- Email: [email protected]
Reliability evaluation of IGCT from accelerated testing, quality monitoring and field return analysis
Microelectronics Reliability 88–90 (2018) 510–513
Contents lists available at ScienceDirect
Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel
Reliability evaluation of IGCT from accelerated testing, quality monitoring and field return analysis
T
⁎
Th. Stiasny , O. Quittard, Ch. Waltisberg, U. Meier ABB Switzerland Ltd, Semiconductors, Lenzburg, Switzerland
A R T I C LE I N FO
A B S T R A C T
Keywords: Power semiconductor IGCT Integrated gate commutated thyristor
The IGCT (Integrated Gate Commutated Thyristor) was launched as a new power semiconductor device 20 years ago, using design features where only few reference was available at that time. Especially the gate path that needs to take the full principal current for micro seconds was significant extension of the state of the art. Inbetween 20 years of field experience with IGCTs is available. In this paper test results from accelerated testing, quality monitoring results over many years, reliability date from evaluated field failure rates and analysis of devices that were in service for 15 years under well-known load are summarized and show an excellent reliability of the power semiconductor.
1. Introduction 20 years ago the first high power applications using IGCT (Integrated Gate Commutated Thyristor) were launched [1]. At that time, the IGCT was a new high power semiconductor device concept combining the ruggedness of press-pack devices, the low on-state losses of thyristors and the turn-off ruggedness of transistors. For this device, a powerful gate-unit and a low impedant gate-cathode path is required. After launching the device, concerns were raised on newly introduced features of the device, especially the long term reliability of the gate circuit path exceeding the experience of precious power semiconductor device generations. In-between, more than 250,000 IGCTs are in service in motor drives, interties, STATCOM, breakers and other demanding applications giving a base for a deep reliability analysis. 2. IGCT device design with respect to reliability The device principle is sketched in Fig. 1, only considering the demanding turn-off circuit of the gate unit. The turn-on of the devices is principally equivalent to the turn-on of a thyristor or GTO. Due to the distributed gate structure (see Fig. 2) a significantly higher turn-on pulse can easily be applied to the power semiconductor leading to a fast latching of the thyristor and with this to high dI/dt capability. In conducting state, the IGCT operates as a latched thyristor leading to a very low on-state voltage drop. For initiating the turn-off of the IGCT, the gate-cathode path is biased in reverse direction. This is reached by closing the switch S in the turn-off circuit of the gate-cathode path to a
⁎
voltage source (see Fig. 1). In the IGCT, the S is represented by tens of MOSFETs operated in parallel, the voltage source by tens of electrolytic capacitors in parallel (see Fig. 2). The commutation of the device principal current from the cathode path to the gate path is taking place during the time tcom ~IG·Lσ/UGU (assuming for simplicity a pure inductive gate cathode impedance). This commutation needs to be finished before anode-cathode voltage is built up across the device (hard drive limit) and is typically well below 1 μs. By this commutation, the active part of the device is transformed into an open base pnp-transistor. The turn-off is then possible without a dV/dt limiting circuit (e.g. snubber circuit that is needed for a GTO). Therefore, the gate-cathode impedance is the decisive parameter to ensure a reliable turn-off. Degradation effects of this parameter would have a negative influence on the long term reliability of the IGCTs. The low impedant gate-cathode path during the blocking phase also ensues the high dV/dt capability of the device compared to standard thyristors. The power device is packaged in a hermetic ceramic press pack housing, ensuring the protection of the sensitive structures of the power semiconductor from environmental influences. Therefore, issues like influence of humidity on the blocking capability, observed for nonhermetic packages [2], are eliminated by design. 3. Reliability assessment For the evaluation of the reliability concerning the main failure modes, results from the device qualification including accelerated tests, quality monitoring results out of continuous production, results from
Corresponding author. E-mail address: [email protected] (T. Stiasny).
https://doi.org/10.1016/j.microrel.2018.07.079 Received 31 May 2018; Received in revised form 6 July 2018; Accepted 6 July 2018 0026-2714/ © 2018 Elsevier Ltd. All rights reserved.
