GB2248965A - Semiconductor testing method - Google Patents

Abstract

In a method of avoiding use of nuclear radiation (5), eg gamma rays, X-rays, electron beams, for testing semiconductor components for resistance to hard radiation, which hard radiation causes data corruption in some memory devices and `latch-up' in others, similar fault effects can be achieved using a xenon or other 'light' flash gun even though the penetration of light is significantly less than that of gamma rays. The method involves treating a device with gamma radiation, measuring a particular fault current at the onset of a fault event, repeating the test with light to confirm the occurrence of the fault event at the same measured fault current, and using the fault current value as a reference for future tests using light on similar devices. <IMAGE>

Description

3 _)2,1 1 +3?S3 CC/4035 Semiconductor Testing Method This invention

relates to a semiconductor testing method and is particularly concerned with testing semiconductor circuit components for resistance to high-energy 'hard' radiation. By 'hard' radiation is meant radiation such as gamma rays, particle radiation, X-rays, electron beams etc which, when absorbed by semiconductor material, cause Compton scattering and resultant ionisation. The term 'semiconductor circuit component' is intended to include circuit components of any non-metallic material which is subject to the above ionisation damage, eg silicon, germanium, quartz etc.

Hard-radiation, eg from nuclear and beam weapons in the form of high dose rates of high energy gamma, X-rays and electrons, causes ionisation in semiconductor components and the materials used in their construction. If a suitable voltage bias is present across the device, charge separation occurs and photo-currents are produced at the interfaces of material discontinuities (junctions). These photocurrents can cause detrimental effects in electronic systems.

Assuming a fixed bias voltage, the photocurrent for a particular radiation energy and junction is proportional to the junction area and radiation dose rate (energy deposited per unit mass per second).

Depending on the type of semiconductor component and technology being considered, the high radiation dose rate can cause:- 1. Transient voltage changes when the currents flow across circuit impedances.

2. Corruption of stored data in semiconductor memory structures.

3. 'Latch-up' of devices fabricated with four layer parasitic SCR structures present. Burn out of junctions, bond wires and metallisation tracks, due to the photocurrent pulse or as a result of 'latch-up'.

Electronic components and circuits which may be exposed to high dose rate radiation sometimes require testing to determine their susceptibility and the adequacy of any protection employed.

The normal accepted test methods involve the use of high cost intricate radiation sources such as: flash X-ray generators, electron accelerators or lasers. The cost of such apparatus is typically millions of pounds and even the hiring cost can be over MOO per day.

An object of the present invention is therefore to provide a method of testing semiconductor circuit components for resistance to hard radiation which does not involve use of a high-energy hard-radiation source with every sample tested.

According to the present invention, in a method of testing semiconductor circuit components for resistance to doses of high-energy particle or Xray, hard-radiation, a sample component is subjected to a dose of the hard radiation of such intensity as just to cause an operational fault, a quantitative fault parameter varying with this radiation intensity is measured at a level corresponding to the occurrence of the operational fault, the hard radiation is replaced by a source of optical radiation and the intensity of optical radiation producing the same level of the quantitative fault parameter is determined, one or more further components of the same type are subjected to the optical radiation at the determined intensity and the incidence of the operational fault is determined.

The quantitative fault parameter may be a photo-current arising from radiation ionisation at a semiconductor junction.

The operational fault may consist of data corruption or 'latch-up' in a pn-p-n component. The quantitative fault parameter may then be measured at levels corresponding to successive operational faults.

Preferably, a plurality of values of the parameter are measured for the sample component under high energy radiation conditions, correspondence with operational faults determined, and values of the optical radiation intensity corresponding to the plurality of values of the parameter are determined for future testing of the further components Where the components are enclosed or covered with a metal layer, the metal layer is removed prior to the component being subjected to the optical radiation.

Where exceptionally high power irradiation is required, the source of optical radiation may comprise a plurality of individual flash gun sources, means for synchronising their operation and focussing means for combining and concentrating their individual outputs.

A method of testing semiconductor components for resistance to hard radiation will now be described, by way of example, with reference to the accompanying drawings, of which:

Figure 1 shows a cross section through a typical semiconductor component (a CMOS) showing the depths of penetration of different kinds of incident radiation; Figure 2 shows a similar cross section including the circuit connections and photocurrent resulting from radiation such as shown in Figure 1; Figures 3(a) and (b) show respectively the diagrammatic equivalent of the device of Figure 2 and the parasitic circuit resulting from the induced photocurrents; Figure 4(a) shows a cross section of a single transistor and the effect of a light pulse from an optical source, and Figure 4(b) shows a standard transistor symbol and current paths; Figure 5 is a diagram of a test arrangement for an integrated circuit subject to a light pulse; Figure 6(a) is a diagram of a standard test pulse of gamma radiation; d v k and Figure 6(b) is a diagram comparing the pulse profiles of two known hard radiation test sources with a light pulse according to the invention.

