CA1199969A - Electrical testing - Google Patents

Abstract

ELECTRICAL TESTING

Abstract of the Disclosure Insulation faults in electrical components are detected by measurement of electrical leakage before and after immersion in a mobile ionising solvent, e.g. a lower alkyl alcohol. A significant increase in leakage is indicative of a fault, e.g. a crack in the insulation.

Description

- R.C~ Chittick-2Y

~LECTRICAL TESTING
_ This invention relates to testing insulation faults such as may occur in electrical components and to techniques for improving the quality control of such components.
A major problem with the manufacture of electrical components incorporating an insulating material is that of detecting insulation faults. Such faults may, at the time o. manufacture, be only minor and thus difficult to detect but can subsequently be a caus2 of failure of the component. It would therefore be a considerable advantage if an incipient insulation failure could be detected by non destructive test proceedure.
For example, Ceramic dielectric capacitors, and particularly multilayer ceramic capacitors are widely used in the electronics industry as they are relatively inexpensive and have a high capacitance/volume ratio. It is usual to employ a multilayer structure when fabricating ceramic capacitors, so that layers of ceramic are interleaved with layers of metal electrode in such a way that an interdigitated two~electrode component of high capacitance value is produced. Various methods are used to make the ceramic layers as thin 'leaves', usually formed from a mix of the finely powdered ceramic material and an organic binder solvent system. For example, in a typical conventional process, a ceramic/binder/solvent mixture is coated on to 9~

polyethylene strip, by a tape drawing process~ After ~ drying, the ceramic/binder film is peeled off and then silk screen printed with electrodes using an ink formed from precious metal powders in an organic binder. A
number of such 'leaves' are stacked and pressed together, diced, heated to remove the binder, then fired at a high temperature. End terminations and leads may be attached following normal practice and such processes as described above are well known in the art of multilayer ceramic capacitor manufacture. Following the present industry trend to decrease dielectric thickness the dielectric film integrity has assumed great importance.
It is desirable to decrease the capacitor size for several reasons, mainly compatibility with micro-eIectronic trends and economy of materials.
A problem that has arisen with presently manufactured ceramic multilayer capacitors is that of occasional cracking of the dielectric, during the manufacturing process. This cracking provides an ~ intrinsic breakdown path between the capacitor electrodes or between an electrode and the opposite plurality end termination and can lead to subsequent failure in service. The mechanism of this failure is not fully understood, but it is thought to involve electrochemical dissolution and migration o~ the materials electrode or end termination which then provides a low resistive breakdown path. This rnigration is thought to occur in those cracked regions to which there is access to the atmosphere.
In most instances cracking of the dielectric cannot be detected by visual inspection and the defect only becomes manifest after the capacitor has been in use for an extended period. It is clearly desirable to reduce such long term failures to a minimum.
Further problems arise with components provided with insulation e.g. in the form of an encapsulating plastics material. Again it is desirable to detect flaws in this-insulation to eliminate the risk of subsequent failure.
Previous attempts to detect insulatioll faults have generally involved some form of electrical destructive testing on a batch basis. Such tests cannot be employed to provide 100 percent screening of components. Furthermore since they generally involve the provision of relatively costly precautions.
A number of techni~ues have for example been proposed for dielectric crack detection in ceramic capacitors. In one process acoustic emissions from the capacitors are monitored. These acoustic emissions are then computer processed to deterrnine the good and bad capacitors. Such a process is however costly and somewhat uncertain.
The term component as employed herein is understood to relate not only to capacitors but also to other electrical components and devices including inter alia integrated circuits, wire products and cables~
The object of the present invention is to minimise or to overcome the disadvantages of the prior art techniques by providing an insulating fault detection process that is inexpensive, reliable and non-destructive.
According to one aspect of the invention there is provided a process for testing for an insulation fault in an electrical component, the process including subjecting the component to a volatile mobile ionising solvent, measuring a parameter associated with the electrical condition of the insulation, and comparing the parameter with a reference value to provide an indication of the presence of an insulation fault.
According to another aspect of the invention there is provided a process for testing for an insulation fault in a component, the process including subjecting the component to a mobile ionising solvent, rneasuring the electrical leakage current of the .
~ ` ~

