Screening procedures for electrolytic capacitors

Abstract

Method for iteratively screening a batch of electrolytic capacitors having a predetermined rated voltage, the method comprising: measuring an initial leakage current in an initial batch of capacitors and calculating an initial average leakage current from the results; Removing capacitors from the first set that have a measured first leakage current greater than or equal to a first predetermined value, thereby creating a second set of capacitors, wherein the first predetermined value is one or more standard deviations above the first mean leakage current; Performing a burn-in treatment with the second set of capacitors, wherein the burn-in treatment includes applying a predetermined test voltage, approximately 0.8 to approximately 1.2 times the rated voltage, to the capacitors; After the burn-in treatment, measure a second leakage current at the second set of capacitors, calculate a second average leakage current from the same; and Removing capacitors from the second set that have a measured second leakage current greater than or equal to a second predetermined value, thereby creating a third set of capacitors, where the second predetermined value is one or more standard deviations above the second mean leakage current.

Description

Background of the invention

[0001] EP 1 978 370 A1 describes a method for conditioning and selecting tantalum capacitors to achieve low failure rates, the method comprising the steps: Applying a series of voltage pulses to the capacitor to cause a series of current pulses through the capacitor, Applying a predetermined voltage to the capacitor and measuring a leakage current flowing through the capacitor at predetermined time intervals after applying the predetermined voltage to the capacitor, based on the leakage current measurements made on the capacitor, determining whether the capacitor is suitable, wherein the capacitor is determined to be acceptable if the capacitor has a leakage current value that decreases over time or is less than a predetermined current value at a selected one of the predetermined time intervals, wherein the capacitor is determined to be unacceptable if the capacitor has a leakage current value that increases over time or is equal to or greater than a predetermined current value at the selected one of the predetermined time intervals.

[0002] US 7 671 603 B2 describes a method for screening electrolytic capacitors, comprising: Applying a DC voltage equal to or greater than a mean breakdown voltage to each capacitor and an associated resistor while each capacitor is connected in series with the associated resistor Rs, monitoring the charging curve of each capacitor and associated resistor until the voltage drop reaches the mean breakdown voltage or until time t is reached, and classifying the electrolytic capacitors based on the observed voltage drop.

[0003] CN 1 01 313 380 A describes a system for shielding a capacitor, comprising: a power supply electrically connected to a connector for receiving at least one capacitor; a controller operatively connected to the power source and the connector, the controller selectively applying an electrical signal from the power source to the at least one capacitor and selectively receiving an electrical input indicative of a state of charge of the at least one capacitor; and logic instructions executable by the controller to compare changes in the state of charge of the at least one capacitor within a predetermined period of time with at least one threshold value to check the at least one capacitor for leakage current and self-discharge of the at least one capacitor.

[0004] US 3 553 805 A describes a method for testing solid electrolytic capacitors to obtain capacitors with specifically graded reliability, which comprises subjecting the capacitors to an accelerated life test in which voltage is applied to the capacitors for specified periods of time; recording the exact elapsed time whenever a failure occurs in a capacitor; accumulating the number of such failures after each failure and calculating the accumulated failures as a percentage of the total test batch; plotting the cumulative failure percentage against the respective failure ages; and rejecting all capacitors that do not survive the life test.

[0005] Electrolytic capacitors are used in various medical, military, aerospace, and commercial applications where reliability and extremely low failure rates are critical. Therefore, various screening methods, such as accelerated aging tests, reflow tests, current peak tests, and breakdown voltage tests, have been developed to screen electrolytic capacitors and eliminate defective parts. However, many of these tests have failure criteria that only look for complete failures (i.e., fuse blows), allowing defective parts to slip through undetected. These screening methods cannot uncover latent defects.For example, a fuse may not have blown under severe stress conditions such as high voltage and temperature, but the tested capacitor may have been damaged during screening, which can lead to long-term instability. Conventional methods for screening and delivering high-reliability electrolytic capacitors involve Weibull calculations based on batch sampling, whereby a small number of capacitors are subjected to highly elevated voltage (e.g., 1.5 times the rated voltage (VR)), temperature (e.g., 85°C), and time (e.g., 40 hours or more) conditions during burn-in. Conventional Weibull burn-in allows parts that are statistically different before burn-in to enter the normal population after burn-in because there is no pre-burn-in screening to weed out parts with early-onset defects.Although many of these parts appear stable in long-term reliability tests (e.g., endurance tests), possibly due to self-healing during burn-in, a proportion of the parts entering the normal population are unstable and may exhibit long-term reliability problems in practice. Weibull statistical calculation encourages the practice of leaving these unstable parts in the population so that a Weibull distribution can be constructed for performance grading purposes, as described in MIL-PRF-55365H. As a result, screening using Weibull tests cannot ensure the removal of unstable or defective capacitors from the population, which may result in a capacitor batch that exhibits an unacceptable level of reliability.Despite the advantages achieved, there is a need for an improved screening method for electrolytic capacitors that can detect and reject capacitors that have latent defects, as well as for a method for determining the predicted failure rate of the selected capacitors that, in contrast to the Weibull method, does not take the rejected capacitors into account. Brief description of the invention

[0006] According to one embodiment of the present invention, a method for iteratively screening electrolytic capacitors having a predetermined voltage rating is disclosed. The method comprises measuring a first leakage current for a first set of capacitors and calculating a first average leakage current from the results, and removing capacitors having a measured first leakage current greater than or equal to a first predetermined value from the first set, thereby creating a second set of capacitors. The first predetermined value is one or more standard deviations above the first average leakage current. The method further comprises performing a burn-in treatment on the second set of capacitors. The burn-in treatment comprises applying a predetermined test voltage, which is about 0.8 to about 1.2 times the rated voltage, to the capacitors.After the burn-in treatment, a second leakage current is measured for the second set of capacitors, and a second average leakage current is calculated from this. Then, the capacitors in the second set that have a measured second leakage current greater than or equal to a second predetermined value are removed from the second set, creating a third set of capacitors. The second predetermined value is one or more standard deviations above the second average leakage current.

[0007] In another embodiment, a method for delivering a batch of capacitors to a customer is disclosed. The method includes determining a voltage rating for the capacitors and iteratively screening the capacitors, wherein, at each iteration, capacitors whose leakage current is above a predetermined value are removed from the batch. The predetermined value is one or more standard deviations above the average leakage current measured during the iteration. After screening, the batch of capacitors can be delivered to the customer without reducing the voltage rating.

[0008] In yet another embodiment, a method for calculating a predicted failure rate for electrolytic capacitors is disclosed. The method includes performing a burn-in treatment on the capacitors at a first temperature and a first voltage for a first period of time; performing a life test on the capacitors at a second temperature and a second voltage for a second period of time; and determining the number of capacitors that failed after the life test based on the number of capacitors that have a leakage current above a predetermined value, wherein capacitors that failed before the burn-in treatment are excluded from the calculation to determine the predicted failure rate.

[0009] Further features and aspects of the present invention are set forth in more detail below. Brief description of the drawings

[0010] In the remainder of the specification and with reference to the accompanying drawings, a full and reproducible disclosure of the present invention, including its best mode of implementation, is particularly set forth for those skilled in the art, in which: Fig. 1 is a block diagram showing an embodiment of the method of the present invention; Fig. Figure 2 is a cross-sectional view of an embodiment of a solid electrolytic capacitor that can be screened by the method of the present invention; Fig. 3 is a cross-sectional view of another embodiment of a solid electrolytic capacitor that can be screened by the method of the present invention; Fig.4 is a cross-sectional view of an embodiment of a liquid electrolytic capacitor that can be screened by the method of the present invention; Fig. Figure 5 is a graph showing the first leakage currents for a sample of capacitors and separating the leakage currents into three zones; Fig. 6 a graph showing the leakage current of the “Zone 1” parts of Fig. 5 shows after assembly; Fig. 7 is a graph showing the leakage current of the “Zone 1” parts of Fig. 5 after a 1000-hour life test at 125 °C and 2 / 3 of the rated voltage; Fig. Figure 8 is a graph comparing the leakage current for each of the ‘Zone 1’ capacitors subjected to a 1000-hour life test at 125 °C and 2 / 3 of the rated voltage at different stages of the life test; Fig.9 is a graph showing the change in leakage current for each of the capacitors of Fig. 6 to Fig. 7 shows; Fig. Figure 10 is a graph showing the leakage current of three different batches of Zone 1 capacitors subjected to a burn-in at 125°C and then a 2000-hour life test at 85°C and the rated voltage, at different stages of the life test; Fig. Figure 11 is a graph showing the leakage current of three different batches of Zone 1 capacitors subjected to a burn-in at 125°C and then a 2000-hour life test at 85°C and rated voltage, at different stages of the life test; Fig. Figure 12 is a graph showing the leakage current of two different batches of Zone 1 capacitors subjected to a burn-in at 125°C and then a 2000-hour life test at 85°C and the rated voltage, at different stages of the life test; Fig. Figure 13 is a graph showing the leakage current of two different batches of Zone 1 capacitors subjected to a burn-in at 125°C and then a 2000-hour life test at 85°C and the rated voltage, at different stages of the life test; Fig. Figure 14 is a graph showing the leakage currents of the capacitors with the ten highest leakage currents after burn-in, but still within the three standard deviation exclusion limit when measuring the first leakage current, while the capacitors were subjected to a 2000-hour endurance test at 85 °C and the rated voltage; Fig. 15 is a graph showing the leakage current behavior of capacitors that failed the screening according to the first leakage current before burn-in and were initially categorized as “Zone 2” parts, but showed self-healing and would have become “Zone 1” parts in a later test; Fig. 16 is a graph showing the leakage current behavior of capacitors that failed the first leakage current screening but were within the 0.01*CV*12 limit; Fig. 17 is a graph comparing the leakage current of a group of capacitors before burn-in with the leakage current of capacitors after burn-in after being subjected to the burn-in method of the present invention and the Weibull burn-in method; Fig. 18 is a graph comparing the leakage current shift of capacitors that passed the pre-burn-in screening method of the present invention with the leakage current shift of capacitors that failed the pre-burn-in leakage current screening method after the capacitors were subjected to a life test at 85°C; Fig.Figure 19 is a graph comparing the leakage current for capacitors tested at 25 °C and 125 °C; Fig. 20 is a graph comparing the leakage current of a batch of capacitors before burn-in with the leakage current of the capacitors after burn-in, after being subjected to the burn-in method of the present invention and the Weibull burn-in method; Fig. 21 is a graph comparing the leakage current of several batches of capacitors before burn-in with the leakage current of the capacitors after burn-in, after being subjected to the burn-in method of the present invention and the Weibull burn-in method; Fig.22 is a graph comparing the life tests of capacitors subjected to the pre-burn-in screening and subsequent burn-in method of the present invention with the Weibull burn-in method; Fig. 23 is a graph showing the shift in leakage current after a lifetime test at 85°C with capacitors subjected to the screening and burn-in method of the present application; Fig. Figure 24 is a graph comparing the leakage currents of capacitors before the life test in a batch with the leakage current of the capacitors after the life test, with the capacitors grouped into Zone 1 units, Zone 1 at limit units and Zone 2 units, as described in Fig. 5 is shown; Fig. 25 the graphic of Fig. 24, which is scaled to show leakage current up to about 0.25 microamperes (µA); Fig.26 is a graph showing a method for determining whether a batch of capacitors is an outlier compared to other batches of capacitors; and Fig. 27 is a cross-sectional view of one embodiment of a hermetically sealed capacitor that can be screened by the method of the present invention.

