Flip-chip wire bondless power device

Abstract

A flip-chip wire bondless power device and method for using a two sided contact bare die power device as a single-connection-level power device. The device uses a top pad solder ball array for connecting a top pad electrically connected to the top contact of the bare die power device and a bottom pad solder ball array for connecting a bottom pad that is electrically through an electrically conductive bottom pad connector that is electrically connected to the bottom contact of the bare die power device using an electrically conductive die-attach material, the top pad and bottom pad, and thereby the top pad solder ball array and the bottom pad solder ball array are planar for flip chip mounting. A trench can be formed between the top pad and bottom pad for isolation and insulation purposes. A method of assembling a flip-chip wire bondless power device is also provided.

Claims

1. A bare die converted to flip-chip power apparatus, the apparatus comprising: a bare die package with first side contacts at the first level and second side contacts at the second level, the bare die package including a silicon carbide power semiconductor device; a solderable plating applied to the second side contacts forming a first die pad; an electrically conductive mounting body defining a first mounting pad at a first level and a second mounting pad at the second level, the second mounting pad being solderable; the electrically conductive pad mounting body including an electrically conductive path from the first mounting pad to the second mounting pad; a die attach material electrically connecting the first side contacts to the first mounting pad.

2. The apparatus of claim 1, further comprising: electrically conductive pad mounting body extending from the bare die package to define a trench between the first mounting pad and the second mounting pad.

3. The apparatus of claim 1, further comprising: a solder mask is applied to the first mounting pad.

4. The apparatus of claim 1, further comprising: a solder mask applied to the second mounting pad.

5. The apparatus of claim 1, further comprising: the electrically conductive pad mounting body made out of metal.

Description

BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

(1) In the following drawings, which form a part of the specification and which are to be construed in conjunction therewith, and in which like reference numerals have been employed throughout wherever possible to indicate like parts in the various views:

(2) FIG. 1A is a schematic showing the flip-chip power device according to some embodiments.

(3) FIG. 1B is a schematic showing the inputs and outputs of the flip-chip power device implemented in a generic power electronics system.

(4) FIG. 1C is a schematic showing how top and bottom heat removal is improved by the flip-chip power device.

(5) FIG. 1D is a schematic comparison of the size of a TO-247 package to a Flip chip package of the present invention.

(6) FIG. 2 is a schematic and flow chart showing exemplary methods to assemble a flip-chip power device using a bare die power device according to some embodiments.

(7) FIG. 3 is a flow chart showing a process flow for electroless plating of a bare power device according to some embodiments.

(8) FIG. 4. is a flow chart showing a process flow for electroless plating of a bare power device according to some embodiments.

(9) FIG. 5 is an optical microscope image showing the electroless plated metal layers on the original aluminum pads of a bare die power device.

(10) FIG. 6 is a graph of results of a ball shear test conducted on randomly picked solder balls bonded to a deposited nickel metallization layer.

(11) FIGS. 7A-7B are photographs of samples secured in place by a Kapton stencil prior to reflow and samples including a pair of finished samples in the bottom right corner.

(12) FIG. 8 are photographs showing optimization steps performed to ascertain a proper exposure time.

(13) FIG. 9 is a photograph showing a solder-ball bumped flip-chip power device.

(14) FIG. 10A shows a patterned substrate in the form of a half bridge.

(15) FIG. 10B shows a module with busbars and gate connection headers

(16) FIG. 11 is a photograph showing three modules tested for comparison.

(17) FIG. 12A is a schematic depicting a power loop and gate loop in a conventional lateral loop design.

(18) FIG. 12B is a schematic depicting a reduced power loop area using a vertical loop layout.

(19) FIG. 13 is a graph of results of a finite element simulation contrasting the critical loop inductances of the flip-chip module and a conventional wire-bonded module.

(20) FIG. 14A shows a schematic of a standard double pulse test setup.

(21) FIG. 14B shows the test setup used for the study.

(22) FIG. 15A is a graph depicting the measured turn-off waveforms of the three modules.

(23) FIG. 15B is a graph depicting a comparison between the estimated and measured loop parasitics.

(24) FIG. 15C is a graph depicting the turn-on waveforms of the three modules.

(25) FIG. 15D is a graph depicting inductor current of the flip-chip module.

