Characterization of Materials and their Interfaces in a DBC - - PowerPoint PPT Presentation

Characterization of Materials and their Interfaces in a DBC Substrate for Power Electronics Applications

ECPE Workshop “Future of Simulation” Aymen BEN KABAAR1, Cyril BUTTAY2, Olivier DEZELLUS3, Rafaël ESTEVEZ1, Anthony GRAVOUIL4, Laurent GREMILLARD5

1SIMaP

, UMR 5266, CNRS, Grenoble-INP , UJF , France

2 Univ Lyon, INSA-Lyon, CNRS, Laboratoire Ampère UMR 5005, F-69621, Lyon 3Univ Lyon, Univ Lyon 1, CNRS, LMI, UMR 5615, F-69622, Lyon 4 Univ Lyon, INSA-Lyon, CNRS, LaMCoS, UMR 5259, F-69621, Lyon 5 Univ Lyon, INSA-Lyon, CNRS, MATEIS Laboratory, UMR 5510, F-69621, Lyon

21/11/18

1 / 29

Outline Introduction Characterization of the copper layers Characterization of the Ceramic Layer Characterization of the Metal-Ceramic Interface Conclusion

2 / 29

Outline Introduction Characterization of the copper layers Characterization of the Ceramic Layer Characterization of the Metal-Ceramic Interface Conclusion

3 / 29

Introduction – Power Electronic Module

Ceramic substrate Ensures ◮ Electrical insulation ◮ Heat conduction

4 / 29

Introduction – Power Electronic Module

Ceramic substrate Ensures ◮ Electrical insulation ◮ Heat conduction Direct Bonded Copper ◮ Ceramic:

◮ Heat conduction ◮ Electrical insulation

◮ Patterned Metal:

◮ Forms circuit ◮ Bonding to module

4 / 29

Introduction – Manufacturing of a DBC substrate

Copper Ceramic Copper Ceramic O2 Copper Oxide Copper Ceramic Eutectic Melt Heating O2 Diffusion and Cooling Copper Ceramic

1080 - 1070 - 1060 - 1050 -

O

2

0.4 0.8 1.2 1.6

Eutectic Concentration in Atom%

Source: J. Schulz-Harder, Curamic [1]

◮ Standard: Al2O3/Cu (AlN also possible, with separate oxidation) ◮ Bonding temperature very close to Cu melting point

5 / 29

Introduction – Manufacturing of a DBC substrate

Copper Ceramic Copper Ceramic O2 Copper Oxide Copper Ceramic Eutectic Melt Heating O2 Diffusion and Cooling Copper Ceramic

1080 - 1070 - 1060 - 1050 -

O

2

0.4 0.8 1.2 1.6

Eutectic Concentration in Atom%

Source: J. Schulz-Harder, Curamic [1]

◮ Standard: Al2O3/Cu (AlN also possible, with separate oxidation) ◮ Bonding temperature very close to Cu melting point Objective: modelling of the DBC for thermo-mechanical simulations

5 / 29

Outline Introduction Characterization of the copper layers Characterization of the Ceramic Layer Characterization of the Metal-Ceramic Interface Conclusion

6 / 29

Copper – Preparation of the samples

Note: the content of this presentation is detailed in [2] and [3] Tests on 3 Copper states: Cu3: Cu sheet prior to any process Cu2: The same after DBC annealing (but not bonded to ceramic)

◮ temperature history ◮ no external mechanical stress

Cu1: Full DBC process, followed by etching of the ceramic

◮ temp. and mech. history

7 / 29

Copper – Preparation of the samples

Note: the content of this presentation is detailed in [2] and [3] Tests on 3 Copper states: Cu3: Cu sheet prior to any process Cu2: The same after DBC annealing (but not bonded to ceramic)

◮ temperature history ◮ no external mechanical stress

Cu1: Full DBC process, followed by etching of the ceramic

◮ temp. and mech. history

7 / 29

Copper – Preparation of the samples

Note: the content of this presentation is detailed in [2] and [3] Tests on 3 Copper states: Cu3: Cu sheet prior to any process Cu2: The same after DBC annealing (but not bonded to ceramic)

◮ temperature history ◮ no external mechanical stress

Cu1: Full DBC process, followed by etching of the ceramic

◮ temp. and mech. history

7 / 29

Copper – Preparation of the samples

Note: the content of this presentation is detailed in [2] and [3] Tests on 3 Copper states: Cu3: Cu sheet prior to any process Cu2: The same after DBC annealing (but not bonded to ceramic)