[0011] Where reference symbols are used multiple times in the present description and the drawings, these are intended to represent the same or analogous features or elements of the invention. Detailed description of representative embodiments

[0012] Those skilled in the art should appreciate that the present discussion is only a description of exemplary embodiments and is not intended to limit the broader aspects of the present invention.

[0013] Generally speaking, the present invention relates to a method for iteratively screening electrolytic capacitors. The method described in the present application for screening a batch or sample of electrolytic capacitors (i.e., two or more capacitors) comprises measuring the leakage current of the capacitors in several iterations of the test procedure and then removing from the sample or batch capacitors that exhibit a leakage current above a predetermined value, determined by statistical analysis at each iteration.For example, the leakage current of a first set of capacitors, which may include all capacitors in the batch, may be measured. Those capacitors in the first set that exhibit a leakage current above a predetermined value after the first leakage current measurement may be removed from the sample or batch, and the remaining capacitors may form a second set of capacitors that may be subjected to further testing. The further testing may include a burn-in heat treatment of the capacitors at a predetermined voltage, such as the capacitors' voltage rating; the latter is a predetermined nominal value related to the DC voltage rating for continuous operation up to 85°C. The voltage rating is based on the thickness of the dielectric layer.The burn-in treatment can be used to subject the capacitors to stress and thereby detect unstable parts when measuring the leakage current of the second set of capacitors.

[0014] After the burn-in process is complete, additional processing, such as reflow soldering, may occur, which may subject the capacitors to further stress. Either before or after reflow soldering, a second iteration of screening may take place, wherein the leakage current is measured for each of the capacitors in the second set, and capacitors in the second set that have a leakage current above a new, second predetermined value are removed from the second set, creating a third set of capacitors that can then be subjected to further testing. Furthermore, at the end of the screening process, a third iteration of screening may take place, wherein the leakage current is measured for each of the capacitors in the third set, and capacitors in the third set that have a leakage current above a third predetermined value are removed from the third set, creating a fourth set of capacitors.The first, second, and third predetermined values ​​are determined by statistical analysis. In addition to screening capacitors based on leakage current, other parameters such as capacitance, equivalent series resistance (ESR), and dissipation factor (DF) can also be measured, and the capacitors can be further screened based on a statistical analysis of the average capacitance, ESR, and DF results.Without wishing to be bound by any particular theory, we believe that subjecting the capacitors to be screened to a burn-in treatment at a predetermined test voltage close to their rated voltage, as opposed to a test voltage 1.5 times their rated voltage, which can permanently damage the capacitors, combined with performing multiple iterations of leakage current tests on the capacitors, screening out all capacitors above a predetermined value at each iteration, can effectively screen out unstable capacitors, producing a batch of capacitors with extremely high reliability and extremely low failure rates. Such high-reliability, low-failure capacitors are critical in certain applications, including medical, military, and aerospace.

[0015] An embodiment of the method of the present invention is shown in the block diagram of Fig. 1, but it should be understood that the steps can be performed in a different order, and additional tests or iterative screening can be added to further remove defective or unstable capacitors. The iterative screening procedure 100 of Fig. For example, Figure 1 shows various process steps 102 and screening steps 104 that are performed to obtain specific output values ​​106. The process steps 102 include: a first iteration (before burn-in) DCL measurement 108, burn-in 114, reflow soldering 120, a second iteration (after burn-in) DCL measurement 126, and a third iteration DCL measurement 132.

[0016] The leakage currents in steps 108, 126, and 132 are measured using a leakage test set that measures the leakage current (DCL) at a predetermined temperature and at the rated voltage after at least 10 seconds. For example, the leakage current may be measured after 3 minutes at a temperature ranging from about 20°C to about 85°C across a charging resistor with a minimum resistance of 1 kΩ connected in series with the capacitor with the rated voltage applied. The leakage current may also be measured at higher temperatures (i.e., hot DCL), although at temperatures greater than about 85°C up to about 140°C, such as about 125°C, the applied voltage may be about 2 / 3 of the rated voltage to normalize the capacitor's stress to the value it experiences, for example, at about 85°C. Leakage current or DCL refers to the current that flows through a capacitor when a DC voltage is applied after the capacitor is charged.In general, leakage current can be used to determine whether a capacitor has defects or is at risk of failure, and a capacitor may need to have a leakage current below a minimum value to be suitable for use in a given application. Leakage current depends on a variety of factors, such as the voltage applied to the capacitor, the temperature conditions, and the type of electrolyte to be used in the capacitor. As discussed below, results from leakage current measurements taken in various iterations of the screening process can be used to determine whether a capacitor is acceptable or should be removed from a given batch or sample.

[0017] We now discuss the iterative screening method of the present invention in terms of the specific method steps, screening steps, and output values. In a first iteration of the screening method, a first leakage current may be measured for the batch or sample of capacitors to be tested (i.e., a first set of capacitors), as in method step 108 of Fig.1. The first leakage current may be determined at a temperature that, in some embodiments, ranges from about 20°C to about 150°C. For example, the DCL may be measured at a temperature that, in some embodiments, ranges from about 20°C to about 30°C, such as about 25°C, in some embodiments from about 75°C to about 95°C, such as about 85°C, and in still other embodiments from about 100°C to about 150°C. For example, the temperature may range from about 110°C to about 140°C, such as about 120°C to about 130°C, such as about 125°C. However, when the leakage current is measured at a temperature above 85°C, as discussed above, the voltage applied during the leakage current measurement may be about 2 / 3 of the nominal voltage, as discussed above.Regardless of the temperature at which the first leakage current is determined, a first average leakage current may be calculated from the data collected regarding the first leakage current measurements for the first set of capacitors. After the first average leakage current is determined, a first predetermined value for the first set of capacitors may be calculated, which may be greater than or equal to one or more standard deviations above the first average leakage current. However, in some embodiments, the first predetermined value may be three or more standard deviations above the first average leakage current. This is demonstrated by method step 110, where a first threshold (i.e., predetermined value) is determined.Then, all capacitors exhibiting a first leakage current above the first predetermined value may be removed from the sample or batch, as they represent potentially unstable capacitors or outliers, as shown in output step 112. Meanwhile, all capacitors exhibiting a first leakage current below the first predetermined value pass the first iteration of screening and may remain in the sample or batch for further screening beyond the first iteration screening, thus forming the second set of capacitors.

[0018] After removing capacitors above the first predetermined value, the second set of capacitors (i.e., the capacitors in the sample or batch that passed the screening of the first iteration 112) are subjected to a burn-in treatment, as shown in process step 114. Burn-in is generally the process by which capacitors may be subjected to harsh conditions to determine whether they are prone to failure early in their lifetime. The burn-in treatment 114 may involve selectively controlling and applying a predetermined test voltage to the capacitors, as shown in step 116. The burn-in treatment may be performed at a temperature that, in some embodiments, may range from about 100°C to about 150°C, in other embodiments, from about 110°C to about 140°C, and in still other embodiments, from about 115°C to about 130°C.For example, the second temperature may be 125°C. Regardless of the temperature at which the heating or burn-in treatment 114 is performed, in one embodiment, the burn-in treatment may be performed for a period of time ranging from about 25 hours to 75 hours. Meanwhile, in other embodiments, the burn-in process time may range from about 35 hours to about 50 hours, such as about 40 hours to about 45 hours. For example, the burn-in process time may be 42 hours.

[0019] Furthermore, regardless of the temperature or time frame of the burn-in process, the burn-in process includes selectively controlling and applying a predetermined test voltage to the second set of capacitors. The applied voltage is generally in a specific relationship to the nominal voltage of the capacitors to be tested. For example, in some embodiments, the applied voltage may range from about 0.7 to about 1.3 times the nominal voltage, in other embodiments from about 0.8 to about 1.2 times the nominal voltage, and in still other embodiments from about 0.9 to about 1.1 times the nominal voltage. For example, the voltage applied during the burn-in process may be about 1.0 times the nominal voltage of the capacitors remaining in the sample or batch to be tested.It has been found that performing a burn-in treatment at approximately 1.0 times the capacitor voltage rating with the capacitors of the second set allows for sufficient screening of unstable or defective capacitors in subsequent screening iterations without causing damage to the capacitors, as occurs at higher voltages used in Weibull tests. In general, the burn-in treatment step 116 results in a reduction in the leakage current of the base population of capacitors and can be used to further debunk unstable units, as shown in output value 118.It should be noted that the reduction in leakage current at this stage may be due to self-healing, and therefore, a second iteration 130 leakage current screening step may be performed for the second set of capacitors that passed the first iteration 112 leakage current screening step to remove any outliers or defective parts after the completion of the burn-in process. The second iteration 130 leakage current screening step may be performed after the burn-in or after the reflow process step 120, as discussed in more detail below.

[0020] If desired, soldering through the reflow process step 120 may be performed on the second set of capacitors before a leakage current of the second iteration 126 is measured for the second set of capacitors. In the reflow process step 120, the capacitors may be subjected to further stresses to detect additional unstable capacitors after the leakage current screening of the second iteration 130. However, as noted above, this specific process sequence is not necessary, and it should be understood that in some cases, the leakage current screening of the second iteration 130 may also be completed after the burn-in treatment 114 but before the reflow step 120. In some cases, it may even be possible to perform the reflow step on the first set of capacitors.

[0021] If the reflow process step 120 is terminated before the second set of capacitors has been subjected to the leakage current screening step of the second iteration 130, as described in Fig. 1, it can be terminated after the burn-in treatment. In general, the reflow step is the process by which capacitors can be soldered to a board. During reflow soldering after burn-in, capacitors that have become damaged as a result of the additional thermomechanical stress that the reflow process exerts on the capacitors, as shown in the output value 124 of Fig. 1, can be exposed when the second set of capacitors is subjected to leakage current screening of the second iteration 130. For this reason, the reflow step can be performed before the leakage current measurement of the second iteration 126 to further screen and remove unstable parts.

