Polyimide Passivation Approaches on Double-Mesa Thyristors

2nd International Conference on Electrical and Electronics Engineering (ICEEE) and XI Conference on Electrical Engineering (CIE 2005) Mexico City, Mexico. September 7-9, 2005 Polyimide Passivation Approaches on Double-Mesa Thyristors Yasuhiro Matsumoto1, Emmanuel Saucedo2 1 2 Departamento de Ingeniería Eléctrica, CINVESTAV del IPN, Av. IPN 2508 Zacatenco, Mexico D.F. Departamento de Ingeniería de Proyectos, CUCEI, Universidad de Guadalajara, Zapopan, Jal. Mexico Phone (52) 55-5061-3783 Fax (52) 55-5061-3978 E-mail: [email protected] Abstract – Polyimide layers have been applied by a simple blade squeeze technique on double-mesa high voltage thyristor devices for its electrical passivation. Reverse and forward electrical blocking characteristics were measured as a function of polyimide application and curing processing conditions. During high temperature reverse bias reliability test, the thyristors showed appreciable leak-current increment, but after three day thermal process, the devices showed lower leakage currents than that devices fabricated by conventional glassivation technique. Mesa etched G: Gate K: Cathode (a) Substrate A: Anode Die form wafer Glass Keywords – Passivation, Polyimide, Thyristor 1. INTRODUCTION I (Current) Glass-based material have been employed during many years as a device junction and surface electrical isolation. Glass is one of the most stable electrical insulator for semiconductor device. However, processing glass induces several drawbacks due to its critical chemical and mechanical properties during the application process. In this sense, a manageable material and process steps are desirable. Polyimide has been considered as one of the promissory materials [1-4]. Among its properties: Low temperature curing process of about 350 oC; Exceptional thermal stability even at temperatures of 450 oC; high dielectric strength of 200KV/mm; Low Coefficient of Thermal Expansion; High chemical and solvent resistance. Their continuous quality improvement have achieved similar properties to replace different semiconductor device coatings. In this work, polyimide were applied by using blade method. This method consists of material direct application on wafer-level devices, using displacement of a simple-blade. Polyimide were applied and cured at different environment, temperature and time. Electrical evaluation before and after wafer dicing and molding are shown for the applied devices. 2. DEVICE DESCRIPTION Fig. 1a shows sort of typical thyristor cross section, indicating its electrical terminals. Fig. 1b shows the thyristor current-voltage forward-blocking characteristic indicating its breakover voltage (Vbo) and the correspondent leakcurrent with open gate terminal. Typical Vbo values in the conventional glassivated thyristor are about 1,000V while the leak-current at this point, less than 10 PAmp. IEEE Catalog Number: 05EX1097 ISBN: 0-7803-9230-2 0-7803-9230-2/05/$20.00 ©2005 IEEE. (b) I-V characteristics Cathode – grounded Anode + bias Leak-current V (volt) Vbo ~ 1,000 V Fig. 1. (a) Typical glassivated, double mesa thyristor cross section schematic with the definition of its electrical terminals and (b) correspondent forward-biased blocking characteristics. The leak-current is measured at the maximum breakover voltage Vbo. 3. EXPERIMENTAL DEVELOPMENT Fig. 2 shows an experimental flow chart, from waferlevel device to final passivated-die for the developed polyimide application method. The final step is an electrical evaluation after molded or encapsulated die devices. A. Procedure steps 1) Best polyimide curing condition were chosen using material provider’s recipe and by using cut and place basis. Annealing furnace and hot-plate were employed for the material curing process. Wafer cleaning process, material 223 Blade (Teflon or plastic) Polyimide application and evaluation for Thyristor Blade displacement direction Devices: wafer-level thyristor (with and without metal-contacts) Wafer Applied polyimide Material to test: polyimide --- Die 3 Surface preparation: acid / alkali etch; junction cleaning Material thickness Dicing-test: wirebondable-wafer Electrical characterization Die- and wire-bond pull- & shear-test Vbo Leak-current Molding Detected problems: walkout, channeling, low-Vbo 4. EXPERIMENTAL DETAILS A. Wafer cleaning, material application and curing As a starting step, the wafer-level processed devices were moat-etched using an acid solution. The acid-etch were performed in the order to separate each wafer-level devices for its electrical isolation and p-n junction