We have developed a consistent process for growing ZnO nanorods. We start by evaporating a thin Au layer (20 Å) and annealing to form islands, which serve as the nucleation sites for the nanorods. The nanorods are then deposited by Molecular Beam Epitaxy (MBE) using high purity Zn metal and an O₃/O₂ plasma discharge as the source chemicals. After ∼2 h growth time at 600 °C, we have grown single-crystal nanorods with a typical length around 2∼10 μm and diameter in the range of 30-150 nm. Figure 1 shows a scanning electron micrograph of the as-grown rods. These alone are sensitive to hydrogen, but in some experiments we attempted to increase the sensitivity by sputtering Pd, Pt, Au, Ni, Ag or Ti thin films (∼100 Å thick) to form catalytic metal clusters.

Contacts to the multiple nanorods were formed using a shadow mask and sputtering of Al/Ti/Au electrodes. The separation of the electrodes was ∼30 μm. Au wires were bonded to the contact pad for current–voltage (I-V) measurements performed at 25 °C in a range of different atmospheres (N₂, O₂ or 10-500 ppm H₂ in N₂). A schematic of the resulting sensor is shown in Figure 2 (top) with an optical image of a wire-bonded device shown below. Note that no currents were measured through the discontinuous Au islands and no thin film of ZnO on the sapphire substrate was observed with the growth condition for the nanorods. Therefore the measured currents are due to transport through the nanorods themselves. The I-V characteristics from the multiple nanorods were linear with typical currents of 0.8 mA at an applied bias of 0.5 V.

Figure 3 shows the time dependence of resistance of either Pt-coated or uncoated multiple ZnO nanorods as the gas ambient is switched from N₂ to various concentrations of H₂ in air (10-500 ppm) as time proceeds. There is clearly a strong increase (approximately a factor of 5) in the response of the Pd-coated nanorods to hydrogen relative to the uncoated devices. The addition of the Pd appears to be effective in catalytic dissociation of the H₂ to atomic hydrogen. In addition, there was no response to the presence of O₂ in the ambient at room temperature, and the relative response of Pt-coated nanorods is a function of H₂ concentration in N₂. The Pd-coated CNTs detected hydrogen down to < 10 ppm, with relative responses of > 2.6% at 10 ppm and > 4.2% at 500 ppm H₂ in N₂ after 10 min exposure. By comparison, the uncoated devices showed relative resistance changes of ∼0.25% for 500 ppm H₂ in N₂ after 10 min exposure and the results were not consistent for lower concentrations. The gas sensing mechanism suggested include the desorption of adsorbed surface hydrogen and grain boundaries in poly-ZnO [83], exchange of charges between adsorbed gas species and the ZnO surface leading to changes in depletion depth [84] and changes in surface or grain boundary conduction by gas adsorption/desorption [85]. It should be noted that hydrogen introduces a shallow donor state in ZnO and this change in near-surface conductivity may also play a role.

Figure 4 shows the time dependence of resistance change of Pt-coated multiple ZnO nanorods as the gas ambient is switched from vacuum to N₂, oxygen or various concentrations of H₂ in air (10-500 ppm) and then back to air. This data confirms the absence of sensitivity to O₂. The resistance change during the exposure to hydrogen was slower in the beginning and the rate resistance change reached a maximum at 1.5 min of the exposure time. This could be due to the some of the Pd becoming covered with native oxide and then removed by exposure to hydrogen. Since the available surface Pd for catalytic chemical absorption of hydrogen increased after the removal of oxide, the rate of resistance change increased. However, the Pd surface gradually saturated with the hydrogen and resistance change rate decreased. When the gas ambient switched from hydrogen to air, the oxygen reacted with hydrogen right away, and the resistance of the nanorods changed back to the original value instantly.

Figure 5 shows the Arrhenius plot of the rate of nanorod resistance change. The rate of resistance change for the nanorods exposed to the 500 ppm H₂ in N₂ was measured at the different temperatures. An activation energy of 12 kJ/mole was extracted from the slope of the Arrehnius plot. This value is larger than that of the typical diffusion process. Therefore the dominant mechanism for this sensing process should be the chemisorption of hydrogen to the Pd surface.

Having established improved sensitivity by coating the nanorods with catalytic metal, we then investigated the effects of different metals to further improve sensitivity and response. Using the same coating procedure described above, we investigated metal coatings of Ti, Ni, Ag, Au, Pt, and Pd. Figure 6 shows the time dependence of relative resistance change of either metal-coated or uncoated multiple ZnO nanorods as the gas ambient is switched from air to 500 ppm of H₂ in N₂. These were measured at a bias voltage of 0.5 V. There is a strong enhancement in response with Pd, and to a lesser extent Pt coatings, but the other metals produce little or no change. This is consistent with the known catalytic properties of these metals for hydrogen dissociation. Pd has a higher permeability than Pt but the solubility of H₂ is larger in the former [86]. Moreover, studies of the bonding of H to Ni, Pd and Pt surfaces have shown that the adsorption energy is lowest on Pt [87].

