PRODUCTION OF RESISTORS BY ARC PLASMA SPRAYING

Arc plasma spraying (APS) is an accepted method of producing coatings for many engineering applications. The wide range of materials that can be used to form the thick film coatings make this technique interesting as an alternative method of producing electrical components and circuits.


INTRODUCTION
The application of arc plasma spraying (APS) for the production of coatings with useful electrical properties has been investigated by several workers in recent years 1-4.
The production of resistors is obviously of prime importance if APS is to become established as an alternative method of producing electrical com- ponents or circuits.This paper reports the results that have been obtained in producing mixed oxide thick film resistors.

ARC PLASMA SPRAYING
In this technique, used for many years in the mechanical engineering industry, a high power dc arc is struck between an annular anode and a rod shaped cathode (Figure 1).A stream of inert gas (argon) blows the arc into the annulus of the anode.This constriction results in the gas flow becoming a high temperature (10,000C), high velocity (60 m/see) plasma jet.A fine powder injected into the plasma is melted and accelerated by the jet so that it flows out of the front of the anode as a spray of semi-molten 135 particles.These, on striking a cold substrate, splatcool to form a coating.
The parameters controlling the nature and quality of the coating produced by a given plasma gun are arc current, gas flow rate, powder particle size, nature of the powder material, gun/substrate distance and nature of the substrate surface.All of these parameters have been examined for their relevance to the electrical properties of the deposited film.The potential advantages of APS as a method of depositing thick film resistors are: 1) Any material th a melting point can be used.
2) The substrate can be kept at low temperature, and usable materials include plastics.
3) The deposited thickness, can readily be con- trolled to quite close limits.4) There is no practical limitation on the area that can be coated.

5)
The system is capable of a high degree of aut.omation.
6) The coatings are robust and resistant to abra- sion.
resist step was used to define the resistors.A pattern of standard resistors, each 18 mm x 2.8 mm was used for most of the work.
In a preliminary investigation 4 a number of pure and mixed oxide compositions was used.The resistors were assessed by measurement of their temperature coefficient of resistance (TCR) and the mixed oxide composition with the lowest TCR was chosen for the general investigation.This consisted of a 55% NiO 45% FeaO4 by weight powder with a particle size 1-20/am. 4

STRUCTURE OF FILMS 3 PRODUCTION METHOD
To produce thick film resistors the substrate is mounted on a holder located in front of the spray gun.The holder is rotated and the gun is traversed laterally, as illustrated in Figure 2. By varying rota- tion, traverse rate and number of traverses, the thickness of the deposit can be controlled for fixed spraying conditions.
In the present experiments the required coating pattern was defined by covering the slides with a photoresist laminate.This was exposed and developed to leave bare the substrate in the region where deposition was required.After spraying the whole substrate area the remaining photoresist was dissolved leaving the deposit only in the areas exposed by the mask.In the first step a pattern of aluminium conductors was sprayed, and then a second photo-Gas cooling FIGURE 2 APS deposition system.
The structure of the films is determined by the build-up of splatted particles.Thus the original particle size is the main factor on which the physical structure depends.The gun parameters of gas flow rate and arc current have little effect on structure over the normal working range but if taken to extremes powdery, non-adherent coating may be produced.Additional factors are substrate roughness and the incident velocity of the particles.In Figure 3 are shown typical scanning micrographs of the films deposited using a nickel oxide powder.Figure 3(a) shows the effect of spraying 1-20/am powder at a gas velocity of mach 0.2 on to a smooth glass substrate and (b) on to a slide grit-blasted to a roughness of 1.3/am CLA.The coating structure appears similar in both cases but a lower magnifi- cation examination of the boundary area as in Figure 3(c) reveals a smoother film on the grit blasted substrate.This is due to better adhesion of particles in the initial stages of deposition giving a more uniform surface coverage.This has been confirmed by high speed cine photography which shows that deposited particles are often mobile on the smooth substrate during deposition, leading to coalescence, and a continuous coating is achieved by island growth rather than the build up of individual particles.This variation is only seen during the initial coating stage, no differences being apparent after the initial con- tinuous coating is formed.
splatted particles, which tend to fill in the interstices between particles giving a denser, smooth film.The effect of increasing particle size from 1--20/am to 20-40/am is shown in Figure 3(g) and (h) respectively, for mach 0.2.The splatted particles are obviously much larger with little evidence of the formation of secondary droplets.The large platelets themselves do, however, show signs of fragmentation as seen in Figure 3(i) in which cracks are visible in the larger particles.

