Highly sensitive hydrogen sensor based on graphite-InP or graphite-GaN Schottky barrier with electrophoretically deposited Pd nanoparticles
© Zdansky; licensee Springer. 2011
Received: 12 May 2011
Accepted: 10 August 2011
Published: 10 August 2011
Depositions on surfaces of semiconductor wafers of InP and GaN were performed from isooctane colloid solutions of palladium (Pd) nanoparticles (NPs) in AOT reverse micelles. Pd NPs in evaporated colloid and in layers deposited electrophoretically were monitored by SEM. Diodes were prepared by making Schottky contacts with colloidal graphite on semiconductor surfaces previously deposited with Pd NPs and ohmic contacts on blank surfaces. Forward and reverse current-voltage characteristics of the diodes showed high rectification ratio and high Schottky barrier heights, giving evidence of very small Fermi level pinning. A large increase of current was observed after exposing diodes to flow of gas blend hydrogen in nitrogen. Current change ratio about 700,000 with 0.1% hydrogen blend was achieved, which is more than two orders-of-magnitude improvement over the best result reported previously. Hydrogen detection limit of the diodes was estimated at 1 ppm H2/N2. The diodes, besides this extremely high sensitivity, have been temporally stable and of inexpensive production. Relatively more expensive GaN diodes have potential for functionality at high temperatures.
Keywordshydrogen sensor metal nanoparticles electrophoresis Schottky barrier InP GaN
Hydrogen gas (H2) monitoring sensors are in demand mainly for detection of H2 leakage in many industry productions such as, H2 filling stations, cryogenic cooling, research labs, etc. The gas is odorless, colorless, and highly inflammable, and therefore, effective H2 sensors are of great need for safety reasons. Highly sensitive and selective (i.e., exclusive to one gas) H2 sensors are needed in forming gas leak detectors for testing leaks in various equipment like vacuum apparatuses, refrigerators, heat exchangers or fuel systems in cars, etc. Such detectors contain highly sensitive H2 sensors and forming gas (noncombustive mixture of H2 in nitrogen) in place of expensive helium (the price of helium has recently risen sharply due to increased demand and limited resources) .
Thus, research on new H2 sensors has been well stimulated. Sensors based on semiconductor Schottky barriers principally exceed in sensitivity over the best results reported by sensors based on other sensing principles. The advantages of these sensors are also long life, low cost, and easy large-scale production. Palladium (Pd)/Si H2 sensors with two to three orders-of-magnitude change in current for 150 ppm of H2 in nitrogen (N2) were published already in 1981 . About twice higher in sensitivity has been achieved with Pd/InP using electrophoretic deposition of Pd . High sensitivity with about six orders-of-magnitude response to 5,000 ppm H2 in N2 has been achieved with porous Pd/GaN Schottky sensors . It has been shown on the Pd/Si Schottky sensor that it responds linearly to H2 concentration in the range of three orders-of-magnitude, while the response starts to saturate above 1% of H2 in N2 and decreases faster below 10 ppm . Similar behavior can be expected at other Schottky barrier sensors as well.
Nanoparticles of palladium or platinum are suitable for making H2 sensors based on Schottky barriers, intended to operate at room temperature. The reason is in the catalytic affectivity of these metal nanoparticles for dissociation of H2 molecules on metal-semiconductor interface. Ionized hydrogen atoms (protons), can form electric interface double layer with free electrons in the semiconductor, changing the height of the barrier which strongly affects barrier's electric properties. It has been shown that Pd/InP H2 sensors made by electrophoretic deposition of Pd nanoparticles are more sensitive than those made by thermal evaporation of Pd or even by electroless plating .
Recently, it was shown in our lab that the best H2 sensitivity of InP- or GaN-based structures could be achieved by combining electrophoresis of Pd nanoparticles with mechanical deposition of colloidal graphite for making Schottky contacts . In this letter, the author reports on further studies of these structures.
