A Gas Sensor Array Using Carbon Nanotubes and Microfabrication Technology Jing Li, * ,z Yijiang Lu, Qi Ye, Lance Delzeit, and M. Meyyappan ** NASA Ames Research Center, Center for Nanotechnology, Moffett Field, California 94035, USA A nanosensor technology has been developed using single-walled carbon nanotubes SWNTscombined with silicon-based microfabrication processes. A sensor array containing twelve interdigitated electrode IDEpairs with different gap sizes has been designed. The IDE fingers were fabricated using photolithography and thin-film metallization techniques. SWNTs were grown directly on the IDEs for trace amounts of gas detection. This sensor array has been exposed to nitrogen dioxide at various concentrations from 10 ppm to 400 ppb, and ammonia from 50 to 5 ppm. The results show a very reproducible sensor perfor- mance from one sensor to the other in the array. © 2005 The Electrochemical Society. DOI: 10.1149/1.2063289All rights reserved. Manuscript submitted June 6, 2005; revised manuscript received July 25, 2005. Available electronically September 16, 2005. Monitoring concentrations of chemicals, gases, and vapors is a critical task in the chemical industry, power plants, automotive ex- haust, environmental protection, planetary science, and numerous other situations. Chemical sensors for specific species with varying sensitivity and discrimination levels are commercially available. Common chemical sensors differ in terms of the sensing material and the nature of the property change such as electrical conductiv- ity, optical characteristics, temperature, etc.. Some of the current sensor technologies include high temperature oxide thin-film sen- sors, polymer-based sensors, catalytic-based sensors, and surface acoustic wave sensors. 1 Key attributes expected of a sensor include sensitivity in the parts per million to billion ppm, ppbrange where trace levels are involved, absolute discrimination, room temperature operation, low power consumption, reasonable size, volume and mass, and low cost for large-scale applications. Given such a broad set of desirable attributes and diverse application fields, sensor de- velopment is a constantly evolving research area. The newest con- tributions to this field are nanoscience and technology where novel nanomaterials, because of their size, large surface to volume ratio, and properties that differ from their bulk counterparts, promise to offer better performance than micro- and macrosensors. Some of the candidate nanomaterials include carbon nanotubes CNTs, 2 inorganic nanowires of high-temperature oxides and semi- conducting elements or compounds, and quantum dots. Of these, CNTs have commanded much attention for physical, chemical, and biosensors due to their interesting physical, electrical, and other properties. 3 The sensor array described in this article consists of carbon nanotubes as sensing material and an interdigitated electrode IDEas a transducer. It is one type of electrochemical sensor that depends on the transfer of charge from one electrode to another electrode. This means that at least two electrodes constitute an elec- trochemical cell to form a closed electrical circuit. Due to the cur- vature of the graphene sheet in single-walled carbon nanotubes SWNTs, the electron clouds change from a uniform distribution around the C–C backbone in graphite to an asymmetric distribution inside and outside the cylindrical sheet of the nanotube. 2 Since the electron clouds are distorted, a rich -electron conjugation forms outside the tube, therefore making the carbon nanotube electro- chemically active. Electron donating and withdrawing molecules such as nitrogen dioxide NO 2 and ammonia NH 3 will either transfer electrons to or withdraw electrons from SWNTs, thereby giving SWNTs more charge carriers, either electrons or holes re- spectively, which increases or decreases the nanotube conductance accordingly. 4,5 Theoretical calculation showed the binding energy of the NO 2 and NH 3 to carbon nanotubes, 6 which indicates that a weak charge transfer can occur. The typical electrochemical interaction may be denoted as CNT + Gas ——CNT e Gas + or CNT + Gas e where is a number that indicates the amount of charge transferred during the interaction. The conductivity change may also be caused by the contact modulation between the metal electrode and carbon nanotube, and/or the contact between carbon nanotube and carbon nanotube. By measuring the conductivity change of the CNT device, the concentration of the chemical species or gas molecules can be measured. In this work, we describe a chemical sensor based on such a conductivity change using SWNTs. While carbon nanotubes can provide the sensitivity and other attributes expected from nano- materials in general, broad commercial acceptance would largely depend on the ability for mass production and cost since the sensor market is extremely cost sensitive. Here, we also describe a micro- fabrication approach to fabricate the sensor platform incorporating the nanotubes. A multilayer sensor array chip was designed and fabricated for optimal gas sensing capabilities with in situ heating and temperature monitoring. A detailed chip configuration schematic is shown in Fig. 1. p-Type boron-doped silicon100wafers with a resistivity of 0.006–0.01 cm and thickness of 500 ± 25 m were used as sub- strates. A layer of 0.5 m silicon dioxide was thermally grown on top of the Si substrates. Patterned platinum 200 nm thicklines were deposited on top of the SiO 2 layer for electrical heating and resistive temperature detection RTD. A layer of 1 m silicon ni- tride was deposited on top of the heaters and the RTDs as the insu- lating layer between the platinum heating elements and the IDE fingers. A layer of 200 nm platinum on top of 20 nm titanium was deposited on the Si 3 N 4 layer onto the designed finger patterns. These patterns consisted of 4, 8, 12, and 50 m finger gaps with 10 and 20 m finger widths see Fig. 1b. Next, SWNTs need to be deposited on the surface of the IDEs. In our previous single-sensor efforts, 7 we have used a solution casting approach from bulk SWNT samples but here we have chosen an in situ chemical vapor deposition CVDprocess to deposit the SWNTs. The SWNT growth process and characterization are de- scribed in detail in Ref. 8 and only a brief outline is given here. The catalytic metals used in these experiments for SWNTs growth are 99.9% pure and are sputtered using a VCR Group Incorporated ion beam sputter IBS, model IBS/TM200s. The deposition of the metal catalyst for growing SWNTs was over the entire area of the IDE fingers. The catalyst used to grow the SWNTs consists of two com- ponents, a metal underlayer Aland an active catalyst Fe. Since the thin aluminum layer breaks up into droplets at the growth tem- perature, this layer does not form the conducting bridge between the two terminals. The Al underlayer helps to increase the nucleation and generation of the catalyst particles. The thickness of the Al underlayer was varied from 1 to 10 nm and the Fe catalyst was var- * Electrochemical Society Active Member. ** Electrochemical Society Fellow. z E-mail: jingli@mail.arc.nasa.gov Electrochemical and Solid-State Letters, 8 11H100-H102 2005 1099-0062/2005/811/H100/3/$7.00 © The Electrochemical Society, Inc. H100 Downloaded 05 Nov 2008 to 131.94.116.191. Redistribution subject to ASCE license or copyright; see http://www.ecsdl.org/terms_use.jsp