The HiPak TM power miconductor module ries is designed for reliable operation under demanding condi-tions throughout the module’s lifetime. The operation conditions and thus the expected module’s lifetime strongly depends on the application. In operation, the modules are subjected to a variety of temperature pro-files, which cau cyclic thermo-mechanical stress in all components and joints of the modules and finally lead to device failure. The magnitude and frequency of the- stress-cycles define the lifetime expectancy. Each specific profile leads to different stress distribution throughout the module, so that the weakest link of the module, which finally leads to failure, can be found in different components or joints. Moreover it is not possi-ble to calculate the exact lifetime of individual modules. Instead the lifetime must be expresd in terms of the B10 lifetime, which is the number of cycles during which 10% of the total number of modules fails.
The aim of this application note is to provide load cy-cling lifetime data for the power electronics designer to estimate the module lifetime for optimisation of the par-ticular application.没什么大不了英文
Table of Contents
1Objective of this application note 3 2Lifetime asssment 3 2.1Power cycling experiments 3 2.2From experiments to lifetime models 3 2.3Lifetime in terms of the B10 lifetime 4 3The load-cycling capability of the HiPak TM power modules 4 3.1Lifetime of the solder joints of the conductor leads and substrates 4 3.2Lifetime of the solder joint of the chips 6 3.3Lifetime of the wire bonds 10 4Lifetime calculation of a traction application example 11 5Revision history 12 6Application support 12 7References 12
1 Objective of this application note
choresPrior to this application note, the load cycling reliability of the HiPak TM power modules was described in Applica-tion Note “5SYA 2043-01 Load-cycle capability of HiPaks”. The lifetime data was given in two lifetime curves. One was for a slow cycle period (t cycle = 2 min) and the other curve for a fast cycle (t cycle = 2 s). The curves were valid for the whole power module including all components and joints.
Since the relea of the old application note, more pow-er cycling data has become available and more sophisti-cated solder and wire bond fatigue models have been created. Therefore, this new ap
plication note is relead. Here, individual lifetime curves are prented for the crit-ical joints, each of which fail due to different failure mechanisms [1, 2] and are described by different life time models. Moreover, for each critical joint veral life-time curves are calculated and plotted for different cycle periods (t cycle) and the absolute temperatures (T j or T c). All the curves reprent a wide matrix of accurate life-time data under numerous cycling conditions.
哥伦比亚大学校训
facts2 Lifetime asssment
readitlater2.1 Power cycling experiments
The lifetime of the power modules is assd by power cycling experiments, in which a given temperature cycle is repetitively applied to a module until it fails. The failure criterion is defined as a 5% increa in V ce or a 20% in-crea in R th of the tested module. The modules’ tem-perature increas as current pass through the chips and they are cooled by the cooler mounted on the ba plate. The temperature cycle is generally defined by the minimum and maximum values of the temperature and the period of the cycle. In order to complete the experi-ment within a reasonable period of time, the power mod-ules are subjected to higher temperature swings than in
a typical application.
2.2 From experiments to lifetime models
建军节用英语怎么说The modules’ lifetime is described using a two parame-ter Weibull distribution. The Weibull shape and scale parameters are fitted to the obtained lifetimes of the indi-vidual modules in the power cycling experiment. The resulting Weibull distribution is ud to determine the B10 lifetime under the given cycling conditions. In order to calculate the lifetime under different cycling conditions than in the power cycling experiment, lifetime models are required. The lifetime models in this applica-tion note are bad on the Coffin-Manson law and fa-tigue of the joints due to plastic deformation [2-4]. Life-time data from power cycling experiments and material creep data from the literature is ud to build the lifetime models. Three different models describe the lifetime of the solder joint of the die attach (chip solder joint), the solder joints of the conductor leads and substrates, and the wire bonds, respectively. The different joints in a power module are depicted in Fig. 1.
Figure 1 Sketch of the different joints in a power module.
The lifetime models for the solder joints are bad on time dependent creep and therefore the cycle period (t cycle) has an influence on the solder joint lifetime. On the other hand the model for the wire bond lifetime is inde-pendent of t cycle, becau this model assumes that im-mediate plastic deformation leads to fatigue instead of time dependent creep.
edy
An example for the temperature profiles ud to calcu-late the fatigue per cycle in the solder joints is shown in Fig. 2. The example shows the temperature profile ud for estimating the chip solder lifetime for t cycle = 120 s, T j,max= 100°C, and T j= 40 K. All the profiles for the solder joint lifetime estimation are of similar shape, de-spite the different cycling conditions.
Figure 2 Temperature profile for the lifetime calculation of the chip solder joint for t cycle = 120 s, T j,max = 100°C, and T j = 40 K.
2.3 Lifetime in terms of the B10 lifetime
The modules’ reliability is defined by the B10 lifetime, which is described as the number of cycles where 10% of the modules of a population fail [5]. The B10 lifetime curves are generated using the lifetime models and the temperature profile in Fig. 2.
cuckooTaking into account that the power modules are heated by the chips and cooled at the ba plate, the length of the heating and cooling periods defines the level of the thermo-mechanical stress at each component or joint. In ca of a short cycling period (i.e. t on = t off = 1 s) the chips and wire bonds are expod to the temperature cycles while the ca temperature (T c) remains fairly constant. In addition the lifetime of the solder joints de-pends on t cycle as explained above. Therefore, individual B10 lifetime curves are generated for the different solder joints for veral t on and T c or T j values in order to cover as many realistic cycling scenarios as possible. 3 The load-cycling capability of the HiPak TM
万国司考power modules
3.1 Lifetime of the solder joints of the conductor leads
and substrates
The lifetime of the solder joints connecting the conduc-tor leads to the substrates and the substrates to the ba plate is described by the same model. Both solder joints are shown in dark blue colour in Fig. 1. The graphs in Figs. 3-6 show the B10 lifetime curves of the joints at various values of t on and T c,min. The B10 lifetime values are also listed in Tables 1 and 2 for simpler ac-cess to the lifetime data.
If necessary, the B5 and B1 lifetimes, which are the total number of cycles during which 5% and 1% of the mod-ules’ population fails, under the given cycling conditions can be calculated by multiplying the B10 lifetime with the factors k5 = 0.90 and k1 = 0.70, respectively. For exam-ple, it can be read from Table 1 that the B10 lifetime is equal to 106’000 cycles for t on = 10s, T c,min = 40°C, and T c = 60 K. The respective B5 and B1 lifetimes under the cycling conditions can be calculated as 95’400 and 74’200 cycles, respectively.
10cycle c,min c 10
Figure 3 The
B10 lifetime curves of the solder joints of the conductor leads and substrates for t cycle equal to 10s.
>hospitalization
