Portable device for dual detection of fluorescence and absorbance for biosensing or chemical sensing applications

Graphical abstract


Hardware in context
Early detection of biomarkers in the bodies or the level of toxic chemicals or pyrogens in foods is an important preventive measure to tackle health problems. However, most of the analyses are performed by highly skilled personal in well-equipped laboratories, which are not easily accessible, especially in remote areas or in developing countries. An interesting alternative is to use biosensors or chemical sensors that offer high selectivity, high sensitivity, low detection limit, ease of use and, quite often, rapid detection compared with conventional methods in laboratories. Despite all of the benefits, a lot of them still rely on benchtop equipment in laboratories. For examples, there are various optical bio-and chemical sensors developed for the detection of key biomarkers or analytes that are important in healthcare and food safety applications. Most of them uses either changes in sensors' fluorescence emission or light absorption to signal the binding and a lot of them still rely on benchtop fluorometers or absorption spectrometer to measure the signals.
To address the problem, several research groups introduced portable fluorescence detectors or absorption spectrometers that can be used outside laboratories [1][2][3][4]. In fact, some of them are commercially available, such as Qubit Fluorometers (Fisher Scientific, USA) or Optizen mini UV-Vis Spectrophotometer (K LAB, Republic of Korea). However, they require teams of experts to develop and the commercial models are still expensive for most of the developing countries. In addition, to the best of our knowledge, none of them is capable of measuring both fluorescence emission and light absorption. In this work, we provide detailed instructions, lists of materials and downloadable STL (stereolithography) and g-code files for constructing a portable device for dual detection of fluorescence and light absorption. It is in our best interest that it should widely reach audience being interested in on-site or point-of-care applications, especially in areas with limited resources. However, it must be emphasized that we do not intend to replace commercial benchtop fluorometers and UV-Vis absorption spectrometers with our device; it is good for applications that has a clear target and do not require spectral measurement.

Hardware description
The device was developed based on opensource technology as follow:

Fluorometer and absorption spectrometer setup
In the current setup, a 450 nm laser diode (PLT5 450B, Osram Opto Semiconductor, USA) is focused by a plano-convex lens (4.87 mm lens diameter, 1.76 mm thickness, 7.15 mm focal length and 3.74 mm aperture stop diameter) into a cuvette containing the sample (Fig. 1). We chose a laser instead of light emitting diodes (LED) to minimize the cross-talk of the excitation light into the fluorescence emission channel and to avoid using excitation filters.
The semi-micro cuvette being used is disposable yet made from high quality polymethyl methacrylate (PMMA) (97000-590, VWR, U.K.) suitable for working with light having wavelengths between 300 nm and 900 nm. The internal dimensions are 10.0 mm Â 4.0 mm (length Â width) while the outer dimensions are 12.5 mm Â 12.5 mm Â 45.0 mm (length Â width Â height). Cheaper alternatives, such as microcentrifuge tubes were tested but the semi-micro cuvette offered a much better reproducibility with only a slight extra cost. Consequently, our portable device accepts a regular full-size (outer dimensions of 12.5 mm Â 12.5 mm Â 45.0 mm (length Â width Â height)) fluorescence cuvette as well. There are three photodiodes (PS1.0-5-TO52S1, First Sensor, Germany) for light detection. Photodiode 1 (PD1) is for measuring light absorption (at 450 nm in the current setup) after passing through the sample's cuvette. If needed, the light intensity can be reduced to the sensitivity range of the photodiode by a low-cost poly (methyl methacrylate) 525 nm cut-on absorption filter which is mounted in front of PD1. Photodiode 2 (PD2) and 3 (PD3) are for detecting fluorescence emission at 90 degrees relative to the direction of the excitation beam in dual wavelength ranges. As a demonstration, we place a 500-550 bandpass filter in front of PD2 to detect green fluorescence (e.g. from fluorescein) and a 592-648 bandpass filter in front of PD3 to detect orange-red fluorescence (e.g. from tetramethylrhodamine (TMR)). Each photodiode is mounted in a preassembled housing with another plano-convex lens (4.87 mm lens diameter, 1.76 mm thickness, 7.15 mm focal length and 3.74 mm aperture stop diameter) to maximize incident light intensity. Even though there are many cheaper photodiodes, such as the BPW34, available, the PS1.0-5-TO52S1 appears to be the best choice for constructing our device. When compared with the BPW34, the responsivity at 550 nm (close to the peak wavelength of fluorescein) from the PS1.0-5-TO52S1 is twice as much. In addition, the PS1.0-5-TO52S1 provided 0.2 nA dark current and 1.5 Â 10 -14 W/Hz ½ noise equivalent power which are 10 and 2.66 times lower than that of the BPW34, respectively. Therefore, the BPW34 is still inferior to the PS1.0-5-TO52S1 in the case of fluorescence measurement applications, but it may be used in absorbance measurement applications.