Microelectronics Reliability 88–90 (2018) 510–513
T. Stiasny et al.
Fig. 1. Sketch of the IGCT turn-off in different phases. Only the turn-off channel of the gate unit is shown.
capacitor bank (turn-off circuit) MOSFETs (turn-off circuit) low impedant coupling of gate unit to power semiconductor
Fig. 3. The specified turn-off capability (5SHY 55L4520 orange line). Last pass turn-off at VDC = 2.8 kV, 125 °C in double pulse test (blue dots). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 2. IGCT (5 kA turn-off capability, 4.5 kV max. blocking voltage, ABB article no.: 5SHY 55L4520) with opened power semiconductor package for demonstration. Indicated are the essential components in the turn-off circuit of the gate-unit.
production variation.
field failures and from device analysis after long time operation in demanding applications are discussed. Special focus is laid on the design feature that were new at the introduction of the device compared to previous device generations like GTOs.
3.1.2. Long term blocking capability (cosmic ray ruggedness) The robustness of the turn-off process of the IGCT allowed to reduce the silicon thickness significantly for power loss reduction compared to e.g. GTOs. This increases the electric field under blocking conditions. Therefore, apart from standard long term blocking stability tests like HTRB, the ruggedness against single event burn-out failures (see [4]) was investigated using long term blocking tests (natural cosmic) and accelerated testing with artificial proton beams. The evaluated failure rate for the different IGCT types is shown in Fig. 4 for different bias voltages showing a significant margin to the data sheet.
3.1. Qualification and quality monitoring testing 3.1.1. Quality monitoring of turn-off safe operating area (SOA) The turn-off mode of the IGCT differentiated this power semiconductor from other switches. Therefore, special focus was put on the turn-off capability. In 2007, the SOA of the IGCTs was significantly improved by design improvements [3]. During the development, many devices were operated in a step stress test to destruction to establish a SOA baseline. The specified turn-off capability is tested on each IGCT as out-going inspection as a go/no-go test. To monitor the turn-off capability of the running production, devices are picked on sample base and tested to destruction in a step-stress test (increased turn-off current). The turn-off capability over the production years is presented in Fig. 3 showing a constant improvement which was achieved by continuous reduction of
3.2. Analysis of field failures ABB offers the possibility to analyse failed devices out of the field. Some customers use this service consequently and return close to all failed devices to ABB for analysis. This allows to close the loop for reliability and quality improvement on the semiconductor manufacturer side as well as on the application side. Additionally, the data allow to 511
Microelectronics Reliability 88–90 (2018) 510–513
T. Stiasny et al.
Pareto Chart of gate unit (GU) fail. pattern 70
1 00
60 Percent
80
50
GU events
error bars indicate 90% confidence
60
40 30
40
20 20
10 0 GU fail. pattern
IC M
F OS
ET U
GU events Percent Cum %
Fig. 4. Cosmic Ray induced failure rate in FIT (device failures in 109 device hours) for 4.5 kV, 5.5 and 6.5 kV devices in L-package (91 mm silicon diameter) from storage tests under high voltage (natural cosmic) and accelerated tests with artificial proton beams. Fat symbols are data sheet values.
41 61 .2 61 .2
ne
in la xp
9 1 3.4 74.6
ed
od di
5 7.5 82.1
e
pt O
4 6.0 88.1
s ic O
4 6.0 94.0
er th
0
4 6.0 1 00.0
Pareto Chart of GCT fail. pattern 70
GCT events
50
80
40
60
30
40
20
20
10 GCT fail. pattern
0
tu
r
ng ni
f of
t
GCT events Percent Cum %
oo
45 70.3 70.3
or sh
f do tt
f un
5 7.8 78.1
ne ai pl x e
d er ov
t ol -v
5 7.8 85.9
e ag
2 3.1 89.1
O
er th
0
7 1 0.9 1 00.0
Fig. 6. Pareto charts of the failure cause of returned field failed IGCTs. Upper chart gate unit failures, bottom chart power semiconductor failures.
Fig. 5. Field failure rate for a 91 mm diameter IGCT with 4.5 kV blocking voltage in FIT.
Table 1 Electrical parameters of an IGCT of outgoing inspection (125 °C) and re-measured after 15 years of operation.
calculate and monitor field failure rates on a solid basis. For a specific application (MVD motor drive) where the dataset is most comprehensive, the calculated field failure rate is shown in Fig. 5. The field failure rate in FIT (device failure in 109 devices hours) was calculated out of the number of device field failures r in the reporting year, the devices being operated in the field at the date of reporting in this type of application N and the estimated uptime of the application per year (assumed 6000 h/year) T, according to Eq. (1).
