)
Poya Yasaei,† Amirhossein Behranginia,† Tara Foroozan,‡ Mohammad Asadi,† Kibum Kim,†,§ Fatemeh Khalili-Araghi, and Amin Salehi-Khojin*,† †
ARTICLE
Stable and Selective Humidity Sensing Using Stacked Black Phosphorus Flakes )
Department of Mechanical and Industrial Engineering, ‡Department of Civil and Materials Engineering, and Department of Physics, University of Illinois at Chicago, Chicago, Illinois 60607, United States and §Department of Mechanical Engineering, Chungbuk National University, Cheongju, 361-763, South Korea
ABSTRACT Black phosphorus (BP) atomic layers are known to undergo
chemical degradation in humid air. Yet in more robust configurations such as films, composites, and embedded structures, BP can potentially be utilized in a large number of practical applications. In this study, we explored the sensing characteristics of BP films and observed an ultrasensitive and selective response toward humid air with a trace-level detection capability and a very minor drift over time. Our experiments show that the drain current of the BP sensor increases by ∼4 orders of magnitude as the relative humidity (RH) varies from 10% to 85%, which ranks it among the highest ever reported values for humidity detection. The mechanistic studies indicate that the operation principle of the BP film sensors is based on the modulation in the leakage ionic current caused by autoionization of water molecules and ionic solvation of the phosphorus oxoacids produced on moist BP surfaces. Our stability tests reveal that the response of the BP film sensors remains nearly unchanged after prolonged exposures (up to 3 months) to ambient conditions. This study opens up the route for utilizing BP stacked films in many potential applications such as energy generation/storage systems, electrocatalysis, and chemical/ biosensing. KEYWORDS: two-dimensional (2D) materials . black phosphorus . phosphorene . humidity sensing . liquid-phase exfoliation
B
lack phosphorus (BP) is the most thermodynamically stable allotrope of phosphorus with an orthorhombic layered structure and highly anisotropic properties.18 Recent exfoliation of BP into atomic layers (called phosphorene) has demonstrated its potential for use in electronics and optoelectronics, owing to its high charge carrier mobility,5,7,912 tunable direct bandgap,9,13,14 large on/off ratios (>105),5,9 and anisotropic properties.5,6,15 BP also holds promise for use in energy generation1619 and storage systems,2023 electrocatalysis,22,24 and chemical/biosensing.25,26 In most of these applications, the material does not necessarily need to be atomically thin, but rather should be in the form of thin films, composites, or embedded structures. In this context, liquid exfoliation has recently been utilized to produce high-quality BP nanoflakes (NFs) with decent control over the concentration and flake thickness.2628 However, the material is yet to be proven appropriate for long-term applications in view of the YASAEI ET AL.
observed ambient instability of the atomically thin flakes.29,30 RESULTS/DISCUSSION In this report, we discovered that films of BP NFs exhibit excellent sensitivity and selectivity for humidity detection with quick recovery characteristics. The impedance spectroscopy and electrical characterizations suggest that the sensing mechanism of the BP film sensors is based on modulation in the leakage ionic current. While previous studies on BP demonstrate the ambient instability for the devices made by atomically thin flakes,28,29,31,32 our study reveals highly stable sensing characteristics of BP films after prolong exposure to humid environments. The BP films used in this study are produced in large quantities by the liquid exfoliation method via sonication of the ground bulk BP (powder) in dimethylformamide (DMF) or dimethyl sulfoxide (DMSO) solvents (for details see the method section and ref 27). Figure 1a shows a uniform and VOL. XXX
’
NO. XX
’
* Address correspondence to
[email protected]. Received for review June 2, 2015 and accepted September 24, 2015. Published online 10.1021/acsnano.5b03325 C XXXX American Chemical Society
000–000
’
XXXX
A www.acsnano.org
stable dispersion of BP NFs in DMF after the sonication. To separate the NFs from the bulky particles, the solutions are centrifuged after the sonication, and their supernatants are collected (shown in Figure 1b). To make films of stacked BP NFs, hydrophilic polytetrafluoroethylene (PTFE) filter papers with 0.1 μm pore size are used in a vacuum filtration setup (Figure 1c), and the filtered NFs are thoroughly rinsed with ethanol and IPA to remove the solvent residues. Electron energy loss spectroscopy and electron diffraction X-ray spectroscopy experiments on similarly produced flakes in our former report have revealed that the surfaces of the flakes are fairly clean.27 Scanning electron microscopy (SEM) images of the NFs on the membrane show densely packed and uniformly distributed BP NFs (Figure 1d). Figure 1e shows the crosssection SEM view of a film made by filtering 3 mL of a typical DMF solution with an estimated thickness of ∼26 μm. The films are characterized by Raman point spectroscopy (Figure 1f), and three typical BP spectral peaks of A1g (out-of-plane mode) and B2g and A2g (in-plane modes) were observed at wavenumbers of ∼360, ∼ 437, and ∼466 cm1, respectively.7,30,32 The peak positions are consistent with the signature spectrum of bulk BP, suggesting that the flakes remain crystalline after exfoliation.30 This is also supported by atomic resolution transmission electron microscopy imaging on similarly produced NFs in our earlier report.27 YASAEI ET AL.