Signal conditioning
The dynamic range of detection can be easily adjusted by varying the laser intensity and the amplifying gain of photodiode. The laser intensity is proportional to the current fed from a low-cost laser driver power supply 100mW-3 W 12VDC. The current setting can be done in two ways. The first was by directly limiting the current via a variable resister. In this mode, the TTL (transistor-transistor logic) port was set to 5 V to enable the current feeding, and the intensity of the laser can be set at a desirable value (e.g., 100 mW in our case) by manually adjusting the resister. The second method uses PWM (pulse width modulation) generated from an Arduino-compatible board (Mega 2560 Pro, RobotDyn, China) at the TTL port for fine light intensity adjustment.
The photodiodes are operated in reverse bias by a laboratory-built circuit. The reverse bias (at À20 V) forces the photodiode to operate in the photocurrent mode, in which the output current is linearly proportional to the incident light intensity. Also, the reverse bias gives a better responsivity compared with the zero bias (photovoltaic mode). The photocurrent output is converted into voltage signal and amplified via an operational amplifier based transimpedance amplifier. Finally, the signals were adjusted to match the analogue to digital circuit (ADC) dynamic range for data acquisition. The conversion rate for the ADC is set at 1,600 times per second or approximately 0.625 ms accumulation time. This ADC conversion rate was chosen so that the accumulation time is well (approximately 25 times) shorter than the laser on time in each PWM cycle. The op-amp being used (LM318) is capable of amplifying the signal with frequency ranging from DC to 15 MHz (i.e., 15 MHz bandwidth). Therefore, the ADC and op-amp are compatible.
The key features of the device are as follow: -Capable of dual detection of fluorescence and light absorption in one portable device.
-Contains no moving parts, inexpensive, customizable and easy to operate with minimal maintenance.
-Works with sample volume as small as 500 ll.

Design files summary
Bill of materials summary

Build instructions
The optical breadboard and device's body frame was 3D printed using a fused deposition modeling (FDM) printer (Chiron, Anycubic, China). The circuit mainboard was engraved by a computer numerical control (CNC) router machine (3018 mini-CNC Grbl Control Systems, CRONOS MAKER, China).
Caution: If using a lead alloy solder, make sure that the soldering is performed with chemical resistance gloves (e.g., PVC gloves) on in an area with good air ventilation. Avoid inhaling solder fume and wash hands thoroughly with soap and water after finish soldering.
2. Connect a 12 V 5A DC power to the 2P Terminal connector J1 (at G4 in Fig. 3C). The power line connects to the right terminal and the neutral line connects to the left terminal, respectively. 3. Connect a multimeter probe to the output terminal of the 12 V to 5 V DC-DC Step Down module (at E4 in Fig. 3C) and adjust the voltage output via the variable resistor on the DC-DC Step Down module until the output voltage is 5 V. 4. Do the same to the other DC-DC Step Down module (at E2 in Fig. 3C) and the DC-to-DC Step Up module (at E3 in Fig. 3C) to set the output voltages to 7 V and 20 V, respectively. 5. Cap the connector J2 (at D3 in Fig. 3C), J3 (at D2 in Fig. 3C) and J4 (at D3 in Fig. 3C) to connect the power source to all components on the PCB. 6. Connect the cathode of PD1, PD2 and PD3 to a pin on the connector J7, J8 and J9 (at B2 in Fig. 3C), respectively. Then connect the anode of PD1, PD2 and PD3 to a pin on the connector J10, J11 and J12 (at B5 to D5 in Fig. 3C), respectively. 7. Connect a pin of the connector J5 and J6 (at C2 in Fig. 3C) to Vin and GND pins of the Arduino Mega. This powers the microcontroller. 8. Connect the I2C port of the ADS1105 to the I2C port of the Arduino Mega.