(26) FIG. 16A is a graph showing the variation of power loop inductance with the loop area.

(27) FIG. 16B is a graph showing the percentage overshoot as a function of parasitic power loop inductance.

(28) FIG. 17A is the measured waveform showing the gate turn-off of the three modules.

(29) FIG. 17B is the gate turn-on voltages of the three modules.

(30) FIG. 18A shows flip-chip Schottky diodes and a wire bonded control sample.

(31) FIG. 18B is a graph of the forward I-V characteristics of the two modules.

(32) FIG. 18C shows the turn-off waveform comparison between the wire bonded and flip-chip module.

(33) FIG. 18D shows the turn-on waveform comparison of the two modules, and

(34) FIG. 18E shows the low-side diode voltage and inductor current plotted to demonstrate operation at 450 VDC, 15 A.

DETAILED DESCRIPTION OF THE INVENTION

(35) As shown in FIG. 1A of the drawings, one exemplary embodiment of the present invention is generally shown as a bare die converted to flip-chip power device 19 according to one embodiment of the present invention. Flip-chip wire bondless power device packages and methods of assembly are described herein which may provide one or more advantages over prior devices. Packages and methods herein leverage advantages that SiC technology has to offer, and does so using an economical manufacturing process.

(36) In a preferred embodiment, a power device in bare die form 10 is re-engineered into a flip-chip package through a series of steps. It is relatively simple to flip-chip bond a device with pads on the same side of the wafer. The situation is significantly more complicated for a silicon carbide power device which has electrical contacts on either side of the wafer. The method described herein provides a viable solution to this issue. FIG. 1A shows a schematic representation of the flip-chip power device 19. The bottom electrical contact 101 of the bare die power device 10 is routed to the top surface at the top surface second contact 103 using a machined and electrically conductive bottom pad connector 11. The die 10 is attached to this connector at the lower level pad 104 using an electrically conductive die-attach material 105. The clearance between the surfaces of the die 10 and the bottom connector 11 is determined by the maximum voltage rating of the bare die device 10. To further increase the creepage distance between these contacts, a trench 106 is milled between the two planar terminal surfaces 103, 107. As in the case of a MOSFET or IGBT, the gate pad 108 and source pads 109 are in very close proximity on the top surface 107 of the die 10. It is necessary to provide additional clearance between the pads to obtain a reliable flip-chip connection. Insufficient spatial clearance may result in the formation of conductive bridges between dissimilar pads leading to system failure. This is achieved by using a solder mask 12 applied to the surface of the package and patterning designated pads for solder ball 13, 14 attachment.

(37) Solder ball arrays have been extensively used in the manufacturing of high-density, high-speed, and highly reliable electronics at low voltages in the past. This package and method leverages this advantage as it applies to power electronics. Multiple solder spheres on the same pad builds redundancy in the system, thus enhancing the reliability. Solder ball arrays are also self-planarizing and self-centering due to the property of surface tension of molten solder. This makes for a highly repeatable manufacturing and assembly process. Top pad solder spheres 13 and bottom pad solder spheres 14 are applied to the respective top and bottom pads. The solder ball array also enables short and low-inductance signal paths. Vertical interconnections are made possible with this device to form 3D stacked power electronics modules.

(38) A preferred embodiment is depicted in the schematic shown in FIG. 1B, where the gate driver circuits may be populated on the top side of an interposer, with the power devices 15 flip-chip bonded to the bottom tier. FIG. 1C depicts a schematic showing how heat may be removed in an exemplary configuration. Heat may be removed from the bottom surface of the metallic connector 16 when the power device 15 is bonded to substrate 17 using top pad solder spheres 13 and bottom pad solder spheres 14 with underfill 18. In high-current applications which do not require ultra-fast switching and ultra-low inductances, heat may be efficiently removed from either side of the package by employing double-sided cooling methods as illustrated. This is impossible in previous state-of-the-art wire bonded silicon carbide power devices and modules.

(39) The resulting package using a preferred method of assembly occupies a 14.24 times smaller footprint as compared with a conventional discrete TO-247 device as depicted in the photograph presented in FIG. 1D. This demonstrates the potential of realizing unprecedented power levels within a given volume.