ELECTRICAL PROPERTIES
With fixed spraying conditions of arc current 400 A, argon flow rate 25 1/min and gun substrate distance 5 cm (standard throughout the work) resistors of different sheet resistivity were produced by varying the thickness of deposit.The material used was 50% NiO 50% Fe304 with particle size in the range 1--20/am and the thickness was varied by varying the number of passes made by the gun.Sheet resistivities from 5 to 500 fZ/sq were produced with the higher values corresponding to thin films which have visible "holes" in them.The coatings are pinhole-free and continuous only for resistivities of 100 f/sq and less.
The variation of resistance with film thickness is shown in Figure 4 in which 2/am corresponds Io the film thickness al which the film becomes continuous.
The parameters chosen for assessment of the electrical properties at different resistivities were resistance, TCR and third harmonic index (THI).The  latter has been defined by Kirby and shown to be correlated with the electrical noise generated by the resistor.Rysanek and Anderson 6 have suggested that THI is also a measure of the probable long-term stability of the resistor.It is defined as 3rd harmonic amplitude in THI 20 log, o \ pp-ii-nnda--ent-al n volts and can be directly measured by an instrument designed for the purpose (Radiometer, Copenhagen, Model CLT.1).TCR is defined, in general by a (l/Oo)/(30/)T), where/90 is the resistivity at a reference temperature To.In this work To was 25C and TCR was taken as (1/Ro)(R Ro)/(T To), where R was the resis- tance at T C (125 C).
Using TCR and THI we have investigated the electrical effects of varying the following parameters: 1) Spraying conditions, (a) gun current, (b) argon flow rate, (c) gun/substrate distance 2) Powder composition 3) Particle size 4) Substrate preparation, (a) smooth, (b) grit blasted; 10,000 hour life tests have also been carried out.The experimental results are now given.

Effect of Spraying Conditions
Figure 5 shows THI as a function of gun current and argon flow rate of NiO resistors having a sheet resistivity of -2 ;Z/sq.It will.be seen that the variation is not large, but that the optimum current and flow rate, corresponding to the lowest THI, is around 200 A with 25 1/min.The effect of substrate roughness was examined at 400 A and showed a slight lowering of THI for the grit blasted surface.In each case the material is NiO powder with.particle size 1--20/am.Figure 6 shows the effect of flow rate and gun current on the resistivity of the deposited films.
Below 30 1/min the resistivity is found to be fairly independent of flow rate.The use of a grit blasted substrate lowers resistivity for given gun conditions.
In Figure 7 the effect on THI of varying gun/ substrate distance is shown for fixed current and flow-rate.It will be seen that distance has little effect between 50 mm and 70 ram.

2 Izffect of Powder Composition
The parameters mainly affected by composition are resistivity and TCR.The variation of resistivity with percent by weight of NiO in an NiO--FeaO4 mixture is shown in Figure 8 for films of equal nominal thickness, whilst the variation of TCR with mean resistance is given in Figure 9 for different composi- tions and different thicknesses.It will be seen that, although the standard deviation is large, the TCR appears to decrease monotonically with increasing film resistance, i.e. with decreasing thickness.At high values the TCR becomes negative in every case, THI 10 -8 O "200 Amps $ 400A Rough sub,trote X 400A Smooth substrote 1 (l -600 AmlDs \.Arc gas flowrte L/min Resistivity of sprayed nickel oxide vs plasma arc gas flow rate for various plasma arc power levels.
crossing the axis at higher values for higher NiO content.
An electron microprobe analysis of a typical fihn, sprayed from a 56% Fe304 44% NiO mixture under standard conditions to a thickness of 15/am ('-100 fZ/sq)., gave an overall iron/nickel ratio of 58.3/41.7%,showing that the sticking coefficient of the two powders at the substrate is much the same.However, individual iron and nickel peaks, appearing together, occasionally show a much higher ratio of Fe to Ni and in some cases, using pure Fe and Ni to calibrate the peaks, it would appear that free iron is present in some groups of particles.Despite this, however, sprayed Fe304 films show a resistivity of -10-1 2/cm, whilst bulk material has a value of -10 -2 2/cm.No free nickel particles were detected.
Borgianni et al. 7 have shown that there is some loss of oxygen from NiO on spraying; these results show that this also occurs with Fe304.It is not possible to make accurate measurements of the bulk resistivity of sprayed material because of the influ- ence of the film structure on measured resistance.However, we may conclude that sprayed NiO has a lower resistivity than sprayed Fe304 and that, for a given thickness of film, increasing the Fe304 per- centage pushes the TCR in the negative direction.boundaries between splatted particles.There will be a localized strain and disorder at these boundaries which may give rise to localized charge and hence to a non-linear current-voltage relationship across the boundary.In such a case it would be expected that THI will increase with decreasing particle size, i.e.
with increase in the number of boundaries.Figure 10 shows that this effect is small.In Figure 11 is shown the variation in resistivity with particle size for NiO f'dms, with smooth sub- strates, produced with the same number of gun passes and fixed spraying conditions.The increase in resis- tivity for the larger particles is due to the increasing porosity of the film as particle size increases.The rise in resistivity at very low particle sizes may be associated with the particles losing heat so that they do not easily bond to each other on arrival at the substrate.

