The detection and measurement of CO using ZnO single crystals

65

65 - 73

Sensors and Actuators,

5 (1984)

THE DETECTION CRYSTALS

AND MEASUREMENT

OF CO USING ZnO SINGLE

B BO’M’, T A JONES and B MANN Health and Safety Executive, Sheffield 53 7HQ {U K )

Research

and Laboratory

Servwes

Dmszon,

Red

Hdl,

(Received January 19, 1983, accepted In rewsed form May 27, 1983)

Abstract

A gas sensor that 1s based on the electrical conductnnty changes effected by the adsorption of gases on smgle crystals of ZnO has been developed The behavlour of the sensor 111CO-, CH4-, Hz- and HaO-m-au mixtures 1s described The effect of temperature m the range 300 - 500 “C on the characterlstlcs of the sensor has been investigated m detal Data are presented describing the variation of conductance with CO concentration, the response and the recovery times when the sensor 1s exposed to step changes m gas concentration and the long-term behavlour of the sensor. The results indicate that the sensors are sensltlve to CO and to Hz but are msensitlve to CH4 and water vapour, have good long-term stability and have response times -2 mm

1. Introduction

The chemlsorptlon of a gaseous species on the surface of a metal oxide semiconductor can be treated as an electronic process m which charge transfer occurs between the adsorbed species and the semiconductor [ 1,2 3 This can result m a change m the electrical conductlvlty of the semlconductor, an effect that 1s widely used m gas-sensmg devices [3,43. The processes Involved m gas adsorption and conductlvlCy changes m oxide semiconductors have been treated m detal [ 5 - 91 In recent years metal oxide semiconductor gas sensors, with a range of &fferent oxides [lo, II], have found many apphcatlons To date, however, they have been used prnnanly to provide quahtatlve mformatlon about the presence and range of concentration of pollutants Only rarely have they been used to provide quantltatlve mformatlon This IS because the sensors are not sufflclently stable or reproducible and lack the reqmred selectlvlty The sensor described here 1s the culmmatlon of work auned at provldmg a gas sensor capable of measurmg carbon monoxide m au at concentrations up to 100 ppm, but which 1s msensltlve to the presence of up to 1% 0250-6874/84/$3

00

@ Elsevler Sequola/Prmted

in The Netherlands

66

CH, All commercrally avwlable oxide gas sensors use polycrystallme oxides, usually Sn02, often with metal ions added [3, 41 The sensor used m this work utilizes a single crystal of ZnO with no deliberate mtroductlon of impunties

2 Expenmental The crystals of ZnO were grown from a melt m KOH at 400 “C followmg the method described by Kashyap [12] Crystals approximately 3 mm long and 0 5 mm m diameter were selected and mounted on heatable alumina substrates The substrates, obtained from Platfllm Ltd, were 3 X 3 X 0 3 mm rectangular alumma blocks with a laser-tnmmed 10 St + 0 1 52 thick-film deposited heater on one face (Fig l(a)) Platmum connecting wires (SO pm dmmeter) were attached to the substrates w&h gold paste The crystal was mounted across the parallel electrode arrangement (Fig l(b)) and held m place with Hanovla gold paste (Engelhard Ltd) The substrate was then mounted on a 4-pm TO5 transistor header by spot welding the connectmg wires to the support posts The temperature of the substrate was kept constant by mcorporatrng the heater m a bridge cn-cult and automatically adjusting the voltage on the bridge to mamtam the resistance of the pIatmum heater at a constant value, the temperature could be varied by changmg the resistance m the senes arm of the bndge The sensor could thus be located either in a dlffuslon-fed gas system or directly m the gas flow without the temperature being affected The conductance of the ZnO crystal was measured by applymg a known voltage (0 25 V) across the crystal and a known resistance m senes The voltage change across the series resistance was used to calculate the change m current and thus the change m conductance of the ZnO crystal The gases used were Bntlsh Oxygen Company certified grade mixtures of 100 ppm CO-m-mr, 100 ppm H,-m-air and 1% CH4-m-air and Bntlsh (IOLD

