- Email: [email protected]
A new method to characterize the thermomechanical response of multilayered structures in power electronics
Microelectronics Reliability 46 (2006) 1844–1847 www.elsevier.com/locate/microrel
A new method to characterize the thermomechanical response of multilayered structures in power electronics P. Zimpricha*, T. Lichtb, B. Weissa a
University Vienna, Institute of Materials Physics, Faculty of Physics, Vienna , Austria b Infineon Technologies AG, 59581 Warstein, Germany
Abstract To record the coefficient of thermal expansion (CTE) of metallized surfaces on Si- substrates with different thickness ratios, a new testing method is introduced. Laseroptical sensors based on the speckle correlation method were applied to determine non-contacting thermal strain values of multilayered structures with high strain resolution. This technique is usually used for the determination of mechanical properties of freestanding foils and wires performing tensile tests. It could be shown, that the thickness of substrate clearly influences the thermal expansion coefficient of the metallization layer.
1. Introduction Power semiconductor modules with semiconductor based IGBTs (insulated gate bipolar transistors) are very convenient devices for controlling current and frequency and the application of power devices like IGBTs and diodes has dramatically increased for industrial and traction applications for the last decade [1]. These power devices are vertical components. This means, the current goes vertical through the device. Due to the development of new chip generations the size and the thickness of silicon base material is reduced, which has a high influence on the electrical parameters like power losses. For the optimization of power losses and the switching parameters the thickness of the devices is reduced by every new chip generation and will be further reduced in the future. Typical thickness for todays power chips are from 70 µm up to some hundreds of micrometers. On both sides of the device metallizations are deposited on the surface. This “sandwich” of metal – Si – metal is a combination of different materials and therefore different CTEs.
* P. Zimprich ([email protected]) Strudlhofgasse 4, 1090 Vienna, Austria Office: +43-1-4277-51335, Fax: +43-1-4277-9513 0026-2714/$ - see front matter Ó 2006 Published by Elsevier Ltd. doi:10.1016/j.microrel.2006.07.067
By the reduction of thickness of the (Si) semiconductor material the influence of the metals on both sides increase and gets a more dominant factor, because the thickness of the metallization remains constant, but not the thickness of the substrate. The today typical joining techniques used for power devices are heavy wire bonding on the top side and soldering on the bottom side. The reliability of these joining techniques are always influenced by the differences in the CTE between the connected partners which lead to thermal stresses resulting in crack initiation and propagation during operation [2,3]. As the exact value of the CTE for a “final” power chip (metal/Si/metal) is very important for the interpretation and simulation of reliability test results, a technique is needed with the opportunity to determine CTE values, which are necessary to calibrate simulation models and to verify the influence on reliability testing. In manufacturing of IGBTs the knowledge of the thermal behavior of metallized substrates plays an important role for functionality and reliability design. Because of the large differences of thermal expansion coefficients between the materials involved (e.g. 24 ppm/°C for Al, 2 - 3 ppm/°C for Si) one important information is the influence of a thick substrate (Si) on the thermal expansion behavior of a thin metallization layer
P. Zimprich et al. / Microelectronics Reliability 46 (2006) 1844–1847
H = 'l/l = (' surf II – ' surf I)/ l
(Al), when the ratio of substrate thickness to layer thickness is varied.
For measurements of the thermal strain of thin multilayered structures a laser speckle based dilatometer (LSBD) was designed and consists of a closed, temperature controlled furnace chamber adapted to a laser speckle extensometer [4]. The technical arrangement consists of an illuminating and two displacement recording systems. The illumination consists of two collimated laser diodes with maximum power output of 5mW and a wavelength of 680 nm providing laser spots each of 1 mm in diameter on the specimen surface. The images of these surface regions obtained by lenses are recorded by CCD cameras. Due to the natural surface roughness of the sample and the scattering at the lens aperture an interference (speckle) pattern in the image occurs which is characteristic for a particular surface region. From the shifts of these patterns in subsequent measurements the surface shifts can be deduced. The displacement recording is performed by two lenses (separated by the base length lo) with fixed focal length and standard chargecoupled device cameras (CCD cameras) feeding the signals of the images into a personal computer-based frame grabber, where the signal processing occurs, see Fig. 1.