Optical setup
In order to minimize the cost and assembly effort, we used pre-assembled housings and lens from commercial 5 mW red laser modules (e.g., SYD1230, Howard Batchen, UK) which are relatively inexpensive and widely available as the housing and focusing lens for our 450 nm laser diode and photodiodes. For the laser, we carefully replaced the red laser diode with our 450 nm laser diode. In the case of photodiodes, the laser housing has to be enlarged to 5 mm to match the photodiode's diameter. After that, the photodiode and laser diode housings were installed in the Part 2 before being mounted on the printed optical breadboard as shown in Fig. 4A.
Next, the optical setup for the absorption measurement was performed as follows: -Cover photodiode 1 (PD1, Fig. 1) to prevent unnecessary light exposure during the setup.
-Connect the laser diode and PD1 to their drivers.
-Re-install the focusing lens to the laser diode housing and adjust the lens position until a parallel laser beam is obtained.
-Set the TTL signal on the laser drive to 5 V before adjusting the variable resistor until the laser power is 100 mW (on the power meter equipped with a neutral density filter) (Laser safety glasses for eye protection are required) -Uncover PD1 so that the laser is incident on the photodiode. Adjust the position and tilt angles of the photodiode until the output signal is maximum. Secure PD1. Caution: Proper and certified laser safety glasses must be used at all time when working with laser. In this work, we used Thorlabs' laser safety glasses (LG3, Thorlabs Inc., U.S.A.) that effectively block the 450 nm blue light (OD 450 > 7.0) from the laser diode while allowing some light (with wavelength longer than 550 nm) emitted from the sample to be observed. In addition, during assembling and testing the device, avoid wearing or using personal accessories with glossy surfaces, such as a wristwatch or a smartphone.

Electrical condition
The output signal from each photodiode has to match the dynamic range of the ADC module, therefore the signal conditioning was required. Photocurrents generated by the photodiodes were converted to voltage and amplified via an op-amp based transimpedance amplifier. The amplifying gain was adjusted from a negatively feedbacked resistor to obtain measurable signal. The voltage signal was passed through an inversing amplifier with an offset to match the dynamic range of the ADC module and stored in the microcontroller. Note that even though the photodiodes (PS1.0-5-TO52S1) and op-amp (LM318) are designed for high-speed measurements, our device operates only in a stationary mode. The rationale behind choosing the photodiodes is based purely on their high responsivity to fluorescence from our model fluorophore (i.e., fluorescein), low dark current, wide availability, and low price. The op-amp was chosen based on the compatibility with the selected photodiodes.

System calibration
For fluorescence measurement Prepare a ''Fluorescent sample" by filling a semi-microcuvette with 500 ml of 1.000 mM fluorescein in 10 mM Tris and 1 mM EDTA (1 Â TE) buffer solution pH 8.0. A ''Blank sample" is prepared in the same way but without fluorescein present in the solution. Next, the calibration of fluorescence detection is performed as follows: -With the device switched off, set the multimeter to measure resistance and connect the positive and negative probes to the inverting input (pin 2) and output (pin 6) of the op-amp U1 (at B6 in Fig. 3C) and adjust the variable resister RV1 (at B5 in Fig. 3C) until the resistance value shown on the multimeter is 500 kX. -Repeat the procedure with the op-amp U2 (at C6 in Fig. 3C) and the variable resistor RV2 (at C5 in Fig. 3C) until the resistance value shown on the multimeter is 500 kX. -Repeat the procedure with the op-amp U3 (at D6 in Fig. 3C) and the variable resistor RV3 (at D5 in Fig. 3C) until the resistance value shown on the multimeter is 200 kX. -Switch the device on.
-Set the D5 pin of the Arduino board to 5 V to turn the laser on. -Set the multimeter to measure DC voltage and connect the positive and negative probes to pin 7 and pin 2 (counting from the top) of the connector J13 (at F7 in Fig. 3C). -Place the cuvette containing the ''Blank sample" in the sample holder.
-Adjust the variable resistor RV7 (at E5 in Fig. 3C) until the voltage displayed on the multimeter is 0.00 V. This sets the zero value in Channel 1 (with PD2 as the detector) of the fluorescence measurements. -Connect the positive and negative probes of the multimeter to pin 8 and pin 2 (counting from the top) of the connector J13 (at F7 in Fig. 3C). -Replace the cuvette containing the ''Blank sample" with the cuvette containing the ''Fluorescent sample" in the sample holder. -Adjust the variable resistor RV4 (at E5 in Fig. 3C) until the voltage displayed on the multimeter is 5.00 V. This sets the maximum value in Channel 1 (with PD2 as the detector) of the fluorescence measurements -To calibrate Channel 2 (with PD3 as the detector) of the fluorescence measurements, first record the resistance values of the variable resistor RV4 and RV7 that have been set during Channel 1 calibration. Then, adjust the variable resistor RV5 (at F5 in Fig. 3C) until its resistance matches with that of RV4 and adjust the variable resistor RV8 (at F5 in Fig. 3C) until its resistance matches with that of RV7.
For absorbance measurement -With the device switched off, set the multimeter to measure resistance and connect the positive and negative probes to the inverting input (pin 2) and output (pin 6) of the op-amp U3 (at D6 in Fig. 3C) and adjust the variable resister RV3 (at D5 in Fig. 3C) until the resistance value shown on the multimeter is 200 kX. -Switch the device on.
-Set the D5 pin of the Arduino board to 5 V to turn the laser on.
-Set the multimeter to measure DC voltage and connect the positive and negative probes to pin 9 and pin 2 of the connector J13 (at F7 in Fig. 3C). Read the voltage displayed on the multimeter; this is called V air . Place the cuvette containing the ''Blank sample" in the sample holder. Read the voltage displayed on the multimeter; this is called V blank. If V blank < V air , go to the next step. If not, keep adding an absorption filter in the mount in front of PD1 and repeat the process until V blank < V air . -Set D5 pin to 0 V to turn the laser off.
-Adjust the variable resistor RV9 (at H6 in Fig. 3C) until the voltage displayed on the multimeter is 0.00 V. This sets the zero value of the measured light intensity impinging on the photodiode that is later used to calculate absorbance measurements. -Set the D5 pin of the Arduino board to 5 V to turn the laser on again.
-With the cuvette containing the ''Blank sample" still in the sample holder, adjust the variable resistor RV6 (at G5 in Fig. 3C) until the voltage displayed on the multimeter is 5.00 V. This sets the maximum value of the measured light intensity impinging on the photodiode that is later used to calculate absorbance measurements.