PASTE \

ZINC OXIDE SINGLE CRYSTAL /

CONNECTING

I

SUBSTRATE

N_UMINA SUBSTRATE

@I

Fig 1 (a) Substrate heater arrangement

(b) ZnO smgle crystai mounted

on substrate

67

Oxygen Company medical grade a.~ The gases were passed through activated charcoal traps to remove residual Impunties and dried by passmg them through magnesium perchlorate When necessary, the gas mixtures were humidified by bubbling part of the gas flow through delomzed water mamtamed at - 22 “C, this gas was mixed vvlth dry gas to give the required water vapour pressure The water vapour pressure mamtamed m most of this work was 10 + 1 Torr (-50% r h at 22 “C) All the data presented m this paper were obtamed from automated experiments using a Research Machme 3802 microcomputer for both experiment control and data processing The system 1s capable of mvestlgatmg 12 sensors sunultaneously, so that data on a large number of sensors could be obtamed under exactly reproducible condltlons The followmg experunents were carned out (a) The sensors were exposed for 20 mm to mixtures of 100 ppm CO, 1% CH4 or 100 ppm H2 m either dry or wet an- at 10 different temperatures between 280 “C and 580 “C After each exposure the sensors were returned to an an environment for 20 mm before being exposed to the next gas mixture Measurements were carned out with both mcreasmg and decreasing temperature mcrements (b) The sensors were exposed to five drfferent concentrations of CO m ar between 20 ppm and 100 ppm and to mixtures of 60 ppm CO + 60 ppm Hz and 60 ppm CO + 1% CH4 m both dry and wet an over the same temperature range (c) The response tunes were measured by placing the sensors directly m the gas flow and mm~mlzmg the dead space m the gas flow system This reduced to 10 s the delay between the gas valves being switched and the step change m the gas concentration reaching the sensor The conductance of the crystal was measured automatically every 20 s following the swltchmg of the gas valves. The gases used were an followed by 100 ppm CO m mr, followed by a return to asr Measurements were made m both dry and humid atmospheres. (d) The effect of varying the water vapour pressure between 6 and 20 Torr on the conductance m sllltand on the sensltlvlty of the sensor to CO was investigated between 280 “C and 470 “C 3. Results The results presented are a summary of data obtamed from 60 sensors The performance of this type of sensor 1s illustrated most usefully by a plot of sensltlvlty against workmg temperature, where sensltlrnty 1s defmed as Qg -

ua1r

u air

x 100,

og being the conductance m gas and (T,,, the conductance m an Figure 2 is typical of the response of the sensors to mixtures of 100 ppm CO, 1% CH4

68

0 DRY 1OOppm H, 0 WET 1OOppm Hz A DRY 1OOppm CO x WET IOOppm CO Al%CH WET&DRY 0 WET (50% RHI AIR

s 8 x

60

x 60

300

400 Temperature

Fig

2

Varhron

condltlons

500

6C

“C

of sensltlvlty to CO, Hz and CH4 with temperature under wet and dry

and 100 ppm H2 m sur under wet (10 Torr HZ0 pressure) and dry condltrons The mean temperature at which the CO sensltlvlty IS a maximum 1s 390 “C, with a standard devlatlon of 30 “C The maxlmum sensltlv&y to CO m dry an- has a skew dlstnbutlon and hence cannot be treated snnply, but the natural log of maximum sensltxvlty to CO does show a normal dlstrlbutlon with a mean of 5 0 (150% mcrease m conductance) and a standard deviation of 0 7 (range 50 - 290%), The sensltlvlty to 1% CH4 at 390 “C IS <20% for >80% of the sensors and approaches zero at temperatures >420 “C The ratio of the sensllxvlty to Hz to the senwkvlty to CO varies between I 0 and 2 0 at