field failure rate = (r·109)/(N ·T)
Percent
1 00
60
Initial After 15 y
On-state @ 4kA
Gate cathode leakage @ −20.5 V
Leakage current @ 4.5 kV
3.26 V 3.13 V
0.13 mA 0.17 mA
17.6 mA 20.1 mA
(1)
The error bars were determined out of the chi-squared distribution with the confidence interval of 90%. The failure analysis showed, a failure ratio power semiconductor to gate unit is in averaged in the order of 1:1. In many cases the failure pattern can be assigned to known patterns from verification testing of the devices. The main failure causes on the power semiconductor parts are turn-off failures (likely due to electrical overstress). As was shown in Section 3.1.1., the turn-off capability is continuously improved. This reflects in the observed field failure rate (see Fig. 5). The main failure causes on the gate unit parts was never found in the high current path (MOSFET or electrolytic capacitor). Especially the most suspicious component the electrolytic capacitor was never found to be the failure root cause of any field return. Here the main failure causes were failed ICs and MOSFETs not in the turn-off channel (see Fig. 6).
Fig. 7. Gate circuit inductivity Lσ of IGCT with different ages that were in operation until now (circles) or not in operation as a reference (squares).
512
Microelectronics Reliability 88–90 (2018) 510–513
T. Stiasny et al.
Fig. 8. Surface scans of cathode segment part of an IGCT after 15 years of operation (top left) and a reference device (unloaded). A cross section of metal layers of the two surface scans are overlaid in the graph on the right.
Comparing the two surface scans and the surface topology of the cathode metal layers clearly shows a degradation, as the cathode metal layer was compressed in height. As consequence, the width increased. Compared to the initial cathode metal height, the degradation is in the 10% range and not regarded as critical for the device performance (see Fig. 8).
3.3. Analysis of IGCTs after 15 years of operation IGCT operated in a lift application (lifting up excavate gravel 800 m during the Gotthard railway tunnel construction) were dismantled after 15 years of operation and analysed. 3.3.1. Power semiconductor analysis The devices were tested with the same electrical outgoing test sequence as 15 years earlier. No significant deviation of electrical parameters was found and all parameters were within the device specifications (see Table 1).
4. Conclusion More than 20 years ago, the IGCT was launched as new power semiconductor device especially for demanding high power electronic applications requiring high reliability. Due to the new demanding requirements on the gate circuit (package, interface to gate unit and gate unit), concerns were raised especially regarding the reliability of the gate circuit path. After 20 years of IGCT applications, the observed failure rate shows state of the art performance. Analysis of devices after being in service in a demanding application for 15 years showed hardly any wear-out. Especially no significant degradation of the gate circuit path with respect to the gate unit components (MOSFETs and the electrolytic capacitors) was found. The mechanical structure was found intact.
3.3.2. Gate path analysis Special focus was set on the analysis of the gate path. From electrical turn-off measurements, out of commutation time tcom, driving voltage UGU and the turn-off current, the Gate circuit impedance Lσ was evaluated. This evaluation was compared for devices that were in operation and those that were just stored. The variation of the values shows no systematic degradation of the gate circuit impedance neither under operating condition nor under storage condition (see Fig. 7). Later electrolytic capacitors were cut out of the gate unit print and measured. No significant degradation of the capacitance and resistivity values was observed.
References
3.3.3. Degradation of the cathode metallization After disassembly of the silicon chip from the package, the surface of the dry interfaces were investigated. Especially the metallization of the cathode segments was analysed as the metal stack on top of the cathode segments is exposed to the highest pressure in the press pack construction (smallest contact area of all parts) and is exposed to high temperature swings during load cycling as it is located close to the Junction. The topology of cathode segments were scanned with a surface scanning tool and compared to devices produced in a similar production period but not exposed to thermal load cycles.
[1] S. Klaka, M. Frecker, H. Grüning, The Integrated Gate-commutate Thyristor: a new High-effiviency, High-power Switch for Series or Snubberless Operation, Proc. PCIM Europe, 1997. [2] N. Tanaka, K. Ota, S. Iura, Y. Kusakabe, K. Nakamura, E. Wiesner, E. Thal, Robust HVIGBT Module Design Against High Humidity, Proc. PCIM Europe, 2015, pp. 368–373. [3] T. Wikström, Th. Stiasny, M. Rahimo, D. Cottet, P. Streit, The corrugared p-base IGCT – a new benchmark for large area SOA scaling, Proc. ISPSD Jeju Korea (2007) 29–32. [4] H. Kabza, H.J. Schulze, Y. Gerstenmeier, P. Voss, J. Wihelmi, W. Schmid, F. Pfirsch, K. Platzöder, Cosmic Radiation as a Cause for Power Devices Failure and Possible Countermeasures, Proc. ISPSD Davos Switzerland, 1994, pp. 9–12.
