Monolithic Integrated OLED–OPD Unit for Point-of-Need Nitrite Sensing

In this study, we present a highly integrated design of organic optoelectronic devices for Point-of-Need (PON) nitrite (NO2−) measurement. The spectrophotometric investigation of nitrite concentration was performed utilizing the popular Griess reagent and a reflection-based photometric unit with an organic light emitting diode (OLED) and an organic photodetector (OPD). In this approach a nitrite concentration dependent amount of azo dye is formed, which absorbs light around ~540 nm. The organic devices are designed for sensitive detection of absorption changes caused by the presence of this azo dye without the need of a spectrometer. Using a green emitting TCTA:Ir(mppy)3 OLED (peaking at ~512 nm) and a DMQA:DCV3T OPD with a maximum sensitivity around 530 nm, we successfully demonstrated the operation of the OLED–OPD pair for nitrite sensing with a low limit of detection 46 µg/L (1.0 µM) and a linearity of 99%. The hybrid integration of an OLED and an OPD with 0.5 mm × 0.5 mm device sizes and a gap of 0.9 mm is a first step towards a highly compact, low cost and highly commercially viable PON analytic platform. To our knowledge, this is the first demonstration of a fully organic-semiconductor-based monolithic integrated platform for real-time PON photometric nitrite analysis.


Introduction
Nitrates and nitrites play an essential role for plant growth in agriculture. Thus, they are a major component of inorganic fertilizers [1]. Their high solubility in water results in a particularly critical exceedance of limits in the ground water [2]. The accumulation of agricultural chemicals in the groundwater is a well-known problem. In 1989, Hallberg reported about 39 pesticides in the groundwater of 34 states or provinces of the United States with nitrate as one of the common agricultural chemicals [3]. The exceedance of nitrates and nitrites limits in environmental and physiological systems have adverse effects on animal and human health. The World Health Organization (WHO) set a guideline value of 3 mg/L nitrite in water. Recent studies reported carcinogenic effects, methemoglobinemia, and detrimental effects on the thyroid gland and other organs, associated with the ingestion of high concentrations of nitrate and nitrite due to the high toxicity. Infants are particularly susceptible to methemoglobin formation due to reduction in nitrite. In contrast to hemoglobin, methemoglobin is unable to transport oxygen to the tissues. This condition causes cyanosis or asphyxia. Thus, the detection of nitrite in samples such as water, urine, saliva or blood plasma is crucial and has been reported by many different research groups [4].
Current methods for nitrite ion detection include electrochemical methods (including voltametric, potentiometric and impedimetric electrodes), spectrophotometry, spectrofluorimetry and ion chromatography [5][6][7][8][9]. Electrochemical sensors are popular due to the potential low cost, portability and simple fabrication. Despite the wide application areas there are still some limitations. Current research on electrochemical sensors focusses on improving the limit of detection and reducing the cross-sensitivity of the electrodes [10]. Here, we focus on spectrophotometry as the most common method for nitrite detection due to its easy procedure, low detection limits, high selectivity and low cost. In particular, the assay based on the Griess reagent for colorimetric nitrite sensing is very popular due to its high stability, selectivity and sensitivity [11]. Early work has been conducted by Petsul et al. using the Griess reaction and an external light emitting diode (LED) with a spectrometer for the absorbance measurements, achieving a limit of detection (LOD) of 0.2 mM [12]. Since then, a variety of different sensor systems with different optical device configurations have been reported. Colorimetric nitrite sensors mostly use LEDs as the light source. Devices such as CCD cameras, spectrometers, smartphone cameras, photomultiplier tubes (PMT) and photodiodes (PD) were recently successfully integrated on the detector side [13][14][15][16][17][18][19][20][21][22]. While very impressive results are achieved with the compact integration of the inorganic optoelectronic devices and microfluidic-based approaches, these systems are still rather bulky. For low-cost portable Point-of-Need (PON) solutions, further miniaturization and a simple fabrication method are needed.
To overcome the aforementioned limitations, the integration of organic light emitting diodes (OLEDs) and organic photodetectors (OPDs) promises compact and low-cost optical detection units as hybrid integrated sensors. Recently, very impressive results have been achieved with fabrication of OLEDs and OPDs on flexible and rigid substrates. The devices have been proposed as wearable sensors for health monitoring and Point-of-Need analysis applications [23][24][25][26][27][28][29]. Due to the thermal evaporation-based device fabrication technique, all sensing elements are permanently aligned, allowing a high degree of miniaturization. These units can be easily laminated to a microfluidic system for sensing applications in a liquid [30][31][32][33][34].
Based on our experience in biomedical lab-on-chip systems for multiplexed detection [35,36] and monolithic integrated organic optoelectronic devices [37,38], we aim at developing a highly integrated nitrite sensor for Point-of-Need analysis. In this work we present, to the best of our knowledge, the first demonstration of a fully organic optoelectronic system for photometric nitrite sensing based on the Griess reagent with a reflection-type architecture. The OLED and the OPD are successfully monolithically integrated and have a device size of 0.5 mm × 0.5 mm each. The gap between the devices is 0.9 mm. Due to the monolithic integration and a simple fabrication process, the scalability of parallel fabricated sensor units is promising, leading to low cost and potentially disposable PON systems. As a consequence, the commercial viability can be significantly increased.
In the Experimental Section 2, we present in Section 2.1 the OLED-OPD device design and the fabrication procedure of the organic stacks. The experimental characterization of the OPD and OLED devices is given in Section 2.3. Section 2.4 presents the design of the nitrite sensing platform followed by the nitrite sensing experiments in Section 2.5. Conclusions are given in Section 3.