[0022] The basic reflow soldering process includes the steps of applying a solder paste to the desired contact points on a printed circuit board (PCB), such as an FR-4 board, placing the capacitors in the paste, and applying heat to the assembly, which causes the solder in the paste to melt (reflow). The solder then wets the PCB and the capacitor terminals, resulting in the desired solder fillet joint. The reflow process can take place in a convection linear oven, as in step 122 of Fig.1. The convection linear furnace may, in some embodiments, have a peak temperature profile of about 200°C to about 280°C, in other embodiments about 205°C to about 270°C, and in still other embodiments about 210°C to about 260°C. For example, in medical, military, and aerospace applications using SnPb-based (lead-based) solders that can melt at a lower temperature, the reflow step may occur at a temperature ranging from about 210°C to about 225°C. Meanwhile, in commercial applications using a lead-free solder that melts at a higher temperature, the reflow step may occur at a temperature ranging from about 245°C to about 260°C.Note that although the reflow process discussed above uses a convection oven, the reflow process can also use an infrared convection oven or a vapor phase oven, and the process can be performed by wave soldering or by using a hot plate.

[0023] After the reflow process is finished, a leakage current of the second iteration for the second set of capacitors can be measured as in Fig. 1 as method step 126, although in some embodiments, the second leakage current may also be measured before the reflow step. The second leakage current may be measured for capacitors that passed the leakage current screening of the first iteration 112 (i.e., for the second set of capacitors) and that were also subjected to the burn-in treatment 114 and the reflow step 120, as shown in Fig.1. The second leakage current may be determined at a temperature that, in some embodiments, ranges from about 20°C to about 150°C. For example, the DCL may be measured at a temperature that, in some embodiments, ranges from about 20°C to about 30°C, such as about 25°C, in some embodiments from about 75°C to about 95°C, such as about 85°C, and in still other embodiments from about 100°C to about 150°C. For example, the temperature may range from about 120°C to about 130°C, such as about 125°C. However, when the leakage current is measured at a temperature above about 85°C, as discussed above, the voltage applied during the leakage current measurement may be about 2 / 3 of the nominal voltage, as discussed above.Regardless of the temperature at which the second leakage current is determined, a second average leakage current may be calculated from the data collected regarding the second leakage current measurements for the second set of capacitors. After the second average leakage current is determined, a second predetermined value may be calculated for the second set of capacitors, which may be greater than or equal to one or more standard deviations above the second average leakage current. However, in some embodiments, the second predetermined value may be three or more standard deviations above the second average leakage current. This is demonstrated by method step 128, where the second threshold (i.e., predetermined value) is determined.Then, all capacitors exhibiting a second leakage current above the second predetermined value may be removed from the sample or batch, as they represent potentially unstable capacitors or outliers, as shown in output step 130. Meanwhile, all capacitors exhibiting a second leakage current below the second predetermined value pass the second iteration of screening and may remain in the sample or batch for further screening beyond the second iteration screening, thus forming the third set of capacitors. Note that if the reflow process 120 was not performed for the second set of capacitors prior to measuring the second leakage current, the third set of capacitors resulting after the second iteration of leakage current screening may be subjected to the reflow process 120, as discussed above.

[0024] As an additional method step, a functional test 132 may be performed to determine standard capacitor characteristics for the capacitors that passed both the first iteration 112 leakage current screening and the second iteration 130 leakage current screening, as discussed above (i.e., for the third set of capacitors). At this stage, additional unstable or defective capacitors may be screened out and removed from the batch or sample based on further statistical analysis. The functional test may be performed at a temperature ranging from about 15°C to about 35°C in some embodiments, or from about 20°C to about 30°C in other embodiments. For example, the test may be performed at 25°C.

[0025] Regardless of the temperature at which the functional test 132 is performed, a third iteration leakage current can be measured for the capacitors remaining in the sample (i.e., the third set of capacitors), as shown in Fig.1. Regardless of the temperature at which the third leakage current is determined, a third average leakage current may be calculated from the data collected regarding the third leakage current measurements for the third set of capacitors. After the third average leakage current is determined, a third predetermined value for the third set of capacitors may be calculated, which may be greater than or equal to one or more standard deviations above the third average leakage current. However, in some embodiments, the third predetermined value may be three or more standard deviations above the third average leakage current. This is demonstrated by method step 134, where the third threshold (i.e., predetermined value) is determined.Then, all capacitors exhibiting a third leakage current above the third predetermined value can be removed from the sample or batch, as they represent potentially unstable capacitors or outliers, as shown in output step 136. Meanwhile, all capacitors exhibiting a third leakage current below the third predetermined value pass the third iteration of screening and can remain in the sample or batch for approval or further screening beyond the third iteration screening, thus forming the fourth set of capacitors. These capacitors have thus undergone at least three iterations of leakage current screening based on statistical analysis to ensure that the capacitors remaining in the batch are highly reliable with an extremely low risk of failure.

[0026] With the 132 capacitors screened in the third iteration, additional tests (not included in Fig.1). In addition to screening based on leakage current, the capacitors that passed the first two iterations of leakage current screening (i.e., the third batch of capacitors) can be tested for their equivalent series resistance (ESR), dissipation factor (DF), and capacitance measurements can be performed. These capacitors can then be further screened to remove any unstable parts based on statistical analysis with the data collected regarding ESR, DF, and capacitance. For example, if capacitance is being measured, the capacitors can be screened based on compliance with a guard band tolerance limit, whereas if ESR and / or DF are being measured, any outliers present that are 1 or more standard deviations above the mean threshold are discarded from the batch or sample to be released for use.

[0027] The screening method as described in the present invention can be carried out with both solid and liquid electrolytic capacitors. The solid or liquid electrolytic capacitor screened according to the method of the present invention can be used in various applications; these include, but are not limited to, medical devices such as implantable defibrillators, pacemakers, cardioverters, nerve stimulators, drug delivery devices, etc., automotive applications, military applications such as radar systems, consumer electronics such as radios, TV sets, etc., etc. In one embodiment, the capacitor can be used, for example, in an implantable medical device configured to provide treatment of a patient with a therapeutic high voltage (e.g.between about 500 volts and about 850 volts, or desirably between about 600 volts and about 800 volts). The device may include a container or housing that is hermetically sealed and biologically inert. One or more terminals are electrically coupled between the device and the patient's heart via a lead wire. Cardiac electrodes are provided to monitor cardiac activity and / or apply a voltage to the heart. At least a portion of the terminals (e.g., an end portion of the terminals) may be provided near or in contact with a chamber and / or an atrium of the heart. The device also includes a capacitor bank, typically including two or more capacitors connected in series and coupled to a battery disposed internally or externally with respect to the device, that supplies power to the capacitor bank.Due in part to the high conductivity, the capacitor screened by the method of the present invention can achieve excellent electrical properties, making it suitable for use in the capacitor array of the implantable medical device. For example, the equivalent series resistance ("ESR")—the extent to which the capacitor acts as a resistor during charging and discharging in an electronic circuit—can be less than about 1500 milliohms, in some embodiments less than about 1000 milliohms, and in some embodiments less than about 500 milliohms, measured with a 2-volt bias and a 1-volt signal at a frequency of 1000 Hz.

[0028] After a sample or batch of capacitors has been screened according to the method of the present invention, the sample or batch of capacitors can be delivered to a customer without first having to derate the voltage at which the capacitors can be used to a level lower than the predetermined voltage rating. In other words, the screening process can filter out unstable capacitors so that the capacitors delivered to the customer can be used at their rated voltage, as opposed to a lower (i.e., derated) voltage.

[0029] To further reduce the risk of latent defects in a batch of capacitors screened according to the iterative screening procedure discussed above, an additional screening step can be performed based on a comparison of the screened batch with other screened batches of capacitors for safety reasons in order to filter out a screened batch with a mean leakage current that is an outlier compared to the mean leakage current of all screened batches as a whole. The mean leakage current of all screened batches considered as a whole is referred to as the large mean leakage current. As described in Fig.As shown in Figure 26, the large mean leakage current of the multiple batches of capacitors may be determined after they have been screened. The mean leakage current for each of the screened batches may be determined using capacitors in the batch that passed the initial leakage current screening, such as capacitors that underwent a burn-in heat treatment or any additional treatments and / or screenings. Then, any batches of capacitors that have a mean leakage current above a predetermined value, such as a leakage current value that is one or more standard deviations above the large mean leakage current for all batches, may be rejected as a batch to be delivered to the customer. For example, a batch that has a mean leakage current that is more than three standard deviations above the large mean leakage current may be discarded.This additional screening step can limit batch-to-batch variations in capacitors delivered to the customer.

[0030] However, since the screening method of the present application involves removing all capacitors with an initial leakage current above a predetermined value from a batch of capacitors prior to the burn-in process, such capacitors being considered "early failure" or "victims of infant mortality," it is not possible to calculate a predicted failure rate using the conventional Weibull model described in MIL-PRF-55365H. Furthermore, it should be noted that the Weibull method does not consider the effects of multi-sided soldering of surface-mount components to substrates on the calculation of the predicted failure rate. Therefore, when using the screening method of the present application, it should be understood that the predicted failure rate can be calculated using a new method, as explained below.

[0031] First of all, it should be understood that to determine the predicted failure rate of a batch of capacitors delivered to a customer, before any calculation is performed, a sample from the population is run through a simulated production routine, where the production routine includes double-sided soldering. Then, a calculation is performed based on the sample's behavior in a simulated production routine. In general, the calculation of the predicted failure rate is based on two main steps: converting the results of endurance tests on a number of parts at an elevated temperature (e.g., 125°C) and voltage (e.g., 2 / 3 of the rated voltage) for a given period of time into an equivalent number of component / part-hours at 25°C.The number of failures and equivalent component hours are then used to calculate a predicted failure rate and mean time between failures. The determination of equivalent component hours is based on reliability prediction according to MIL-HDBK-217 using the Arrhenius model as applied to solid tantalum capacitors. The Arrhenius model is used to predict the failure rate increase due to temperature increases. The MIL-HDBK-217 manual is expressly incorporated herein by reference for all purposes. The calculation of equivalent component hours also considers the activation energy of tantalum capacitors, which can range from approximately 1.08 eV to approximately 1.15 eV.