513
Contents lists available at ScienceDirect
Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel
Reliability evaluation of IGCT from accelerated testing, quality monitoring and field return analysis
T
⁎
Th. Stiasny , O. Quittard, Ch. Waltisberg, U. Meier ABB Switzerland Ltd, Semiconductors, Lenzburg, Switzerland
A R T I C LE I N FO
A B S T R A C T
Keywords: Power semiconductor IGCT Integrated gate commutated thyristor
The IGCT (Integrated Gate Commutated Thyristor) was launched as a new power semiconductor device 20 years ago, using design features where only few reference was available at that time. Especially the gate path that needs to take the full principal current for micro seconds was significant extension of the state of the art. Inbetween 20 years of field experience with IGCTs is available. In this paper test results from accelerated testing, quality monitoring results over many years, reliability date from evaluated field failure rates and analysis of devices that were in service for 15 years under well-known load are summarized and show an excellent reliability of the power semiconductor.
1. Introduction 20 years ago the first high power applications using IGCT (Integrated Gate Commutated Thyristor) were launched [1]. At that time, the IGCT was a new high power semiconductor device concept combining the ruggedness of press-pack devices, the low on-state losses of thyristors and the turn-off ruggedness of transistors. For this device, a powerful gate-unit and a low impedant gate-cathode path is required. After launching the device, concerns were raised on newly introduced features of the device, especially the long term reliability of the gate circuit path exceeding the experience of precious power semiconductor device generations. In-between, more than 250,000 IGCTs are in service in motor drives, interties, STATCOM, breakers and other demanding applications giving a base for a deep reliability analysis. 2. IGCT device design with respect to reliability The device principle is sketched in Fig. 1, only considering the demanding turn-off circuit of the gate unit. The turn-on of the devices is principally equivalent to the turn-on of a thyristor or GTO. Due to the distributed gate structure (see Fig. 2) a significantly higher turn-on pulse can easily be applied to the power semiconductor leading to a fast latching of the thyristor and with this to high dI/dt capability. In conducting state, the IGCT operates as a latched thyristor leading to a very low on-state voltage drop. For initiating the turn-off of the IGCT, the gate-cathode path is biased in reverse direction. This is reached by closing the switch S in the turn-off circuit of the gate-cathode path to a
⁎
voltage source (see Fig. 1). In the IGCT, the S is represented by tens of MOSFETs operated in parallel, the voltage source by tens of electrolytic capacitors in parallel (see Fig. 2). The commutation of the device principal current from the cathode path to the gate path is taking place during the time tcom ~IG·Lσ/UGU (assuming for simplicity a pure inductive gate cathode impedance). This commutation needs to be finished before anode-cathode voltage is built up across the device (hard drive limit) and is typically well below 1 μs. By this commutation, the active part of the device is transformed into an open base pnp-transistor. The turn-off is then possible without a dV/dt limiting circuit (e.g. snubber circuit that is needed for a GTO). Therefore, the gate-cathode impedance is the decisive parameter to ensure a reliable turn-off. Degradation effects of this parameter would have a negative influence on the long term reliability of the IGCTs. The low impedant gate-cathode path during the blocking phase also ensues the high dV/dt capability of the device compared to standard thyristors. The power device is packaged in a hermetic ceramic press pack housing, ensuring the protection of the sensitive structures of the power semiconductor from environmental influences. Therefore, issues like influence of humidity on the blocking capability, observed for nonhermetic packages [2], are eliminated by design. 3. Reliability assessment For the evaluation of the reliability concerning the main failure modes, results from the device qualification including accelerated tests, quality monitoring results out of continuous production, results from
Corresponding author. E-mail address: [email protected] (T. Stiasny).
https://doi.org/10.1016/j.microrel.2018.07.079 Received 31 May 2018; Received in revised form 6 July 2018; Accepted 6 July 2018 0026-2714/ © 2018 Elsevier Ltd. All rights reserved.
Microelectronics Reliability 88–90 (2018) 510–513
T. Stiasny et al.
Fig. 1. Sketch of the IGCT turn-off in different phases. Only the turn-off channel of the gate unit is shown.
capacitor bank (turn-off circuit) MOSFETs (turn-off circuit) low impedant coupling of gate unit to power semiconductor
Fig. 3. The specified turn-off capability (5SHY 55L4520 orange line). Last pass turn-off at VDC = 2.8 kV, 125 °C in double pulse test (blue dots). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 2. IGCT (5 kA turn-off capability, 4.5 kV max. blocking voltage, ABB article no.: 5SHY 55L4520) with opened power semiconductor package for demonstration. Indicated are the essential components in the turn-off circuit of the gate-unit.
production variation.
field failures and from device analysis after long time operation in demanding applications are discussed. Special focus is laid on the design feature that were new at the introduction of the device compared to previous device generations like GTOs.