69

390 “C and between 1 0 and 1 3 at 440 “C The maxunum sensltlvlty to CH4 occurs at a mean temperature of 300 “C with a standard devlatlon of 30 “C and to H, at 360 “C with a standard devlatlon of 30 “C The sensltlvlty to Hz0 (10 Torr H,O pressure) at 390 “C 1s 80% of the sensors and approaches zero at temperatures above 400 “C The effect of the presence of water vapour on the sensltlvlty to CO ISmore pronounced, the mean ratios of sensltlvlty to CO m a wet atmosphere to sensltlvlty m a dry atmosphere at 390 “C and 440 “C are 1 3 with a standard devlatlon of 0 2 and 1 0 with a standard deviation of 0 1 respectively The effect of water vapour on the conductance m an and on the sensltlvlty to CO and Hz IS, however, confined to the mlt1a.l change from dry to wet condltlons vmatlons m water vapour pressure have little effect The presence of 1% CH4 has no effect on the sensltlvlty of the sensor to CO at temperatures above 400 “C Mixtures of CO and H2 result m an increase m conductance greater than the change effected by the mdlvldual gases, but less than the sum of the two changes The conductances of the ZnO crystals m an at 390 “C also have a skew dlstnbutlon, but agam the natural log of this parameter shows a normal dlstnbutlon with a mean of -7 6 (4 7 X 10e4 a-‘) and a standard deviation of 1 5 (1 2 x lo- 4 - 2 3 X 10F3 SY’) for the sensors mvestlgated However, these figures do not fully reflect the very wide differences observed for crystals prepared m different batches Conductances as low as IO-’ Sz-’ and as high as 5 X 10F2 $2-l have been obtamed There was no correspondmg mde vmatlon m the other charactenstlcs between the batches There was no slgmflcant difference m the data obtained for any of the sensors between measurements made with temperature increasing or decreasing Figure 3 shows calibration curves (l e , plots of sensltlvlty agamst concentration) for CO m ar at four different temperatures The sensltlvltles are normahzed to 100 at 100 ppm at all temperatures The curves show an increased non-1memt.y at the higher temperatures The shapes of the responses as a function of time following a step change from an to 100 ppm CO m ar at different temperatures are shown m Fig 4 The response times decrease with mcreasmg temperature The return tunes followmg the change back to an are -20% faster than the response times at all temperatures, the response time being defined as the tune taken to achieve 90% of the total change Followmg the step change resulting from the change in enmonment, the conductance remams stable The sensors were exposed to 100 ppm CO and 100 ppm H2 m a~ for periods of several days, and there was no evidence over this penod of time of contmuous tift m conductance The long-term stablhty of the sensor 1s good Over a penod of SEXmonths of continuous operation at 425 “C, the conductance of the sensor in a~ dmfts up through the equivalent of the change obtamed for 100 ppm CO However, the sensitivity to CO expressed as percentage change m conductance vanes by less than 20% of the mltlal SensltlvAy, determmed after a brief mitral runnmg-xn penod of about three days m sLlrat 425 “C

70

sE a a 8

Tii 8

e 028OOC l 3lO~C q 39OT X 420°C

40

60

Concentration

Fig 3

Response

60 CO ppm

to CO m the concentration

range 0 - 100 ppm m au-

In the work reported here the potential applied to the crystal was 0 25 V When higher voltages >l V were used, the stability of the sensor detenorated with more pronounced &fts m conductance bemg observed Scannmg electron microscope and energy-dlsperslve X-ray exammatlon showed that there was no change m the topography of the crystal, but there was evidence of gold migration from the electrodes, this was not observed when the applied voltages were GO 25 V

71

EO-

60 -

0470% l 44OT x 42OT r3900c D 360°C

I 2

I 4

I 8

I 6 Time

Fig 4 The change m conductance function of time

I 10

1 12

I 14

‘I

minutes

followmg a step change from 0 to 100 ppm CO as a

4. I)lscusslon The ZnO crystals are very easy to produce and sensor fabncatlon usmg the Platfilm substrates or other sunllar devices IS not dlfflcult The success rate achieved m producmg workmg devices (z e , devices that show sensltlvltles of >25% change m conductance for 100 ppm CO at 400 “C) 1s high (>80%) The selectlvlty of the device m terms of sensltlvlty to CO relative to the sensltlvlty to CH4 1s very good and also reproducible The cross sensltlvlty to Hz may be unportant m some apphcatlons The lack of sensltlvlty to methane extends to other alkanes and methyl-substituted alkanes, but the