(40) The process flow for a preferred embodiment wherein a commercially available bare die MOSFET is reconfigured to a flip-chip MOSFET is described with reference to the flowchart shown in FIG. 2. The top-side metallization of commercially available bare die SiC power devices is aluminum. This is to facilitate the aluminum wire bonding process. The bottom side contacts, however, are always solderable to support the die attachment process. The only other viable method to achieve a metallurgical bond with aluminum is by using conductive epoxy. However, this runs the risk of increasing the contact resistance. The favorable effect of low parasitic flipchip interconnections may be completely offset by using an electrically, and thermally, inferior attachment material. Hence, achieving a solderable top contact is a first step in the process.

(41) An exemplary optimized process flow used for re-metallizing the top surface is described in FIG. 3. In step 31, the bare die power devices are mounted on a chip carrier. Prior to mounting, the samples are rinsed thoroughly in isopropyl alcohol and dried. The part is subject to a rigorous cleaning procedure in step 32. The mounted samples are cleaned a second time for five minutes in an argon plasma. A third cleaning cycle is performed using a detergent-based cleaner. The sample is then subjected to a three-cycle rinse. This ensures the deposition of highly adherent homogenous metal films.

(42) The native oxide film on the surface of aluminum requires is removed in step 33 before plating. This can be achieved in two steps. During the de-oxidation step, the sample is immersed and mildly agitated in a de-smutting solution. In the next step 34, the sample is immersed in a solution of aluminum micro-etchant. This insures an oxide-free and pristine surface is presented for the deposition of the first zinc film. Failure to do this will result in insufficient adhesion and an inconsistent deposition of the zinc film. After rinsing the sample, the first zincate layer is deposited in step 35. The first layer is then stripped in 50% nitric acid in step 36. A second zincate step ensures a smooth and uniform zinc layer. In step 37, the second zincate process is deposited. The sample is subjected to a three-cycle rinse before the nickel plating step to prevent any possible bath contamination. Nickel plating is then performed in step 38. Finally, a layer of gold may be deposited in step 39 for better solder wettability and reliability.

(43) One of the major benefits using electroless plating 21 as a metallization technique in the embodiment shown in FIG. 2 is that it does not require any additional masking, photolithography, or lift-off step. It is completely self-patterning and saves on the cost and complexity of the re-metallization process. It also forms the zinc, nickel, and gold layers directly atop the aluminum layer, thereby adding negligible contact resistance. The bottom connector 11 may be milled out of a copper sheet using a CNC machine in step 22. A wet etching process may also be employed to achieve similar results.

(44) In step 23 in the embodiment in FIG. 2 is the die attachment to the bottom connector, which Attachment of the bottom connector, which may be performed using high-temperature solder, like SAC305, or sintered silver. The solder balls 13, 14 used for flip-chip attachment are of lower reflow temperature than the die-attach material 105. Using a die-attach with a higher reflow temperature for the die attachment process will ensure that the die remains securely attached to the bottom connector during the flip-chip bonding step. A stencil may be used to squeegee a thin layer of die-attach paste on a die-sized area on the drain connector. The die may then be placed using a vacuum pickup tool, and a custom fixture may be used to secure the die in position while it passes through the reflow step. The reflow may be performed using a well-controlled profile set on a reflow oven or hot plate. After the reflow process the sample will be rinsed thoroughly with isopropyl alcohol and blow-dried.

(45) A dry film solder mask, which is a photosensitive material akin to dry film photoresist, may be used for the solder masking process shown in step 24 of this prefer-ed embodiment. A thermal laminator may be used for laminating a thin layer of dry film solder mask on the surface of the flip-chip device. Alternatively a photo-patternable isolation material like benzocyclobutene (BCB) or polyimide (PI) may also be used to the same end if the application so demands. The next step in the process is photolithography. A mask containing the pattern of the solder ball array is mounted on a transparent glass plate and aligned to the bond pads of the flip-chip package on an optical aligner. The solder ball bond pads may be undersized by 20% to help the solder ball retain a spherical shape after reflow. After UV light exposure, the sample may be developed in a suitable bath. Once the bond pads are sufficiently developed, the sample will be rinsed thoroughly in DI water and dried.

(46) In step 25, solder balls are attached to the solder masked flip-chip power device. In order to secure the solder balls in position, a low temperature solder may be used. The melting point of this solder must be lower than the material of the solder ball. The ball attachment solder paste may be stenciled on to the open pads on the surface and the solder balls placed on the wet paste.