[0032] Meanwhile, the failure rate calculation is based on a chi-square calculation for time-limited sample tests, where the number of degrees of freedom is equal to the sum of the number of failures and 1 multiplied by 2. The parameters to be entered into the failure rate calculation include the rated voltage of the capacitors, the number of capacitors tested, the number of hours during which the capacitor parts were tested, the test temperature, the test voltage, the number of failures, the desired confidence level, the desired application temperature, and the desired application voltage.The resulting calculated output values ​​include the equivalent number of component hours at 25°C, as demonstrated by the life tests on a specified number of samples at a specified voltage and temperature during a specified test period, taking into account a test temperature acceleration factor and a test voltage acceleration factor, as well as a predictive calculation of the failure rate for the total population of capacitors from which the samples were selected, based on the number of failures that occurred during the life tests and the total number of equivalent component hours, further taking into account an application voltage acceleration factor. Finally, the predicted failure rate calculation can be used to determine the mean time between failures (MTBF) in hours.

[0033] To determine the equivalent component hours at the end-use application temperature of the screened capacitors, a test / screening temperature acceleration factor and a test / screening voltage acceleration factor can first be determined. The following formula 1 shows how the test / screening temperature acceleration factor (TTAF) can be determined and includes a conversion of the application and test temperatures to Kelvin: TTAF=e(activation energyBoltzmann constant)x((1application temperature+273)−(1test temperature+273))

[0034] Meanwhile, the following formula 2 shows how the test / screening voltage acceleration factor (TVAF) can be determined: TVAF=(test voltage nominal voltage)3

[0035] Furthermore, the following formula 3 shows how the equivalent component hours are determined at the application temperature of the capacitors: equivalent component hours at application temperature = (Number of capacitors tested)(Test duration in hours)(TTAF)(TVAF)

[0036] Then, if desired, the equivalent component hours can be converted into equivalent component years. An application stress acceleration factor (AVAF) can then be determined, as shown in the following formula 4: AVAF=(application voltageNominal voltage)3

[0037] Now, using the calculations from the above formulas, the failure rate can be calculated, where the failure rate is shown as a percentage of failures per 1000 hours. The failure rate is based on a chi-square distribution and involves determining the inverse function of the one-tailed probability of the chi-square distribution. Formula 5 shows the equation for determining the failure rate: Failure rate (% per 1000 hours) = (CHIINV (1 − (% confidence level 100)) ⋅ (2 × (number of failures + 1)) (2 × equivalent component hours)) × (AVAF2) × ((1000 hours) × 100)

[0038] The "CHIINV" function calculates the chi-square value of two factors: the confidence level factor and the degrees of freedom factor. The confidence factor is 1 minus the entered confidence level expressed as a decimal. The degrees of freedom factor is twice the sum of the number of failures observed during the capacitor life test and one. This factor represents sample tests, which are performed regardless of the number of failures during a given period. The chi-square value is then divided by twice the equivalent component hours determined in Formula 3. The result is then multiplied by the application voltage acceleration factor (AVAF) determined in Formula 4 after first dividing the AVAF by two. This result is then multiplied by a factor of 1000 times 100, or 100,000, to obtain the final predicted failure rate in "percent failures per 1,000 hours."

[0039] After determining the predicted failure rate in percent failures per 1000 hours, as shown above in Formula 5, the failure rate can be converted into the mean time between failures (MTBF) in hours, as shown below in Formula 6: MTBF=(1((Failure rate in % per 1000 hours) / 100))×(1000 hours)

[0040] When Formula 5 shown above is used after the life test to determine the predicted failure rate of the capacitors screened according to the method of the present application, the predicted failure rate of the capacitors can be in the range of about 0.000005% failures per 1000 hours to about 0.01% failures per 1000 hours, such as about 0.000008% failures per 1000 hours to about 0.009% failures per 1000 hours, such as about 0.00001% failures per 1000 hours to about 0.008% failures per 1000 hours, when determined with a confidence level of about 50% to about 99.9%, such as about 55% to about 95%, such as about 60% to about 90%. In a particular embodiment, the predicted failure rate of the capacitors may range from about 0.00001% failures per 1000 hours to about 0.008% failures per 1000 hours at a confidence level of about 90%.

[0041] As discussed above, the capacitors screened by the method of the present invention can be solid or liquid electrolytic capacitors. A solid electrolytic capacitor generally includes a capacitor element comprising an anode body, a dielectric layer, and a solid electrolyte. The capacitor may also include an anode lead (e.g., ribbon, wire, plate, etc.) electrically connected to the anode body for connection to an anode terminal. The valve metal composition may include a valve metal (i.e., a metal capable of oxidation) or a compound based on a valve metal, such as tantalum, niobium, aluminum, hafnium, titanium, alloys thereof, oxides thereof, nitrides thereof, etc.For example, the valve metal composition may include an electrically conductive oxide of niobium, such as a niobium oxide having a niobium to oxygen atomic ratio of 1:1.0 ± 1.0, in some embodiments 1:1.0 ± 0.3, in some embodiments 1:1.0 ± 0.1, and in some embodiments 1:1.0 ± 0.05. The niobium oxide may be NbO. 0,7 , NbO 1,0 , NbO 1,1 and NbO2. In one embodiment, a valve metal-based powder is used, which is compacted using a conventional powder mold to form a porous anode body. The porous anode body is then sintered to form a porous integral body.

[0042] Once constructed, a dielectric layer can be formed by anodizing the sintered anode body. This results in the formation of a dielectric layer on and / or within the anode. For example, a tantalum (Ta) anode can be anodized to tantalum pentoxide (Ta2O5). Typically, anodization is performed by first applying an electrolyte to the anode, such as by immersing the anode in the electrolyte. The capacitor element also contains a solid electrolyte, which acts as the cathode for the capacitor.

[0043] In one embodiment, the cathode of a solid electrolytic capacitor may consist primarily of manganese dioxide and is then formed by a process commonly called "manganizing." In this process, a conductive counterelectrode coating is formed over the dielectric formed by anodizing. The manganizing step is typically performed by immersing the anodized component in a solution of manganese(II) nitrate and heating the impregnated component in a humid atmosphere to convert the nitrate into solid, conductive manganese dioxide. In other words, a manganese dioxide solid electrolyte can be formed by the pyrolytic decomposition of manganese(II) nitrate (Mn(NO3)2).Such capacitors having a cathode formed of manganese dioxide can operate at high temperatures, such as up to about 250 °C, for example up to about 230 °C if the capacitor is a hermetically sealed capacitor, which will be discussed in more detail below.

[0044] In another embodiment, the solid electrolyte can also be formed from one or more conductive polymer layers. The conductive polymer includes, for example, polypyrroles, polythiophenes, poly(3,4-ethylenedioxythiophene) (PEDT), polyanilines, polyacetylenes, poly-p-phenylenes, polyphenolates, etc., and derivatives thereof. The anode part can also optionally be coated with a carbon layer (e.g., graphite) or silver layer. The silver coating can act, for example, as a solderable conductor, contact layer, and / or charge collector for the capacitor, and the carbon coating can limit the contact of the silver coating with the solid electrolyte. Such coatings can cover part or all of the solid electrolyte.

[0045] Regardless of the particular manner in which the capacitor is formed, it may be connected to end portions, as is well known in the art. For example, the anode and cathode end portions may be electrically connected to the anode terminal (e.g., a sheet or wire) and the cathode, respectively. Generally speaking, it is desirable to electrically isolate the anode terminal from the cathode terminal so that the capacitor functions in the desired manner. A variety of techniques may be used to achieve such isolation. In one embodiment, for example, any oxide and / or cathode layers formed on the terminal may be simply removed by an etching process (e.g., chemical, laser, etc.).

[0046] As previously mentioned, the solid electrolytic capacitor that can be screened using the method of the present invention includes an anode end portion to which the anode terminal of the capacitor element is electrically connected, and a cathode end portion to which the cathode of the capacitor element is electrically connected. Any conductive material, such as a conductive metal, can be used to form the end portions. The end portions can be joined using any method known in the art, such as welding, gluing, refractory metal paste, etc. Once the capacitor element is attached, the lead frames / terminals can be encased in a housing, which can then be filled with silicon oxide or another known encapsulating material. The width and length of the housing can vary depending on the intended use.Suitable packages include, for example, packages "A", "B", "C", "D", "E", "F", "G", "H", "J", "K", "L", "M", "N", "P", "R", "S", "T", "V", "W", "Y" or "X" (AVX Corporation). Regardless of the package size used, the capacitor element is embedded in such a way that at least a portion of the anode and cathode end sections remain exposed. After embedding, exposed portions of the anode and cathode end sections can be aged, screened and trimmed to the desired size.

[0047] As discussed above, the anode terminal may be in the form of a sheet or wire, etc., and may be formed from a valve metal compound such as tantalum, niobium, niobium oxide, etc. For example, the screening method of the present invention may be used to screen numerous embodiments of electrolytic capacitors, such as the one shown in the Fig. 2 and Fig.3 electrolytic capacitors, to undergo a screening. As shown in Fig. 2, the solid electrolytic capacitor can be provided with an anode terminal in the form of a sheet. As shown in Fig. 3, the solid electrolytic capacitor may further include an anode terminal in the form of a wire.

[0048] For example, in one embodiment, a sheet (e.g., sheet metal, foil, etc.) bonded to the anode body may be used for the capacitor to be screened, as shown in Fig. 2. Various examples of such capacitors are described, for example, in U.S. Patent Nos. 5,357,399 (Salisbury), 6,751,085 (Huntington), 6,643,121 (Huntington), 6,849,292 (Huntington), 6,673,389 (Huntington), 6,813,140 (Huntington), and 6,699,767 (Huntington), which are expressly incorporated herein by reference for all purposes. Fig.2, one embodiment of a solid electrolytic capacitor 200 that can be screened according to the method of the present invention is shown, which includes an anode lead 210 in sheet form. The anode lead 210 is bonded to a compressed anode body 213 made of a valve metal composition (e.g., tantalum). Although other bonding methods may be used, the anode lead 210 is bonded to the anode body 213 using an adhesive 212. The adhesive 212 may first be applied to a surface of the anode lead 210. Thereafter, the compressed anode body 213 may be placed over the adhesive 212. The anode body 213 and the anode lead 210 may then be sintered, causing the formation of a bond between the adhesive and the metal of both the anode body and the anode lead.Once attached, the anode body 213 can then be anodized and coated with a solid electrolyte, as described above. If desired, additional layers can be employed, such as a carbon layer 227 and / or one or more silver layers 221 or 222, as also discussed above. The capacitor 200 can also include an encapsulating resin on the sidewalls 224 surrounding the anode body 213. The end caps 228 and 229 are provided as the cathode and anode ends of the capacitor 200, respectively.