72

device 1s sensltlve to unsaturated hydrocarbons and gases such as SO2 and NH, Sensltlvlty to H,S 1s particularly high With the exception of Hz, the gases that could interfere wrth CO measurement can be eliminated by operating the sensor behind a charcoal screen This also protects the sensor from the surface polsonmg effects of materials contammg catalytic poisons such as sLzlcon, phosphorus or lead When exposed to a mrxture of gases, the total response of the sensor 1s less than the sum of mdlvldual responses It has also been shown that when one of the gases m a mixture does not itself affect the sensor, It does not affect the sensltlvlty of the sensor to the other components Some of the ma,m operating parameters of the sensor, namely the optimum operating temperature, the effects of water vapour, the lack of sensltlvlty to CH4, the shape of the cahbratlon curve and the response tune are reproducible The parameters which show the greatest spread are the conductance m air at a sven temperature and the sensltlvlty The variation m conductance 1s particularly apparent between crystals from different batches This 1s to some extent inevitable since the conductance ISlikely to be affected by small differences such as the variation m concentration of surface lmperfectlons m the crystal surfaces and m the concentration of cation unpurltles introduced during the crystal preparation stage Further work 1s necessary m order to isolate the reasons for this wide spread However, since the other parameters are reproducible and largely independent of the conductance, it should be possible to accommodate the spread m conductance and sensltlvlty by simply building sufficient adJustment into the associated clrcultry As would be expected of a single crystal, the electrical stablhty of the sensor over a period of months 1s very good The sensltlvlty to gases 1s also stable over slmllar penods, although this parameter could be affected by catalytic poisons The sensor has therefore many properties that make It a promlsmg device for detecting and measuring CO m the presence of CH4 Its use IS not hmlted to this apphcatlon, but the nature and concentration of the gases present need to be known and their effects ascertamed before it can be used quantltatlvely m any other apphcatlon

References 1 Th Wolkenstem, Electromc theory of catalysis on semiconductors, Adv Catalysq 12 (1960) 189 - 264 2 J Lagowskl, E S Sproles and H S Gatos, Quantltatlve study of the charge transfer m chemlsorptlon, qxygen chemlsorptlon on ZnO, J Appl Phys , 48 (8) (1977) 3566 -

3575 3 N Taguchl, Gas detection

device, Br Pat 1 280 809 (1972) B Bott, J G F&h, A Jones and T A Jones, Br Pat 1 374 575 (1974) 5 G Helland, E Mollwo and F Stockman, Electromc processes m ZnO m F Seltz and D Turnbull (eds ), Sohd State Physrcs 8, Academic Press, New York, 1959

4

73 E Mohnarl and F Cramarossa, Oxygen chemlsorptlon and surface p-type behavlour of ZnO powders, J Catalysu, 2 (4) (1963) 315 - 323 7 H Watanabe, M Wada and T Takahashl, Effect of oxygen on the electrlcal conductlvlty of ZnO powder, Denshl Shashm (Electroph$ography), 7 (1) (1966) 1 - 14 8 W Gopel, Reactions of oxygen with ZnO - 10 10 -surfaces, J Vuc Scl Technol , 15 (4)(1978) 1298 - 1310 9 A Jones, T A Jones, B Mann and J G Forth The effect of the physical form of the oxide on the conductlvlty charges produced by CH4, CO and Hz0 on ZnO, Sensors

6 A Clmmo,

and Actuators, 5 (1984) 75 - 88 10 J G Forth, A Jones and T A Jones, Sohd state sensors for gas detectlon, Proc IERE Conf Env Sensors and Applmatlons, London, 1974, pp 57 - 65 11 J G Fnth, A Jones and ‘I’ A Jones, Solid state sensors far carbon monoxide, Ann Occ Hyg, 18 (1975) 63 - 68 12 S C Kashyap, Growth of ZnO needles from molten hydrous KOH solutions, J Appl Phys, 44 (10) (1973) 4381 - 4384

Biographies Thomas Alwyn Jones graduated from the Umverslty of Wales with a B SC m Physics m 1963 and a Ph D m 1966 He joined the Safety m Mines Research Establishment in 1967, where he worked m various areas of gas detection with particular mterest m spectroscopic and semiconductor techniques He 1s currently head of the Semiconductor Sensors Section of the Research and Laboratory Semces Dlvlslon of the Health and Safety Executive m Sheffield Barry Bott Joined the Safety m Mines Research Estabhshment m 1960 Between 1971 and 1975 he took a degree course m Chemistry at Sheffield Polytechmc, graduating with a B SC (Hons) m 1975 He 1s currently a Senior Sclentlflc Officer m the Semiconductor Sensors Sectlon HLScurrent mterests mclude the apphcatlon of microcomputer systems to control and data processing in laboratory experunents Brenda Mann Joined the Safety m Mines Research Establishment m 1966 where she worked mltlally on the development of Semiconductor Gas Sensors and IS at present a Higher Sclentlfic Officer m the Semiconductor Sensors Section