Colloid solutions of Pd nanoparticles (NPs) in isooctane were prepared by reverse micelle technique with surfactant of sodium bis-(2-ethylhexyl) sulfoccinate (AOT) from water solutions of palladium chloride (PdCl2) and reducing agent hydrazine . Shapes of Pd NPs in the colloid solution were monitored by a transmission electron microscope and/or by a scanning electron microscope (SEM). The Pd NPs were spherical, 7 nm in diameter, with 10% dispersion. Optical absorption peak due to surface-plasmon-resonance of Pd NPs in isooctane at 280 nm wave length was monitored by a split-beam photospectrometer. Pure chemicals and polished n-type InP and GaN (doping levels 2.5 × 1015 and 2 × 1017 cm-1) crystal wafers were purchased from recognized commercial companies as stated previously . Each crystal wafer of 10 × 10 mm2 size was first shortly treated in boiled methanol and then the unpolished side was procured with the all-area ohmic contact at room temperature by rubbing liquid solution of tin in gallium with a tin rod and a cotton-wool swab. Electrophoretic depositions (EPD) of Pd NPs onto polished InP or GaN were performed from the colloid solution by applying an electric field of 2,000 V/cm for 50 ms, with a 100-ms period for sufficient time. The field was held with the negative pole on the semiconductor wafer (sample) and the positive pole on the plane-parallel graphite electrode built in the tightly closed teflon cell . The deposited layers on InP and GaN crystal wafers with Pd NPs were observed by SEM.
Schottky contacts were provided on the polished sides of the wafers (Pd NPs deposited or not deposited) by painting droplets of colloidal graphite in separate spots using a soft teflon needlepoint. The contacts were photographed on an optical microscope with Nomarski contrast and the photos were converted to the digital form for estimating contact areas. For that purpose, the digitized photos were modified to get the image with black background and white contact area, converted to the matrix form, and the contact area was integrated using a program on a computer.
Each Schottky contact and the all area ohmic contact on the other side of the wafer formed a rectifying Schottky diode. For measuring electrical properties, the diode was placed with the ohmic contact on a conducting platform, and a golden needlepoint on a bonze spring was touched on the Schottky contact, in the measuring cell. The cell was constructed with two holes to enable gas through-flow with free outlet to ambience for measuring gas sensitivity of electronic devices like H2 sensors.
All forward current-voltage characteristics show distinct linear parts in the semi-log scale in Figure 5. Using these linear parts, the Schottky barrier heights and ideality factors (IF) were evaluated as described in Ref. . The height value was 0.873 eV and the ideality factor was 1.08 for InP-Pd-C diode, giving evidence that the thermionic emission primarily governed the electron transport in this case. In the case of GaN-Pd-C diode, the IF exited at 1.74 showing that a generation-recombination current (IF = 2) added to the thermionic emission current (IF = 1). When the linear part of the current-voltage curve of GaN-Pd-C diode was fitted with both currents added, the barrier height was estimated at 1.14 eV. The values of Richardson constants used in the evaluation were 9.24 and 26.4 A/(K·cm)2 for InP and GaN, respectively.
The diode InP-C, made by graphite on the plain InP, was also tested on the hydrogen sensitivity. However, there was no change of current when such voltage-biased diode was exposed to a gas containing hydrogen.
The Schottky diodes obtained by application of colloidal graphite on n-type InP and n-type GaN, marked InP-C, InP-Pd-C, and GaN-Pd-C are inexpensive but of very high quality, having low reverse leakage currents and high rectification ratios. Schottky barrier heights of 0.873 and 1.14 eV of InP-Pd-C and GaN-Pd-C diodes are much higher than those obtained by other methods, like thermal evaporation, which was, e.g., 0.55 eV in the case of Pd onto InP . It shows on a very small or virtually negligible Fermi level pinning in these diodes so that any change in the Schottky barrier height should be equal to the change of the work function caused by an external charge appearing at the interface. Indeed, the measured values of Schottky barrier heights of 0.873 or 1.14 eV are close to differences between the electron work function of palladium metal 5.12 eV  and the electron affinity of InP, 4.38 eV  (0.74 eV) or of GaN, 4.1 eV  (1.02 eV). In principal, Fermi level pinning is caused by interface states in a semiconductor near the intimate contact with the metal. There are two basic ways creating interface states, physical breakings  and chemical reactions . I believe that elimination of chemical reactions, due to forming Schottky barriers with colloidal graphite and surfactant wrapped Pd NPs, is the main reason for the absence of Fermi level pinning in the prepared diodes.