Device Characterization
The characteristics of one OPD are shown in Figure 2a

Device Characterization
The characteristics of one OPD are shown in Figure 2a The OLED emission spectrum was measured with a spectrometer (iHR320, Horiba, Kyoto, Japan) by using an optical fiber placed right above the OLED pixel. For optical and electrical characterization, we used a commercial calibrated Si-photodiode (FDS1010, Thorlabs GmbH, Lübeck, Germany) and two source measure units (SourceMeter 2450 and 2400, Keithley Instruments, Cleveland, OH, USA). The bottom-emitting OLED was placed on top of the large-area photodiode to collect the outcoupled light. The current density and the luminance were calculated with the recorded OLED current and the Si-photodiode photocurrent induced by the OLED. Figure 3a shows the spectrum of the green-emitting OLED. The peak wavelength is located at ~512 nm. The inset shows a photograph of The OLED emission spectrum was measured with a spectrometer (iHR320, Horiba, Kyoto, Japan) by using an optical fiber placed right above the OLED pixel. For optical and electrical characterization, we used a commercial calibrated Si-photodiode (FDS1010, Thorlabs GmbH, Lübeck, Germany) and two source measure units (SourceMeter 2450 and 2400, Keithley Instruments, Cleveland, OH, USA). The bottom-emitting OLED was placed on top of the large-area photodiode to collect the outcoupled light. The current density and the luminance were calculated with the recorded OLED current and the Si-photodiode photocurrent induced by the OLED. Figure 3a shows the spectrum of the green-emitting OLED. The peak wavelength is located at~512 nm. The inset shows a photograph of an OLED pixel operated at 9 V. Figure 3b shows the electrical and optical characterization of the Ir(mppy) 3 OLED. The onset voltage is approximately 3.3 V. The measurement was performed 4 times in sequence and showed good repeatability.
an OLED pixel operated at 9 V. Figure 3b shows the electrical and optical characterization of the Ir(mppy)3 OLED. The onset voltage is approximately 3.3 V. The measurement was performed 4 times in sequence and showed good repeatability. It is stressed that the spectrometer is only used for device characterization and reference measurements. No spectrometer is needed for the final nitrite-measurement system.

Design of the Test Chamber
For proof-of-principle demonstration of nitrite sensing with the OLED-OPD unit, we realized a test setup for contacting and fluid application. The OLED-OPD units were contacted with a printed circuit board (PCB) and elastomeric ZEBRA ® connectors. The proof of concept was performed with a custom rubber seal covered with a cap to ensure tightness and avoid leakage of the chemicals. The setup was used with a drilled cap ( Figure  4III) and the liquids were filled into the chamber with a pipette. The measurements were performed with two source measure units for OLED supplying and recording the photocurrent of the OPD in a darkened environment.