VOL. XXX
’
NO. XX
’
000–000
’
ARTICLE
Figure 1. Photograph of the BP dispersions in DMF solvent (a) after sonication and (b) after centrifugation and supernatant collection. (c) Optical image of a film of stacked BP NFs prepared by vacuum filtration on a PTFE membrane. (d) SEM image of the filtered NFs, showing tightly stacked structure of the film. Scale bar is 2 μm. (e) SEM image of the film cross-section. Scale bar is 20 μm. (f) A representative Raman point spectrum obtained from the vacuum filtered film.
Next, the BP films are cut into desired sizes, and twoprobe electrical contacts are established using either galliumindium eutectic (inset of Figure 2a) or conductive gold paste (see section S1 in SI file). The pulse injection method (dynamic sensing) was used to test the sensing performance of the fabricated films33,34 (see Methods/Experimental section). Figure 2a shows a typical current modulation of the sensor in response to injection of 2 nL of selected chemicals including water vapor, alcohols (ethanol, isopropanol), ketones (toluene, acetone), and benzenes (dichlorobenzene) at a constant (DC) applied bias of 0.5 V in identical experimental conditions. Interestingly, we observed ∼5-fold enhancement in the drain current upon injection of the water vapor, while the response to all other tested analytes was at least 2 orders of magnitude smaller. Also, the sensors show negligible response upon exposure to direct flow of hydrogen (H2), oxygen (O2), and carbon dioxide (CO2) gases, tested up to 20 standard cubic centimeters (sccm). The selective response of the BP film against water vapor is highly important for practical humidity detection without cross-sensitivity and false-positive issues. The response of the films also fully recovers within 15 s of the injection, depending on the volume of the water vapor pulse. We compared the water vapor sensing characteristics of BP sensors to that of polycrystalline monolayer graphene and molybdenum disulfide (MoS2) grown by chemical vapor deposition33,3538 as well as graphene and MoS2 films of stacked flakes made by liquid exfoliation3941 (see section S2 in SI file). Figure 2b shows the sensitivity (defined as S = ((I I0)/I0)%) for these sensors with respect to reciprocal recovery time (1/T) upon injection of 0.512.5 nL of water vapor. The results for stacked film of MoS2 NFs are not included in Figure 2b as we did not observe a noticeable response for the range of concentrations used in this study. Under identical conditions, BP sensors exhibit more than 2 orders of magnitude higher sensitivity and more than 2-fold faster recovery in comparison to all other tested nanomaterial-based sensors. The performance of BP films was also tested by stepwise increasing of the humidity in a custom-made environmental chamber equipped with reference temperature and humidity sensors (static sensing experiments) (see section S3 in SI file). As shown in Figure 2c, the drain current of the sensor increases by ∼4 orders of magnitude as the relative humidity (RH) varies from 10% to 85%. To the best of our knowledge, this level of sensitivity stands among the highest ever reported values for humidity detection.4247 Similar sensitivity levels have also been observed in multiple tested devices; however, the calibration curves of the sensors show a slight device-to-device variation. Thus, for precise humidity measurements the sensors shall be individually calibrated.
B
XXXX www.acsnano.org
ARTICLE Figure 2. (a) Response of the stacked BP NFs to different analytes. The inset (right) magnifies the same curves. The curves have the same baseline, but are offset for clarity. The inset (left) shows a typical BP film sensor fabricated on PTFE film on a scotch tape for mechanical support and with GaIn eutectic contacts. (b) Sensitivity vs reciprocal of recovery time (1/T) for 4 different sensors upon exposure to different concentrations of water vapor in identical experimental conditions. The error bars represent the standard deviation of multiple experiments. (c) The current of a typical BP sensor vs RH at 25 °C. The inset shows the scheme of the custom-made chamber that is used for the experiment. (d) Effect of temperature on the response of the sensor in an isolated environment. The upper plot shows the RH and temperature obtained from a reference sensor. The lower plot shows the calculated absolute humidity and the response of the device. This plot shows that the sensor is only sensitive to the absolute humidity and is almost insensitive to temperature.
