’ resonance frequency. At this critical frequency, the interaction of the majority free carriers via the trapping–detrapping process with empty defect states in the energy gap of amorphous silicon can be monitored under low-bias illumination and low-bias voltage.

After a brief description of the developed method, materials used, and proposed experimental details in Section 2, the obtained results and discussions are presented in Section 3. Finally, the conclusions are given.

2. Materials and Methods

2.1. Cell Culturing

The human cell line HeLa (cervical carcinoma) was cultured as a monolayer and maintained in Dulbecco Modified Eagle Medium (DMEM, LONZA, Basel, Switzerland) supplemented with 10% foetal bovine serum (FBS, Gibco, New York, NY, USA), 2 mM L-glutamine (Gibco, USA), 100 U/mL penicillin, and 100 µg/mL streptomycin (LONZA, Basel, Switzerland), 0.25% trypsin (LONZA, Basel, Switzerland) in a humidified atmosphere with 5% $\mathrm{C}{\mathrm{O}}_2$ at 37 °C. This complemented DMEM is the cell culture medium (CCM). The HeLa cells are purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA). The concentration of cells, as well as the determination of dead cells after blue LED light irradiation, was determined by a Neubauer chamber (Glaswarenfabrik Karl Hecht, Sondheim vor der Rhön, Germany).

2.2. Measurement Set-Up

A blue LED (RGB LED Lamp, Kingbright, China) emitting at a peak wavelength of 430 nm was used as a constant and pulsed monochromatic excitation light source. The chosen LED light wavelength is 430 nm, and the pulse period is 3 ms with a 50% duty cycle. The commercial c-Si p-i-n photodiode, Osram, Austria BPW 21, is used as a control photodiode positioned laterally relative to the analyte (Figure 1). We develop a label-free method that does not immobilise HeLa cells in the culture medium (CCM) on the sensor surface.

2.2.1. Long-Time Photocurrent Measurements

In the long-time transient conduction measurement, the photodiodes from a 5 × 6 array were used. The photodiode array was fabricated at the Department of Information Engineering, Electronics and Telecommunications, Sapienza University of Rome, Italy. Further details on the photodiode array fabrication can be found in [25].

The measurement setup includes a dark metallic box where the chemiluminescence reaction takes place. Inside a plastic well positioned on the surface of the a-Si:H photodiode array, the photodiode is connected in series with a load resistor, ${\mathrm{R}}_{\mathrm{L}}$, of 10 kΩ, a DC voltage power supply (Keysight Power Supply E36313A, Penang, Malaysia), and a digital multimeter (DMM, Keysight 6 1/2 Digital Multimeter Truevolt 34461A). The current and voltage were measured using a DMM (Keysight 6 1/2 Digital Multimeter Truevolt 34461A). All instruments were connected to the PC, and the simultaneous measurement process was automated using test flow sequences created in Keysight BenchVue.

The initial photocurrent conditions were established by first exposing the photodiodes to constant white LED illumination for about 1 min. In this way, the defect density concentration initially increases, and consequently, the sensitivity to a weak biological signal increases. Then, with the plastic well on the surface, the open-circuit voltage was measured under constant blue LED illumination. The sample, containing 3 mL of solution with the appropriate concentration of HeLa cells in cell culture medium (CMM), was added using a pipette before applying the reverse-bias voltage. Subsequently, a constant reverse-bias voltage was applied to the photodiode. This process first establishes the initial point of the long-time current transient and serves as a calibration. Measurements were conducted for one hour under constant blue LED illumination, with the LED current set to 20 mA. The photodiode current was monitored for the entire hour and recorded on a PC using Keysight BenchVue software. Characteristic current curves were observed during the first 15–20 min.

This procedure is first applied to the photodiode, then to the photodiode with a plastic well on its active surface, and finally to the photodiode with the analyte, as described above and previously published for various HeLa cell concentrations [26]. Here, the results of long-time current measurements of the photodiode with HeLa cells in CCM are presented.

2.2.2. The Transient Response Measurements

In a series of experiments, a Keysight 35510B function generator (Malaysia) was used to drive a pulsed blue LED. A pulsed signal with a frequency of 100 Hz–500 Hz and an amplitude of 0–4 V was used. The DC voltage (Keysight Power Supply E36313A, Malaysia) was 0–2 V. A step voltage was applied to the blue LED, and the transient response of the photodiodes positioned below and lateral to the analyte was recorded with an oscilloscope (Keysight DSOX2014A Infini Vision 4CH 100 MHz 2 Gsa/s, Malaysia).