3.1.2. Long term blocking capability (cosmic ray ruggedness) The robustness of the turn-off process of the IGCT allowed to reduce the silicon thickness significantly for power loss reduction compared to e.g. GTOs. This increases the electric field under blocking conditions. Therefore, apart from standard long term blocking stability tests like HTRB, the ruggedness against single event burn-out failures (see [4]) was investigated using long term blocking tests (natural cosmic) and accelerated testing with artificial proton beams. The evaluated failure rate for the different IGCT types is shown in Fig. 4 for different bias voltages showing a significant margin to the data sheet.
3.1. Qualification and quality monitoring testing 3.1.1. Quality monitoring of turn-off safe operating area (SOA) The turn-off mode of the IGCT differentiated this power semiconductor from other switches. Therefore, special focus was put on the turn-off capability. In 2007, the SOA of the IGCTs was significantly improved by design improvements [3]. During the development, many devices were operated in a step stress test to destruction to establish a SOA baseline. The specified turn-off capability is tested on each IGCT as out-going inspection as a go/no-go test. To monitor the turn-off capability of the running production, devices are picked on sample base and tested to destruction in a step-stress test (increased turn-off current). The turn-off capability over the production years is presented in Fig. 3 showing a constant improvement which was achieved by continuous reduction of
3.2. Analysis of field failures ABB offers the possibility to analyse failed devices out of the field. Some customers use this service consequently and return close to all failed devices to ABB for analysis. This allows to close the loop for reliability and quality improvement on the semiconductor manufacturer side as well as on the application side. Additionally, the data allow to 511
Microelectronics Reliability 88–90 (2018) 510–513
T. Stiasny et al.
Pareto Chart of gate unit (GU) fail. pattern 70
1 00
60 Percent
80
50
GU events
error bars indicate 90% confidence
60
40 30
40
20 20
10 0 GU fail. pattern
IC M
F OS
ET U
GU events Percent Cum %
Fig. 4. Cosmic Ray induced failure rate in FIT (device failures in 109 device hours) for 4.5 kV, 5.5 and 6.5 kV devices in L-package (91 mm silicon diameter) from storage tests under high voltage (natural cosmic) and accelerated tests with artificial proton beams. Fat symbols are data sheet values.
41 61 .2 61 .2
ne
in la xp
9 1 3.4 74.6
ed
od di
5 7.5 82.1
e
pt O
4 6.0 88.1
s ic O
4 6.0 94.0
er th
0
4 6.0 1 00.0
Pareto Chart of GCT fail. pattern 70
GCT events
50
80
40
60
30
40
20
20
10 GCT fail. pattern
0
tu
r
ng ni
f of
t
GCT events Percent Cum %
oo
45 70.3 70.3
or sh
f do tt
f un
5 7.8 78.1
ne ai pl x e
d er ov
t ol -v
5 7.8 85.9
e ag
2 3.1 89.1
O
er th
0
7 1 0.9 1 00.0
Fig. 6. Pareto charts of the failure cause of returned field failed IGCTs. Upper chart gate unit failures, bottom chart power semiconductor failures.
Fig. 5. Field failure rate for a 91 mm diameter IGCT with 4.5 kV blocking voltage in FIT.
Table 1 Electrical parameters of an IGCT of outgoing inspection (125 °C) and re-measured after 15 years of operation.
calculate and monitor field failure rates on a solid basis. For a specific application (MVD motor drive) where the dataset is most comprehensive, the calculated field failure rate is shown in Fig. 5. The field failure rate in FIT (device failure in 109 devices hours) was calculated out of the number of device field failures r in the reporting year, the devices being operated in the field at the date of reporting in this type of application N and the estimated uptime of the application per year (assumed 6000 h/year) T, according to Eq. (1).