Referring now to Figure 1, this shows a semiconductor device having an ntype substrate 1, and a p-type well 3 and source and drain areas S and D. Contacts and connections are not shown. The depth of penetration of a high energy gamma beam is shown by the path 5 to be completely through the device, the beam causing ionisation throughout its path.

In contrast, an optical beam 7 of visible light emitted by say, a photographic xenon flash gun, will penetrate only 80-100 microns. However, this is commonly sufficient to include the whole of the active regions of the device, eg the well 3. A xenon light source operates in the region of 800nm wavelength and xenon photons deposit their energy in silicon and silicon dioxide within the above depth of, say, 100 microns. Ionisation is produced within this region by the photoelectric effect.

It has been realised therefore that a xenon flash gun can be an effective substitute for a hard radiation source such as a pulsed nuclear radiation source, flash X-ray, high power laser etc.

The photoelectric effects of xenon radiation are similar to those produced by gamm/X-ray/electron radiation and resulting Compton scattering in so far as they cause data corruption in memory devices and latch-up of p-n-p-n elements. These effects are shown in Figures 2 and 3.

Figure 2 shows a section through a CMOS device having n-type substrate 1, p-type well 3, a p-channel MOS element 9 and an n-channel MOS element 11. A rectifying junction, indicated by a diode 13, exists inherently at the interface of the p-well and the n-type substrate 1. This junction is normally reverse biased by the positive and negative voltages Vdd and Vss. Any ionisation occurring in the region of this junction causes a photocurrent I p and possible latch-up of the device.

Figure 3(a) shows the intended circuit diagram for the device, a standard CMOS element. Figure 3(b) shows the parasitic circuit resulting from the ionisation. A transistor 15 is simulated by the p source region of MOS device 9, the n substrate 1 and the p well 3. Rn is the inherent resistance of the substrate. A further, n-p-n, transistor is simulated by the n substrate, the p well and the n source of MOS device 11. This, together with the biasing resistance R p of the p well forms the circuit of Figure 3(b). A path for the photo current Ip is shown. This situation is self sustaining, with resultant 'latch-up' of the CMOS device.

Figure 4 shows the basic photocurrent effect for a single transistor and also the shadowing effect of metal contact areas. In operation the basecollector junction 15 is reverse biased and is thus the main seat of photo currents Ip. The current I p is amplified, as shown in figure 4(b) to produce a current of j6 I p which may cause transistor burn-out.

Figure 4 also shows shadow regions 17 under the metal layer contacts of the emitter, base and collector. The metal layer is impervious to xenon radiation and thus these regions 17 are not subjected to the ionisation effects that would arise with gamma radiation. This is a weakness in the use of light instead of hard radiation but can be compensated to some extent by correlation with gamma radiation tests.

A quantitative measurement of the fault arising from ionisation is obtained by measurement of the emitter currentAI P This parameter is first obtained for gamma irradiation of a device, and then a substantially identical device is subjected to increasing levels of xenon radiation until the same fault current value is obtained. If the fault event (data loss, latch-up etc) occurs at the same fault current value in the xenon case then the xenon radiation provides a good radiation substitute suitable for future tests in the same device type.

The subsequent testing is clearly more cost effective, requiring no hard radiation sources. A typical xenon flash gun, as used commonly in photographic application, is safer, portable and of course enormously cheaper than the hard source. A very simple method is thus provided with which to undertake many of the mundane tasks such as: device batch, wafer lot or procurement lot screening for radiation hardness acceptance.

If the xenon radiation does not produce the same fault event at the same fault current then shadowing effects or deep silicon effects need to be considered to obtain, if possible, xenon levels providing the fault event. A correlation between xenon fault currents and gamma fault currents may thus be obtained.

Figure 5 shows in skeleton form an integrated circuit having input and output connections and collector and emitter terminations. In such a case the metal casing, or lid, of the device must be removed before light testing since, as explained above, while metal layers are largely transparent to gamma rays they are largely opaque to xenon light. As explained above, test data from a gamma dose rate source is correlated with that from a light source with respect to the output response and current drawn/generated. This latter parameter is the factor which determines when the two radiation sources are equivalent and it is at this point that the output response of the device should be the same for the two kinds of source. Thus the occurence of a fault event with xenon radiation is monitored at the same value of curent drawn/generated: this value, for the light source, is then a reference value for future operation of light source tests.

Figure 5 shows diagrammatically the measurement of the photocurrent I p with the use of an oscilloscope.