(1)
can be determined. The achievable strain resolution is of the order 10-5 depending on the selected baselength of a few millimeters. Since thermal loading of specimens may cause surface oxidation and plastic deformation, distortions of the speckle patterns occur resulting in decorrelation effects. To overcome such effects a repetitive reinitialization of the image acquisition system is performed >5]. For faster testing a similar optical system using Fourier –filtering in combination with a CCD line camera was applied for measurement rates up to 40 pics/s [6]. Calibration tests determining the thermal expansion of Copper Standard Reference Material NIST SRM 736 (rod material with 6 mm diameter) showed high accuracy and reliability of the measured strain values (Fig.2). Successive runs show high reproducibility (1.5 %) of the thermal expansion results and very good agreement with the reference data. Up to 250° C the deviation of the measured thermal strains is lower than 7% compared to the reference material.
Thermal strain, ppm
2. Experimental
1845
4000 3500 3000 2500 2000 1500 1000 500 0 0
50
100
150
200
250
Tem perature °C
Fig. 2. Thermal expansion of NIST Copper SRM 736 (black dots) and three successive runs with the laser speckle dilatometer.
Fig. 1. The optical arrangement of the laserspeckle correlation system. In order to determine thermal strain H of the specimen the two surface elements of defined distance l are tracked as a function of temperature. By determining the surface element displacements ' surf, taking their spatial difference 'l and dividing by a selectable baselength l the strain value (see Eq. 1).
Measurements of thermal expansion of a thin Almetallization layer of 3 µm thickness were carried out on Si- substrates with 300 and 70 µm thickness for different industrial chips as shown in Fig.3, which were prepared from the module after removing the bonding wires. The typical chip size is 10x10 mm.
1846
P. Zimprich et al. / Microelectronics Reliability 46 (2006) 1844–1847
3. Results and Discussion As indicated in Fig. 5 and 6, the coefficient of thermal expansion of a thin metallization layer is clearly influenced by the thickness of the substrate. The thermal expansion coefficient of the aluminum layer on substrates with 70 µm shows almost only minor reduction compared with the CTE known from bulk aluminum (Fig.5). 10 mm
Fig. 3. Typical chip-configuration of an IGBT module. These samples were heated up from RT to 150°C maximum and temperature of the specimen was determined within an error of +/- 0,5°C. The heating rate was set to less than 1 degree per minute. The heating chamber was designed also for low-temperature measurements down to about -100°C using a liquid nitrogen temperature buffer which can be adjusted to achieve different rates for heating up the samples. The sample chamber can be floated with argon gas to avoid surface oxidation of the samples which is especially important beyond 200°C. With this setup, like shown in Fig. 4, thermal strain data of thin multilayered structures can be measured with a reproducibility of 1.5 % between different runs on one sample. For typical measurements three successive runs were performed. Specimen chamber
Fig. 5. CTE behaviour of three Si chips (70 µm thick) with 3µm Al-layer. The dotted line indicates the CTE value of bulk aluminum.
Laser sensor Laser diodes
Argon floating
Thermoelements
Heating stage
Position unit
Fig. 4. Practical realization of the laser speckle dilatometer (LSBD).
Fig. 6. CTE behavior of a 3µm Al-layer on a 300 µm Si Chip substrate (a). For comparison the measured bulk CTE of Al is indicated (b). The observed decrease of CTE with increasing temperature as shown in Fig.5 could be attributed to a possible bowing of the thin chips during the heating period. The observed differences in CTE behavior between the three different chips may be a result of their dissimilar state of internal stresses and bowing, because of slightly variations during production of the chips. For the thick substrate (300 µm) the aluminum layer was clearly reduced in CTE down to about 12 ppm/°C, see Fig.6. It can be assumed, that the thicker substrate restricts the thin layer in its thermal expansion because of higher stiffness
P. Zimprich et al. / Microelectronics Reliability 46 (2006) 1844–1847
and therefore more pronounced mechanical constraint effects lead to a reduced CTE value on the aluminum surface. From the theory of multilayered structures it can be shown, that the influence of a thin layer deposited on a thick substrate will become important for layer thickness of about 8% of the substrate thickness [7]. This has interesting implications for the understanding of thermal driven processes during fabrication of these layers and also for the simulation of multilayer behavior. 4. Conclusions A new laser speckle based dilatometer (LSBD), originally designed for strain measurements of thin films and wires was successfully applied for non-contacting CTE measurements of thin metallic layers on relatively thick and brittle substrates. The non-contacting strain sensor, based on the laser speckle correlation technique, is due to the simple optical arrangement easily adaptable to a temperature chamber and characterized by high strain resolution in the order of 2 x 10-5. This method achieves sufficient degree of accuracy and precision for CTE evaluation in the temperature range -100°C up to 300°C. Various applications in material science like CTE evaluation of metals, polymers, ceramic components and multilayer-structures without the need of any surface preparation and on non-standardized geometries. With this system it could be shown, that the thermal expansion behavior of metallization layers is significantly influenced by the thickness of the substrate. Holding the metallization thickness constant, an increasing substrate thickness results in a reduction of the observed CTE values of the aluminum metallization layer. The knowledge of this effectively occurring thermal expansion values are crucial for processes like ultrasonic bonding and the estimation of joint reliability. With this new method is seems possible to improve the understanding and simulation of thermal mismatch problems in multilayered structures, especially in the case of thin metallization films and printed wiring boards.