Nitrite Sensing with OLED-OPD Matrix
The photometric nitrite sensing was performed with a reagent based on the Griess reaction. Figure 5a shows the chemical transformation of N-(1-naphthyl)ethylenediamine dihydrochloride (NED) and sulfanilamide in an acidic solution. Sulfanilamide reacts with the nitrite ion and forms a diazonium salt. The azo dye formation occurs due to the reaction with NED. The absorbance around 540 nm strongly depends on the initial nitrite concentration. It is stressed that the spectrometer is only used for device characterization and reference measurements. No spectrometer is needed for the final nitrite-measurement system.

Design of the Test Chamber
For proof-of-principle demonstration of nitrite sensing with the OLED-OPD unit, we realized a test setup for contacting and fluid application. The OLED-OPD units were contacted with a printed circuit board (PCB) and elastomeric ZEBRA ® connectors. The proof of concept was performed with a custom rubber seal covered with a cap to ensure tightness and avoid leakage of the chemicals. The setup was used with a drilled cap ( Figure 4III) and the liquids were filled into the chamber with a pipette. The measurements were performed with two source measure units for OLED supplying and recording the photocurrent of the OPD in a darkened environment.

Nitrite Sensing with OLED-OPD Matrix
The photometric nitrite sensing was performed with a reagent based on the Griess reaction. Figure 5a shows the chemical transformation of N-(1-naphthyl)ethylenediamine dihydrochloride (NED) and sulfanilamide in an acidic solution. Sulfanilamide reacts with the nitrite ion and forms a diazonium salt. The azo dye formation occurs due to the reaction with NED. The absorbance around 540 nm strongly depends on the initial nitrite concentration.
As depicted in Figure 5b, we chose the organic composition of the OLED-OPD matrix for the highest absorbance of the green OLED light on the OPD. Due to total internal reflection (TIR) the light propagated partly inside the glass substrate and induced photocurrent in the OPD, since the OLED is an extended device with isotropic emission. The following Figure 6 shows a schematic of the expected azo dye concentration dependent signal change of the OLED-OPD matrix unit. In Figure 6a the light propagated partly inside the glass substrate and was partly reflected back onto the OPD after traveling inside the analyte. For lower sample concentration the partly absorbed light was reflected multiple times at the surfaces inside the liquid chamber and was partly reflected back onto the OPD. For higher concentration (Figure 6b) of nitrite standard sample and, consequently, higher concentration of the azo dye, the absorbance of the light inside the chamber increased. Thus, only a fraction of initial emitted amount of light was reflected back onto the OPD and the photocurrent consequently decreased.   As depicted in Figure 5b, we chose the organic composition of the OLED-OPD matrix for the highest absorbance of the green OLED light on the OPD. Due to total internal reflection (TIR) the light propagated partly inside the glass substrate and induced photocurrent in the OPD, since the OLED is an extended device with isotropic emission. The   As depicted in Figure 5b, we chose the organic composition of the OLED-OPD matrix for the highest absorbance of the green OLED light on the OPD. Due to total internal reflection (TIR) the light propagated partly inside the glass substrate and induced photocurrent in the OPD, since the OLED is an extended device with isotropic emission. The  reported a reaction time of 15 min to achieve the most stable results [39]. Pai and Yang evaluated the azo forming kinetics for different concentration of NED and the final acidity [40]. The evaluated t(90%) values were measured to be in the region of <1 min. The OLED was operated at constant current mode at 60 µA during all measurements. The photocurrent of the OPD was recorded at zero bias. Figure 7a,b shows the measurement of the premixed Griess reagent and nitrite standard solutions. Prior to the measurements with the OLED-OPD unit we tested the nitrite samples with a UV-Vis spectrometer (Lambda 800, PerkinElmer, Waltham, MA, USA). Figure 7a shows a calibration plot with a linearity of 99% as a reference. Figure  following Figure 6 shows a schematic of the expected azo dye concentration dependent signal change of the OLED-OPD matrix unit. In Figure 6a the light propagated partly inside the glass substrate and was partly reflected back onto the OPD after traveling inside the analyte. For lower sample concentration the partly absorbed light was reflected multiple times at the surfaces inside the liquid chamber and was partly reflected back onto the OPD. For higher concentration (Figure 6b) of nitrite standard sample and, consequently, higher concentration of the azo dye, the absorbance of the light inside the chamber increased. Thus, only a fraction of initial emitted amount of light was reflected back onto the OPD and the photocurrent consequently decreased.
(a) (b) Figure 6. Schematic of the azo dye concentration-dependent absorbance of OLED light inside the analyte. (a) The light propagated partly inside the glass substrate and was partly reflected back onto the OPD; (b) for higher concentrations of nitrite standard sample and consequently higher concentrations of the azo dye the light showed increased absorption inside the chamber at the azo dye pigments. Thus, only a fraction of initial amount of light was reflected back onto the OPD. The photocurrent decreased as a consequence.