[0049] Another embodiment of a solid electrolytic capacitor that can be screened according to the method of the present invention is described in Fig.3. The solid electrolytic capacitor 300 may utilize an anode terminal 360 in the form of a wire embedded in the porous anode body. In such an embodiment, the anode and cathode terminals may be electrically connected to the anode wire 360 ​​and the solid electrolyte layer 354 after the formation of the capacitor element 330. The specific configuration of the terminals may vary, as is well known in the art. Fig. For example, FIG. 3 shows an embodiment that includes an anode end portion 370 and a cathode end portion 380. In this particular embodiment, the cathode end portion 380 includes a portion 382 in electrical contact with the bottom surface 339 of the capacitor element 330. To secure the capacitor element 330 to the cathode end portion 380, a conductive adhesive may be used, as is known in the art.

[0050] The anode end portion 370 includes a first section 376 positioned substantially perpendicular to a second section 374. The second section 374 includes a region 351 that supports the anode wire 360. If desired, the region 351 may have a "U-shape" to further enhance the surface contact and mechanical stability of the wire 360. The anode wire 360 ​​may then be welded to the region 351 with a laser or other suitable method. Once the capacitor element is attached to the end portions, it is enclosed within a resin housing, which may then be filled with silicon oxide or another known encapsulant. Again, Fig.For example, in Figure 3, a specific embodiment of such an encapsulation housing for a capacitor 300 is shown as element 388. The encapsulation housing 388 provides additional structural and thermal protection for the capacitor 300. After encapsulation, the exposed portions of the respective anode and cathode end portions can be aged, screened, and trimmed. If desired, the exposed portions can optionally be bent twice along the outside of the housing 388 (e.g., at an angle of approximately 90°).

[0051] Another embodiment of a solid electrolytic capacitor that can be screened according to the method of the present invention is a hermetically sealed capacitor such as that described in Fig. 27 shown capacitor. As in Fig.27, a solid electrolytic capacitor element 520 is hermetically sealed within a housing 522 to form the capacitor 500. A variety of materials may be used to form the housing, such as metals, plastics, ceramics, etc. In one embodiment, for example, the housing comprises one or more layers of a metal such as tantalum, niobium, aluminum, nickel, hafnium, titanium, copper, silver, steel (e.g., stainless steel), alloys thereof (e.g., electrically conductive oxides), composites thereof (e.g., metal coated with electrically conductive oxide), etc. In another embodiment, the housing may comprise one or more layers of a ceramic material such as aluminum nitride, alumina, silicon oxide, magnesium oxide, calcium oxide, glass, etc., as well as combinations thereof.

[0052] The housing can have any shape, such as cylindrical, D-shaped, rectangular, triangular, prismatic, etc. In Fig.For example, in Figure 27, an embodiment of a capacitor assembly 500 is shown that includes a housing 522 and a capacitor element 520. To increase volumetric efficiency, the capacitor element 520 may have a length (excluding the length of the anode terminal 560) that is relatively similar to the length of an internal cavity 526 defined by the housing 522. Furthermore, it should be understood that in Fig.27, only a single capacitor element 520 is shown, but the housing 522 may also include multiple capacitor elements 520. It should also be understood that each of the capacitor elements 520 may be separately screened according to the method of the present invention before being hermetically sealed within the housing. It should also be understood that, alternatively, the capacitor elements 520 may be hermetically sealed within the housing 522, and thereafter, the capacitor assembly 500 itself may be screened according to the method of the present disclosure.

[0053] Although by no means required, the capacitor element may be attached to the housing such that an anode end portion and a cathode end portion are formed outside the housing for subsequent integration into a circuit. The particular configuration of the end portions may depend on the intended use. In one embodiment, for example, the capacitor assembly may be formed to be surface-mountable yet mechanically robust. For example, the anode terminal may be electrically connected to external, surface-mountable anode and cathode end portions (e.g., pads, sheets, plates, frames, etc.). Such end portions may extend through the housing to terminate the capacitor. The thickness or height of the end portions is generally chosen to minimize the thickness of the capacitor assembly. If desired, the surface of the end portions may be plated with nickel, silver, gold, tin, etc., as is known in the art.Electroplated to ensure that the final part can be mounted on the circuit board. In a particular embodiment, the end parts are coated with nickel or silver protective layers, and the mounting surface is also coated with a tin solder layer. In another embodiment, thin outer metal layers (e.g., gold) are deposited on a base metal layer (e.g., copper alloy) on the end parts to further increase conductivity.

[0054] In certain embodiments, connecting elements can be used within the interior of the housing to facilitate the connection to the end parts in a mechanically stable manner. For example, if we return to Fig.27, the capacitor assembly 500 may include a connecting element 562 consisting of a first part 567 and a second part 565. The connecting element 562 may be formed of conductive materials similar to the outer end parts. The first part 567 and the second part 565 may be formed as one piece, or they may be separate parts that are connected to each other, either directly or via an additional conductive element (e.g., metal). In the embodiment shown, the second part 565 is in a plane that is substantially parallel to a lateral direction in which the terminal 560 extends (e.g., -y direction). The first part 567 is "upright" in the sense that it is in a plane that is substantially perpendicular to the lateral direction in which the terminal 560 extends.In this way, the first portion 567 can restrict the movement of the terminal 560 in the horizontal direction to enhance surface contact and mechanical stability during use. If desired, an insulating material 570 (e.g., a Teflon™ ring) can be inserted around the terminal 560.

[0055] The first portion 567 may also include a mounting region (not shown) connected to the anode terminal 560. The region may have a "U-shape" to further enhance surface contact and mechanical stability of the terminal 560. The connection of the region to the terminal 560 may be accomplished using a variety of known techniques, such as welding, laser welding, conductive adhesives, etc. In a particular embodiment, the region is attached to the anode terminal 560 by laser welding, for example. Regardless of the technique chosen, however, the first portion 567 may hold the anode terminal 560 in a substantially horizontal orientation to further enhance the dimensional stability of the capacitor assembly 500.

[0056] If we turn to Fig.27, an embodiment of the present invention is shown in which the interconnecting member 562 and the capacitor element 520 are connected to the housing 522 via anode and cathode terminations 527 and 529, respectively. The anode termination 527 includes a first portion 527a located within the housing 522 and electrically connected to the interconnecting member 562, and a second portion 527b located outside the housing 522 and providing a mounting surface 201. Similarly, the cathode termination 529 includes a first portion 529a located within the housing 522 and electrically connected to the solid electrolyte of the capacitor element 520, and a second portion 529b located outside the housing 522 and providing a mounting surface 503. It should be noted that not all of such areas need to be inside or outside the enclosure.

[0057] In the embodiment shown, a conductive trace 527c extends within a base 523 of the housing to interconnect the first portion 527a and the second portion 527b. Similarly, a conductive trace 527c extends within the base 523 of the housing to interconnect the first portion 527a and the second portion 527b. The conductive traces and / or portions of the end portions may be separate or integral. The traces may extend not only through the outer wall of the housing but may also be located at other locations, such as outside the outer wall. Of course, the present invention is by no means limited to the use of conductive traces to form the desired end portions.

[0058] Regardless of the particular configuration employed, the connection of end portions 527 and 529 to the capacitor element 520 can be accomplished using any known technique, such as welding, laser welding, conductive adhesives, etc. In a particular embodiment, for example, a conductive adhesive 531 is used to connect the second portion 565 of the connecting element 562 to the anode end portion 527. Similarly, a conductive adhesive 533 is used to connect the cathode of the capacitor element 520 to the cathode end portion 529. The conductive adhesives can consist of conductive metal particles contained in a resin composition.

[0059] The metal particles may be silver, copper, gold, platinum, nickel, zinc, bismuth, etc. The resin composition may include a thermosetting resin (e.g., epoxy resin), a curing agent (e.g., acid anhydride), and a coupling agent (e.g., silane coupling agent). Suitable conductive adhesives are described in U.S. Patent Application Publication No. 2006 / 0038304 (Osako et al.), which is expressly incorporated herein by reference for all purposes.

[0060] Optionally, a polymeric restraint may also be in contact with one or more surfaces of the capacitor element, such as the back surface, front surface, top surface, bottom surface, side surfaces, or any combination thereof. The polymeric restraint may reduce the likelihood of the capacitor element delaminating from the housing. In this regard, the polymeric restraint may have a certain degree of strength that allows it to hold the capacitor element in a relatively fixed position even when subjected to vibrational forces, but is not so strong that it cracks. For example, the restraint may have a tensile strength of about 1 to about 150 megapascals ("MPa"), in some embodiments about 2 to about 100 MPa, in some embodiments about 10 to about 80 MPa, and in some embodiments about 20 to about 70 MPa, measured at a temperature of about 25°C.It is usually desirable that the clamping is not electrically conductive.

[0061] While a variety of materials that exhibit the desired strength properties mentioned above may be employed, curable thermosetting resins have been found to be particularly well-suited for use in the present invention. Examples of such resins include epoxies, polyimides, melamine resins, urea-formaldehyde resins, polyurethanes, silicone polymers, phenolic resins, etc. In certain embodiments, one or more polyorganosiloxanes, for example, may be employed in the restraint. Epoxy resins are also particularly well-suited for use in the present invention. Still other suitable conductive adhesive resins are also described in U.S. Patent Application Publication No. 2006 / 0038304 (Osako et al.) and U.S. Patent No. 7,554,793 (Chacko), which are expressly incorporated herein by reference for all purposes.

[0062] If desired, curing agents may also be employed in the polymeric consumable to help promote curing. The curing agents typically comprise about 0.1 to about 20 wt.% of the consumable. Other additives may also be employed, such as photoinitiators, viscosity modifiers, suspension aids, pigments, stress relaxers, coupling agents (e.g., silane coupling agents), non-conductive fillers (e.g., clay, silica, alumina, etc.), stabilizers, etc. When employed, such additives typically comprise about 0.1 to about 20 wt.% of the total composition.

[0063] For example, if we return to Fig. 27, an embodiment is shown in which a single polymeric clamp 597 is in contact with a top surface 581 and a back surface 577 of the capacitor element 520. While in Fig.27, it should be understood that separate restraints may be employed to perform the same function. Indeed, more generally, any number of polymeric restraints may be employed and in contact with any desired surface of the capacitor element. When multiple restraints are employed, they may be in contact with each other or remain physically separate. For example, in one embodiment, a second polymeric restraint (not shown) may be employed that contacts the top surface 581 and the front surface 579 of the capacitor element 520. The first polymeric restraint 597 and the second polymeric restraint (not shown) may or may not be in contact with each other.In yet another embodiment, a polymeric restraint may also contact a bottom surface 583 and / or one or more side surfaces of the capacitor element 520, either in conjunction with or in place of other surfaces.