;~

component, and comparing the leakage current with a - reference value to provide an indication of the presence of an insulation fault.
According to another aspect of the invention there is provided an apparatus for measuring and testing for insulation faults in components, the apparatus including means for applying a mobile ionising solvent to the component so as to penetrate any discontinuity in the insulation, means for measuring a parameter associated with the electrical condition of the solvent treated component, and a comparator whereby the leakage condition is compared with a reference value corresponding to the leakage condition of an untreated component.
Advantageously the process is used in the on-line testing or screening of capacitors, e.g. of the ceramic or mica type. The test may be applied at the fabrication stage to detect dielectric faults and/or at the encapsulation stage to detect insulation faults.
We have found that treatment of an insulator with a mobile ionising solvent, for example a lower alkyl alcohol, provides an efficient non-destructive means for fault detection. The electrical leakage current of the insulator is measured prior to treatment with the liquid and is measured again after treatment.
The two measurements are then compared. Alternatively a single measurement is made after solvent treatment and compared with a reference value. A significant increase in the leakage is indicative of the presence of one or more potentially active faults in the dielectric.
A variety of solvents may be used for this process. In general the solvent is of the type that produces an increase in leakage current when applied to a capacitor having a cracked dielectric.
An embodiment of the invention will now be described with reference to the accompanying drawing in which:

- Fig. 1 is a schematic diagram of a capacitor measurement and test apparatus;
- and Fig. 2 is a cross-section of a multilayer capacitor illustrating typical dielectric faults.
It should be understood that this description is by way of example only and ~hat the techniques described herein are in no way limited to capacitor applications.
Referring to Fig. 1, capacitors 11 to be treated are mounted on a conveyor 12 and are carried through a first test station 13 where the electrical leakage of each capacitor 11 is measured and the result stored in a memory 14. The capacitors are then carried via a treatment station 15, where they are immersed in or sprayed with a mobile ionising solvent, to a drying station 16 whereby excess solvent is removed. Typically drying is effected in a current of air at ambient temperature. In some applications the capacitors may be preheated e.g. to 100C prior to immersion to enhance penetration by the solvent.
The treated capacitors are carried from the drying station 16 to a second test station 17. The electrical leakage current of each capacitor is again measured and is compared, via comparator 18, with the measurement for that capacitor recalled from the memory 14. If the second leakage current measurement is significantly higher than the first, i.e. the measurement differ by a predetermined magnitudeJ then that capacitor is directed to a reject bin. In this way ceramic capacitors, and particularly multilayer ceramic capacitors, may be screened to remove those whose dielectric is imperfect~
In a further preferred embodiment the first test station 13 is dispensed with and the leakage current of each solvent treated capacitor is compared with a reference value corresponding to the leakage of an untreated good capacitor. Those capacitors whose leakage current is significantly greater than the ~ reference val~e are rejected.
In some applications a further screening of the capacitors by this technique ma~ be effected after encapsulation. Defects in the encapsulation allow the ingress of the solvent causing a subsequent size in leakage. Again the test process is non destructive and can thus be applied on a lOOg6 basis.
Ceramic dielectrics are of course not the only dielectrics that can be tested in this way. We have successfully applied the technique to the non-destructive testing of mica capacitors.
In a typical test process a batch of 0.1 microfarad multilayer capacitors formed from an X7R
dielectric were tested for electrical leakage at 10 volts. In each case the leakage current was less than or e~ual to 10 amps. The capacitors were then immersed in methanol for 10 minutes, air dried and remeasured for leakage. The majority maintained a leakage of 10 9 amps but a few showed an increase in leakage to 5 x 10 9 amps or greater. These latter capacitors when subsequently sectioned and microscopically examined were found to exhibit dielectric cracking.
A variety of liquids may be employed in the technique. Typically we employ methanol, but ethanol, isopropyl alcoholJ industrial methylated spirit or mixtures of any of these solvents may be used. ~ess advantageously water containing a wetting agent can be employed, although in certain applications the use of water is undesirable. This list of sol~ents is given by way of example only and is not to be considered as li~iting. Preferably the solvent should be mobile, i.e.
a low viscosity and surface tension to allow rapid penetration. The so]vent should also be polar and of the ionising type. Where methanol is employed it is - preferred that the water content is less than 0.1% and the conductivity is about 2 rnicromhos.