exposure (mesaetch). Before polyimide application, wafers were cleaned using HF10% in water to remove surface oxide. The wafer surface and its streets must be cleaned adequately before polyimide application, because a possible impurity or oxide induces higher leak-current. Polyimide were applied to wire-bondable thyristors. Curing temperature, curing time are important parameters to achieve maximum adherence. Polyimide were cured on hotplate at 90 and 150oC for few seconds and then, annealed with temperature increment of 5oC to 10 oC/min, from 150 to 350oC. Approximately 30 min were required to cure completely at 350 oC in high purity nitrogen environment [5]. Vbo: Breakover voltage HTRB: High Temperature Reverse Bias HTRB test Detected problems: Leak-current, stability Fig.2 Block diagram characterization. for experimental flow and application procedure, curling temperature with different rising ratio and annealing time are emphasized. 2) Vbo measurements using needle probe on waferlevel devices were performed: Vbo, were measured with the applied dc bias between device K-A terminals. 3) Dicing test for polyimide applied wafers were performed; physical wafer observation for a possible chipping or fracture as well as dicing-blade properties were evaluated. 4) Separated die from the wafer were bonded to leadframes and molded for thermal-electrical evaluation by means of High Temperature Reverse Bias (HTRB). Dielevel devices were mounted to the correspondent commercial base TO-220 lead-frame. The HTRB process was performed at 85oC, 168hrs at reverse voltage of 320V for the device reliability test. Leak-current was considered as a main parameter under this experiment. Polyimide (Pyralin HD Microsystems) an enterprise of Hitachi Chemical and Du Pont Electronics were considered for the present experiment. IEEE Catalog Number: 05EX1097 ISBN: 0-7803-9230-2 Die 1 Fig. 3. “Blade-squeeze” procedure for polyimide application on wafer. The applied material covers completely the mesaetched space (wafer streets). Material application and curing conditions: temperature, time,environment Material adherence Die 2 B. Polyimide Thickness and adherence The blade pressure on the wafer and its displacement velocity are important parameters during polyimide application to control polyimide thickness. After polyimide curing process, the remained thickness were in the order of 10 Pm, while for thicker region were about 15ҏPm. The calculated weight loss for polyimide after 150oC, 5min annealing were evaluated in the order of 50%. For further annealing at 350oC during 30min, the total weight loss were about 86% as the provider indicates. In the order to obtain thicker polyimide layer on device-wafer, polyimide were applied twice or three times. For each application, a soft-curing were carried out at 90 to 150 oC on the hot-plate. However, thicker layer does not always resulted in a higher Vbo, due to channeling degradation as is shown in Fig. 4a. 224 voltage. The device-polyimide interfaces seems to be stabilized after this high-bias application. (a) I (Current) C. Wafer dicing and molding K-A terminal Characteristics Separation in dies from the wafer by dicing have been performed for wire-bondable devices cured with polyimide. As a result, there were no presence of fractures or chipping on silicon dies and the polyimide layer remained stable. At the diced die-edge region, some small peeling were observed, but does not affect p-n junction region as revealed by using a simple microscope. Also dicing blade were not affected after the wafer dicing, because may be the polyimide-layer was thin enough of about 10Pm to 15Pm. Several dies have been mounted on lead-frames, and wirebonded. A standard pull- and shear-test have been made for die-bonded devices and resulted in an acceptable adherence for the tested samples. Channeling effect with very high leak current ~mA order V (Volt) (b) I (Current) D. High Temperature Reverse Bias (HTRB) characterization Finally an electrical characterization were performed after molding the thyristor. Several thyristor samples were used for HTRB test. The test duration were 168 hrs, and the reverse biased leak-current were measured at the temperature of 85oC. Several samples were compared with the conventional glass processed wire-bondable thyristor. Most of the samples showed high leak-current at the starting reverse biasing condition as is shown in Fig. 5a. However, within some hours, the leak was decreased considerably, and after 4 days of continuous test, all the samples turned to stable and relatively small leak-current. From this test, we have observed