In a particular example, in a brief radiation trial using a linear accelerator, for a particular CMOS memory device, data corruption occurs just above 1.00 cGy/sec and 'latch-up' at 5.0e7 cGylsec. In these figures e7 represents '10 to the power V, and cGy is a "Centi-Gray" ie one "rad".

With the knowledge of these two 'spot' operational faults a range of 6 values of radiation intensity covering the two critical values is used to obtain corresponding transient supply photocurrents. These points are: 5. 0e6, 1.0e7, 5.0e7, 1.0e8, 1.Oeg and 6.0e9 cGylsec. The measured photocurrents and resulting device output effects are then reproduced with the xenon source and equivalent light output dose rates for the memory device are recorded as a function of device distance from the face of the flash gun, this distance being adjusted to control the flash gun power. Any future batches of the same device structure can then be tested using the xenon source to verify whether latch-up and data corruption are still the same or whether a change in response with distance from the flash gun has occurred. The light output from the flash is monitored using a previously calibrated photodiode to ensure that the source conditions are unaltered.

- 6 A suitable xenon light source is that designated "Pulse Xenon 110V manufactured by Pulse Photonics Limited.

Radiation testing at strategic nuclear weapon levels requires very high radiation levels that cannot be achieved even on some multi-million pound research facilities in the UK. The present invention provides at least a partial solution to this problem by providing a xenon light source simulator comprising an array of flash guns and a focussing system to combine the powers of the individual sources to irradiate semiconductors and electronic circuits at these high dose rates. Such an array requirs flash trigger co-ordination, but the reduction in cost, size, maintenance and risk and the ease of transportation to different test locations amply compensates for the extra complexity.

While the above example has employed a standard xenon photographic flash gun, other sources of light may be employed, preferably operating within the visible light range from about 400 nanometres (nm) to 1000 nm.

Figure 6(a) shows a standard gamma radiation test pulse showing power against time. Maximum power is achieved at about 30 nanoseconds after the initiation of the pulse, this interval being used as a time unit for the remainder of the pulse. Since both the power and time scales are logarithmic the fall-off can be seen to be approximately linear.

Figure 6(b) shows a comparison of the pulse forms of (19) a pulse from a linear accelerator, (21) a flash X-ray pulse, and (23) a xenon flash gun as employed in the present invention.

The peak powers available are as shown, but have been rationalised to the linear accelerator level for comparison purposes. It will be appreciated that the high power particle accelerators and X-ray flash guns are not commonly available (only from Government Departments generally) and considerable advantage is to be gained by avoiding at least some use of such equipment.

Claims (12)

1. A method of testing semiconductor circuit components for resistance to doses of high-energy particle or X-ray, hard-radiation, in which:

a sample component is subjected to a dose of said hard radiation of such intensity as just to cause an operational fault, a quantitative fault parameter varying with said intensity is measured at a level corresponding to the occurrence of said operational fault, said hard radiation is replaced by a source of optical radiation and the intensity of optical radiation producing the same said level of said quantitative fault parameter is determined, one or more further said components of the same type are subjected to said optical radiation at said determined intensity and the incidence of said operational fault determined.

2. A method according to Claim 1, wherein said quantitative fault parameter is a photo-current arising from radiation ionisation at a semiconductor junction.

3. A method according to Claim 1 or Claim 2, wherein said operational fault consists of data corruption.

4. A method according to Claim 1 or Claim 2, wherein said operational fault consists of 'latch-up' in a p-n-p-n component.

5. A method according to Claim 1 or Claim 2, wherein said quantitative fault parameter is measured at levels corresponding to successive operational faults.

6. A method according to any preceding claim, wherein a plurality of values of said parameter are measured for said sample component under high energy radiation conditions, c6rrespondence with operational faults determined, and values of said optical radiation intensity corresponding to said plurality of values of said parameter are determined for future testing of said further components.

7. A method according to any preceding claim, for testing components which are enclosed or covered with a metal layer, wherein the metal layer is removed prior to the component being subjected to said optical radiation.

8. A method according to any preceding claim, wherein said hard radiation is derived from a gamma-ray, X-ray or electron beam source.

9. A method according to any preceding claim wherein said source of optical radiation comprises a plurality of individual flash gun sources, means for synchronising their operation and focussing means for combining and concentrating their individual outputs.

10. A method according to any preceding claim wherein said source of optical radiation operates at wavelengths between about 400 and 1000 nanometres.

11. A method according to any preceding claim wherein said source of optical radiation is a photographic flash gun.

12. A method of testing semiconductor circuit components for resistance to high energy hard radiation, substantially as hereinbefore described with reference to the accompanying drawings.