Signal conversion
The conversion from fluorescence intensity impinging on the photodiode to the final voltage output followed by the digitization by the ADC may be explained as follow: Step 1: The conversion from the impinging fluorescence intensity to the photocurrent: Where I photo is the photocurrent (A) F 520 is the fluorescence intensity measured at 520 nm, where fluorescein usually emits most brightly (W/m 2 ) c is the responsivity of the photodiode at 520 nm (approximately 0.27 A/W, according to the data sheet from the manufacturer) A is the active area of the photodiode (1 mm 2 ) I D is the dark current (0.2 nA, according to the data sheet from the manufacturer) Step 2: The conversion from the photocurrent to a voltage by the transimpedance amplifier Where V photo is the voltage converted from the photocurrent (V) R is the transimpedance gain Step 3: The final amplification Where V out is the output voltage before digitization (V) V offset is the offset voltage set during the calibration process (V) G is the gain of the inverting amplifier Step 4: The digitization is done by a 12-bit ADC. One level of the digital signal equals to 0.125 mV of V out .

Maintenance
The developed device is designed to be simple to operate and easy to maintain. It is recommended that when not in use the lid of the sample holder compartment should be kept closed to prevent dust accumulation. If needed, a simple rubber dust blower can be used to remove dust particles from the sample compartment. In case of normal spillages, first switch the device off, remove the power cord from the outlet and then remove the spilled liquid with a piece of lint free cloth (or lint free tissue paper) followed by rubbing the spilled area with another piece of lint free cloth damped with clean water and let dry. In the case of persistent stain, rubbing the surface with a piece of lint free cloth damped with mild detergent or ethanol may be used before rubbing with another piece of lint free cloth damped with clean water and let dry. Avoid excessive washing.

Software
The software of the device works according to the flowchart shown in Fig. 5 and was developed in Arduino IDE (integrated development environment) (Arduino, Italy) and Nextion Editor (Nextion, China. The installation process is as follows. a. Download and extract ''Dual_Portable_Fluorescence_and_Absorption_Device.rar". The file is located in the ''Software" folder at https://doi.org/10.17632/bdg2c7tsxm.1 b. Download and install Arduino IDE (https://www.arduino.cc/en/software) and Nextion Editor (https://nextion.tech/ nextion-editor/).  For example, the port for our Arduino connection is shown as ''USB-SERIAL CH340". g. Click ''Upload" and wait until the ''upload completed" message is shown. h. Open file ''Dual_Portable_Fluorescence_and_Absorption_Device_GUI.HMI" in Nextion Editor.
i. In the command panel, click ''Device ID", set the device to ''basic" series and choose the model ''NX8048T07_011". j. Connect the USB to serial UART module (e.g., FT232RL) to the UART port of the Nextion HMI display. k. Click ''Upload". l. Disconnect the UART wires from the USB to serial UART module and then make UART connection between the Arduino Mega 2560 pro and the Nextion HMI display. m. After finishing the software installation, assemble all parts as shown in Fig. 6. The device was designed to be simple and easy to use. Both absorbance and fluorescence measurements can be done using the same procedure as follows.