Only with small Fermi level pinning can Schottky diodes form effective gas sensors. Hydrogen sensing mechanism works as follows. H2 molecules penetrate through the porous graphite contact to the surface of the semiconductor where they are dissociated to hydrogen atoms due to the catalytic effect of the present Pd NPs. Positively charged hydrogen atoms (protons) after dissociation are attracted by electrons in the n-type semiconductor and form dynamically changing electric double layer. This electric double layer decreases the work function of the Schottky contact material and, consequently, it decreases the Schottky barrier height and increases the current of the voltage-biased diode. There are several favorable factors explaining the high H2 sensitivity of the diodes. First factor is the high Schottky barrier height and low leakage current of the interface between the graphite and InP or GaN semiconductor. Second factor is surfactant wrapped Pd NPs in concentration just partly covering the semiconductor surface, which does not lead to a serious decrease of the Schottky barrier height formed by the graphite and simultaneously it is sufficient for dissociation of penetrated hydrogen molecules. Third factor is the porous state of the graphite contact which allows easy penetration of hydrogen molecules to the interface with the Schottky barrier. Due to the abovementioned favorable factors, the current change ratio of the prepared diodes after exposure to 0.1% H2/N2 was 7 × 105 which represents more than two orders-of-magnitude improvement over the best result reported previously by H2 nanosensors .
Both types of diodes, based on InP and GaN, show about the same response and recovery time developments. The recovery time development consists of two parts: faster, just after the change from H2 to air exposure and a slower tail consisting of about 10% of recovering current change at the end, caused probably by slow release of H2 from the crystal lattice of Pd NPs . The slow release shows that H2 in the crystal lattice of Pd is chemically bound in the form of palladium hydride (PdHx) . This notion is supported by our further observation that the recovery tail is suppressed in cognate diodes with Pt NPs in place of Pd ones, which is in agreement with known experimental observations of PdHx and no observation of PtHx.
The current of InP-based diodes is more than one order-of-magnitude larger than the current of GaN-based diodes (see Figures 6 and 7), but the ratio of current change due to exposure to H2 is about the same for both types of diodes. InP diodes are advantageous for H2 measurements at room temperature due to their lower cost and larger current needing less laborious electronics. Beyond, more expensive GaN diodes are predicted for H2 measurements at high temperatures.
It should be noted that besides approximately 7 nm also approximately 10 nm Pd NPs were used to fabricate hydrogen sensors, and no demonstrable difference in their sensitivity was observed. It is important for prospective applications that the diodes are temporally stable as it is seen on the curve (2) in Figure 7. Good temporal stability of both, current-voltage characteristics and current transient responses to H2 exposure, has been proven for InP-Pd-C and GaN-Pd-C diodes.
The Schottky diodes were prepared on polished single crystals of n-type InP or n-type GaN by painting colloidal graphite on the surface previously partly deposited with Pd NPs. The Pd NPs were deposited by electrophoresis from colloid solutions in isooctane prepared by chemical reduction of Pd-salt water solution in reverse micelles. The Schottky diodes showed current-voltage characteristics of low leakage currents and large rectification ratios, and they were much sensitive to H2 exposure with more than two orders-of-magnitude improvement over the best result reported previously . Hydrogen detection limit of reported diodes was estimated at 1 ppm H2/N2.