We placed an a-Si:H p-i-n photodiode (PD1) [25], ref. [26] below the analyte inside a dark metallic box, along with the a-Si:H p-i-n (PD2) and c-Si p-i-n photodiodes (PD3) positioned laterally to the analyte. The 10 kΩ resistor was in series with PD1, and a 1V reverse-bias voltage was applied. The 220 kΩ resistors were in series with PD2 and PD3, respectively, and both were at zero-bias voltage. Initial photocurrent conditions were established by exposing the photodiodes to constant white LED illumination for about 1 min, as was done in the long-time current experiment.

The a-Si:H p-i-n photodiodes PD1 and PD2 used in the transient response measurements were fabricated similarly to a standard solar cell, deposited on a transparent conductive oxide (TCO) coated glass as described in [27], along with the preparation conditions and detailed structure of the photodiodes following the method developed in [27,28].

The transient response measurement setup included a dark metallic box in which the chemiluminescence reaction took place at ambient atmospheric conditions and room temperature. The open circuit voltage of PD1 was measured first. Then, 3 mL of solution containing the appropriate concentration of HeLa cells in CMM was added using a pipette. The analyte was placed on a microscope slide positioned on the surface of the a-Si:H photodiode (PD1). Finally, the reverse-bias voltage was increased to 1 V to bring PD1’s initial voltage close to 0 V.

Photovoltage/photocurrent measurements of the a-Si:H and c-Si p-i-n photodiodes were carried out by exposing the analyte to pulsed blue LED light (peak wavelength 430 nm). The LED driver voltage amplitude, ${\mathrm{V}}_{\mathrm{p}-\mathrm{p}}$ was 4 V, with a period of 3 ms and a duty cycle of 50%. The photocurrent transient (rise to a quasi-steady state and decay to unilluminated dark-current level) was recorded on a digital oscilloscope (Keysight 2000 X Series, Malaysia) over several decades of time, ${10}^{-6} \mathrm{s}\le t\le 3·{10}^{-3} \mathrm{s}$. The steady-state PD1 was measured in the absence of the analyte. It was assumed that during this experimental time, all trapped charges had sufficient time to interact with the bands [29]. The total time taken to measure the transient response for each sample was one hour.

The details of the long-time transient conduction and transient response experiments are described in [26,27,28]. The procedures outlined above are initially applied to the photodiode, then to the photodiode with a plastic well or a microscopic glass on its active surface, both with and without CCM, and finally to the photodiode with the analyte, HeLa cells in CCM, as described above. The results for selected HeLa cell concentrations are presented here. It is observed that the a-Si:H p-i-n photodiodes can distinguish CCM from HeLa cells in CCM.

2.3. Theoretical Models

The defect pool model [30] describes the density of states (DOS) in amorphous silicon. It consists of exponential band tails of monovalent states and deep defect states of bivalents. The valence band-tail states are considered acceptor-like, being neutral when empty and negatively charged when occupied. The conduction band-tail states are donor-like, being neutral when occupied and positively charged when empty. The dangling-bond states have three possible charge states: positive when unoccupied, $D^{+}$, neutral when singly occupied, $D^0$, and negative when doubly occupied, $D^{-}$. Due to coulombic repulsion, the energy of the doubly occupied state is higher than that of the singly occupied state.

It is well known [31] that after exposing a-Si:H material to constant illumination when the generation rate G remains steady, the total concentration N of metastable dangling bonds (MDBs) increases with exposure time. It is assumed that the electrons determine the photoconductivity. Their retention time in the conduction band decreases as N increases. As a consequence, the photodiode current shows a well-known exponential transient decay when a reverse-bias voltage is applied under constant blue LED illumination, with and without a plastic well on the photodiode surface.

When the luminescent reaction occurs, light from HeLa cells (CL) and DMEM (fluorescence, FL) increases the photodiode current. In our experiment, light emitted by living cells has unknown intensity, wavelength, and modulation frequency. The luminescence performance of CCM and HeLa cells under continuous blue-light illumination was investigated. The data have been normalised to the photocurrent, ${\mathrm{I}}_{\mathrm{p}\mathrm{h}}\left({\mathrm{t}}_0\right)$ (Figure 2) at the start of the increase at ${\mathrm{t}=\mathrm{t}}_0$, i.e., the first minimum, ${\mathrm{I}}_{\mathrm{m}\mathrm{i}\mathrm{n}}={\mathrm{I}}_{\mathrm{p}\mathrm{h}}\left({\mathrm{t}}_0\right)$ observed immediately after the applied reverse-bias voltage to the photodiode. Therefore

$$I_{ph}^{norm}(t)={\displaystyle \frac{I_{ph}\left(t\right)}{I_{ph}\left(t_0\right)}},$$

(1)

where ${\mathrm{I}}_{\mathrm{p}\mathrm{h}}\left(\mathrm{t}\right)$ is the photocurrent at time t.