field failure rate = (r·109)/(N ·T)
Percent
1 00
60
Initial After 15 y
On-state @ 4kA
Gate cathode leakage @ −20.5 V
Leakage current @ 4.5 kV
3.26 V 3.13 V
0.13 mA 0.17 mA
17.6 mA 20.1 mA
(1)
The error bars were determined out of the chi-squared distribution with the confidence interval of 90%. The failure analysis showed, a failure ratio power semiconductor to gate unit is in averaged in the order of 1:1. In many cases the failure pattern can be assigned to known patterns from verification testing of the devices. The main failure causes on the power semiconductor parts are turn-off failures (likely due to electrical overstress). As was shown in Section 3.1.1., the turn-off capability is continuously improved. This reflects in the observed field failure rate (see Fig. 5). The main failure causes on the gate unit parts was never found in the high current path (MOSFET or electrolytic capacitor). Especially the most suspicious component the electrolytic capacitor was never found to be the failure root cause of any field return. Here the main failure causes were failed ICs and MOSFETs not in the turn-off channel (see Fig. 6).
Fig. 7. Gate circuit inductivity Lσ of IGCT with different ages that were in operation until now (circles) or not in operation as a reference (squares).
512
Microelectronics Reliability 88–90 (2018) 510–513
T. Stiasny et al.
Fig. 8. Surface scans of cathode segment part of an IGCT after 15 years of operation (top left) and a reference device (unloaded). A cross section of metal layers of the two surface scans are overlaid in the graph on the right.
Comparing the two surface scans and the surface topology of the cathode metal layers clearly shows a degradation, as the cathode metal layer was compressed in height. As consequence, the width increased. Compared to the initial cathode metal height, the degradation is in the 10% range and not regarded as critical for the device performance (see Fig. 8).
3.3. Analysis of IGCTs after 15 years of operation IGCT operated in a lift application (lifting up excavate gravel 800 m during the Gotthard railway tunnel construction) were dismantled after 15 years of operation and analysed. 3.3.1. Power semiconductor analysis The devices were tested with the same electrical outgoing test sequence as 15 years earlier. No significant deviation of electrical parameters was found and all parameters were within the device specifications (see Table 1).
4. Conclusion More than 20 years ago, the IGCT was launched as new power semiconductor device especially for demanding high power electronic applications requiring high reliability. Due to the new demanding requirements on the gate circuit (package, interface to gate unit and gate unit), concerns were raised especially regarding the reliability of the gate circuit path. After 20 years of IGCT applications, the observed failure rate shows state of the art performance. Analysis of devices after being in service in a demanding application for 15 years showed hardly any wear-out. Especially no significant degradation of the gate circuit path with respect to the gate unit components (MOSFETs and the electrolytic capacitors) was found. The mechanical structure was found intact.
3.3.2. Gate path analysis Special focus was set on the analysis of the gate path. From electrical turn-off measurements, out of commutation time tcom, driving voltage UGU and the turn-off current, the Gate circuit impedance Lσ was evaluated. This evaluation was compared for devices that were in operation and those that were just stored. The variation of the values shows no systematic degradation of the gate circuit impedance neither under operating condition nor under storage condition (see Fig. 7). Later electrolytic capacitors were cut out of the gate unit print and measured. No significant degradation of the capacitance and resistivity values was observed.
References
3.3.3. Degradation of the cathode metallization After disassembly of the silicon chip from the package, the surface of the dry interfaces were investigated. Especially the metallization of the cathode segments was analysed as the metal stack on top of the cathode segments is exposed to the highest pressure in the press pack construction (smallest contact area of all parts) and is exposed to high temperature swings during load cycling as it is located close to the Junction. The topology of cathode segments were scanned with a surface scanning tool and compared to devices produced in a similar production period but not exposed to thermal load cycles.
[1] S. Klaka, M. Frecker, H. Grüning, The Integrated Gate-commutate Thyristor: a new High-effiviency, High-power Switch for Series or Snubberless Operation, Proc. PCIM Europe, 1997. [2] N. Tanaka, K. Ota, S. Iura, Y. Kusakabe, K. Nakamura, E. Wiesner, E. Thal, Robust HVIGBT Module Design Against High Humidity, Proc. PCIM Europe, 2015, pp. 368–373. [3] T. Wikström, Th. Stiasny, M. Rahimo, D. Cottet, P. Streit, The corrugared p-base IGCT – a new benchmark for large area SOA scaling, Proc. ISPSD Jeju Korea (2007) 29–32. [4] H. Kabza, H.J. Schulze, Y. Gerstenmeier, P. Voss, J. Wihelmi, W. Schmid, F. Pfirsch, K. Platzöder, Cosmic Radiation as a Cause for Power Devices Failure and Possible Countermeasures, Proc. ISPSD Davos Switzerland, 1994, pp. 9–12.
513