Acknowledgement The authors thank the Austrian National Science Foundation (P 14732 TEC) for financial support. The help and continuous support from Prof. B. Zagar is gratefully acknowledged.
References [1] G. Lefranc, B. Weiss, C. Klos, J. Dick, G. Khatibi, H. Berg Aluminum bond-wire properties after 1 billion mechanical cycles, Microelectronics Reliability, Vol.43, (2003), pp.1833-1838. [2] H. Berg, E. Wolfgang, Advanced IGBT modules for railway traction testing for thermal fatigue effects due to traction cycles, Microelectronics reliability, Vol.38, (1998), pp. 1353-1359 [3] S. Ramminger, P. Türkes, G. Wachutka, Crack mechanism in wire bonding joints, Microelectronic Reliability, Vol. 38, (1998), pp. 1301-1305 [4] B. Weiss, P. Zimprich and G. Khatibi, Thermomechanical properties of small-scaled structures using laser techniques, Int. Workshop on Materials Issues for MEMS/MST Devices, ATEM`03, JSME-MMD (2003), pp. 22-27. [5] M. Anwander, B. Zagar, B. Weiss and H. Weiss, Non contacting strain measurements at high temperatures by laser speckle technique. Exp. Mech., Vol. 40, March (2000), pp. 1-8 [6] B. Zagar, B. Trummer, M. Anwander, B. Weiss and H. Weiss, Ein schnell messendes Laser-Speckle-Extensometer zur Bestimmung von statischen und dynamischen Materialparametern, Elektrotechnik und Informationstechnik, 117 Jg. (2000) 4, pp. 273278 [7] Brezina S., Anwendung eines Laserinterferometers zur berührungslosen Messung der lokalen Wärmeausdehnung von verschiedenen Materialien, Diploma Thesis, University Vienna, Austria, 1996
1847
A new method to characterize the thermomechanical response of multilayered structures in power electronics P. Zimpricha*, T. Lichtb, B. Weissa a
University Vienna, Institute of Materials Physics, Faculty of Physics, Vienna , Austria b Infineon Technologies AG, 59581 Warstein, Germany
Abstract To record the coefficient of thermal expansion (CTE) of metallized surfaces on Si- substrates with different thickness ratios, a new testing method is introduced. Laseroptical sensors based on the speckle correlation method were applied to determine non-contacting thermal strain values of multilayered structures with high strain resolution. This technique is usually used for the determination of mechanical properties of freestanding foils and wires performing tensile tests. It could be shown, that the thickness of substrate clearly influences the thermal expansion coefficient of the metallization layer.
1. Introduction Power semiconductor modules with semiconductor based IGBTs (insulated gate bipolar transistors) are very convenient devices for controlling current and frequency and the application of power devices like IGBTs and diodes has dramatically increased for industrial and traction applications for the last decade [1]. These power devices are vertical components. This means, the current goes vertical through the device. Due to the development of new chip generations the size and the thickness of silicon base material is reduced, which has a high influence on the electrical parameters like power losses. For the optimization of power losses and the switching parameters the thickness of the devices is reduced by every new chip generation and will be further reduced in the future. Typical thickness for todays power chips are from 70 µm up to some hundreds of micrometers. On both sides of the device metallizations are deposited on the surface. This “sandwich” of metal – Si – metal is a combination of different materials and therefore different CTEs.