The nitrite sensing experiments were performed with nitrite standard solutions at room temperature prepared 10 min prior to use. Although the azo dye formation was near instantaneous, according to the datasheet the reaction was accomplished after an incubation time of 10 min. The signal was stable for 1 h after adding the reagents. Brizzolari et al. reported a reaction time of 15 min to achieve the most stable results [39]. Pai and Yang evaluated the azo forming kinetics for different concentration of NED and the final acidity [40]. The evaluated t(90%) values were measured to be in the region of < 1 min. The OLED was operated at constant current mode at 60 µA during all measurements. The photocurrent of the OPD was recorded at zero bias. Figure 7a,b shows the measurement of the premixed Griess reagent and nitrite standard solutions. Prior to the measurements with the OLED-OPD unit we tested the nitrite samples with a UV-Vis spectrometer (Lambda 800, PerkinElmer, Waltham, MA, USA). Figure 7a shows a calibration plot with a linearity of 99% as a reference. Figure 7b shows the measurement with the investigated OLED-OPD sensing unit. We started each measurement with an empty fluid chamber and subsequently added samples of 200 µL ranging from 0.2 to 1.2 mg/L nitrite with a pipette. The Griess complex recorded the signal response in real time. The signal was a superposition of waveguided light, stray light and The system shows the highest photocurrent induced by the OLED light in the absence of the analyte in the fluidic chamber. Injecting the azo dye concentrated sample inside the chamber resulted in decreasing photocurrent due to the superposition of the refractive index change and the absorbance of the light at azo dye pigments. Since the critical angle for total internal reflection (TIR) inside the glass substrate increases for liquid analytes on the OLED-OPD matrix surface, the amount of light entering the analyte also increased and resulted in less guided stray light inside the glass substrate. For constant refractive indices of the analytes, the offset of the photocurrent was also constant and was consequently neglected during the measurement. The OPD signal at high absorption is determined with a highly concentrated nitrite sample (200 mg/L) to be 2.17 nA. As a further verification of the absorption inside the analyte, we evaluated the influence on the OPD photocurrent by adding a mirror on top of the OLED-OPD measurement unit. Due to reflection of the light back to the OPD we observed a small increase in the photocurrent for low concentrations of the nitrite samples, whereas adding a mirror for high nitrite concentrations resulted in no enhancement of the photocurrent. This is due to the high absorption of the high azo dye concentration. The system shows the highest photocurrent induced by the OLED light in the absence of the analyte in the fluidic chamber. Injecting the azo dye concentrated sample inside the chamber resulted in decreasing photocurrent due to the superposition of the refractive index change and the absorbance of the light at azo dye pigments. Since the critical angle for total internal reflection (TIR) inside the glass substrate increases for liquid analytes on the OLED-OPD matrix surface, the amount of light entering the analyte also increased and resulted in less guided stray light inside the glass substrate. For constant refractive indices of the analytes, the offset of the photocurrent was also constant and was consequently neglected during the measurement. The OPD signal at high absorption is determined with a highly concentrated nitrite sample (200 mg/L) to be 2.17 nA. As a further verification of the absorption inside the analyte, we evaluated the influence on the OPD photocurrent by adding a mirror on top of the OLED-OPD measurement unit. Due to reflection of the light back to the OPD we observed a small increase in the photocurrent for low concentrations of the nitrite samples, whereas adding a mirror for high nitrite concentrations resulted in no enhancement of the photocurrent. This is due to the high absorption of the high azo dye concentration. Table 1 shows a comparison between our investigated OLED-OPD unit and other recently reported photometric based nitrite sensing platforms. It shows that our proof-ofconcept detection system already has a promising limit of detection. Taking the step towards a fully integrated organic-semiconductor chip, our approach has the potential for parallel mass fabrication. Whereas the proof of concept was successfully performed, the analyte consumption was still rather high due to the large fluid chamber. The filling of the chamber requires approximately 100 µL to avoid light scattering due to air-liquid boundaries. Furthermore, operating the current setup in sunlight results in a significant photocurrent offset. This background signal can be subtracted in the post processing of the raw data or simply suppressed by using a non-transparent microfluidic device. Recently, we suggested employing a black absorptive material combined with a PDMS microfluidic for successful suppression of stray light [30]. Consequently, we aim to integrate the OLED-OPD matrix with a microfluidic device as the next step. Since the requirements for PON applications implicate a portable and wireless system, we also aim to upgrade the OLED-OPD sensing platform to a battery-powered system. A reliable hardware setup can be  Table 1 shows a comparison between our investigated OLED-OPD unit and other recently reported photometric based nitrite sensing platforms. It shows that our proofof-concept detection system already has a promising limit of detection. Taking the step towards a fully integrated organic-semiconductor chip, our approach has the potential for parallel mass fabrication. Whereas the proof of concept was successfully performed, the analyte consumption was still rather high due to the large fluid chamber. The filling of the chamber requires approximately 100 µL to avoid light scattering due to air-liquid boundaries. Furthermore, operating the current setup in sunlight results in a significant photocurrent offset. This background signal can be subtracted in the post processing of the raw data or simply suppressed by using a non-transparent microfluidic device. Recently, we suggested employing a black absorptive material combined with a PDMS microfluidic for successful suppression of stray light [30]. Consequently, we aim to integrate the OLED-OPD matrix with a microfluidic device as the next step. Since the requirements for PON applications implicate a portable and wireless system, we also aim to upgrade the OLED-OPD sensing platform to a battery-powered system. A reliable hardware setup can be achieved by integrating low-cost, state-of-the-art hardware, i.e., a transimpedance amplifier for an I/V conversion and an analog-to-digital converter (ADC) that translates the photocurrent in a digital dataset representing the magnitude of the induced current. The results can be processed with a microcontroller and transmitted to a monitor. Wang et al. recently reported a comprehensive design of a hardware circuit for an optoelectronic sensor [41].
Further performance improvements of the system can be achieved by tuning the organic stacks of the OLED-OPD unit. Green OLEDs with higher efficiency have been recently proposed [42].