To analyse the photocurrent associated with HeLa cell emission without CCM influence and to eliminate noise, the measured current is normalised and smoothed using a Savitzky–Golay (SG) filter [32]. In the Supplemental Materials, a section describes in depth description of the photocurrent data processing workflow.

The post-minimum photocurrent increase was modelled using a stretched exponential function,

$$\mathrm{I}\left(\mathrm{t}\right)={\mathrm{I}}_{\mathrm{m}\mathrm{i}\mathrm{n}}+\mathrm{A}\left[1-\mathrm{e}\mathrm{x}\mathrm{p}\left(-{\left({\displaystyle \frac{\mathrm{t}-{\mathrm{t}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}{\mathsf{\tau }}}\right)}^{\mathsf{\beta }}\right)\right],$$

(2)

where A denotes the photocurrent recovery amplitude, $\mathrm{t}$ the time, τ the characteristic relaxation time, ${\mathrm{t}}_{\mathrm{m}\mathrm{i}\mathrm{n}}$, the time at the current minimum ${\mathrm{I}}_{\mathrm{m}\mathrm{i}\mathrm{n}}$, and β is the stretching exponent accounting for heterogeneous or distributed relaxation processes commonly observed in complex photo-responsive systems [33]. The obtained curve is subtracted from the measured curve and then processed further.

After switching off the illumination, the time evolution of the photovoltage/photocurrent is determined by the occupation of valence band-tail states, conduction band-tail states, and dangling-bond states, which may have different capture and emission coefficients [29]. The balance between thermal capture, emission, and recombination determines the concentration of excess carriers in these states. An occupied gap state can thermally emit an electron to the conduction band and a hole to the valence band.

Kruger and Sax [31] have shown that localised entities are precursors A of a type of MDBs and can be neutral closed-shell systems. They well describe that exposure to light with an energy at least the band gap energy (blue light in our case) generates an electron–hole pair, where the electron in the conduction band (CB) is responsible for photoconductivity. At the same time, the precursor A is excited and is able to trap the CB electron. Hydrogen motion is the precursor for defect creation. The excited complex $A^{*}+e^{-}\to B^{*}$ is formed, which, through internal conversion—radiation-less deactivation—ends up as a metastable defect B. B is a radical anion, $D^{-}$, localised within the a-Si:H band gap at about 1.7 ± 0.15 eV. The original state of the material is restored by thermal activation of the reaction between B and the initially created weakly coupled hole present in the silicon frame. The assumed energy difference between B and A is (0.4–0.6 eV), a typical metastable defect energy difference. This process of reaction kinetics describes the changes in electron concentration of electrons in CB by

$${\displaystyle \frac{\mathrm{d}\mathrm{n}}{\mathrm{d}\mathrm{t}}}=\mathrm{G}-{\mathrm{k}}_1\left[{\mathrm{h}}^{+}\right]\left[{\mathrm{e}}^{-}\right]-{\mathrm{k}}_2\left[{\mathrm{A}}^{\mathrm{*}}\right]\left[{\mathrm{e}}^{-}\right],$$

(3)

where G is the carrier generation rate, $\left[h^{+}\right]$ holes and $\left[e^{-}\right]$ electrons concentration, $k_1$ is the rate constant of direct electron–hole recombination, and $k_2$ is the rate constant of the trapping of CB electrons by excited precursors $A^{\mathrm{*}}$. The following sequence of reactions, the time dependence of the concentrations of excited precursors, excited complexes $B^{\mathrm{*}}$, respectively, formation of B, and the defect precursor concentration is described in depth in the above-mentioned model of T. Krüger and A.F. Sax in [31]. This complex process of carrier trapping and release controls the photocurrent as it decays over time [34].

The initial steady-state photocurrent [34] may be expressed in a-Si:H, as

$${\mathrm{I}}_{\mathrm{p}\mathrm{h}}\left(0\right)=\mathrm{q}\mathrm{A}\mathrm{G}\mathsf{\mu }\mathsf{\tau }\overrightarrow{\mathsf{\xi }},$$

(4)

where q is the fundamental electronic charge, A the photodiode’s cross-sectional area for current flow, G the carrier generation rate, and the applied electric field $\overrightarrow{\mathsf{\xi }}=\mathrm{U}/{\mathrm{d}}_{\mathrm{i}}$ in the depleted intrinsic photodiode i-layer of width ${\mathrm{d}}_{\mathrm{i}}$, $\mathsf{\mu }$ the mobility of free charge carrier electrons, and $\mathsf{\tau }$ the lifetime of the free charge carriers. Determining the underlying mobility (μ) and lifetime (τ) quantities is essential to understand the tail trap and MDB’s recombination energy levels involved in current overshoot and decay applicable in the case of light-induced HeLa cells CL process.

