1.5: High Voltage MEMS Platform for Fully Integrated, On-Chip, Vacuum Electronic Devices

Srividya Natarajan1*,Charles B. Parker1, Jeffrey T. Glass1 ,Christopher A. Bower2, Kristin H. Gilchrist2, Jeffrey R. Piascik2, and Brian R. Stoner1,2

1) Department of Electrical and Computer Engineering Duke University

129 Hudson Hall, Durham, NC, USA, 27708 2) Center for Materials and Electronic Technologies

RTI International 3040 Cornwallis Road, Durham, NC, USA, 27709, stoner@rti.org

Abstract: We demonstrate a fully integrated, on-chip, vacuum microtriode capable of handling voltages up to 800 V. The ability to operate at such high voltages is achieved by the addition of a 10 μm-thick silicon dioxide layer to the device. The device is fabricated using MEMS fabrication principles and utilizes carbon nanotubes as field emitters. A dc amplification factor of 600 was obtained. To the best of our knowledge, this is the highest value reported for CNT-enabled microtriode devices. The high voltage capability of these microscale devices will enable their use in a wider variety of applications such as miniature ion sources and x-ray sources.

Keywords: MEMS; field emission; Carbon Nanotubes; microtriode.

Introduction Previously, Bower et al. [1] reported an on-chip vacuum microtriode using carbon nanotube field emitters. The technique combined the unique properties of carbon nanotubes with solid state fabrication technology to create miniaturized power amplifying vacuum devices. Conventional vacuum devices that use thermionic emission to generate electrons are limited by high power consumption and performance degradation due to high operating temperatures. Cold cathodes, on the other hand, require much lower operating power, because the cathode is not heated and hence, are naturally suited to miniaturization. Carbon nanotubes lend themselves to this application perfectly because of their size and field emission properties [2]. Further, the device used a 3-layer polycrystalline silicon MEMS process that enables complex structures to be built with relatively easy and reliable fabrication processes [3]. A silicon nitride insulating layer that is standard to the 3-layer polysilicon MEMS process was used in the device. In this abstract, we demonstrate the high voltage performance of an on-chip microtriode built using a similar MEMS process, with the addition of an insulating, 10 μm-thick silicon dioxide (SiO2) layer.

Materials and Methods The microtriode devices were built using a three-layer poly-Si micromachining process as described in earlier publications [1, 3]. A significant enhancement in these

devices was the inclusion of an insulating SiO2 layer for improved leakage performance. A 10 µm layer of high quality SiO2 was grown using high pressure thermal oxidation (HiPOx) on the wafers prior to the standard MEMS process. The microtriode consists of micro-scale poly-Si panels as the cathode, grid and anode. The panels were initially parallel to the substrate and buried in a highly doped sacrificial SiO2 layer. This SiO2 layer was etched away after fabrication to release the electrode panels. 50 Å of iron catalyst for multi-walled carbon nanotube (MWNT) growth was selectively patterned onto the cathode using an integrated poly-Si shadow mask. Vertically aligned MWNTs were selectively grown on the cathode by Microwave Plasma-enhanced CVD (MPECVD) using methane as the feedstock at 850 °C [4]. Following CNT growth, the panels were rotated into an upright position using micromanipulators and locked in place using integrated latches to secure each panel. The device was then mounted and wire bonded to a ceramic substrate for testing. A scanning electron micrograph of the assembled device is shown in Figure 1. The MWNT patterns were squares with 4 μm sides and a 10x10 grid of such patterns comprised the cathode. The cathode-to-grid spacing of this particular device was 40 μm before MWNT growth, and was reduced to 20 μm after a 20 μm-long MWNT film was deposited on the cathode. The MWNTs had an average diameter of 30-50 nm. The grid was a 3x3 array of 20 μm x 20 μm slits with 2.5 μm grid wire. The grid-to-anode spacing was 285 μm and the anode had an area of 150 μm by 150 μm.

Results The standard silicon nitride insulating layer used in the earlier device resulted in significant leakage at 100 V and complete breakdown at 200 V [1]. Therefore, an external anode was required for measurements beyond 200 V [5]. The improved devices reported here show a marked improvement in leakage and breakdown voltage characteristics. The leakage currents for the thick oxide devices are in the 10-100 pA range for voltages up to 500 V. Breakdown was observed at approximately 850 V for

978-1-4244-1715-5/08/$25.00 © 2008 IEEE 24

the thick oxide devices, suggesting that these devices could be operated at voltages up to 800 V without breaking down.

