Tweet

Posted on 29 June 2019

Root Causes of Failures of Disc Type Bipolar Power Semiconductor Devices

Free Bodo's Power Magazines!

 

 

 

In spite of simplicity in schematic design of the circuit with power thyristors and diodes application, practically a customer deals with a lot of details that should be taken into account to ensure reliable and continuous operation of power semiconductor devices (PSD). Basically all details of PSD application are determined by design features of the devices. Lack of knowledge and ignorance of application peculiarities of PSD as on the stages of circuit design and during mounting with heat sink as well lead to abnormal operation of the devices and consequently to parametric failure. The analysis of failures is generally aimed at identification of causes connected with them.

By Alexander Stavtsev, Technical Director “Proton-Electrotex” JSC and Alexey Surma, Head of R&D Center “Proton-Electrotex” JSC

 

The root causes of failures of PSD produced according to modern principals of design and technological standards can be nominally classified as following:

- Incorrect mounting in equipment
- Electrical and thermal operational conditions going above safe operation of PSD that is indicated by a manufacturer
- Abnormal and emergency operation conditions of equipment, due to which the operation condition of thyristor goes above safe operation
- Using of PSD with worked-out resource
- Ignorance of design and technological peculiarities of definite types of PSD choosing operational conditions

In the article the most popular types of failures according to the groups noted above are described, the possibilities of identification of failure causes in accordance with any findings of emergency diodes and thyristors testing are discussed.

Incorrect mounting in equipment

The most popular cases of incorrect mounting of PSD is the one with heat sink. Damage of semiconductor element may occur as a result of such mounting, as well as of insufficient dissipation of heat from device in operation process. As a rule, incorrect mounting with heat sink can be detected during visual examination of damaged PSD. Some case study is described below.

1. Poor or incorrect prepared for mounting contact surfaces of PSD and heat sink. Hold-down print with uneven surface is shown on the Figure 1. Such mounting, as a rule, cannot guarantee secure heat (sometimes electrical) contact of PSD with heat sink, and with increased mounting force can lead to damage of semiconductor element.

Uneven surface

2. Covering contact surfaces of disc type devices with heat-conducting paste wanting electroconducting characteristics, Figure 2. On the picture some leftovers of insulating heat-conducting paste can be seen in the loop for centring of device case. Presence of thin and spotty layer of insulating paste between surfaces of PSD case and heat sink or mounting electroconducting bar lead to electric contact get pointy, resistance of contact repetitively increases, and voltage loss of PSD on the contact equals or even exceeds the on-state voltage loss. It leads to additional lost power, inadmissible PSD overheat and its thermal breakdown. Herewith, due to localization of electric contact, overheat areas of semiconductor element have local character as well.

Leftovers of insulating heat-conducting paste

3. Incorrect selection of contact surface diameter of heat sink, Figure 3. Clamping machine print of smaller diameter than in contact package base is shown on the picture. Contact bases of disc type packages are made of soft material (oxygen-free copper), that’s why in spite of relatively high thickness they do not transmit pressure load to the associated parts of semiconductor element. Herewith, for such areas thermal and electrical contacts of semiconductor element with package bases may get disrupted.

Clamping machine print of smaller diameter than in contact package

Loss of thermal contact leads to local overheat of the defined areas of semiconductor element. However, even for device operating in overlapping of single current pulses mode which doesn’t require effective heat dissipation, incorrect mounting may lead to break- down due to insecure electrical contact. During high peak current pulse overlapping with high rise speed in front, in local areas of PSD, which don’t have any electrical contact between cathode metallization of semiconductor element and conducting cathode package base (or cathode layer), electrical arc discharge may appear (in case the clearance between contact surfaces is few micron). Herewith gradual degradation of semiconductor element surface occurs (pitting) what as a result leads to break-down.

4. Incorrect selection of center element dimensions or its axial offset of the proper mouth of PSD, Figure 4. The above mentioned mounting blemish leads to rapid increase of local pressure at the semiconductor element in its center. As a result, the element can get mechanical damages, or (speaking about thyristors) short-circuiting of amplifying electrode by cathode base.