Assuming emission from the deep traps, all weak bonds below the thermalisation energy ${\mathrm{E}}_{\mathrm{t}\mathrm{h}}$, neglecting in the initial approximation the barrier lowering in the case of electron accumulation, have been entirely converted to defects. The energy an electron needs to overcome the barrier for the forward reaction rate depends on the time t it attempts to escape

$$\mathrm{t}={\mathrm{\nu }}_0^{-1}\mathrm{e}\mathrm{x}\mathrm{p}\left({\mathrm{E}}_{\mathrm{t}\mathrm{h}}/\mathrm{k}\mathrm{T}\right),$$

(5)

where ${\nu }_0={10}^{12} {\mathrm{s}}^{-1}$ is the attempt to escape frequency, ${\mathrm{E}}_{\mathrm{t}\mathrm{h}}$ is the trap activation energy or trap depth, k is Boltzmann’s constant, and T is the absolute temperature. The photocurrent depends directly on the mobility-lifetime product, $\left(\mathsf{\mu }{\mathsf{\tau }}_{\mathrm{n}}\right)$. Knowing ${\tau }_n$, help us better understand of light amplification measured through photocurrent time evolution in a-Si:H.

J.-H. Zollondz et al. in [35] investigate the appearance of primary photocurrent amplification in reverse-biassed a-Si:H p-i-n devices illuminated with a two-beam, low-intensity ‘probe’ beam and a higher-intensity ‘bias’ beam from the other side. The probe beam introduces space charge, which modulates the internal fields that release charge produced by the bias beam for collection. The modulation of the bias beam induces a low-field region at the p-i interface through hole capture in the i-layer and reduces the $D^{-}$ density of states, but increases the electric field minimum at the transition in the i-layer from space charge dominated with trapped electrons in ${\mathrm{D}}^{-}$ and holes trapped in valence band-tail states. M. Pomoni and P. Kounavis [36] proposed the out-of-phase modulated photocurrent (MPC) signal to determine trapping–detrapping events, recombination processes, and gap-state parameters in amorphous silicon. They found that at lower frequencies, recombination occurs through centres of the ${\mathrm{D}}^{+/0}$ and ${\mathrm{D}}^{0/-}$ levels of normal silicon dangling bonds. Recombination through the recombination centres of the ${\mathrm{D}}^{0/-}$ level dominates at low frequencies.

The distribution and temporal activity of unsaturated structural defects, Si dangling bonds, localised states in the mobility gap of a-Si:H, which act as recombination centres, are crucial for the detected characteristic shapes of light-induced CL and FL responses, respectively. To analyse the connection between the a-Si:H metastable defects and the HeLa cells light-induced changes in the material, the photocurrent decays were fitted using the model of Fehr et al. [37] proposed by [38] to describe the signal y as a superposition of two exponentials:

$$\mathrm{y}={\mathrm{A}}_1·{\mathrm{e}}^{-\left(2\mathrm{t}/{\mathsf{\tau }}_1\right)}+{\mathrm{A}}_2·{\mathrm{e}}^{-{\left(\mathrm{t}/{\mathsf{\tau }}_2\right)}^2},$$

(6)

where ${\mathrm{A}}_1$ and ${\mathrm{A}}_2$ represent the densities of fast-relaxing dangling-bond state (DBS) and slowly decaying DBS, respectively, and ${\tau }_1$ and ${\mathsf{\tau }}_2$ are their associated relaxation times. The quantitative evaluation of the characteristic times was performed on measured transient responses in accordance with the model given in [28].

Finally, to better evaluate the evolution of HeLa cell activity, the normalised photovoltage/photocurrent transient response, h(t) [31], on pulsed blue LED light of a-Si:H p-i-n photodiodes, is calculated by

$$\mathrm{h}\left(\mathrm{t}\right)={\displaystyle \frac{\mathrm{i}\left(\mathrm{t}\right)-{\mathrm{i}}_{\mathrm{m}\mathrm{a}\mathrm{x}}}{{\mathrm{i}}_{\mathrm{m}\mathrm{i}\mathrm{n}}-{\mathrm{i}}_{\mathrm{m}\mathrm{a}\mathrm{x}},}},$$

(7)

where ${\mathrm{i}}_{\mathrm{m}\mathrm{i}\mathrm{n}}=\mathrm{i}\left({\mathrm{t}}_0\right)$ when the light pulse is switched on, ${\mathrm{i}}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ is the maximum photocurrent reached, and $\mathrm{i}\left(\mathrm{t}\right)$ is the measured photocurrent.

3. Results and Discussion

3.1. Long-Time Current

The measured normalised photodiode long-time current (Equation (1)) for samples containing HeLa cells in the cell culture medium (CCM) is shown in Figure 2. The current transients indicate blue LED-induced HeLa cell chemiluminescence, and the increase in current caused by CCM fluorescence, detected by a p-i-n a-Si:H photodiode, was presented and analysed.