Conclusions
We developed the first photometric nitrite sensor based on a fully organic optoelectronic chip. The organic composition of the OLED-OPD matrix was designed to have suitable peak positions for high absorbance measurements of the azo dye (Abs.~540 nm). We demonstrated a measurement of premixed nitrite standard solutions of different concentrations utilizing the Griess reagent. The calibration plot obtained from these measurements was found to be linear up to 1.2 mg/L, offering a linearity of 99%. The limit of detection (LOD) was calculated to be 46 µg/L (1.0 µM). This nitrite sensing unit is now ready for toxic threshold testing of drinking water or aquaculture systems. For quantification of concentrations below the current LOD, further improvements shall be investigated. As the next step, tuning of the organic stacks may serve for improvement of the system performance. The analyte consumption may be reduced by integrating the organic chip with a microfluidic unit. For Point-of-Need applications it is crucial to integrate the device with a portable and battery-powered system.
On the fabrication side, we already demonstrated the monolithic integration of 8 OLED-OPD pairs on a 12.5 mm by 12.5 mm substrate. Thus, this chip holds the potential for highly integrated multiplex sensing as well as the implementation of redundancy. In summary, employing organic optoelectronic devices in optical sensing approaches led to several benefits. The sensors can be manufactured on rigid or flexible substrates. The thermal evaporation technique enables a wide variety of geometrical shapes and device sizes leading to a high degree of design freedom. Particularly, the fabrication on flexible substrates is highly promising for large-scale roll-to-roll fabrication. The OLED-OPD unit holds great promise for use in a variety of applications ranging from bio-photonic sensors for human disease detection to environmental monitoring. Funding: This project has received funding partly from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (BEAMOLED, grant agreement No. 899861) and partly from the European Regional Development Fund (EFRE) by the European Union (OPTOCHIP, LPW-E/1.2.2/1303). We also acknowledge financial support by DFG within the funding programme Open Access Publizieren.

Conflicts of Interest:
The authors declare no conflict of interest.