Incorrect selection of center element dimensions

5. Uneven distribution of mounting force on the contact surfaces of PSD, Figure 5. Such mounting blemish can lead to semiconductor element destruction as a result of inadmissible increase in local pressure at hold-down, as well as overheat and following it thermal breakdown due to insufficient heat dissipation of element.

Uneven distribution of mounting force

6. Mounting overforce that leads to mechanical damage of semiconductor element. If overforce is evenly distributed over the contact base surface, visual examination of the device as a rule doesn’t give any results. However, after decapsulation (for thyristors with subdivided amplifying electrode) a characteristic deep imprint of amplifying electrode image can be seen on the cathode base of the package. Mounting overforce leads to destruction of semiconductor element, to splitting of the semiconductor wafer which can be determined only by etching of the element with special etchants and at certain conditions, Figure 6.

 

Mounting overforce that leads to mechanical damage

7. Insufficient mounting force or its total absence. It leads to damage of operating thermal conditions and parametric failure of PSD. In operating process overheat of PSD as a result of insufficient mounting force can be determined, for example, with help of infrared imaging, Figure 7.

Insufficient mounting force or its  total absence

Electrical and Thermal Conditions Overrunning the Safe Operating Areas

These conditions can be divided into several groups.

Excess of the permissible rates in average current (with allowance for cooling condition) or in surge current peak.

Failed semiconductor element of PSD in this case has characteristic “large” weld penetration areas ( Figure 8) located in the area of power current flow. Since the device failure as a rule is caused by thermal breakdown or (in overlapping of surge current mode) by filament formation of high peak current, the weld penetration areas usually appear in the most overheated areas of the element as well as in the areas with poor heat dissipation.

Failed semiconductor element

When thermal filament formation appears in overlapping of standby high peak current mode the temperature in thyristor or diode base increases rapidly and as a result intrinsic carrier density increases as well, which in the most heated area of the semiconductor element almost equals the concentration of injected carriers. Resistance in this area is getting lower, current through it is getting higher, which triggers the retroaction motion, and that leads to current constriction into local thermic path and as a result to destruction of the semiconductor element. Pitting of cathode area appears during smelting of one of the layers of the semiconductor element (compound of aluminum and silicon has the lowest melting point), and as a result one of the semiconductor junctions gets damaged, usually the one responsible for direct blocking voltage, or it leads to total destruction of semiconductor element.

Low anode current of thyristor

Thyristor is a semiconductor consisting of auxiliary and main thyristors, Figure 9. Anode circuit current Ia is divided into two currents: anode current of the auxiliary thyristor (Ia aux) also as amplifying electrode current of main thyristor (Iae main) and anode current of the main thyristor (Imain). If anode current is low, current in the amplifying electrode circuit of the main thyristor cannot reach the level when the main thyristor opens completely all over the amplifying electrode perimeter. Conduction area will be formed in local point where high local power allocates capable of bringing it to local overheat and destruction of the semiconductor element. If anode current cannot switch the main thyristor at all even in the local point, during long running of thyristors at such conditions there is possibility of overheat of amplifying electrode area of the main thyristor and as a result the destruction of the semiconductor element.

Anode circuit current Ia is divided into two currents

During circuit designing anode currents of the main power currents as well as of the snubber circuits should be considered. Such effect should be thoroughly studied in case of fast thyristors which have subdivided amplifying electrode of the main element and as a result high gate current of the main thyristor.

Problems connected with amplifying are complicated by the fact that structurally amplifying area has one-way heat dissipation. Heat dissipation is carried out from anode side, from cathode side there is a wafer with center mouth which unshorts the amplifying and cathode area. Hence, possibility of failing is growing with temperature rise of device.

Undue processes in thyristor amplifying circuit.

If any undue signals in the amplifying circuit, the area of semiconductor element damage usually is close to the amplifying electrode of the auxiliary thyristor, Figure 10.

Damage usually is close to the amplifying electrode

In this case the following reasons may lead to failure:

Poor amplifying signal or switching thyristor by interfering signal.