3 Effect of Particle Size
In the previous section (Figure 3(g) and (h)) are shown scanning micrographs of deposits with NiO particles of different sizes prepared with gun con- ditions of 200 A and 25 l/cm.Figure 10 shows THI and surface roughness for the same films.It is reasonable to suppose that the third harmonic com- ponent arises from non-linearity associated with grain 5.4 Effect of Substmte Preparation (a) Grit blasting The principal effect of grit blasting is to increase the number of particles adhering to the substrate in the initial stages of deposition.This is evidenced by the micrographs of

5
FIGURE 10 THI and surface roughness (CLA) as a function of particle size; NiO powder, 200 A, 25 l/min, smooth substrate.For each particle size all smaller particles are included.a slight lowering of its value, due to denser packing of the particles, and the resistivity for a given thickness is lowered for the same reason.
(b) Heating Films were deposited with the sub- strates heated to 200C.No significant improvement in properties on either smooth or grit-blasted sub- strates was observed.
These are, respectively, for 55/45 NiO-FeaO4 resis- tors of 85 D, and 340 2 prepared under standard conditions with powder particle size in the range 1-20 and smooth substrates.It will be seen that the change in resistance is always less than _+ 12% with a mean of "-5% for most resistors; in fact, they are remarkably stable.Each histogram refers to the mean values for 160-180 resistors.

Life Tests
The results of 10,000 hour life tests carried out at    a single production run.The resistors were prepared from 55/45 NiO-FeaO4 powder of particle size 1-20/a with standard gun conditions.The substrates were grit blasted and the mean resistance was varied by changing the number of spraying passes made by the gun.
tors on low-cost substrates.Standard trimming tech- niques, now in use with screen-printed thick films, could be employed to upgrade tolerances but the system also offers the advantage that resistance can be monitored during deposition, so that production to close tolerance specifications using automatic control of gun traverses is relatively straightforward.
6 DISCUSSION AND CONCLUSIONS

ACKNOWLEDGEMENTS
The properties of APS resistors can be understood in terms of the mechanics of the deposition process.The films are necessarily particulate in structure and include many grain boundaries and defects.For this reason their THI values do not compare favourably with screen-printed and fired thick film resistors, or cracked carbon film resistors.With relatively crude control equipment, as used in this work, it is possible to produce 95% of resistors within +20% tolerance in a single run.Whilst cheap glass substrates have been used in the present project, plastic or laminate substrates can be used, and resistors have even been successfully sprayed on paper.The deposition process is fast and the photoresist masking system allows complex patterns to be readily produced.It is concluded that APS offers a technology for cheap quantity production of resistors and conduc- The assistance of the Science Research Council, in the form of a research grant, is acknowledged with gratitude.

FIGURE 5
FIGURE 5 TH! as a function of flow rate and un current on mooth and rit-blatcd ubtratc.powder, sheet resistivity 4 /sq.

Figure 3 (
Figure 3(a), (b) and (c).The effects on electrical properties are consistent with this model and areshown in Figures5 and 6.In the case of THI there is

FIGURE 11
FIGURE 11Resistivity as a function of particle size for the films of Figure10.
150C in air, are shown as histograms of fractional change in resistance, AR/R, in Figures12 and 13 .

5. 6
ReproducibilityIn Figure14is shown the standard deviation in resistivity for different batches of resistors having particular mean values, each batch being deposited The resolu- tion is summarized, for the present work, as Minimum width of conducting path Minimum separation for parallel paths materials used are cheap and the resistors show satisfactory long-term stability. The