This paper describes the operation of a novel gas sensor and how hybrid 
circuit methods have been exploited to manufacture the active element. The
sensor is made with conventional tape-cast substrates and commercially
available conductive and resistive inks, with the addition of newly
developed "sensor" inks.  The finished chip differs from many conventional
encapsulated hybrid circuits because the sensor ink is intentionally
reactive and it must remain exposed to air in order to perform the sensing 
function.  This sensing concept is quite new and it is expected that many
applications and design improvements will become apparent as research and
development continue.

The IGAS Chip (Figure 1) plugs into a 20-pin dual inline socket (0.1 inch
spacing, 0.4 inches wide).  Two pins supply power to the heater (about 2 to
10 watts), and the other eighteen are used to map the pattern of resistances
in the oxide film when the sensor is exposed to a gas sample.  The pattern
of values for the ration of resistance in the gas to resistance in air at
the various points along  the length of the chip is to be processed by a
dedicated microprocessor to analyze and identify the gas.  As an example,
with the IGAS Model TOS-2, alcohols, ketones, and alkanes produce distinctly
different signatures.

The reactions occurring in the IGAS chip are the same as in the commercial
sensors.  The active material, tin oxide or some other semiconducting oxide,
becomes reduced by combustible gases and the resistance of the oxide
decreases.  When the gas is removed the oxide layer is re-oxidized in air
and its resistance increases to its original value.  More information can be
extracted from the IGAS sensor than from the conventional tin oxide sensor
because of the multi-sensor effect created in the manufacture of the IGAS
chip.  In some of the sensors, such as that shown in Figure 1, the heater
strip is located at one end to establish a thermal gradient along the length
of the chip.  The reactivity of a particular gas will dictate how far down
the thermal gradient the reaction can occur.  For relatively simple organic
compounds, the reactivity is determined by the most active functional group,
so that similar compounds (e.g., methanol, ethanol, isopropanol) give
similar reactions.  Figure 2 shows the response of the sensor to two ketones
(acetone and MEK) at two heat settings.  We observe that as the absolute
temperature is increased on the chip, the shape of the curves of the two
ketones remain the same but the reaction, following the thermal gradient,
proceeds further along the chip.

Alternatively, heat may be distributed uniformly over the chip and the
catalytic activity can be altered in different locations by doping the
different areas with various catalytic materials.  Again, a particular
compound will be more reactive to one region than another depending on how
the individual catalysts affect the important functional groups in the gas.
With variations in both temperature and catalyst distribution, the
complexity of the signature is further enhanced, potentially increasing our
ability to discriminate among similar gases.

Eight IGAS chips can be made on one GE Ceramics type 614 or 860 alumina
substrate.  Each chip is 1.00" x 0.42".  The IGAS chip is manufactured by
first printing the conductor paths, and then the resistor pattern that forms
the heater using commercially available ink compositions (DuPont 9910 and
1711) and firing them in a belt furnace following the manufacturer's
recommended firing cycles.  Eighteen of the twenty gold conductor pads
provide contact to the signal processing equipment; the other two provide
power to the resistor strip.  Thick-film resistors, to provide the necessary
heat for the reaction process, are printed in either of two configurations
on the back of the chip (Figure 3); one provides an even temperature across
the entire sensor, the other is a narrow strip at one end of the chip 
that establishes a thermal gradient along the length of the chip.  The heater
strips have a nominal value of 70-100 ohms.  After applying the active
material, as described below, leads are attached to the conductive pads on
the chip either by soldering or with a conductive adhesive.

The design requirements for our sensor ink deviate from normal thick-film
composition development mainly in that the fired sensor ink must be porous
and reactive with its environment; conventional thick-film resistor
compositions are formulated to be dense, inert, and stable over a reasonably
wide range of humidity or temperature.

The first "ink" was a mixture of tin chloride and stearic acid (1) that was
painted on the substrate and fired at 700 C in air.  This process was
repeated several times to build up an adequate film thickness on the
substrate.  The organic component was intentionally burned off, creating
porosity in the tin oxide layer.  On top of the tin oxide layer, an alcohol
solution of one or more precious metal chlorides was then painted in
selected areas and fired at 700 C in air.  Films with sheet resistances in
the range of 5 to 30 kiloohms per square were deposited by this technique.

Figure 4 shows the active layer of one sensor made by this method.  The tin
oxide has a crenulated surface with a distribution of fine platinum catalyst
particles.  It will be shown alter that this sensor is fast and quite
responsive to many organic compounds.  However, the stearic acid method had
several important shortcomings.  First, it was difficult to control and
reproduce in the laboratory (although the technique can, of course, be done
reproducibly with specialized manufacturing equipment).  The tin
oxide/stearic acid layers were not uniform enough, and the material tended
to liquefy readily in the initial firing and partially overlap the heater
strips, rendering the first measurement point on the chip unusable.  Second,
the strength and adhesion of the tin oxide layer was quite poor, and its
initial resistance varied greatly from point to point on the chip.  Third,
the distribution of catalyst was difficult to control; it is likely that the
sensor shown in Figure 4 contains many times more precious metal than
necessary.