Initially, the measured long-time a-Si:H p-i-n photodiode photocurrent under constant illumination by a 430 nm blue LED was analysed to determine luminescence from HeLa cells. The HeLa cells concentration was $5.8\times {10}^5 \mathrm{cells/mL}$, in this work.

Figure 2 shows significant changes in the long-time photocurrent in the first 10–40 min, after the minimum current value was reached due to the well-known Staebler–Wronski effect. The observed continuous photocurrent increase is attributed to fluorescence in the CCM. The concentration of defects in a-Si:H energy gap increases due to continuous blue LED light and variable HeLa cells light absorption. The observed pulse-like photocurrent behaviour is attributed to HeLa cells chemiluminescence reaction (CL), and the activated trapping–detrapping process and recombination process in i-layer of p-i-n a-Si:H photodiode. The extracted HeLa cell’s signal was calculated as the difference between the denoised experimental photocurrent and the fitted model (Equation (S2)). The resultant curve is shown in Figure 2.

The fitting procedure and results are described in more detail in Supplemental Material. This approach reduces the influence of CCM and photodiode dark leakage current.

The numerical evaluation of the carrier lifetimes from the modulated photocurrent and decay transient data are done in the Supplemental Material (Figure S1). We calculated the energy of DOS using Equation (2) in first approximation.

The two calculated energy levels from the long-time photocurrent increase and decrease (Figure 3) are located between 0.78 eV and 0.84 eV in the first three pulse-like photocurrent shapes, and they increase to between 0.84 eV and 0.9 eV over time in the subsequent pulse-like photocurrent shapes. This is in good agreement with obtained values of $0.7\left(+0.2\right)$ eV from transient current measurements when the light pulse is switched off as described below and in accordance with theory described above by other authors [29,30,31,34,35,36,37,38].

3.2. The Transient Response Current

To further investigate the biological activity of HeLa cells, the basic idea was found in the method of two-beam photogating in a-Si:H p-i-n structures [35]. The transient response of an a-Si:H p-i-n photodiode, when illuminated with pulsed blue LED light, gives information about the fundamental defect kinetics processes in the a-Si:H material. When illuminating the analyte, as a consequence of the changes in the light emission intensity originating from the analyte, HeLa cells in CCM (as a ‘probe’ beam) in our case, the transient photocurrent may exceed the expected one. We propose here exciting the analyte with pulsed light at a specific critical frequency of 333 Hz to detect a weak biological light signal. We found that this pulse frequency at which current overshoot occurs is associated with cell activity. We suppose that this critical frequency is the cell’s resonance frequency.

In our experiment, HeLa cells in a CCM solution and CCM were illuminated with blue LED light using square-wave radiation incident on the p-side of the PDs, unlike in [35], where the beams were incident on the opposite sides of the p-i-n junction. The time evolution of the measured transient photovoltage/photocurrent responses of a-Si:H p-i-n PD1 below the analyte is shown in Figure 4, and those posted laterally to the analyte, PD2 and PD3 are shown in Figure 5 and Figure 6, respectively.

Figure 4 shows the time evolution of the PD1 transient photocurrent responses, sampled 13 times in 40 min. The measured transient response shows a small decay or slight overshoot after reaching its maximum, followed by a quasi-linear decay under blue LED illumination. The transient-off responses show a characteristic shoulder-like increase and return to the typical exponential form after 40 min. The shape and magnitude change in time following the activity of HeLa cells (CL signal) and consequently the occupancy of the metastable defects in a-Si:H. The transient response of PD2 when the light turns on (Figure 5) shows current overshoot, which is more pronounced after 15 min of illumination, followed by a quasi-linear decay. The transient off response shows quasi-typical exponential decay. Also, here, the transient on and off response becomes a typical exponential form after 40 min of illumination. Figure 6 shows that the secondary peak in the transient response of the c-Si photodiode disappears after 40 min. The first peak (Figure 6 no. 1) is the c-Si photodiode response to the light switching off. The amplitude of the first peak varies over time following HeLa cell activity (CL). The second peak (Figure 6 no. 2) is a response to the chemiluminescence process and light emission from HeLa cells. The second peak disappears after cca. 40 min. The last curves of transient responses of all photodiodes returned to typical exponential form, when 43% of the HeLa cells were dead, in our experiment.

To gain a deeper understanding of the experimental transient response time evolution, the data have been normalised to the maximum photovoltage difference using Equation (7). The third curves in Figure 4, Figure 5 and Figure 6, which show the normalised measured photovoltage of PD1, the photocurrent of PD2, and the photocurrent of PD3, respectively, and represent the most deformed transient responses, are presented in Figure 7. The anomalous current response form is observed to depend on the transit through the extended states, trapping–detrapping, and recombination processes of photogenerated free carriers by the absorption of pulsed blue light, and by the light resulting from the HeLa cell’s CL process, as the probe light beam in our experiments.