The power requirements for the sensors were very low, which is a key requirement for a competitive marketable sensor. Figure 7 shows the I-V characteristics measured at 25 °C in both a pure N₂ ambient and after 15 min in a 500 ppm H₂ in N₂ ambient. Under these conditions, the resistance response is 8% and is achieved for a power requirement of only 0.4 mW. This compares well with competing technologies for hydrogen detection such as Pd-loaded carbon nanotubes [60,61].

The device layer structures were grown on C-plane Al₂O₃ substrates by Metal Organic Chemical Vapor Deposition (MOCVD). The layer structure included an initial 2 μm thick undoped GaN buffer followed by a 35 nm thick unintentionally doped Al_0.28Ga_0.72N layer. Mesa isolation was achieved by using an inductively coupled plasma system with Ar/Cl₂ based discharges. The Ohmic contacts were formed by lift-off of sputtered Ti/Al/TiB₂/Ti/Au, followed by annealing at 850 °C for 45 sec under a flowing N₂ ambient. A thin (100 Å) Pt Schottky contact was deposited by e-beam evaporation for the Schottky metal. The final step was deposition of e-beam evaporated Ti/Au interconnection contacts. The individual devices were diced and wire-bonded to carriers. These were then placed in our test chamber. Figure 8 shows a schematic of the completed devices and an optical image of a wire bonded device. Mass flow controllers were used to control the gas flow through the chamber, and the devices were exposed to either 100% pure N₂, or H₂ concentrations of 500 ppm down to 1 ppm in N₂ and temperatures from 25 to 500 °C.

Figure 9 shows the linear (top) and log scale (bottom) forward current-voltage (I-V) characteristics at 25 °C of the HEMT diode, both in air and in a 1% H₂ in air atmosphere. For these diodes, the current increases upon introduction of the H₂, through a lowering of the effective barrier height. The H₂ catalytically decomposes on the Pt metallization and diffuses rapidly to the interface where it forms a dipole layer. The differential change in forward current upon introduction of the hydrogen into the ambient is ∼1 mA over the voltage range examined.

To test the time response of the sensors, a 10% H₂/90% N₂ ambient was switched into the chamber through a mass flow controller for periods of 10, 20 or 30 seconds and then switched back to pure N_2. Figure 10 shows the time dependence of forward current at a fixed bias of 2 V under these conditions. The response of the sensor is rapid (< 1 sec), with saturation taking almost the full 30 seconds. Upon switching out of the hydrogen–containing ambient, the forward current decays exponentially back to its initial value. This time constant is determined by the volume of the test chamber and the flow rate of the input gases and is not limited by the response of the diode itself.

To further study response and detection limits, devices were tested under both forward and reverse bias conditions at room temperature (25 °C) in a nitrogen atmosphere at hydrogen concentrations ranging from 500 to 5 ppm, controlled by diluting the gas mix with nitrogen using mass flow controllers. There was again an increase in current under both forward and reverse bias conditions upon exposure to hydrogen, as shown in Figure 11. This is consistent with previously discussed mechanisms in which the hydrogen molecules dissociate into hydrogen atoms through the catalytic action of the Pt gate contact, and diffuse to the Pt/AlGaN interface [88,89]. The hydrogen atoms form a dipole layer, lower the Schottky barrier height, and increase net positive charges on the AlGaN surface as well as negative charges in the 2DEG channel. The calculated barrier height decrease for 500 ppm and 100 ppm hydrogen is 5 meV and 1 meV, respectively. The ideality factors were calculated to be 1.25 and 1.23 in 500 and 100 ppm hydrogen, respectively, compared to 1.26 in 100% nitrogen.

However, a plot of hydrogen sensitivity (defined as the drain current change over the initial drain current) versus bias voltage shows different characteristics for forward and reverse bias polarity conditions at 500 ppm of H₂, as shown in Figure 12. For the forward bias condition, there is a maximum sensitivity obtained around 1 V and further increase of bias voltage reduces the sensitivity. The sensitivity for the reverse bias condition is quite different and it increases proportionally to the bias voltage. We have proposed the following mechanism for the change in sensitivity under forward and reverse bias conditions: (1) The initial increase in the sensitivity is due to the Schottky barrier height reduction. (2) Further increase in forward bias allows electrons to flow across the Schottky barrier. These excess electrons bind with H⁺, form atomic hydrogen, and gradually destroy the dipole layer at the interface, therefore losing the hydrogen detection sensitivity. (3) For the reverse bias condition, electrons given away by the hydrogen atom may be swept across the depletion region. At higher reverse bias voltage, a higher driving force is applied to the electrons to move across the depletion region. Thus the dipole layer is amplified at the Pt/AlGaN interface for higher reverse bias voltage. Due to this dipole layer amplification, the detection sensitivity is enhanced at higher reverse bias voltage.