that the temperature induces electrical insulation for the device in contact with polyimide, reducing the leak-current. This means an instability at the polyimidedevice interface, may be due to the curing process deficiency or environmental humidity absorption during molding process. Conventional glass passivated wire-bondable thyristor shows almost stable leak curve during the performed HTBR test as is shown in Fig. 5b. At the beginning, the leakcurrent at 85oC were 9~18 µA increasing to 16~40 µA at second day but after 7 days stabilizes at 17~34 ҏµA. The polyimide-cured devices diverged from 2 to 367µA at the initial stage of same test, 3~70ҏ µA on second day and after the week, 3~7ҏµA, as can see in Fig. 5a. The final leakcurrent of the polyimide cured devices, showe almost one order less leak-current than that of standard ones. Walk-out phenomena” Voltage displacement while biasing Vbo1 Vbo2 V (Volt) Fig. 4a shows channeling effect, a very high leak-current due to a possible silicon-polyimide interfacial current. Fig. 4b describes the “walk-out” phenomena. Also very thick polyimide application induces peeling during curing process, may be due to the stress induction. The used polyimide, requires nitrogen environment annealing, because otherwise (air-anneal) resulted in a poor Vbo. A slow temperature ramp of ~5oC/min from 150 oC to 350oC/min, had better polyimide adherence and device electrical properties than at ~10 oC/min ramp. The obtained maximum Vbo were around 880V with leak-current of less than 20ҏPA. Even though, some of the devices showed channeling effect with about 1mA leakcurrent magnitude as is shown in Fig. 4a. This phenomena is similar to that of thicker polyimide applied devices. Depending of polyimide application process, some of the devices achieved 950 to 1,000V, but after walking-out phenomena as is shown in Fig. 4b, which could be interpreted as an electric field-induced self-curing phenomena. After this phenomena, the measured leakcurrent were less than 10ҏPA at the mentioned applied IEEE Catalog Number: 05EX1097 ISBN: 0-7803-9230-2 5. CONCLUSIONS In general, polyimide resin demonstrated excellent properties as an electrical isolator. Forward blocking voltage of 880V were achieved for thyristors. However, polyimide 225 been observed during HTRB-test before thyristor heatrelated stabilization. Further study may have done in the order to clarify polyimide-related leak-current components as well as their stability under thermal action. HTRB-IROM-Test (85C-320V) MCR12M-Polyimide 1.00E-03 24 24 24 24 25 27 28 29 30 31 Current (A) 1.00E-04 REFERENCES (a) [1] Optimization of photosensitive polyimide process for cost effective packaging, Ultratech Stepper, Inc. San Jose CA. Surface Mount Tech. Seminar, SPIE1996. [2] L.B.Pothman, “Properties of Thin Polyimide Films “, J. Electrochem. Soc.: Solid State Sc. And Tech. Vol 127, October 1980. pp 2216-2220 [3] Handbook of Polymer Coatings for Electronics, Chemistry, Technology and Applications (Second Edition) by James J.Licari; Noyes Publications (1990), pp55-65. [4] Plastic for Electronics; Materials, Properties and Design Applications, William M.Alvino; McGraw-Hill, Inc, pp155-161. [5] Pyraline Polyimide Coatings for Electronics PI2610 Series HD MicroSystems July, 1998, or http: // 209.238.157.215 / 9PDF / product %20pdf/PI2610.pdf 1.00E-05 1.00E-06 One day Six days 1.00E-07 Tim e HTRB Test, (85°C-320V) Conventional Glass 1.00E-03 30 31 03 04 05 06 Current (A) 1.00E-04 1.00E-05 Seven days 1.00E-06 (b) 1.00E-07 Tim e Fig. 5. HTRB reliability test for both, (a) polyimide and (b) conventional glass-passivated thyristors. The test elapsed days are indicated in the upper-side of the figure. As can see, the time-scales are not linear. application condition and the device surface has to be strictly controlled. Mostly surface cleaning, environment are critic. Too thick polyimide application provokes peeling and channeling. The mentioned effect is probably due to the strong tensile action at the polyimide/silicon interface. For the used material, Pyralin HD of Microsystems, it required a nitrogen controlled environment annealing due to its relatively higher curing temperature of 350 oC compared to other commercial based polyimide materials. No wafer-dicing problem were detected. Wafer with applied polyimide were cut with no mechanical problem. No damage or deformation were detected for dicing-blade after cutting wafers with polyimide. No adherence problem were detected for polyimide on wafer streets. High reverse leak-current characteristics have IEEE Catalog Number: 05EX1097 ISBN: 0-7803-9230-2 226