Fitting the measured transient responses of photodiodes PD1 and PD2 with CCM on the surface to the exponential function with one and two exponential terms, as described in [27], respectively, gives the value of activation energies of trap states by means of Equation (5). The transient response of PD1 when the blue LED light is switched on shows an exponential rise with a one-time component with a calculated trap activation energy level of ${\mathrm{E}}_1=0.5143 \mathrm{e}\mathrm{V}$. The photocurrent decay transient after the light switched off shows two time components with calculated energy levels of energies ${\mathrm{E}}_1=0.4858 \mathrm{e}\mathrm{V}$ and ${\mathrm{E}}_2=0.5806 \mathrm{e}\mathrm{V}$. While PD2 shows on activation energy ${\mathrm{E}}_1=0.5018 \mathrm{e}\mathrm{V}$, and off energies ${\mathrm{E}}_1=0.5095 \mathrm{e}\mathrm{V}$ and ${\mathrm{E}}_2=0.4975 \mathrm{e}\mathrm{V}$.

The last curve in Figure 4, Figure 5 and Figure 6, and in Figure 7 (black line), is measured after 40 min and shows minor differences in energies and transient response shape compared to the one obtained with CCM under the same illumination conditions. This means the HeLa cell is no longer active after 40 min of illumination. HeLa cells stop absorbing the blue LED light. Light transmission of the analyte increases, and more light falls on the PD 1 active surface. The a-Si:H photodiodes PD1 and PD2, as well as the c-Si photodiode (PD3), with and without CCM do not show any distortion in transient responses during one hour of illumination.

3.2.1. Switch-On Response

The shapes of PD1 and PD2 photodiodes’ photovoltage/photocurrent transient responses with analyte, after switching on the light pulse ($\mathrm{t}={\mathrm{t}}_0$), show specific changes such as current amplification observed within two time intervals, and current decay during the fourth time interval of the first half-period of pulsed LED light. On the other hand, at time ${\mathrm{t}}_3$, a second smaller peak appears in the PD3 transient response. When the light pulse is switched on, during the first time interval $t_0\le t\le t_1=0.77 \mathrm{m}\mathrm{s}$, the photovoltage of PD1 rises to its maximum value (Figure 7), characterised by two time components: the time that an electron (hole) spends in the trap and the transit time through the extended states.

The time (${\mathrm{t}}_{{\mathrm{v}}_{\mathrm{p}\mathrm{h}\mathrm{m}\mathrm{a}\mathrm{x}}}{=\mathrm{t}}_1)$ in which the photovoltage reaches the maximum differs for any sampled transient response. The calculated activation energies of PD1 in Figure 7 are ${\mathrm{E}}_1=0.4923 \mathrm{e}\mathrm{V}$ and ${\mathrm{E}}_2=0.48927 \mathrm{e}\mathrm{V}$. The calculated two energy levels for other curves in this time interval are between 0.48 eV and 0.63 eV. In the same time the calculated PD2 activation energy is ${\mathrm{E}}_1=0.50184 \mathrm{e}\mathrm{V}$. At time ${\mathrm{t}}_0$ the PD3 transient response like pulse is evident.

During the time interval ${\mathrm{t}}_0\le \mathrm{t}\le {\mathrm{t}}_1$, the observed decrease in the slope of the signal increase is due to the absorption of blue light mainly by HeLa cells and CCM. It is evident that the free carrier photogeneration rate, G, determined by the free carriers (photocurrent) gradient at initial time of current increase, $\mathrm{G}=\mathrm{d}\mathrm{n}/\mathrm{d}\mathrm{t}=0$, depends on the incident photon flux on the surface of PD1. The activation energy is inversely related to this flux. When pulsed illumination and the HeLa probe beam are activated, the shallow tail states and the deeper metastable dangling-bond states (MDBs) fill until the quasi-Fermi energy reaches a new quasi-equilibrium, and the photocurrent attains steady-state.

After that, in the time interval ${\mathrm{t}}_1=0.77 \mathrm{m}\mathrm{s}<\mathrm{t}<{\mathrm{t}}_2=0.8575 \mathrm{m}\mathrm{s}$, a minimal photovoltage decrease is observed, especially visible on the third curve in Figure 3 and Figure 7. The calculated energy level (third curve) in this time interval is ${\mathrm{E}}_1=0.4534 \mathrm{e}\mathrm{V}$. In the same time PD2 current remain constant. The c-Si PD3 has small detectivity, D and shows very small changes in current. This means the HeLa cells start to absorb more LED light and with this the photon flux reaching the PD1 surface decreases, as well as the density of $D^{-}$ defects by capturing the holes throughout the i-layer reduce and convert to $D^0$ states. With this the local electric field decrease in vicinity of the p/i interface. The calculated energy levels of other curves in this time interval are located between 0.49 eV and 0.54 eV.