Figure 1 SEM of microtriode device

The microtriode was characterized in a vacuum chamber with a base pressure of 10-8 Torr. The field emission current-voltage characteristics of the microtriode are shown in Figure 2. The voltage on the grid was varied while the anode was held at 350 V. The turn-on field (for a current of 1 nA) for this device was 6 V/μm. The anode current followed the grid current, and the ratio of anode current to grid current, (Ia/Ig), ranged from 3 to 5. The inset shows the Fowler-Nordheim behavior of the cathode current, confirming field emission from the MWNTs. The highest total emission current, (Ia+Ig), measured for this device was 40 μA. For a cathode of area 1.6x10-5 cm2, this corresponds to a current density of 2.5 A/cm2. The transconductance, gm = δIa/δVg, a measure of the change in anode current for a small change in grid voltage, was calculated to be 2 μS at 29 μA of anode current.

The microtriode clearly exhibited cut-off, linear and saturation regions of operation. In the earlier silicon nitride devices, an external anode was required to achieve current saturation because the devices were not capable of handling the voltages necessary for saturation. The maximum output power measured at the anode was Pa = IaVa = (350 V)(29 μA) = 10.2 mW, which corresponds to an output power density of 45 W/ cm2. This is almost 5 times the maximum output power measured with the silicon nitride devices, and is expected to be a significant advantage in amplifier and general triode device applications. The internal anode resistance, Ra = δVa/δIa, was calculated to be 100 MΩ in the saturation region of the triode. The dc amplification factor, μ, which is the ratio of the power delivered at the anode to the power lost at the grid, was calculated using μ = Pa/Pg = (Ia

2Ra)/ (IgVg) = (Ia/Ig)gmRa. Using (Ia/Ig) = 3, gm = 2 μS and Ra = 100 MΩ, we obtained μ = 600. To the best of our knowledge, this is the highest value reported in literature for CNT-enabled microtriode devices. In comparison, a dc amplification factor of μ = 27 was obtained with the silicon nitride device. It should be noted

that the internal anode resistance for the earlier devices [1] was 10 MΩ, and they were not able to achieve saturation. The improvements in dc amplification factor and output power are attributed primarily to the presence of the SiO2 insulating layer.

In summary, we have demonstrated an integrated, on-chip vacuum microtriode capable of handling voltages up to 800 V. This MEMS high voltage platform is applicable to a wide variety of fully integrated multi-electrode devices, such as micro-scale x-ray sources, ion sources and hall thrusters. The authors have recently reported a microscale ion source achieving ion currents over 1 μA, which was built using this technology [6].

Figure 2 Anode current and grid current as a function

of grid voltage with the anode held at 350 V. The inset shows the Fowler-Nordheim behavior of the

cathode current

References 1. C. Bower, W. Zhu, D. Shalom, D. Lopez, L. H. Chen,

P. L. Gammel and S. Jin, Applied Physics Letters, Vol. 80, 3820 (2002).

2. W. Zhu, C. Bower, O. Zhou, G. P. Kochanski and S. Jin, Applied Physics Letters, Vol. 75, 873 (1999).

3. J. Carter, A. Cowen, B. Hardy, R. Mahadevan, M. Stonefield, and S. Wilcenski, The PolyMUMPS Design Handbook, MEMSCAP Inc., Durham, NC, 2005

4. H. Cui, O. Zhu and B. R. Stoner, Journal of Applied Physics, Vol 88, 6072 (2000).

5. C. Bower, D. Shalóm, W. Zhu, D. López, G. P. Kochanski, P. L. Gammel and S. Jin, IEEE Trans. Elect. Dev., Vol. 49, 1478(2002).

6. C. Bower, K. H. Gilchrist, J. R. Piascik, B. R. Stoner, S. Natarajan, C. B. Parker and J. T. Glass, Applied Physics Letters, Vol. 90, 124102 (2007).

978-1-4244-1715-5/08/$25.00 © 2008 IEEE 25