◮ temperature history ◮ no external mechanical stress

Cu1: Full DBC process, followed by etching of the ceramic

◮ temp. and mech. history

Preparation and test: ◮ Copper sheets supplied by Curamik ◮ samples formed by electro-erosion ◮ Uniaxial and cycling tensile tests

7 / 29

Copper – Tensile test

0.00 0.05 0.10 0.15 0.20 0.25 0.30 Log(strain) 50 100 150 200 250 300 350 Cauchy Stress [MPa]

Cu3 (no annealing) Cu2 (annealing, free cooling) Cu1 (Full DBC process)

◮ Dramatic change caused by annealing (yield stress) ◮ Also, effect of mechanical stress on yield ➜ Further characterization on Cu1, more representative

8 / 29

Copper – Cycling test

0.00 0.01 0.02 0.03 0.04 0.05 Log(strain) 20 40 60 80 100 120 Cauchy Stress [MPa] 0.051 0.052 0.053 25 50 75 100

◮ Tests on Cu1, repetitive stress 0–120 MPa

◮ No compressive stress to prevent sample from buckling

◮ Ratchet effect caused by kinematic hardening of copper ➜ Need for a suitable model (Armstrong-Fredericks [4])

9 / 29

Copper – Modelling

E ν σy C γ 127 GPa 0.33 60 MPa 1.7 GPa 14.6

0.00 0.01 0.02 0.03 0.04 0.05 Log(strain) 20 40 60 80 100 120 Cauchy Stress [MPa]

Experiment Model

0.051 0.052 0.053 25 50 75 100

◮ Satisfying modelling of

◮ Elastic ◮ Plastic ◮ Hardening

Behaviours ◮ Parameters identification:

◮ E, ν, σy: uniaxial tests ◮ C and γ: cycling tests

10 / 29

Outline Introduction Characterization of the copper layers Characterization of the Ceramic Layer Characterization of the Metal-Ceramic Interface Conclusion

11 / 29

Ceramic – Preparation of the samples

◮ 2 grades of Al2O3 tested:

◮ standard, thickness=635 µm ◮ “HPS” (Zr-reinforced), thickness=250 µm

◮ Material supplied by Curamik ◮ Samples cut using a wafer saw ◮ Sample size: 4 mm×40 mm ◮ 3-point bending test.

12 / 29

Ceramic – Bending Tests

5 10 15 20 25 30 Specimen # 300 320 340 360 380 400 420 440 Young's Modulus [GPa]

Al2O3 Zr Al2O3

E = FL3 48σwt3 ◮ E: Young’s Modulus ◮ F: maximum load ◮ w: sample width ◮ L: support span ◮ σ: deflection ◮ t: sample thickness ◮ good consistency in the results

◮ few defects caused by the sample preparation ◮ good quality of the base material

13 / 29

Ceramic – Bending Tests (2)

Weibull Analysis ◮ Considers the sample as a series of elementary volumes ◮ Each volume has a statistical defect probability ◮ PSi: probability of survival ◮ σw: Weibull stress

5.4 5.6 5.8 6.0 6.2 6.4 6.6 log(

W)

4 3 2 1 1 2 log(log(1/Psi)) 16.03x-92.59 R2=0.97 18.96x-121 R2=0.99

Al2O3 Zr Al2O3

14 / 29

Ceramic – Modelling

Model used ◮ Purely elastic behavior ◮ Considers rupture Identification of model parameters: ◮ E: from bending test ◮ ν: from literature [5] ◮ m, σ0 and Veff: from Weibull analysis. E ν m σ0 Veff Al2O3 403 GPa 0,22 16.03 322 MPa 0.103 mm3 Zr-Al2O3 330 GPa 0.22 18.95 590 MPa 0.501 mm3

15 / 29

Outline Introduction Characterization of the copper layers Characterization of the Ceramic Layer Characterization of the Metal-Ceramic Interface Conclusion

16 / 29

Interface – Test Principle

◮ DBC sample with a notch in top Cu ◮ 4-point bending test ◮ Monitoring of fracture propagation ◮ Parameter identification with FE simulation

17 / 29

Interface – Preparation of the samples

◮ DBC configuration: 500 µm Cu / 250 µm Zr-Al2O3 / 500 µ Cu ◮ Chemical etching of copper patterns ◮ Ceramic cutting with a wafer saw ◮ Sample size: 10 × 80 mm2

18 / 29

Interface – Bending Tests

1 2 3 4 Displacement [mm] 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 Force [N] A

19 / 29

Interface – Bending Tests

1 2 3 4 Displacement [mm] 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 Force [N] A B

19 / 29

Interface – Bending Tests

1 2 3 4 Displacement [mm] 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 Force [N] A B