Since RH strongly correlates with temperature, the temperature-dependent response of the humidity sensors needs to be carefully characterized. Figure 2d (top) shows the temperature and relative RH relationship as measured by a commercial reference sensor. To study the temperature effect, the absolute humidity of the chamber was kept constant (sealed chamber), and the temperature was varied from 20 to 50 °C. Our results clearly indicate that the BP film is insensitive to temperature. Moreover, we found a strong correlation between the absolute humidity and the response of the sensors as shown in Figure 2d (bottom), further confirming the selectivity of BP film against humidity. The performance of BP films was also tested with respect to the pressure of humid air. Figure 3a shows the time-dependent response of our sensor loaded in a vacuum chamber (red) as well as the pressure of the chamber measured by a reference digital vacuum gauge (blue) during an evacuation and refill cycle. As shown in Figure 3a, the drain current of the sensors exponentially decreases as one evacuates the test chamber. Through these simultaneous measurements, the response of the sensor with respect to pressure is extracted and shown in Figure 3b for two RH levels of YASAEI ET AL.
27% and 67%. The extrapolated intersection of these two curves at ∼2 105 mbar offers a wide-range of detection capability (∼2 105 to 103 mbar) for the pressure. This unique feature of our sensors together with their remarkable sensitivities makes the BP films potentially an ideal structure for trace-level humidity measurements in low pressure systems. In the next phase, we analyzed the sensing data and performed additional experiments to gain insight into the operation principle of the BP film sensors. First, we noted that the BP film sensors are selective to only humidity and the responses upon exposure to polar analytes (tested in Figure 2a) are negligible. This selective response to water molecules implies that the effect of a typical charge-transfer mechanism (as in the case of typical chemical sensors) can be ruled out. Otherwise, the tested analytes with high polarity would have shown significant responses, which indeed is not the case here. It is also important to study the effect of contacts on the sensing response of BP films as the electrical transport properties of BP devices is known to be affected by the work function of the contacts.48 For this purpose, we tested the sensing response of the BP films using different contact electrodes (e.g., gold) VOL. XXX
’
NO. XX
’
000–000
’
C
XXXX www.acsnano.org
instead of the InGa eutectic. As discussed in section S4 of the SI file, similar sensitivities were obtained using different metal electrodes which implies that the response is not dominated by the modulation of contact resistances (i.e., Schottky barrier modulation). We also performed pulse injection sensing experiments on BP film sensors with passivated either electrodes or channel. These results (shown in Figure 4a) further confirm the minor role of the contacts. Considering the above-mentioned arguments, we speculate that modulation in ionic conduction mainly governs the operation principle of the BP film sensors. Similar mechanism has also been proposed for previously reported humidity sensors such as the stack of graphene oxide (GO) NFs, reduced graphene oxide (rGO) nanocomposites, and nanoporous polymer membranes.42,44,46,49 To test this hypothesis, we performed cyclic currentvoltage (IV) experiments in different RH levels and at different scan rates. Interestingly, we observed a scan-rate dependency (Figure 4b) which is a characteristic of capacitive behavior (charge storage). As the humidity increases, the slope of the IV trend increases, but the hysteresis becomes less dependent on the scan rate (see section S6 in SI file). This behavior perfectly resembles the characteristics of nonideal (commercial) electrochemical capacitors consisting of a pure capacitance and a residual (leakage) resistance. Moreover, the response of the film to a step YASAEI ET AL.
VOL. XXX
’
NO. XX
’
000–000
’
ARTICLE
Figure 3. (a) Response of the sensor in the transient time of evacuating and refilling the vacuum chamber. The pressure (blue line) is measured by a reference digital vacuum gauge. (b) Response of the sensor with respect to pressure at initial humidity levels of 27% and 67%. The inset shows the black curve. The sensor is sensitive to changes in the pressure for a broad range from 105 to 103 mbar.
voltage is an exponential decay of the current with a residual leakage which is consistent with typical behavior of electrochemical systems (see Figure S6f in SI file). In such a case, the capacitive response is associated with the formation of the electric double layer and the residual leakage current corresponds to the ionic charge transport. To better understand the effects of resistive and capacitive contributions on the overall response of the BP film sensors, we performed impedance spectroscopy (IS) experiments by sweeping the frequency from 100 Hz to 10 MHz. As shown in Figure 4c, a semicircular trend was observed in the Nyquist representation of the results which can be fitted by the response of a parallel RC circuit (inset of Figure 4d). The equivalent series resistance compared with the leakage (charge transfer) resistance in our IS results is notably smaller, hence a parallel RC circuit (inset of Figure 4d) can sufficiently model the experimental data. Figure 4d shows the modulation of the resistance and capacitance in the modeled equivalent circuit as a function of RH. It is observed that as the RH increases from 20% to 90%, the resistance decreases more than 450 times, while the capacitance only varies by