Following the further PD1 photovoltage exponential decay is a small shoulder in interval ${\mathrm{t}}_2=0.8575 \mathrm{m}\mathrm{s}<\mathrm{t}<{\mathrm{t}}_3=1.2125 \mathrm{m}\mathrm{s}$. The calculated two energy levels in this time interval are for curve three ${\mathrm{E}}_1=0.5721 \mathrm{e}\mathrm{V}$ and ${\mathrm{E}}_2=0.5166 \mathrm{e}\mathrm{V}$, and for other curves between 0.43 eV and 0.54 eV. In the same time, the PD2 current further exponentially rise with the calculated energy levels ${\mathrm{E}}_1=0.49084 \mathrm{e}\mathrm{V}$ and ${\mathrm{E}}_2=0.48927 \mathrm{e}\mathrm{V}$. This observed photovoltage decrease in PD1 transient response after time $t_1$ is present only when HeLa cell activity reaches a critical value. Otherwise, on other curves, a small exponential decay as a small shoulder is observed, or even a slight overshoot is present in some of them. When HeLa activity is below the critical value or is not present, the photovoltage exponentially increases to a new quasi-steady state, and those two intervals appear as one (${\mathrm{t}}_1<\mathrm{t}<{\mathrm{t}}_3$). In these two time intervals, we suppose the HeLa cells begin to emit light with a longer wavelength than the excitation LED blue light. This is also observable on the photocurrent overshoot of PD2 correspond to higher light flux. This light was provided by HeLa cells, as a second probe beam, which was absorbed deeply in the i-layer and activated deeper band-tail states or even DOS (${\mathrm{D}}^{0/+}$ and/or ${\mathrm{D}}^{0/-}$) by detrapping the electrons to higher conduction band-tail states or conduction-band states near n/i interface. As a consequence, the trapping reaction is accelerated, and the local electric field at the p-i interface and deeper in the i-layer is perturbed.

In this second time interval, ${\mathrm{t}}_1<\mathrm{t}<{\mathrm{t}}_2$, and especially in the third interval, $t_2<t<t_3$, the blue LED light triggers the generation of nitric oxide ($\mathrm{N}\mathrm{O}$) and hydrogen peroxide ${(\mathrm{H}}_2{\mathrm{O}}_2)$. Chemiluminescence (CL) is observed between these two molecules, where chemical energy is converted into light through a series of redox reaction and detected by photodiodes [39]. As a consequence of this reaction, singlet oxygen, $({}^1{\mathrm{O}}_2)$ and peroxynitrite, ${(\mathrm{O}\mathrm{N}\mathrm{O}\mathrm{O}}^{-})$ are formed and those further, in interval ${\mathrm{t}}_4<\mathrm{t}<{\mathrm{t}}_5$, in cascade of reactions with aminoacides proteins, and lipids produce other chemiluminescence, CL [40,41,42].

At time $t_3=1.2125 \mathrm{m}\mathrm{s}$ the PD1 photovoltage and PD2 photocurrent drop abruptly after which follow the quasi-linear decrease since time $\mathrm{t}=1.5 \mathrm{m}\mathrm{s}$, when the light pulse switches off. At this time the small second peak appear in c-Si PD3 photocurrent.

The recorded quasi-linear changes in photovoltage correspond to the CL reaction in HeLa cells described above. At the moment $\mathrm{t}={\mathrm{t}}_3$, it looks like a small intensity probe light beam is switched off, and a very short, sudden photovoltage drop appears in a few microseconds. The calculated energy level, proposed by Goldie [29,34], from extracted instantaneous lifetime is $\mathrm{E}=2.11 \mathrm{m}\mathrm{e}\mathrm{V}$. This indicates that the concentration of excess carriers in extended states increases to maintain space charge balance. Following this, during the time interval ${\mathrm{t}}_3<\mathrm{t}<{\mathrm{t}}_4=1.5 \mathrm{m}\mathrm{s}$, the steeper quasi-linear decrease occurs until the moment when light pulse is switched off. This implicates the main changes in the primary emission of HeLa cells and point to bivalent [16]. In the i-layer of a-Si:H the recombination of free charge carriers prevails in previously formed deeper defect states ${\mathrm{R}}_{\mathrm{r}\mathrm{e}\mathrm{c}}^{0/-}$, ${\mathrm{R}}_{\mathrm{r}\mathrm{e}\mathrm{c}}^{\mathrm{v}\mathrm{b}\mathrm{t}}$ and ${\mathrm{R}}_{\mathrm{e}\mathrm{m}}^{+/0}$ in i-layer of PD1 with energy ${\mathrm{E}}^{\mathrm{P}\mathrm{D}1}=0.8133 \mathrm{e}\mathrm{V}$, and ${\mathrm{E}}^{\mathrm{P}\mathrm{D}2}=0.5046 \mathrm{e}\mathrm{V}$, respectively. The calculated energy levels for other curves with less distortion in this time interval are located between 0.48 eV and 0.55 eV.