. , .

9~

In sorne applications the efficiency of the - solvent can be enhanced by the addition of a small quantity, e.g. 0.01%, of an ionoqen or mixtures thereof.
Materials of this nature, such as triethylamine, enhance the generation of ions in the-solvent and hence increase the electrical conductivity.
It is preferred, although not essentialr that the ionogen is relatively volatile so that subsequent removal of the material from the component can be readily effected.
The prefered liquid for testing ceramic dielectrics is methanol since it has a higher conductivity than the other alcohols and a suitably low viscosity~ The relative sensitivities of the above mentioned liquids can be demonstrated by the following example. A multilayer ceramic chip capacitor had a reference value of insulation resistance equal to 101 ohms measured 10 seconds after applying lOVDC. The device, remeasured after 15 minutes immersion in methanol followed by sur~ace drying, had an insulation resistance of 108 ohms indicating the presence of a crack between two or more electrodes that had a path to the external environment. The corresponding values of insulation resistance after immersion in ethanol and 25 isopropyl alcohol were 5X108 ohms and 5xlO9 ohms respectively. For the testing of mica capacitors we prefer to employ ethyl alcohol or isopropyl alcohol as, for this particular dielectric, methanol can be over-sensitive.
Addition of an ionogen, e.g. triethylamine at a concentration of about 0.01~, can significantly increase the conductivity of the solvent thus enhancing the sensitivity of the technique. The use of an ionogen is not however essential and we have successfully employed the technique usinq untreated solvents.
The mechanism of the effect is simply a shunting of electeodes connected by a crack or other flaw by the penetrant liquid that remains in the crack ~ after the s~rface liquid has evaporated. The rapid evaporation of this surface liquid is necessary to prevent mas~ing of the effect by surface conductivity.
It has ~een found possible to differentiate -between types of defect in multilayer ceramic chip capacitors according to the relative behaviour of the post-immersion insulation resistance and the reference value. These defects are illustrated in Fig. 2. A
simple crack 21 between two opposing internal electrodes 22 will have an effect as described in the above example. The insulation resistance recovers to the reference value as the liquid evaporates from the crack.
The time to recovery is dependent on the dimensions of the crack but is usually of the order of minutes.
If the defect bridging the electrodes is linked fine porosity, as is often the case in the ZSU
dielectric, the recovery time is very long and in extreme cases there may be no noticeable recovery after several hours.
Another type of defect, known as a knit line fault 23, can also be detected. This is in effect a non-lamination of successive dielectric layers and is seen as a crack extending from one end termination of the device to an internal electrode of the opposite polarity~ Since the end termination material (usually silver) is normally different from the internal electrode material the result of the test has a polarity dependence. If, for example, the end termination (silver) is positive and the internal electrodes, say Palladium/silver, are negative, silver migration can readily occur in the presence of methanol along the knit line crack ~rom the end termination and a silver dendrite can grow back from the internal electrode to 35 the end termination. This results in an insulation resistance that decreases with time. If the polarity is reversed this rapid migration does not occur and the 6~a g test behaviour is the same as that for a sim?le crack between internal electrodes.
The presence of these deEects in capacitors with the above described test behaviour has been confirmed by destructive physical analysis. The relevance of this test with regard to multilayer ceramic capacitors is its use for example as a screening technique for potential low voltage failure. This failure mode is believed to be due to the electrochemical dissolution and migration of electrode materials under an applied d.c. electric field in the presence of atmospheric moisture and various impuritiesO
An essential feature for the occurrence of this failure mode would then be a flaw connecting two or more opposing electrodes and a path to the outside environment from this flaw. The above procedure can detect such a flaw.
The following test procedure is a simple manual implementation that has been found suitable for 100 nF, 100 V ceramlc capacitors.
1. 10 V dc is applied to the capacitor and the current tI1) is measured after 10 seconds.