To overcome these difficulties, it was clear that printable compositions
were needed.  The current sensor ink contains three major components:  a
metal oxide, which is the active agent; glass frit, for adhesion; and
organic vehicles that burn off during firing.  The unfired ink that we have
developed typically contains only 30% solids by volume so that the firing
process will burn off the organic material and leave a highly porous thick
film.  The glass content (about 10% of the fired deposit) provides strength
and adhesion but increases the sheet resistivity.  The resistances of the
fired layers were in the range of 5-100 megohms, depending on the
temperature at which they were measured.  The catalyst can either be
applied after firing, as before, or incorporated directly into the 
ink in the form of a soluble organometallic compound.

Figure 7 shows the IGAS Model TOS-2 responses to seven gases:  methanol,
ethanol, isopropanol, hexane, heptane, acetone, and methyl-ethyl-ketone.
This chip has a tin oxide layer deposited by the tin chloride/stearic acid
method described above.  Platinum catalyst was applied with an ethanol
solution of chloroplatinic acid.  The substrate had a 156 ohm heater at the
end of the chip with approximately 2.5 watts of applied power.  Resistances
were measured for each gas at pin pairs across the chip.  These were
compared with the resistances at the same points in air with the same power
applied to the heater.  The resistance of the oxide while exposed to each
gas divided by the resistance of the oxide in the air is plotted against
distance with temperature decreasing from left to right.  The reaction of
each gas decreases toward the cooler end of the chip with the most reactive
gases being the alcohols (methanol, ethanol, and isopropanol).  Note that
the alcohol curves are similar in shape and essentially differ only in
magnitude of the reaction.  The difference in magnitude is probably a
concentration effect due to vapor pressure differences.  The least reactive
compounds were hexane and heptane, which showed a reaction only at the
hottest end of the chip.  Ketones, exemplified by acetone and
methyl-ethyl-ketone, are slightly more reactive than the alkanes but less
reactive than the alcohols.  All three gas types form patterns that are
distinctly different from each other.

Figure 8 shows results from IGAS Model TOS-7 which has a uniform heater of
about 90 ohms on the back of the chip (with 18 volts applied) and six
different catalyst areas fired onto the tin oxide layers.  The tin oxide
layer for this chip was also formed by the tin chloride/stearic acid method.
Overlapping has occurred in some of the catalyst regions and this has
probably caused some intermixing of the catalyst effects.  The presence of
ruthenium.  The alkanes (hexane and heptane) were the least reactive group,
showing the most reaction in the platinum and rhodium areas and showing
little or no reaction in the osmium and iridium areas.  The ketone group as
most efficiently catalyzed by the platinum, rhodium, ruthenium, and
palladium areas, with much less response in the osmium and iridium regions.
Alcohols, demonstrating the most overall reaction, responded equally
strongly to the palladium, ruthenium, platinum, and rhodium areas and
exhibited a lesser response to osmium and iridium.


Figure 9 shows IGAS Model TOS-9 responses to ethanol and hextane.  The
active area of this sensor was printed with tin oxide - 10% glass ink
containing palladium resinate solution.  An end heater, with about 8 watts
applied, established a thermal gradient along the length of the chip.
Ethanol, characteristic of the alcohol group already described, exhibited a
strong reaction at the hot end.  The reaction of the chip to hexane was
generally confined to higher temperatures, as expected.  Because of the
higher sheet resistivity of the printed sensor ink, measurements at the
lower temperatures were difficult.  Work is in progress to develop printed
ink formulations with lower resistivity.

1.  Naoyoshi Taguchi, "Method for Making a Gas-Sensing Element", United
States Patent, 3,625,756, December 7, 1971.

2.  Paul K. Clifford, "Microcomputational Selectivity Enhancement of
Semiconductor Gas Sensors", @i[Proceedings of the International Meeting on
Chemical Sensors], Fukuoka, Japan, September 19-22, 1983, pp. 153-158.

3.  Takashi Oyabu, Toshiji Kurobe, and Takeshi Hidai, "Development of Tin
Oxide Gas Sensor and Monitoring System", @i[Proceedings of the International
Meeting on Chemical Sensors], Fukuoka, Japan, September 19-22, 1983, pp.
12-17.

Figure 1.  IGAS chip Model TOS-2, showing thick-film resistor (on left) to
provide a thermal gradient along the length of the chip.  The active layer,
covering most of the surface, consists of tin oxide deposited by oxidation
of a tin chloride-stearic acid mixture.  The platinum catalyst was applied
by decomposition of a solution of chloroplatinic acid in ethanol.