Poor amplifying signal initiates switching of thyristor only in the local point close to the auxiliary amplifying electrode, not all over the perimeter of amplifying electrodes, as it happens in case of standard control pulse (Figure 11). Thyristor starts to conduct anode current exactly in this area and at the same time direct blocking voltage starts to decrease. Thus at this moment peak power driven out of the thyristor has maximum value and is localized in one or several spots, which lead to local overheat and destruction of the semiconductor element.

Poor amplifying signal initiates switching of thyristor only in the local point

Signal in amplifying circuit

Signal in amplifying circuit overrunning the safe operating areas, for example high peak control current as well as voltage and reversed polarity current in amplifying circuit can lead to device failure with characteristic local area of damage similar to the one on Figure 10.

Entry into high-voltage amplifying circuit

Entry into high-voltage amplifying circuit and as a result unlimited current flow through the amplifying circuit leads to local overheat and destruction of semiconductor element. Characteristic for such failure “large” weld penetration spot in the area of the auxiliary thyristor is shown on Figure 12.

Characteristic for such failure “large” weld penetration spot

Simultaneous presence of amplifying and reverse anode voltage signals.

In such case at the expense of transistance reverse anode current (leakage current) increases rapidly which extremely exceeds the admissible values for thyristor (Figure 13). As a result substantial power is accumulated close to the amplifying electrode, which may lead to the device destruction, and the higher the amplifying signal the higher leakage current, and as a result the driven out power. The following should also be taken into consideration – the higher temperature, the higher leakage current, and the higher possibility of parametric failure of PSD. Such conditions are not recommended for applying but possible in some cases when inclusive studying certain circuit and choosing conditions for thyristor operation, which would guarantee its durable and proper functioning.

Substantial power is accumulated close to the amplifying electrode, which may lead to the device destruction

Excess of the limit value of rate of anode current rise or low value of control current rate of rise (di/dt effect)

In case of standard control pulse thyristor switches all over the amplifying electrode perimeter of the auxiliary thyristor and longitudinal propagation of on-state has end rate, Figure 14. Therefore at high rates of rise of anode current and limitation of spreading speed in on-state of thyristor, local current density close to the amplifying electrode of the auxiliary element can exceed its limit value, which can cause overheat and destruction of the semiconductor element. Same thing happens when rate of rise of control current is too low.

Local current density close to the amplifying electrode of the auxiliary element

Character of damage area as a rule is similar to the one shown on Figure 10.

Such mechanism may appear at switching of the main thyristor, and in this case area of destruction is close to the amplifying electrode of the main thyristor, Figure 15. As a rule, failure similar to the one shown on Figure 15 happens at pulse-frequency conditions, i.e. at exceeding the limit of the recurrent value of admissible di/dt, and failure similar to the one on Figure 10 happens at exceeding the limit of single value of admissible di/dt.

Exceeding the limit of the recurrent value of admiss di/dt

Undue anode switching over of thyristors without control signal.

In this case the most characteristic reasons leading to failures are as follows.

Switching over due to critical rate of rise of off-state voltage (dv/dt effect).

For such failure the destruction spot is located within the cathode area of the main and auxiliary thyristors. The most characteristic is its location close to the amplifying electrode of the auxiliary thyristor (similar to the one on Figure 10), or of the main thyristor (similar to the one on Figure 15) since these areas as a rule are the most sensitive to switching over initiated by capacitance current.

Switching over at applying direct voltage in the end of turn-off process during time less than tq.

Such undue switching mode is shown on Figure 16. The most probable destruction area is cathode area of the main thyristor. Such damage character is typical in case when concentration of electron-hole plasma in the base layers of the device is not enough to initiate switching over process over most part of the element surface. This is a local process in one or several spots, which have the highest carriers lifetime, or which are characterized by the lowest performance of distributed cathode diversion.

Undue switching mode

Failure at overvoltage in direct and reverse directions

Strong electric field causes avalanche breakdown, and the electric field on the surface is getting stronger as well. As a rule in this case failure happens at the periphery of the semiconductor element of thyristor or diode, Figure 17.