[0064] Regardless of how it is applied, it is typically desirable for the polymeric clamp to also be in contact with at least one surface of the housing to help further mechanically stabilize the capacitor element against potential delamination. For example, the clamp may be in contact with an inner surface of one or more side walls, the outer wall, the lid, etc. Fig.For example, in Figure 27, the polymeric constrain 597 is in contact with the interior surfaces 507 and 509 of the housing 522. While in contact with the housing, it is still desirable for at least a portion of the interior space defined by the housing to remain clear to allow the inert gas to flow through the interior space and limit contact of the solid electrolyte with oxygen. For example, typically at least about 5% of the interior space volume remains clear of the capacitor element and the polymeric constrain, and in some embodiments, from about 10% to about 50% of the interior space volume.

[0065] Once connected in the desired manner, the resulting package is hermetically sealed. For example, Fig.27, the housing 522 includes a base 523 and a lid 525, between which the interior space 526 is formed. The lid 525 and the base 523 can be made of ceramic, metal (e.g., iron, copper, nickel, cobalt, etc., as well as alloys thereof), plastic, etc. In one embodiment, for example, the base 523 is formed of a ceramic material, and the lid 525 is formed of a metal material. The lid 525 includes an outer wall 521 that is integrally formed with at least one side wall 524. For example, in the embodiment shown, two opposing side walls 524 are shown in cross-section. The height of the side walls 524 is generally such that the lid 525 does not come into contact with any surface of the capacitor element 520, so that it is not contaminated.The outer wall 521 and the base 523 both extend in a lateral direction (-y direction) and are substantially parallel to each other and to the lateral direction of the anode terminal 560. The side wall 524 extends from the outer wall 521 in a longitudinal direction that is substantially perpendicular to the base 523. A distal end 506 of the lid 525 is defined by the outer wall 521, and a proximal end 501 is defined by a lip 553 of the side wall 524.

[0066] In particular, the lip 553 extends from the sidewall 524 in a lateral direction that may be substantially parallel to the lateral direction of the base 523. The lip 553 also defines a peripheral edge 551 that may be substantially perpendicular to the lateral direction in which the lip 553 and the base 523 extend. The peripheral edge 551 is located beyond the outer perimeter of the sidewall 524 and may be substantially coplanar with an edge 571 of the base 523. The lip 553 may be sealed to the base 523 using any known technique, such as welding (e.g., resistance or laser welding), soldering, glue, etc. For example, in the embodiment shown, a sealing element 587 (e.g., glass-to-metal seal, Kovar ®-ring, etc.) are used between the components to facilitate their attachment. Regardless, the use of a lip as described above can provide a more stable connection between the components and improve the sealing and mechanical stability of the capacitor assembly.

[0067] Hermetic sealing typically occurs in the presence of a gas atmosphere containing at least one inert gas to inhibit oxidation of the solid electrolyte during use. The inert gas may include, for example, nitrogen, helium, argon, xenon, neon, krypton, radon, etc., as well as mixtures thereof. Typically, inert gases make up the majority of the atmosphere within the housing, such as about 50 wt.% to 100 wt.%, in some embodiments about 75 wt.% to 100 wt.%, and in some embodiments about 90 wt.% to about 99 wt.% of the atmosphere. If desired, a relatively small amount of non-inert gases may also be employed, such as carbon dioxide, oxygen, water vapor, etc. In such cases, however, the non-inert gases typically make up 15 wt.% or less, in some embodiments 10 wt.% or less, in some embodiments about 5 wt.% or less, in some embodiments about 1 wt.%-% or less, and in some embodiments, about 0.01 wt.% to about 1 wt.% of the atmosphere within the housing. For example, the moisture content (expressed as relative humidity) may be about 10% or less, in some embodiments about 5% or less, in some embodiments about 1% or less, and in some embodiments about 0.01 to about 5%.

[0068] As discussed above, the screening method of the present invention can also be used to screen liquid electrolytic capacitors. A liquid electrolytic capacitor generally comprises a porous anode body containing a dielectric layer, a cathode containing a metal substrate coated with an electrochemically active coating (e.g., a conductive polymer), and an aqueous electrolyte. The ionic conductivity of the electrolyte is selectively adjusted within a specific range so that the capacitor can be charged to a high voltage. The physical arrangement of the anode, cathode, and working electrolyte of a liquid electrolytic capacitor screened using the method of the present invention can generally vary, as is well known in the art. Fig.For example, FIG. 4 shows an embodiment of a liquid electrolytic capacitor 400 that includes a working electrolyte 440 located between an anode 450 and a cathode 430. The anode 450 includes a dielectric film 460 and is embedded with a terminal 420 (e.g., tantalum wire). The cathode 430 may be formed from a cathode substrate 410 and an electrochemically active material 490. Although not shown, a separator may be located between the cathode 430 and the anode 450 to prevent direct contact between the anode and cathode but allow the flow of an ionic current of the working electrolyte 440 to the electrodes. A seal 470 (e.g., glass-to-metal) that connects and seals the anode 450 to the cathode 430 may also be employed.Although not shown, capacitor 400 may also include a spacer (not shown) that holds anode 450 within cathode 430. The spacer may be made of plastic and be disc-shaped, for example.

[0069] The present invention will be better understood by reference to the following examples, which relate to the Fig. 5-25 and demonstrate the effectiveness and reliability of the electrolytic capacitor screening method of the present invention. Examples 1-3 and 10 use Fig.5 as a starting point, in which the first leakage currents are plotted for all tested capacitors in 10 batches, separated into zones. "Zone 1" includes parts that have a first leakage current within three standard deviations of the first average leakage current, which in this case is measured at a temperature of 125 °C and at a voltage equal to two-thirds of the nominal voltage. "Zone 1 at limit" includes parts that have a first leakage current within three standard deviations of the first average leakage current, but close to the limit. "Zone 2" includes parts that have a first leakage current above three standard deviations of the first average leakage current. However, the capacitors in "Zone 2" have a first leakage current that is less than the hard-cut limit of 0.001*C*V. R *12. Note that the hard-cut limit is calculated using the following equation: DCL−Limit=0.001*Capacitance(C)*Rated Voltage(VR)*Temperature Factor(TF), where TF is 12 for 125 °C. Thus, the constant used in the calculation of the Fig. 5 calculated leakage current limit is 0.001, with the product of the capacity and the rated voltage (V R ) and is further multiplied by a factor of 12 to account for an elevated temperature of 125 °C. Note that the constant used in commercial or military applications may be 0.01, while the constant used in medical or aerospace applications may be 0.001. In Fig.5, the resulting hard-cut limit based on the variables described above or the product design capability is 0.225 µA. Meanwhile, "Zone 3" includes capacitors with a first leakage current that is both above three standard deviations of the first mean leakage current and greater than the hard-cut limit of 0.001*C*V R *12 is. Example 1Life test at 125 °C and 2 / 3 of the rated voltage

[0070] We refer to the Fig. 6-9. 100 capacitors per batch from 10 batches, which are Fig.5, which fall into "Zone 1," were subjected to a burn-in at 125°C for 42 hours at 1.0 times the rated voltage of the capacitors. The parts were then mounted on FR-4 boards and subjected to a 1000-hour endurance test at 125°C and 2 / 3 of the rated voltage. The leakage current (DCL) was then determined at a temperature of 25°C and at the rated voltage for the parts after the 1000-hour test.

[0071] How to Fig. 6, the “Zone 1” capacitors of Fig. 5 overall a slightly higher leakage current in Fig. 6 after assembly compared with Fig. 5, which, as discussed above, represents the determination of the first leakage current. However, the capacitors still have a leakage current that is below the predetermined hard-cut / failure limit of 0.225 µA for the capacitors. Also, as can be seen in Fig. 7, the “Zone 1” capacitors of Fig.5 at the end of the 1000-hour life test at 125 °C and 2 / 3 of the rated voltage a slightly higher leakage current in Fig. 7 compared to Fig. 5 and Fig. 6. However, the capacitors exhibit a leakage current that is still below the capacitor failure limit of 0.225 µA.

[0072] Meanwhile, Fig.Figure 8 shows the leakage current measurements for the "Zone 1" capacitors as determined at various times during the life tests. One capacitor failed because it exhibited a leakage current above the 0.225 µA limit throughout the test, as indicated by the dashed oval, while three other capacitors exhibited leakage currents above the 0.225 µA limit in the 24-hour DCL leakage current measurement but stabilized during further life testing up to 1000 hours. Note that these failed capacitors would have been removed using the screening method of the present invention, but testing continued with the failed capacitors to see how they would perform in the long-term test. As shown, the leakage current for some of the capacitors was above the threshold limit during the test but then fell below the threshold limit at the end of the test.Therefore, not following the screening procedure of the present invention and looking only at the endpoint life test data could result in the acceptance of capacitors with unstable characteristics.

[0073] Then show Fig. 9 the change in leakage current (DCL) for each of the “Zone 1” parts subjected to a 1000-hour endurance test at 125 °C and 2 / 3 of the rated voltage, from Fig. 7 compared to the leakage current after mounting of Fig. 6. As shown, the graph of the DCL shift observed during the 1000-hour test at 125 °C is represented by a well-behaved population with a negligible mean DCL shift. Example 2Life test at 85 °C and nominal voltage

[0074] We refer to the Fig. 10-13. 10 capacitors per batch from 10 batches, which in Fig.5, falling into "Zone 1," were subjected to a burn-in at 125°C for 42 hours at 1.0 times the rated voltage of the capacitors. The parts were then mounted on FR-4 boards and subjected to a 2000-hour endurance test at 85°C and the rated voltage. The DCL was then determined for the parts at various stages of the 2000-hour endurance test. As can be seen from the 10 graphs representative of the 10 batches, all capacitors were below the DCL threshold (shown by the thick line) after 2000 hours of endurance testing, indicating that the iterative screening method of the appended claims effectively removes unstable capacitors from the tested batches. Note that the threshold is calculated using the following equation: DCL limit = 0.0025 * capacity (C) * rated voltage (VR) * temperature factor (TF), where TF is 1 for 25°C, 10 for 85°C and 12 for 125°C. Life test at 85 °C and rated voltage, non-standard population

[0075] Next, individual capacitors with borderline or anomalous behavior during the burn-in at 125 °C were recorded and categorized as “Zone-1-at-Limit” (ie Fig. 14), “Zone 2 to Zone 1 changer” (ie Fig. 15) or “Zone 2” (ie Fig. 16) and subjected to a life test at 85 °C and nominal voltage.