* P. Zimprich ([email protected]) Strudlhofgasse 4, 1090 Vienna, Austria Office: +43-1-4277-51335, Fax: +43-1-4277-9513 0026-2714/$ - see front matter Ó 2006 Published by Elsevier Ltd. doi:10.1016/j.microrel.2006.07.067
By the reduction of thickness of the (Si) semiconductor material the influence of the metals on both sides increase and gets a more dominant factor, because the thickness of the metallization remains constant, but not the thickness of the substrate. The today typical joining techniques used for power devices are heavy wire bonding on the top side and soldering on the bottom side. The reliability of these joining techniques are always influenced by the differences in the CTE between the connected partners which lead to thermal stresses resulting in crack initiation and propagation during operation [2,3]. As the exact value of the CTE for a “final” power chip (metal/Si/metal) is very important for the interpretation and simulation of reliability test results, a technique is needed with the opportunity to determine CTE values, which are necessary to calibrate simulation models and to verify the influence on reliability testing. In manufacturing of IGBTs the knowledge of the thermal behavior of metallized substrates plays an important role for functionality and reliability design. Because of the large differences of thermal expansion coefficients between the materials involved (e.g. 24 ppm/°C for Al, 2 - 3 ppm/°C for Si) one important information is the influence of a thick substrate (Si) on the thermal expansion behavior of a thin metallization layer
P. Zimprich et al. / Microelectronics Reliability 46 (2006) 1844–1847
H = 'l/l = (' surf II – ' surf I)/ l
(Al), when the ratio of substrate thickness to layer thickness is varied.
For measurements of the thermal strain of thin multilayered structures a laser speckle based dilatometer (LSBD) was designed and consists of a closed, temperature controlled furnace chamber adapted to a laser speckle extensometer [4]. The technical arrangement consists of an illuminating and two displacement recording systems. The illumination consists of two collimated laser diodes with maximum power output of 5mW and a wavelength of 680 nm providing laser spots each of 1 mm in diameter on the specimen surface. The images of these surface regions obtained by lenses are recorded by CCD cameras. Due to the natural surface roughness of the sample and the scattering at the lens aperture an interference (speckle) pattern in the image occurs which is characteristic for a particular surface region. From the shifts of these patterns in subsequent measurements the surface shifts can be deduced. The displacement recording is performed by two lenses (separated by the base length lo) with fixed focal length and standard chargecoupled device cameras (CCD cameras) feeding the signals of the images into a personal computer-based frame grabber, where the signal processing occurs, see Fig. 1.
(1)
can be determined. The achievable strain resolution is of the order 10-5 depending on the selected baselength of a few millimeters. Since thermal loading of specimens may cause surface oxidation and plastic deformation, distortions of the speckle patterns occur resulting in decorrelation effects. To overcome such effects a repetitive reinitialization of the image acquisition system is performed >5]. For faster testing a similar optical system using Fourier –filtering in combination with a CCD line camera was applied for measurement rates up to 40 pics/s [6]. Calibration tests determining the thermal expansion of Copper Standard Reference Material NIST SRM 736 (rod material with 6 mm diameter) showed high accuracy and reliability of the measured strain values (Fig.2). Successive runs show high reproducibility (1.5 %) of the thermal expansion results and very good agreement with the reference data. Up to 250° C the deviation of the measured thermal strains is lower than 7% compared to the reference material.
Thermal strain, ppm
2. Experimental
1845
4000 3500 3000 2500 2000 1500 1000 500 0 0
50
100
150
200
250
Tem perature °C
Fig. 2. Thermal expansion of NIST Copper SRM 736 (black dots) and three successive runs with the laser speckle dilatometer.
Fig. 1. The optical arrangement of the laserspeckle correlation system. In order to determine thermal strain H of the specimen the two surface elements of defined distance l are tracked as a function of temperature. By determining the surface element displacements ' surf, taking their spatial difference 'l and dividing by a selectable baselength l the strain value (see Eq. 1).
Measurements of thermal expansion of a thin Almetallization layer of 3 µm thickness were carried out on Si- substrates with 300 and 70 µm thickness for different industrial chips as shown in Fig.3, which were prepared from the module after removing the bonding wires. The typical chip size is 10x10 mm.
1846
P. Zimprich et al. / Microelectronics Reliability 46 (2006) 1844–1847
3. Results and Discussion As indicated in Fig. 5 and 6, the coefficient of thermal expansion of a thin metallization layer is clearly influenced by the thickness of the substrate. The thermal expansion coefficient of the aluminum layer on substrates with 70 µm shows almost only minor reduction compared with the CTE known from bulk aluminum (Fig.5). 10 mm
Fig. 3. Typical chip-configuration of an IGBT module. These samples were heated up from RT to 150°C maximum and temperature of the specimen was determined within an error of +/- 0,5°C. The heating rate was set to less than 1 degree per minute. The heating chamber was designed also for low-temperature measurements down to about -100°C using a liquid nitrogen temperature buffer which can be adjusted to achieve different rates for heating up the samples. The sample chamber can be floated with argon gas to avoid surface oxidation of the samples which is especially important beyond 200°C. With this setup, like shown in Fig. 4, thermal strain data of thin multilayered structures can be measured with a reproducibility of 1.5 % between different runs on one sample. For typical measurements three successive runs were performed. Specimen chamber
Fig. 5. CTE behaviour of three Si chips (70 µm thick) with 3µm Al-layer. The dotted line indicates the CTE value of bulk aluminum.