19 / 29

Interface – Bending Tests

1 2 3 4 Displacement [mm] 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 Force [N] A B C

19 / 29

Interface – Fracture Observation Ceramic Copper

Cross section (SEM) ◮ Crack length measurement accuracy: ±50µm ◮ Crack occurs at interface ◮ No Al2O3 remaining on Cu ◮ ≈ 20µm bonding defects ➜ To be considered in simulation

20 / 29

Interface – Fracture Observation

Delaminated copper surface (SEM) ◮ Crack length measurement accuracy: ±50µm ◮ Crack occurs at interface ◮ No Al2O3 remaining on Cu ◮ ≈ 20µm bonding defects ➜ To be considered in simulation

20 / 29

Interface – Cohesive model

Cohesive model ◮ Once TMax has been reached, degradation occurs ◮ Gradual reduction in stiffness ◮ Eventualy, separation at interface

TMax δ0 δcr δ K ΦSep T (1-D)K [MPa] [mm]

21 / 29

Interface – Cohesive model

Cohesive model ◮ Once TMax has been reached, degradation occurs ◮ Gradual reduction in stiffness ◮ Eventualy, separation at interface Implementation [6] ◮ Simulation of the 4-point test ◮ Cohesive zone between Al2O3 and bottom Cu ◮ Two parameters: TMax and ΦSep

TMax δ0 δcr δ K ΦSep T (1-D)K [MPa] [mm]

Copper Copper Ceramic Cohesive zone 21 / 29

Interface – Model Identification

2 sources of data for model identification

1 2 3 4 Displacement [mm] 2 4 6 8 10 Force [N] 0.0 0.2 0.4 0.6 0.8 1.0 crack length [mm] 1 2 3 4 5 10 15

Force-Displacement ◮ “Macro” observation ◮ focus on “peeling” region

22 / 29

Interface – Model Identification

2 sources of data for model identification

1 2 3 4 Displacement [mm] 2 4 6 8 10 Force [N] 0.0 0.2 0.4 0.6 0.8 1.0 crack length [mm] 1 2 3 4 5 10 15

Force-Displacement ◮ “Macro” observation ◮ focus on “peeling” region

22 / 29

Interface – Model Identification

2 sources of data for model identification

1 2 3 4 Displacement [mm] 2 4 6 8 10 Force [N] 0.0 0.2 0.4 0.6 0.8 1.0 crack length [mm]

Force-Displacement ◮ “Macro” observation ◮ focus on “peeling” region

22 / 29

Interface – Model Identification

2 sources of data for model identification

1 2 3 4 Displacement [mm] 2 4 6 8 10 Force [N]

Force Crack length

0.0 0.2 0.4 0.6 0.8 1.0 crack length [mm]

Force-Displacement ◮ “Macro” observation ◮ focus on “peeling” region Crack length ◮ “Local” observation

22 / 29

Interface – Model Identification (2)

1 2 3 4 Displacement [mm] 4 6 8 10 12 Force [N]

Sep = 32 J/m2

no defect

Measurement Tmax=350 MPa Tmax=300 MPa Tmax=250 MPa

23 / 29

Interface – Model Identification (2)

1 2 3 4 Displacement [mm] 4 6 8 10 12 Force [N]

Sep = 32 J/m2

no defect

Measurement Tmax=350 MPa Tmax=300 MPa Tmax=250 MPa

2.0 2.5 3.0 3.5 4.0 Displacement [mm] 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 crack length [mm]

Sep = 32 J/m2

no defect

Measurement Tmax=250 MPa Tmax=300 MPa Tmax=350 MPa

23 / 29

Interface – Model Identification (2)

1 2 3 4 Displacement [mm] 4 6 8 10 12 Force [N]

Sep = 32 J/m2

no defect

Measurement Tmax=350 MPa Tmax=300 MPa Tmax=250 MPa

2.0 2.5 3.0 3.5 4.0 Displacement [mm] 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 crack length [mm]

Sep = 32 J/m2

no defect

Measurement Tmax=250 MPa Tmax=300 MPa Tmax=350 MPa

1 2 3 4 Displacement [mm] 4 6 8 10 12 Force [N]

Sep = 10 J/m2

20 µm defect

Measurement Tmax=350 MPa Tmax=400 MPa Tmax=450 MPa Tmax=500 MPa

23 / 29

Interface – Model Identification (2)

1 2 3 4 Displacement [mm] 4 6 8 10 12 Force [N]