The detection sensitivity as a function of hydrogen concentration is shown in Figure 13. It is clear that the diodes are much more sensitive under reverse bias conditions. A detection limit of 100 ppm is achieved under forward bias, but the reverse bias detection limit is an order of magnitude lower, 10 ppm. The change in current at 10 ppm is 14% and over 200% at 500 ppm under reverse bias conditions, where forward bias operation results in changes of 25-75% over the 100-500 ppm range. This is consistent with published reports indicating improved sensitivity under reverse bias [90]. The reliability of the hydrogen sensor may be quite different under the two bias voltage polarities, since different degradation mechanisms in GaN devices are accelerated by either the presence of high voltage depletion regions (reverse bias) or current injection (forward bias) [91].

The bulk ZnO crystals from Cermet, Inc. showed electron concentration of 9 × 10¹⁶ cm^-3 and an electron mobility of 200 cm²/V. s. at room temperature from van der Pauw measurements. Ohmic contacts were formed on the back (O-face) of the substrates by depositing full area Ti/Al/Pt/Au by e-beam evaporation followed by annealing at 200 °C for 1 min in N₂ ambient. The front face was coated with plasma-enhanced chemical vapor deposited SiN_X at 100°C and windows opened by wet etching so that a thin (20 nm) layer of Pt could be deposited by e-beam evaporation to serve as the sensing metal. The final Ti/Au interconnect contacts were deposited, followed by dicing, wire-bonding, and mounting to the test apparatus. Figure 18 shows a schematic of the completed device.

Figure 19 shows the I-V characteristics at 50 and 150 °C of the Pt/ZnO diode both in pure N₂ and in ambients containing various concentrations of C₂H₄. At a given forward or reverse bias, the current increases upon introduction of the C₂H₄, through a lowering of the effective barrier height. The sensing mechanism is once again the decomposition of the C₂H₄ on the catalytic Pt surface, followed by diffusion to the underlying interface with the ZnO. As in hydrogen sensors, the atomic hydrogen forms an interfacial dipole layer that will reduce the Schottky barrier height, which will be manifested by a change in the DC I-V characteristics of the diode. The recovery was many orders of magnitude longer than for Pt/GaN diodes measured under the same conditions in the same chamber, suggesting that hydrogen diffuses into the ZnO lattice much more than it does for GaN. The changes in current at fixed bias or bias at fixed current were larger for the ZnO diodes than for AlGaN/GaN MOS diodes, which will be discussed in the next section, because of this additional detection mechanism, as shown in Figure 20. Note that the changes in these parameters are approximately an order of magnitude larger at 150 °C, suggesting that testing at even higher temperature will further improve sensitivity, but the ZnO diodes were not thermally stable above ∼300 °C.

AlGaN/GaN layer structures were grown on C-plane Al₂O₃ substrates by Metal Organic Chemical Vapor Deposition (MOCVD). The layer structure included an initial 2 μm thick undoped GaN buffer followed by a 35 nm thick unintentionally doped Al_0.28Ga_0.72N layer. The sheet carrier concentration was ∼1×10¹³ cm^-2 with a mobility of 980 cm²/V-s at room temperature. Device isolation was achieved with 2,000 Å plasma enhanced chemical vapor deposited SiNx. The ohmic contacts was formed by annealing e-beam deposited Ti/Al/Pt/Au. 400 Å Sc₂O₃ was deposited as a gate dielectric through a contact window of SiNx layer. Before oxide deposition, the wafer was exposed to ozone for 25 minutes. It was then heat in-situ at 300 °C cleaning for 10mins inside the growth chamber. 100 Å Sc₂O₃ was deposited on AlGaN/GaN by rf plasma-activated MBE at 100 °C using elemental Sc and plasma O₂ [109,110]. The thin Pt Schottky contact was deposited on the top of Sc₂O₃. Then, Ti/Au interconnection contacts were deposited, followed by dicing and wire-bonding. The devices were then mounted in the test chamber previously described. Figure 21 shows a schematic of the completed device.

Figure 22 shows the forward diode current-voltage (I-V) characteristics at 400 °C of the Pt/Sc₂O₃/AlGaN/GaN MOS-HEMT diode both in pure N₂ and in a 10% C₂H₄/90%N₂ atmosphere. The reason for testing at 400 °C will be discussed later. At a given forward bias, the current increases upon introduction of the C₂H₄. The sensing mechanism follows the same path as the hydrogen sensor - Hydrogen either decomposed from C₂H₄ in the gas phase or chemisorbed on the Pt Schottky contacts then decomposed and released hydrogen. The hydrogen diffused rapidly though the Pt metallization and the underlying oxide to the interface where it forms a dipole layer and lowered the effective barrier height.

Figure 23 shows both the change in current at fixed bias as a function of temperature for the MOS diodes when switching from a 100% N₂ ambient to 10% C₂H₄/90% N_2. As the detection temperature is increased, the response of the MOS-HEMT diodes increases due to more efficient cracking of the hydrogen on the metal contact. Note that the change in current is quite large and readily detected.