2. The capacitor is pre-heated to 85C and immersed in methanol at room temperature for a period of 15 minutes. This immersion time is not critical and can be extended indefinitely. However, immersion times of less than one minute can be insufficient to allow methanol penetration in fine cracks or porosity.

3. The capacitor is removed from the methanol, dried on a tissue and blow-dried with air at room temperature until all traces of methanol have been removed from the surface. The total drying time should be as short as possible and not exceed one minute since the methanol can evaporate from large cracks in a very short time.

4. Step 1 is repeated immediately after drying and the current (I2) is measured.

A capacitor should be rejected if I2 exceeds Il significantly. In practice, if a significant defect is present, then the current I2 is usually greater than 10 8 amps and the ratio I2/Il can be several orders of magnitude. In general, lower value capacitors give higher ratios for similar size defects.
It is therefore possible, in the case of low value capacitors and those with a close insulation resistance tolerance1 to omit step 1 and use a predetermined reference value as the failure criterion.
Table 1 lists the results at 2600 hours of a test at g7% R~, 85C on chip capacitors with 4.5 V dc across the components each with 100 kilohms representative of typical production lots but have been selected to contain a relatively large number of screen rejects. Before testi~g, the capacitors were subjected to a 20 hour 'burn in' at rated volts and 85C. The 'burn in' failures, assessed as a greater than 50% drop in insulation resistance, are not included as screen rejects or life test failures. The insulation resistance of each capacitor was monitored throughout the test and the failure criterion was set at 500 megohms. The one failure that was not rejected by screening was tested again after removal from the life test and was found to have developed a flaw during the life test.
The failure characteristics of capacitors in accelerating environments are similar to real time failures, exhibiting both transient and permanent low insulation resistance.

Screened_Chip C P-cito _ ln 97% RH, 85oC at_2600 _lours Capacitor Nu~ber- Burn-inScreen Life Test ~ailures Type- on test Failures ~ejects Screen Screen . Rejects Passes .
X7R 50 V 100 nF 20 5 5 2 0 X7R 100 V ~ 14 1 3 0 0 _ Table 2 lists the results of a life test carried out according to MIL~C-123 on two lots of resin dipped capacitors. The test was extended to 1200 hours.
Capacitors were screened by the test described herein at both the leaded chip stage and after epoxy dipping.
Those included in the life test were selected to contain a large number of rejects from chip screening. The results of screening shown in Table 2 refer to the leaded chip stage. No screen rejects were recorded after encapsulation indicating that the test would be unlikely to detect a defective chip if the encapsulation was mechanically sound~
In this group, the capacitors were screened before the 'burn-in' and, although it is not claimed that screening will detect normal load test failures, it is worth noting that most of the burn in failures had been rejected.

. ~, ~9~6't-~

.

Resin Dipped Capacitors Screened at Chip Stage in 85% RH, 85C at 1200 hours .
CapacitorNumber Screen Burn-in Failures Life Test Failures Type On Test Rejects Screen Screen Screen Screen - -Rejects Passes Rejects Passes .
X7R 100 V 100 nF97 49 9 1 5 0 Z5~ 73 34 8 0 21 2 g~

- The final group, details of which are shown in - Table 3, were subjected to the cylic MII,-STD-810C
moisture resistance test. The capacitors, which were moulded X7R components, were measured at 5 V dc 2~ hours after removal from the test environment. Owing to circuit board leakage only currents above 10 nA were considered as a failure condition. The components on this test differ from the previous two groups in that they are from normal bought-in production lots and the chips themselves are of high quality. Two of the lots had a thermoplastic encapsulant, while the remainder had the more usual thermosetting type. The failures on this test, which were predominantly in the capacitors with the thermoplastic encapsulant, were due to moisture trapped at the ceramic/encapsulant interface through which an ionic current passes between the end terminations giving rise to silver dendrite growth.