Failure happens at the periphery of the semiconductor element

Served out PSD

The most “sensitive” characteristics which can be considered as criterion to serving out are:

Off-state leakage current (for thyristors), reverse current (for diodes and thyristors). Served out PSD have high leakage current (reverse current) even at room temperature changes. This effect is caused by deterioration of protective compound at the periphery of the PSD element and degradation processes on the surface of the semiconductor in the area of the p-n junctions (bevel area). Leakage current at room temperature is just a few milliampere, and at maximum temperature it doesn’t exceed adjusted by manufacturer limits. Nevertheless, high leakage current at room temperature is direct reference that the device is potentially unreliable and should be replaced.

- On-state voltage drop, direct voltage drop. Increase in these characteristics of PSD is caused by degradation of the contact surfaces inside of PSD - metallized area of the semiconductor element cathode, surface of the anode thermal compensator, contact wafers. During continuous service the voltage drop may also be caused by decrease of carriers’ lifetime in the semiconductor element layers under influence of cosmic rays.
- Thermal resistance. Reason for its increase is degradation of the contact surfaces as well.

Measuring of the above mentioned characteristics is a nondestructive method, which helps to determine serving out of some devices. And fixing these characteristics go over the limit adjusted by manufacturer is necessary only as sufficient condition, which determines that the device is served out, because new devices are usually produced with technical allowance, and time history of characteristics during service is not tracked as a rule.

That’s why the final decision concerning serving out of the device is made as a rule only after decapsulation of the device (or screening group of same type devices) and visual examination of its contact surfaces condition.

Necessity of taking into consideration the structural and technological features of certain PSD

Manufacturers as a rule try to fully describe the areas of reliable performance for the produced devices considering the structural and technological features. However it’s impossible to take into account all features at different conditions. And sometimes the sequence of reliable conditions can cause the devices failure.

Let’s describe the above mentioned using the real case of semiconductor failure. Highvoltage thyristor was used in inverter at condition characterized by durable service with low anode current (angle of ignition is 180°), later the angle was abruptly decreased and anode current increased rapidly, however its average and peak values didn’t exceed permissible limit. Low current operation mode also didn’t exceed the limits despite that only the auxiliary thyristor was switching and its overheat didn’t lead to exceeding the maximum admissible temperature.

Nevertheless the sequence of these two reliable modes lead to failures, which had systematic and massive character. The destruction was in the area of the auxiliary thyristor cathode and was similar to the one on Figure 12.

The reason for failure was the following. During operation in low current mode the auxiliary thyristor was locally heated, though the main thyristor was still cold. After switching to high current mode, due to the difference in temperature of the main and auxiliary elements, the auxiliary thyristor was still in operation mode after switching of the main one, and the current level of the auxiliary thyristor exceeded the safe limit ( Figure18). This overload current as a result lead to overheat of the auxiliary element and following it thermal breakdown.

This overload current as a result lead to overheat of the auxiliary element

Therefore the thyristor failure was caused by ignoring certain features of such device: level of minimum anode current necessary to switch the main thyristor, and on-state temperature characteristics of current-voltage diagram of the main and auxiliary thyristor. Herewith the user didn’t have any information about the above mentioned characteristics, because this information is not included in the standard documentation. And the manufacturer didn’t have complete information about the features of the end user application.

It’s clear that tracing the reasons of failure caused by lack of information of specific features of the used semiconductors, which are determined by their structure and production technology, is quite a difficult task that cannot be solved without close cooperation between manufacturer and end user.

Conclusion

Therefore to avoid device failure it is necessary to strictly follow the recommendations of exploitation described in technical requirements, datasheets, and information materials.

Identification of reasons for PSD failures is very complicated task, which can be solved only by close cooperation between manufacturer and end user.

Exhibitor at PCIM Europe 2011

 

 

VN:F [1.9.17_1161]
Rating: 6.0/6 (2 votes cast)
Root Causes of Failures of Disc Type Bipolar Power Semiconductor Devices, 6.0 out of 6 based on 2 ratings

This post was written by:

- who has written 791 posts on PowerGuru - Power Electronics Information Portal.


Contact the author

Leave a Response

You must be logged in to post a comment.