[0076] Fig. Figure 14 shows a graph that traces the leakage currents of the capacitors with the ten highest initial leakage current measurements obtained during the leakage current measurement of the first iteration of Fig.5 still fell within the three standard deviation cutoff limit (i.e., "Zone 1"). Although these 10 capacitors exhibit initial leakage current measurements close to the three standard deviation cutoff limit, the capacitors remained stable throughout the lifetime test, indicating the relative effectiveness of the three standard deviation leakage current cutoff limit used in the iterative screening procedure of the appended claims.

[0077] Fig.Figure 15 shows a graph tracing the leakage current behavior of capacitors that failed the initial leakage current screening prior to burn-in because they exhibited leakage currents above the three standard deviation cutoff limit, and were thus initially categorized as "Zone 2" parts. However, during burn-in, these "Zone 2" parts exhibited self-healing, such that their second-iteration leakage current was reduced to within the three standard deviation cutoff limit of the second leakage current (post-burn-in). While these capacitors pass endurance tests by being within the three standard deviation limit after 2000 hours, the iterative screening method of the appended claims rejects these capacitors. Although the capacitors would technically pass the endurance test, instability during the endurance test warrants the removal of such capacitors from their respective batches.

[0078] Then Fig. 16 a graph showing the leakage current behavior of capacitors that exceeded the first leakage current threshold limit of three standard deviations but were within the hard-cut leakage current limit, which is 0.001*C*V R *12. Although the majority of capacitors appear to be stable during the 85 °C life test, it is likely that this population contains unstable capacitors, as demonstrated by the three failed units represented by the three dashed lines.

[0079] As can be seen from Examples 1-3, the iterative screening method of the present invention is highly reliable in that a life test with a sample size of 1000 screened capacitors at 125 °C at 2 / 3 of the rated voltage for 1000 hours and at 85 °C at the full rated voltage for 2000 hours up to a qualification leakage current limit of 0.005*C*V R , which is half of the current military standard requirements for life tests, which are 0.01*C*V R results in zero failures. Example 4Comparison of the DCL before and after burn-in

[0080] Furthermore, the leakage current of capacitors was determined at 125 °C and 2 / 3 of the operating voltage after 15 seconds of soaking before burn-in, using the method of the present invention and the Weibull method for comparison. In the method of the present invention, burn-in was performed at 125 °C, while the Weibull method was performed at 85 °C. As described in Fig. 17, the leakage current before burn-in was generally higher than both leakage currents after burn-in and ranged from about 0.7 µA to about 1.2 µA. Meanwhile, the leakage current after burn-in in the Weibull method ranged from about 0.6 µA to about 1.3 µA, and the leakage current after burn-in in the method of the present invention ranged from about 0.4 µA to about 1.0 µA. While Fig.17 demonstrates the parameter shift in the DC leakage current resulting from the burn-in process of the present application, where the total DC leakage current is significantly lower, Fig. 17 also shows that the DC leakage current of potentially damaged parts was increased, such as the parts having a leakage current of about 0.8 µA to about 1.0 µA, which can improve the effectiveness of the statistical screening method of the present invention. Example 5: Lifetime test at 85 °C with capacitors that passed or failed the initial DCL screening

[0081] Next, the leakage current shift after a lifetime test at 85°C was compared for two batches of capacitors from one lot, with one batch passing the initial leakage current screening (before burn-in) and one batch failing the initial leakage current screening (before burn-in). The lifetime test was conducted for 2000 hours at 85°C, and then the leakage current shift was determined for each batch of capacitors. The results are shown in Fig. 18, where it is indicated that capacitors that would have been removed in the process of the present application during the first leakage current screening (before burn-in) show a significant shift in leakage current after the lifetime test, while the capacitors that passed the first leakage current screening (before burn-in) and were accepted for later screening showed only a small shift in leakage current after the lifetime test.

[0082] Fig. Figure 18 thus shows that the method of the present application can eliminate inhomogeneous defects in the tested capacitors prior to the burn-in process through a screening step based on leakage current, thereby removing potentially unstable capacitors. Without this screening step, these potentially unstable capacitors can slip into the DCL distribution representing good capacitors after the burn-in during screening due to the healing process induced during the burn-in. However, after the lifetime test, these potentially unstable capacitors can be eliminated, as shown in Fig.18, show a significant shift in leakage current, indicating that these capacitors have defects uncharacteristic of the rest of the capacitors in the batch. On the other hand, the use of statistical screening prior to burn-in at 125°C, as described in the present application, reduces or eliminates the possibility of this small amount of potentially parametrically unstable capacitors making it into the released batch. Example 6 Comparison of leakage current determination at 25 °C and at 125 °C

[0083] As discussed above, the screening method of the present application determines the leakage current of the screened capacitors at elevated temperatures. Fig.Figure 19 shows how determining leakage current at elevated temperatures increases the ability to detect individual leakage current variations in capacitors within a batch of capacitors that would normally go undetected when tested at 25 °C (room temperature). As shown, the capacitors tested at 25 °C show no parameter shift in leakage current, while a portion of the capacitors tested at 125 °C show a parameter shift in leakage current, with the leakage current of these capacitors ranging from approximately 2 µA to approximately 10 µA. If the temperature at which leakage current was measured had not been increased, these outlier capacitors would not have been detected and would have been screened out of the batch, meaning that potentially unstable capacitors would have passed the screening. Example 7Proven DCL improvement with burn-in at elevated temperature

[0084] The Fig. 20 and Fig.21 show the improvement in the total leakage current when the burn-in temperature is increased compared to the conventional burn-in at 85 °C associated with the Weibull method. For example, Fig. Figure 20 shows the leakage current before burn-in for a batch of capacitors compared to the leakage current after the burn-in procedure at 125°C described in the present application and the burn-in procedure at 85°C associated with the Weibull method. The leakage current of capacitors subjected to burn-in at the elevated temperature of 125°C is generally lower than that of capacitors subjected to burn-in using the Weibull method, but at the same time, any outliers that may be present can be more easily detected. Fig. 21 that the reduced DCL obtained by the burn-in process at 125 °C described in the present application is repeatable over multiple batches. Example 8Life test at 125 °C and 2 / 3 of the working voltage for 1000 hours

[0085] Next, compare Fig. 22 compares the leakage current of capacitors that underwent the initial leakage current screening and the burn-in procedure at 125°C as described in the present application with the leakage current of capacitors that underwent a burn-in procedure at 85°C according to the Weibull method after 1000 hours of endurance testing at 125°C and 2 / 3 of the working voltage, with the specified hard-cut leakage current limit set at 0.225 µA. The leakage current was determined at 25°C at the working voltage of the capacitors after an exposure time of 30 seconds. Ten capacitors from 10 batches screened according to the procedure described in the present application were tested together with 170 capacitors that had undergone a burn-in according to the conventional Weibull method at 85°C.

[0086] As shown, two capacitors subjected to Weibull burn-in failed the endurance test because their leakage currents exceeded the predetermined limit of 0.225 µA. Specifically, the two failed capacitors exhibited post-burn-in leakage currents of approximately 0.4 µA and 0.5 µA, respectively. On the other hand, none of the capacitors subjected to burn-in using the method of the present application failed because none of the capacitors exhibited a leakage current above the 0.225 µA limit. Example 9Effect of the screening procedure on the leakage current

[0087] Furthermore, the leakage currents of 10 capacitors, sampled from 10 batches of capacitors subjected to the screening method of the present application, after the life test at 85 °C and the rated voltage for 2000 hours were compared with the corresponding leakage currents before the life test. As in Fig. As shown in Figure 23, after 2000 hours of life test, the shift of the leakage current before and after the life test was negligible. Example 10: Endurance test with Zone 1, Zone 1 at the limit, and Zone 2 parts at 85 °C for 2000 hours

[0088] In another example, the leakage current of capacitors in 10 batches was determined after a burn-in at 125 °C. The capacitors were then grouped into “Zone 1” capacitors, “Zone 1 at limit” capacitors, and “Zone 2” capacitors, as described above using Fig.5. "Zone 1" included capacitors with a leakage current within three standard deviations of the mean leakage current, which in this case is measured at a temperature of 125 °C and at a voltage of 2 / 3 of the rated voltage. "Zone 1 at limit" included capacitors with a leakage current within three standard deviations of the mean leakage current, but which also had leakage currents close to the three standard deviation limit (i.e., the capacitors with the 10 highest leakage currents after burn-in were within the three standard deviation limit). "Zone 2" included capacitors with a leakage current above three standard deviations of the mean leakage current, but which was also less than the hard-cut limit of 0.225 µA.After grouping the capacitors into the appropriate zones, the capacitors were subjected to a 2000-hour endurance test at 85°C. The leakage current was then measured for each capacitor in each zone at 25°C.

[0089] As in Fig.As shown in Figure 24, three capacitors grouped in "Zone 2" that were within the hard-cut leakage current limit of 0.225 µA before the life test failed the life test, exhibiting leakage currents of approximately 0.75 µA, 1.5 µA, and 2.75 µA. Meanwhile, all "Zone 1" parts and "Zone 1 at limit" parts exhibited leakage currents below the 0.225 µA limit after the life test. This suggests that conventional hard-cut limits do not effectively eliminate parts that later experience reliability problems, such as the capacitors that exhibited leakage currents initially within the hard-cut limit but exhibited leakage currents above the hard-cut limit after the life test.