,.

Screened Moulded Capacitors Tested to MIL-STD-810C

.
Lot Moulding Number Screen MIL-STD-810C Failures on Test Rejects Screen Re~ects. Screen Passes 1 thermoplastic 100 99 96 2 thermoplastic 100 97 80 3 thermosetting 100 0 0 0 4 thermosetting 100 1 0 0 thermosetting 100 13 4 6 thermosetting 100 1 2 .

In a further example a life test was conducted in an environment at 85 deg C and 100% RH (non-condensing) on 40 chip capaci~ors 17 of which had failed the above screening technique. After 1~00 hours, 8 of these had failed at an insulation resistance below their specification. None of those that had passed the screening failed on life test.
The test can also be extended to encapsulated capacitors and other components. In the case of encapsulated devices a test failure is indicative of an encapsulant defect that allows the outside environment access to an internal defect bridging two or more of the opposing electrodes that are connected to the circuit.
The bridging defect need not be within the body of the device but can be across the surface along a path formed by lack of bonding between the encapsulant and the surface of the device. This type of defect is particularly important in encapsulated multilayer ceramic capacitors with silver end terminations as silver migration can readily occur in a humid environment resulting in the growth of a shorting silver dendrite between end terminations on the cera~ic surface. If the circuit impedance is sufficiently low, the heat generated when the dendrite shorts out the end terminations may even cause the encapsulant to carbonise and catch fire.
These examples demonstrate the feasibility of the techniques described herein for the non destructive treating of ceramic capacitors prior to use.

Claims (14)

CLAIMS:-

1. A process for testing for an insulation fault in an electrical component, the process including subjecting the component to a volatile mobile ionising solvent, measuring a parameter associated with the electrical condition of the insulation, and comparing the parameter with a reference value to provide an indication of the presence of an insulation fault.

2. A process for testing for an insulation fault in a component, the process including subjecting the component to a mobile ionising solvent, measuring the electrical leakage current of the component, and comparing the leakage current with a reference value to provide an indication of the presence of an insulation fault.

3. A process as claimed in claim 1, wherein the liquid is methanol, ethanol, isopropyl alcohol or mixtures thereof.

4. A process for testing for a dielectric fault and/or an insulation fault in a multilayer capacitor, the process including subjecting the capacitor to a mobile ionising solvent, measuring the electrical leakage current of the capacitor, and comparing the leakage current with a reference value to provide an indication of the integrity of the dielectric and/or the insulation.

5. A process as claimed in claim 4, wherein the capacitor is a multilayer ceramic capacitor.

6. A process as claimed in claim 5, wherein the solvent is methanol.

7. A process as claimed in claim 6, wherein the methanol has a water content of less than 0.1%.

8. A process as claimed in claim 4, wherein the capacitor is a multilayer mica capacitor.

9. A process as claimed in claim 7, wherein the solvent is isopropyl alcohol.

10. A process as claimed in claim 4, 5 or 6, wherein the capacitor is tes-ted both prior to and after the application of an insulating coating.

11. A capacitor tested by a process as claimed in claim 4, 5 or 6.

12. A process as claimed in claim 1, 2 or 4, wherein said solvent contains an ionogen or mixtures thereof.

13. A process as claimed in claim 1, 2 or 4, wherein said solvent contains an ionogen or mixtures thereof and the ionogen is triethylamine.

14. An apparatus for measuring and testing for insulation faults in compon-ents, the apparatus including means for applying a mobile ionising solvent to the component so as to penetrate any discontinuity in the insulation, means for meas-uring a parameter associated with the electrical condition of the solvent treated component, and a comparator whereby the leakage condition is compared with a ref-erence value corresponding to the leakage condition of an untreated component.