Sep = 32 J/m2

no defect

Measurement Tmax=350 MPa Tmax=300 MPa Tmax=250 MPa

2.0 2.5 3.0 3.5 4.0 Displacement [mm] 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 crack length [mm]

Sep = 32 J/m2

no defect

Measurement Tmax=250 MPa Tmax=300 MPa Tmax=350 MPa

1 2 3 4 Displacement [mm] 4 6 8 10 12 Force [N]

Sep = 10 J/m2

20 µm defect

Measurement Tmax=350 MPa Tmax=400 MPa Tmax=450 MPa Tmax=500 MPa

2.0 2.5 3.0 3.5 4.0 Displacement [mm] 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 crack length [mm]

Sep = 10 J/m2

20 µm defect

Measurement Tmax=350 MPa Tmax=400 MPa Tmax=450 MPa Tmax=500 MPa

23 / 29

Interface – Model Identification (3)

200 250 300 350 400 TMax [MPa] No defect 10 20 30 Separation energy

Sep [J/m2]

300 350 400 450 500 TMax [MPa] With 20 m defect

Fits force/displacement measurement Fits optical measurement

◮ Simulation for various:

◮ ΦSep (separation energy) ◮ TMax (crack initiation stress) ◮ With or without defects

◮ A suitable parameter set fits

◮ “Macro” measurements (Force/Displacement) ◮ “Micro” measurements (Crack length)

24 / 29

Outline Introduction Characterization of the copper layers Characterization of the Ceramic Layer Characterization of the Metal-Ceramic Interface Conclusion

25 / 29

Example of simulation results

Delaminated area after 100 cycles (-50/+250° C)

1 2 3 4 tcu/tcera 0.00 0.02 0.04 0.06 0.08 0.10 0.12 Fractured surface [mm²]

Cu thickness=500 µm Cu thickness=500 µm, with dimples

◮ Simulation predicts a strong effect of dimples ◮ Weakest configuration expected to be tCu = tCera ➜ Results compatible with existing data, especially for tCu >> tCera

26 / 29

Simulation of the behaviour of a DBC structure

◮ We identified models for

◮ Copper: behaviour very specific because of bonding process ◮ Ceramic: must take into account material grades ◮ Interface: innovative approach with identifications at macro and micro scales

◮ Theses models have been used for

◮ Evaluation of impact of stress-relaxation effects ◮ Identification of robust Cu/Al2O3/Cu configurations ◮ Evaluation of robustness to thermal cycling

➜ These simulations must be validated against measurements

27 / 29

Simulation of the behaviour of a DBC structure

◮ We identified models for

◮ Copper: behaviour very specific because of bonding process ◮ Ceramic: must take into account material grades ◮ Interface: innovative approach with identifications at macro and micro scales

◮ Theses models have been used for

◮ Evaluation of impact of stress-relaxation effects ◮ Identification of robust Cu/Al2O3/Cu configurations ◮ Evaluation of robustness to thermal cycling

➜ These simulations must be validated against measurements

27 / 29

Simulation of the behaviour of a DBC structure

◮ We identified models for

◮ Copper: behaviour very specific because of bonding process ◮ Ceramic: must take into account material grades ◮ Interface: innovative approach with identifications at macro and micro scales

◮ Theses models have been used for

◮ Evaluation of impact of stress-relaxation effects ◮ Identification of robust Cu/Al2O3/Cu configurations ◮ Evaluation of robustness to thermal cycling

➜ These simulations must be validated against measurements

27 / 29

Bibliography I

• J. Schulz-Harder, “Ceramic substrates and micro channel cooler,” in ECPE

Seminar: High Temperature Electronics and Thermal Management, (Nürnberg), nov 2006.

• A. Ben Kabaar, C. Buttay, O. Dezellus, R. Estevez, A. Gravouil, and L. Gremillard,

“Characterization of materials and their interfaces in a direct bonded copper substrate for power electronics applications,” Microelectronics Reliability, 2017.

• A. Ben Kaabar, Durabilité des assemblages métal céramique employés en

électronique de puissance. PhD thesis, 2015.

• J. Lemaitre, J.-L. Chaboche, and J. Lemaitre, Mechanics of Solid Materials.

CAMBRIDGE UNIV PR, 2002.

• T. J. Ahrens, Mineral physics and crystallography: a handbook of physical

constants. American Geophysical Union, 1995. P . P . Camanho and C. G. Dávila, “Mixed-mode decohesion finite elements for the simulation of delamination in composite materials,” tech. rep., NASA, 2002.

28 / 29

Thank you for your attention.

This work was supported through the grant SuMeCe (Institut Carnot I@L, Lyon). cyril.buttay@insa-lyon.fr

29 / 29