In the case of CCM and non-active HeLa cells, the photovoltage reaches quasi-steady state during time intervals ${\mathrm{t}}_1<\mathrm{t}<{\mathrm{t}}_4$. The calculated two energy levels are located between 0.53 eV and 0.5 eV.

3.2.2. Switch-Off Response

In time after switching off the illumination ($\mathrm{t}={\mathrm{t}}_4=1.5 \mathrm{m}\mathrm{s}$), the photovoltage in a very short time of a few $\mathsf{\mu }\mathrm{s}$ $\left({\mathrm{t}}_4<\mathrm{t}<{\mathrm{t}}_5\right)$ drops abruptly ($\mathrm{E}=0.958 \mathrm{m}\mathrm{e}\mathrm{V}$), indicating the higher trapping through valence band-tail states and following recombination through ${\mathrm{D}}^{0/-}$ and ${\mathrm{D}}^{0/+}$, as described by Schmidt and Goldie [29]. The gradient of photovoltage immediately after switching off the illumination varies over time, indicating changes in generation and emission from conduction band-tail states. We suppose that optical generation also occurs during this time interval due to HeLa cells CL. This increase in the generation rate increases the splitting of the quasi-Fermi levels, turning more states from traps to recombination centres. Initially, the curve shows negative photovoltage gradient ${\left.\mathrm{d}{\mathrm{v}}_{\mathrm{p}\mathrm{h}}\left(\mathrm{t}\right)/\mathrm{d}\mathrm{t}\right|}_{\mathrm{t}\to 0}<0$, meaning that capture and probably recombination are larger than emission. Then the larger shoulder refers to the increased emission from the deeper ${\mathrm{D}}^{+/0}$ states. Variation in the shoulder over time is observed. This emission could be attributed to secondary HeLa cell light emission in the time interval ${\mathrm{t}}_4<\mathrm{t}<{\mathrm{t}}_5$, as described above. The two calculated energy levels in this time interval are ${\mathrm{E}}_1=0.61272 \mathrm{e}\mathrm{V}$ and ${\mathrm{E}}_2=0.65631 \mathrm{e}\mathrm{V}$ for PD1 and located between 0.43 eV and 0.63 eV for other curves and for CMM and non-active Hela cells. The PD2 transient response current decay with one exponential curve and a corresponding calculated energy level of ${\mathrm{E}}^{\mathrm{P}\mathrm{D}2}=0.50184 \mathrm{e}\mathrm{V}$, which correspond to recombinations through depper energy levels throughout i-layer.

The calculated band-tail trap and metastable dangling-bond energy from other curves of a-Si:H photodiode photovoltage time-evolved response to blue LED pulsed light and probe beam from the analyte are in the range 0.47 eV to 0.6 eV. The calculated data are presented in the Supplemental Materials in Tables S1 and S2. The more in-depth analysis will be done in future work.

The signals of the side PDs are one to two orders weaker than those of a-Si:H PD below the analyte. The more evident changes in a-Si:H PD signals can be faster and more easily separated and detected than those with c-Si PDs. For that, we propose the a-Si:H photodiodes as a detector of choice for the detection of weak biological signals.

4. Conclusions

We demonstrated the detection of HeLa light emission resulting from their intrinsic chemiluminescence, excited by a blue LED light. We propose using an a-Si:H p-i-n photodiode as a better transducer choice than a c-Si photodiode. Furthermore, the photodiode, a-Si:H, and c-Si, placed laterally to the analyte, detect the small HeLa cell signal and provide a new structural solution for a transducer worth investigating. Illuminating the HeLa cells with pulsed blue LED light, we identified the “resonance frequency” at which the a-Si:H p-i-n photodiode exhibits overshoot. Simultaneously, smaller second peaks emerge in the c-Si photodiode transient response. After switching off the light, the PD transient below the analyte shows a larger shoulder. Additionally, we established that after one hour of illuminating HeLa cells with blue LED light at 430 nm, the cells become inactive. The main finding of this research is the pulse frequency at which current overshoot occurs, linked to the HeLa cell’s intrinsic chemiluminescence. We believe this critical frequency corresponds to the cell’s resonance frequency. This relatively simple method can be integrated into optical biosensors for label-free detection and treatment of tumour cells using a blue LED. This approach offers the advantage of label-free detection in HeLa cells and utilises an LED light source with less adverse effect on tissue than a laser, making it applicable in microfluidic PoCT.