Laser sensor Laser diodes
Argon floating
Thermoelements
Heating stage
Position unit
Fig. 4. Practical realization of the laser speckle dilatometer (LSBD).
Fig. 6. CTE behavior of a 3µm Al-layer on a 300 µm Si Chip substrate (a). For comparison the measured bulk CTE of Al is indicated (b). The observed decrease of CTE with increasing temperature as shown in Fig.5 could be attributed to a possible bowing of the thin chips during the heating period. The observed differences in CTE behavior between the three different chips may be a result of their dissimilar state of internal stresses and bowing, because of slightly variations during production of the chips. For the thick substrate (300 µm) the aluminum layer was clearly reduced in CTE down to about 12 ppm/°C, see Fig.6. It can be assumed, that the thicker substrate restricts the thin layer in its thermal expansion because of higher stiffness
P. Zimprich et al. / Microelectronics Reliability 46 (2006) 1844–1847
and therefore more pronounced mechanical constraint effects lead to a reduced CTE value on the aluminum surface. From the theory of multilayered structures it can be shown, that the influence of a thin layer deposited on a thick substrate will become important for layer thickness of about 8% of the substrate thickness [7]. This has interesting implications for the understanding of thermal driven processes during fabrication of these layers and also for the simulation of multilayer behavior. 4. Conclusions A new laser speckle based dilatometer (LSBD), originally designed for strain measurements of thin films and wires was successfully applied for non-contacting CTE measurements of thin metallic layers on relatively thick and brittle substrates. The non-contacting strain sensor, based on the laser speckle correlation technique, is due to the simple optical arrangement easily adaptable to a temperature chamber and characterized by high strain resolution in the order of 2 x 10-5. This method achieves sufficient degree of accuracy and precision for CTE evaluation in the temperature range -100°C up to 300°C. Various applications in material science like CTE evaluation of metals, polymers, ceramic components and multilayer-structures without the need of any surface preparation and on non-standardized geometries. With this system it could be shown, that the thermal expansion behavior of metallization layers is significantly influenced by the thickness of the substrate. Holding the metallization thickness constant, an increasing substrate thickness results in a reduction of the observed CTE values of the aluminum metallization layer. The knowledge of this effectively occurring thermal expansion values are crucial for processes like ultrasonic bonding and the estimation of joint reliability. With this new method is seems possible to improve the understanding and simulation of thermal mismatch problems in multilayered structures, especially in the case of thin metallization films and printed wiring boards.
Acknowledgement The authors thank the Austrian National Science Foundation (P 14732 TEC) for financial support. The help and continuous support from Prof. B. Zagar is gratefully acknowledged.
References [1] G. Lefranc, B. Weiss, C. Klos, J. Dick, G. Khatibi, H. Berg Aluminum bond-wire properties after 1 billion mechanical cycles, Microelectronics Reliability, Vol.43, (2003), pp.1833-1838. [2] H. Berg, E. Wolfgang, Advanced IGBT modules for railway traction testing for thermal fatigue effects due to traction cycles, Microelectronics reliability, Vol.38, (1998), pp. 1353-1359 [3] S. Ramminger, P. Türkes, G. Wachutka, Crack mechanism in wire bonding joints, Microelectronic Reliability, Vol. 38, (1998), pp. 1301-1305 [4] B. Weiss, P. Zimprich and G. Khatibi, Thermomechanical properties of small-scaled structures using laser techniques, Int. Workshop on Materials Issues for MEMS/MST Devices, ATEM`03, JSME-MMD (2003), pp. 22-27. [5] M. Anwander, B. Zagar, B. Weiss and H. Weiss, Non contacting strain measurements at high temperatures by laser speckle technique. Exp. Mech., Vol. 40, March (2000), pp. 1-8 [6] B. Zagar, B. Trummer, M. Anwander, B. Weiss and H. Weiss, Ein schnell messendes Laser-Speckle-Extensometer zur Bestimmung von statischen und dynamischen Materialparametern, Elektrotechnik und Informationstechnik, 117 Jg. (2000) 4, pp. 273278 [7] Brezina S., Anwendung eines Laserinterferometers zur berührungslosen Messung der lokalen Wärmeausdehnung von verschiedenen Materialien, Diploma Thesis, University Vienna, Austria, 1996
1847