The author thanks O. Cernohorsky for preparing colloid solutions and SEM imaging, and J. Zelinka for the help. The works were financially supported by the Academy of Sciences of the Czech Republic, grant KAN401220801 in the program Nanotechnology for Society and by program COST EU, Action MP0805, grant OC10021 of the Ministry of Education, Czech Republic.
- Block M: Hydrogen as tracer gas for leak testing. Proceedings of the 9th European Conference on Nondestructive Testing (ECNDT): Sept. 25–29, 2006; Berlin. p.Tu.2.6.1 http://www.schoonoverinc.com/; http://en.foerch.com/; http://www.adixen.com/.
- Ruths PF, Ashok S, Fonash SJ, Ruths JM: A study of Pd/Si MIS Schottky barrier diode hydrogen detector. IEEE Trans Electron Devices 1981, 28: 1003.View Article
- Chou YI, Chen CM, Liu WC, Chen HI: A new Pd-InP Schottky hydrogen sensor fabricated by electrophoretic deposition with Pd nanoparticles. IEEE Electron Device Lett 2005, 26: 62.View Article
- Chiu SY, Huang HW, Liang KC, Huang TH, Liu KP, Tsai JH, Lour WS: High sensing response Pd/GaN hydrogen sensors with a porous-like mixture of Pd and SiO2. Semicond Sci Technol 2009, 24: 045007. 10.1088/0268-1242/24/4/045007View Article
- Skucha K, Fan Z, Jeon K, Javey A, Boser B: Palladium/silicon nanowire Schottky barrier-based hydrogen sensors. Sens Actuators B 2010, 145: 232. 10.1016/j.snb.2009.11.067View Article
- Zdansky K, Yatskiv R, Grym J, Cernohorsky O, Zavadil J, Kostka F: Study of electrophoretic deposition of Pd metal nanoparticles on InP and GaN crystal semiconductors for H2-gas sensors. In Proc 2nd NANOCON International Conference: Oct 12–14, 2010; Olomouc, Czech Republic. Edited by: Radek Zboril. TANGER Ltd, Ostrava, Czech Republic; 2010:182. ISBN: 978–80–87294–19–2 ISBN: 978-80-87294-19-2
- Zdansky K, Kacerovsky P, Zavadil J, Lorincik J, Fojtik A: Layers of metal nanoparticles on semiconductors deposited by electrophoresis from solutions with reverse micelles. Nanoscale Res Lett 2007, 2: 450. 10.1007/s11671-007-9085-1View Article
- Zdansky K, Zavadil J, Kacerovsky P, Lorincik J, Vanis J, Kostka F, Cernohorsky O, Fojtik A, Reboun J, Cermak J: Electrophoresis deposition of metal nanoparticles with reverse micelles onto InP. Int J Mat Res 2009, 100: 1234.View Article
- Hasegawa H: Fermi level pinning and Schottky barrier height control at metal-semiconductor interfaces of inp and related materials. Jpn J Appl Phys 1999, 38: 1098. 10.1143/JJAP.38.1098View Article
- Periodic Table of Elements[http://environmentalchemistry.com/yogi/periodic/Pd.html]
- InP - Indium Phosphid Basic Parameters at 300 K[http://www.ioffe.ru/SVA/NSM/Semicond/InP/basic.html]
- GaN - Gallium Nitride Basic Parameters for Wurtzite Crystal Structure[http://www.ioffe.ru/SVA/NSM/Semicond/GaN/basic.html]
- Brillson LJ, Mosbacker HL, Hetzer MJ, Strzhemechny Y, Look DC, Cantwell D, Zhang J, Song JJ: Surface and near-surface passivation, chemical reaction, and Schottky barrier formation at ZnO surfaces and interfaces. Appl Surf Sci 2008, 254: 8000. 10.1016/j.apsusc.2008.03.050View Article
- Skoskiewicz T: Superconductivity in palladium-hydrogen and palladium-nickel-hydrogen systems. Phys Stat Sol (a) 1972, 11: K123. 10.1002/pssa.2210110253View Article
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.