Operation instructions
a. Fill at least 500 ll of sample solution in the microcuvette.
b. Set the desired intensity by input a value from 1 to 100 (mW of laser power) in the intensity slide bar (Fig. 7). c. Open the sample holder cover, place the microcuvette inside and close the cover. d. Press the ''Measure" button to obtain absorbance at 450 nm or fluorescence emission in the wavelength range from 500 to 550 nm and 592 to 648 nm. The results are shown in the text boxes. (Fig. 6).

Validation and characterization
The developed device was validated by comparing to results with that obtained from a UV-Vis spectrophotometer (U-2900, Hitachi, Japan) for absorbance measurement and a fluorescence spectrometer (F-2700, Hitachi, Japan) for fluorescence measurement using the same sets of samples.

Absorbance measurement
We chose solutions of silver nanoparticle (average size around 28 nm) as a model sample because nanoparticle-based optical sensors are very popular in which the concentration of targeted analytes were often quantified by changes in the absorbance of the solution [5,6]. The solutions appear yellow indicating that they have strong absorption toward blue light. We varied the nanoparticle concentrations from 0.000 to 0.140 nM and the results from both the commercial UV-Vis spectrophotometer (Fig. 8A) and that from our device (Fig. 8B) increase with increasing nanoparticle concentration, as expected from the Beer-Lambert Law. Moreover, there is a good correlation between the two sets of results (Fig. 8C).

Fluorescence measurement
For the fluorescence measurement, we chose fluorescein, a popular fluorescent dye that is widely used in fluorescent biosensors and chemical sensors [7][8][9] as our model sample. We prepared the dye in 1x TE buffer pH 8.0 at various concentrations ranging from 0.000 to 4.000 mM. On the fluorometer, the excitation wavelength was set at 450 nm with 5 nm excitation slit. The emitted fluorescence intensity was summed from 500 nm to 550 nm (in order to match to the measurement range of the developed device) and the results are shown in Fig. 9B. Then the fluorescence intensity from the same set of solutions was measured by our device (from the PD2 output). By comparing the two results obtained from the commercial fluorometer and that from our device, it can be seen that they are highly correlated as shown in Fig. 9C.
To demonstrate the full capacity of our device, a DNA based fluorescent sensor doubly labeled with fluorescein and TMR was chosen as the second model sample. The sensor comprises two DNA hairpins, named H1 and H2, as shown in Fig. 10A. They were designed to be in a metastable state and will not interact with each other. However, the right single stranded DNA target will catalyze the H1:H2 duplex formation via the catalyzed hairpin assembly (CHA) reaction [10]. Therefore, the presence of the target ssDNA even at a small amount could result in a large formation of H1:H2 duplex that is easy to detect by, e.g., measuring the Förster resonance energy transfer (FRET) between two fluorophores being labeled on H1 and H2. This FRET based DNA sensor is very sensitive and is found to be widely used in various applications [11][12][13]. However, most of the work still rely on commercial fluorometers to measure FRET. In this work, we demonstrate that our device is also cap- able of measuring FRET. We label H1 with fluorescein (a FRET donor) and H2 with TMR (a FRET acceptor), as shown in Fig. 10A. It is expected that in the absence of target ssDNA, the signal from the green channel (PD2) is high (low FRET when the dyes are well separated) while that on the red channel (PD3) is low (high FRET when the dyes are in close proximity) and vice versa once the ssDNA is introduced. Preliminary results from the commercial fluorometer (Fig. 10B) show that the fluorescence emission from fluorescein (peaked at around 520 nm) is high while that from TMR (peaked at around 578 nm) is low in the absence of the target ssDNA (i.e., low FRET). The opposite is true in the presence of the target ssDNA. Interestingly, our device also recorded the same events as shown in Fig. 10C.
In conclusion, the results from Section 7.1 and 7.2 indicate that our portable device is capable of measuring both light absorption and fluorescence emission from model samples that are widely used in biosensor and chemical sensor develop-  ment. When compared with commercial equipment (Table 3 and 4), our device offers a smaller, lighter and cheaper alternative with less power consumption.Therefore, it is in our best interest that our device can be beneficial to researchers whose works require a portable and economical device to detect light absorption or fluorescence emission from optical biosensors or chemical sensors.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Table 3 Comparison between a commercial fluorometer (Hitachi F-2700) and the developed device.

Properties
Commercial fluorometer(Hitachi F-2700) The developed device