[0090] Fig. 25 is an exploded view of Fig.24 and shows that the leakage current of the “Zone 2” parts is increased compared to the “Zone 1” and “Zone 1 at limit” parts, indicating possible reliability problems of the “Zone 2” parts. Example 11Calculating the failure rate for a confidence level of 60%

[0091] In Example 11, Table 1 below shows the initial values ​​and results for a failure rate calculation using the formulas discussed above. In Example 11, a confidence level of 60% was chosen, and 30 capacitors with a nominal voltage of 10 volts were tested for 6 hours at a temperature of 125°C at 2 / 3 of the nominal voltage, or 6.6 volts. Then, the predicted failure rate was calculated assuming that the customer would use the capacitors at 25°C and a voltage of 5 volts. As shown in Table 1, testing the capacitors at 125°C for 6 hours at 6.6 volts was equivalent to approximately 2,000,000 hours at 25°C and 5 volts and resulted in a predicted failure rate of approximately 0.0029% failures per 1,000 hours. Example 11: Calculating the failure rate for the iterative screening procedure Initial values (10 V part) Results Nominal voltage (V) 10 Component hours 2 (Equivalent at application temperature) 1 978 593 Number of capacitors tested 30 Component years (equivalent at application temp.) 225,71 Test duration (hours) 6 Test acceleration factor 3 (Temperature) 38234,21 Test temperature (°C) 125 Test acceleration factor 4 (Tension) 0,287496 Test voltage (V) 6,6 Application acceleration factor 5 (Tension) 0,1250 Number of failures 0 Confidence level (%) 60 Failure rate 6 (Percentage of failures per 1000 hours) 0,002894 Application temperature (°C) 25 MTBF (mean time between failures) (hours) 34 549 607 Application voltage (V) 5 Activation energy 1 a tantalum lid (eV) 1,08 Assumptions and factors in the formulas: 1 The activation energy is adjustable. The equivalent component hours are based on the MIL-HDBK-217 model for solid tantalum capacitors, and the total component hours tested at the test temperature are multiplied by the test temperature acceleration factor and the test voltage acceleration factor to obtain the equivalent component hours used in calculating the failure rate. 3 The test temperature acceleration factor is based on the Arrhenius model; temperatures are in Kelvin, and the Boltzmann constant is 8.63 × 10 -5 eV / K. 4 The test voltage acceleration factor is the test voltage divided by the nominal voltage, to the power of three 5 The application voltage acceleration factor is the application voltage divided by the rated voltage, to the power of three 6Failure rate predictions are based on a chi-square distribution; the degrees of freedom when using the chi-square distribution are the number of failures plus 1 multiplied by 2; and the calculated failure rate is multiplied by the application stress acceleration factor to obtain the final failure rate. Table 1Example 12Calculation of the failure rate for a confidence level of 90%

[0092] In Example 12, Table 2 below shows the initial values ​​and results for a failure rate calculation using the formulas discussed above. In Example 12, a confidence level of 90% was chosen, and 30 capacitors with a nominal voltage of 10 volts were tested for 6 hours at a temperature of 125°C at 2 / 3 of the nominal voltage, or 6.6 volts. Then, the predicted failure rate was calculated assuming that the customer would use the capacitors at 25°C and a voltage of 5 volts. As shown in Table 1, testing the capacitors at 125°C for 6 hours at 6.6 volts was equivalent to approximately 2,000,000 hours at 25°C and 5 volts and resulted in a predicted failure rate of approximately 0.0072% failures per 1,000 hours. Example 12: Calculating the failure rate for the iterative screening procedure Initial values (10 V part) Results Nominal voltage (V) 10 Component hours 2 (Equivalent at application temperature) 1 978 593 Number of capacitors tested 30 Component years (equivalent at application temp.) 225,71 Test duration (hours) 6 Test acceleration factor 3 (Temperature) 38 234,21 Test temperature (°C) 125 Test acceleration factor 4 (Tension) 0,287496 Test voltage (V) 6,6 Application acceleration factor 5 (Tension) 0,1250 Number of failures 0 Confidence level (%) 90 Failure rate 6 (Percentage of failures per 1000 hours) 0,007273 Application temperature (°C) 25 MTBF (mean time between failures) (hours) 13 748 671 Application voltage (V) 5 Activation energy 1 a tantalum lid (eV) 1,08

[0093] Assumptions and factors in the formulas: 1 The activation energy is adjustable. 2The equivalent component hours are based on the MIL-HDBK-217 model for solid tantalum capacitors, and the total component hours tested at the test temperature are multiplied by the test temperature acceleration factor and the test voltage acceleration factor to obtain the equivalent component hours used in calculating the failure rate. 3 The test temperature acceleration factor is based on the Arrhenius model; temperatures are in Kelvin, and the Boltzmann constant is 8.63 × 10 -5 eV / K. 4 The test voltage acceleration factor is the test voltage divided by the nominal voltage, to the power of three 5 The application voltage acceleration factor is the application voltage divided by the rated voltage, to the power of three 6Failure rate predictions are based on a chi-square distribution; the degrees of freedom when using the chi-square distribution are the number of failures plus 1 multiplied by 2; and the calculated failure rate is multiplied by the application stress acceleration factor to obtain the final failure rate. Table 2

[0094] These and other modifications and variations of the present invention may be practiced by those skilled in the art without departing from the spirit and scope of the present invention. Furthermore, it should be understood that aspects of the various embodiments may be substituted, in whole or in part, for one another. Furthermore, those skilled in the art will appreciate that the above description is merely exemplary and is not intended to limit the invention, which is particularly described in the appended claims.

Claims

[1] A method for iteratively screening a batch of electrolytic capacitors having a predetermined voltage rating, the method comprising: measuring a first leakage current for a first quantity of capacitors in the batch and calculating a first average leakage current from the results; removing capacitors having a measured first leakage current greater than or equal to a first predetermined value from the first set, thereby creating a second set of capacitors, the first predetermined value being one or more standard deviations above the first average leakage current; performing a burn-in treatment on the second set of capacitors, the burn-in treatment comprising applying a predetermined test voltage that is about 0.8 to about 1.2 times the rated voltage to the capacitors; after the burn-in treatment, measuring a second leakage current in the second set of capacitors, calculating a second average leakage current therefrom; and Removing capacitors having a measured second leakage current greater than or equal to a second predetermined value from the second set, thereby creating a third set of capacitors, the second predetermined value being one or more standard deviations above the second mean leakage current. [2] The method of claim 1, wherein the first leakage current is determined at a temperature in a range of about 20°C to about 150°C. [3] The method of claim 1, wherein the first predetermined value is three or more standard deviations above the first mean leakage current. [4] The method of claim 1, wherein the burn-in heat treatment is performed for a period of time ranging from about 25 hours to about 75 hours. [5] The method of claim 1, wherein the predetermined test voltage is in the range of about 0.9 to about 1.1 times the predetermined nominal voltage. [6] The method of claim 1, wherein the burn-in heat treatment is carried out at a temperature ranging from about 100°C to about 150°C. [7] The method of claim 1, wherein the second leakage current is determined at a temperature ranging from about 20°C to about 150°C. [8] The method of claim 1, wherein the second predetermined value is three or more standard deviations above the second mean leakage current. [9] The method of claim 1, further comprising soldering the first set, the second set and / or the third set of capacitors by a reflow process. [10] The method of claim 9, wherein the reflow process takes place in a convection oven at a peak temperature profile in the range of about 200°C to about 280°C. [11] The method of claim 9, wherein the second set of capacitors is soldered by a reflow process, the soldering occurring between the burn-in heat treatment and the second leakage current treatment. [12] The method of claim 9, wherein the third set of capacitors is soldered by a reflow process. [13] The method of claim 1, further comprising the steps of: Measuring a third leakage current at the third set of capacitors and calculating a third average leakage current therefrom; Removing capacitors having a measured third leakage current greater than or equal to a third predetermined value from the third set, wherein the third predetermined value is one or more standard deviations above the third mean leakage current. [14] The method of claim 13, wherein the third leakage current is determined at a temperature in a range of about 15°C to about 35°C. [15] The method of claim 13, wherein the third predetermined value is three or more standard deviations above the third mean leakage current. [16] The method according to claim 1, wherein the capacitors are solid electrolytic capacitors. [17] The method according to claim 16, wherein the solid electrolytic capacitors are hermetically sealed. [18] The method according to claim 1, wherein the capacitors are liquid electrolytic capacitors. [19] A method according to claim 1, wherein the capacitors comprise an anode containing tantalum or niobium oxide. [20] The method of claim 1, wherein the first predetermined value is less than the product of 0.01 times the capacitance times the rated voltage of the capacitors. [21] The method of claim 1, wherein the first quantity comprises all capacitors in the batch. [22] The method of claim 1, wherein multiple batches of capacitors are subjected to iterative screening. [23] The method of claim 22, further comprising using a mean leakage current for each of the plurality of batches of capacitors to calculate a large mean leakage current therefrom, wherein the mean leakage current for each of the plurality of batches of capacitors is the mean leakage current for the second set, the third set, or any subsequent set of iteratively screened capacitors in the batch; and removing all batches having a mean leakage current greater than or equal to a fourth predetermined value, the fourth predetermined value being one or more standard deviations above the large mean leakage current. [24] A method according to claim 23, wherein the fourth predetermined value is three or more standard deviations above the large mean leakage current. [25] The method of claim 1, further comprising determining a predicted failure rate for the batch of capacitors, wherein capacitors removed from the first group are excluded from the calculation for determining the predicted failure rate. [26] A method according to claim 25, wherein the calculation of the predicted failure rate uses a voltage acceleration factor based on the voltage applied to the batch of capacitors during a lifetime test procedure and a temperature acceleration factor based on the temperature at which the lifetime test procedure takes place. [27] The method of claim 25, wherein the predicted failure rate is in the range of about 0.00001% failures per 1000 hours to about 0.008% failures per 1000 hours when determined with a confidence level of about 99.9%. [28] A method for supplying a batch of capacitors to a customer, the method comprising: determining a nominal voltage for the capacitors; the iterative screening of the capacitors, whereby at each iteration capacitors whose leakage current is above a predetermined value are removed from the batch; wherein the predetermined value is one or more standard deviations above the mean leakage current measured during the iteration; and Delivering the batch of capacitors to the customer without derating the rated voltage. [29] The method of claim 28, further comprising calculating a large mean leakage current for a plurality of batches of capacitors, wherein the batch to be delivered to the customer is included in the calculation of the large mean leakage current, and verifying that the mean leakage current for the batch is within one or more standard deviations of the large mean leakage current. [30] The method of claim 29, wherein the mean leakage current for the batch is within three standard deviations of the large mean leakage current. [31] The method of claim 28, further comprising informing the customer of a predicted failure rate for the batch of capacitors, wherein capacitors removed from the batch during a first iteration screening prior to a burn-in heat treatment are excluded from the calculation for determining the predicted failure rate. [32] A method according to claim 31, wherein the calculation of the predicted failure rate uses a voltage acceleration factor based on the voltage applied to the batch of capacitors during a life test and a temperature acceleration factor based on the temperature at which the life test takes place. [33] The method of claim 31, wherein the predicted failure rate is in the range of about 0.00001% failures per 1000 hours to about 0.008% failures per 1000 hours when determined with a confidence level of about 90%. [34] A method for calculating a predicted failure rate for electrolytic capacitors, the method comprising: performing a burn-in treatment on the capacitors at a first temperature and a first voltage for a first period of time; performing a life test on the capacitors at a second temperature and a second voltage for a second period of time; and determining the number of capacitors that failed after the life test based on the number of capacitors that have a leakage current above a predetermined value, whereby capacitors that failed before the burn-in treatment are excluded from the calculation to determine the predicted failure rate. [35] A method according to claim 34, wherein the calculation of the predicted failure rate further comprises using a voltage acceleration factor based on the voltage applied to the batch of capacitors during a life test and a temperature acceleration factor based on the temperature at which the life test takes place. [36] The method of claim 34, wherein the predicted failure rate is in the range of about 0.00001% failures per 1000 hours to about 0.008% failures per 1000 hours when determined with a confidence level of about 90%.

Images