(47) A flip-chip bonder may be used for the attachment of the power device in step 26. The sample is mounted on the top arm of the flip-chip bonder. The substrate has matching pads on to which the die maybe placed. The substrate may be designed such that it accommodates multiple flip-chip power devices. The power devices may also be hand-placed on the substrate by aligning to a patterned package outline on the substrate surface. The self-centering property of solder offsets any minor inconsistencies in alignment due to surface tension. No-clean tacky flux may be used on the substrate surface to help secure the flip-chip MOSFETs in position during the placement and reflow process. The flux aids in cleaning and de-oxidizing the substrate bond pads during reflow. The flip-chip devices may be underfilled to help bolster the mechanical strength of the solder balls and to provide additional voltage isolation under operating conditions. A person having skill in the art will recognize and appreciate that different configurations, materials, and processes can be substituted.

EXAMPLE

(48) The following example provides a detailed account of the manufacturing process and electrical testing of a preferred embodiment of the invention. First the process of fabricating a flip-chip MOSFET package, starting from a commercial off the shelf bare die power device is described following by a description of a test setup and test results of the fabricate samples.

(49) In the first step of this example, the re-metallization of the top side contact of bare power devices were performed as shown in FIG. 4. Prior to mounting, the samples were rinsed thoroughly in isopropyl alcohol and dried in step 40. The bare die power devices were mounted on 11 alumina plates using double sided Kapton tape in step 41. The mounted samples were cleaned a second time for five minutes in an argon plasma in step 42a. A third cleaning cycle was conducted using an alkaline etch cleaner, MCL-16, purchased from UYEMURA, 3990 Concours, #425 Ontario, CA 91764 in step 42b. The sample was then subjected to a three-cycle rinse in step 42c.

(50) The native oxide film on the surface of aluminum was removed before plating. This was achieved in two steps. During the de-oxidation step 43, the sample was immersed and mildly agitated in a de-smutting solution purchased from Caswell. In the next step 44a, the sample was immersed in a solution of aluminum etchant purchased from Transene Company, Inc. The solution temperature was maintained at 40 oc for 2 minutes. After rinsing the sample, the first zincate layer was applied for 90 seconds in step 44b. In step 45, the first layer was then stripped in 50% nitric acid for 30 seconds and rinsed. In step 46a, the sample was again immersed in aluminum etchant, this time for 1 minute at 40 C. After rinsing, a second zincate layer was applied for 60 seconds in step 46b. The sample was then subjected to a three-cycle rinse in step 47 before the nickel plating step 48 to prevent any possible bath contamination. Nickel plating was performed for 15 minutes at a solution temperature of 82 C.

(51) FIG. 5 shows an optical microscope image of the second zincate layer and the nickel layer. The fine granular zincate layer can be clearly observed and the probe marks on the gate pad are still visible. The probe marks are no longer visible once the thick nickel plating is complete. The photographs show a uniform nickel layer which is clearly defined without any dendritic formations posing the risk of an electrical short. This illustrates one of the major benefits using electroless plating as a metallization technique. It does not require any additional masking, photolithography, or lift-off step. It is completely self-patterning and saves on the cost and complexity of the re-metallization process. It also forms the zinc and nickel layers directly atop the aluminum layer, thereby adding negligible contact resistance.

(52) Ball shear tests were also conducted on the deposited metallization. The results, as presented in FIG. 6, showed a very high bonding strength. In most cases, the failure occurred as a result of the ball being flattened by the shear tip, and not by a ruptured bond. This served as a satisfactory qualification of the metallization process.

(53) The attachment of die to the drain connector was performed using SAC305 solder.

(54) The bottom connector was milled using a CNC machine. The reflow temperature of SAC305 solder is 249 C. The solder balls used for flip-chip attachment were 62Pb36Sn2Ag which has a reflow temperature of 220 C. Using SAC 305 for the die attachment ensured that the die would remain securely attached to the drain connector during the flip-chip reflow step. A stencil fashioned from a 3 mil Kapton sheet was used to squeegee a thin layer of SAC305 on a die-sized spot on the drain connector. The die was then placed using a vacuum pickup tool, and the same Kapton stencil was used secure the die in place while it was passed through reflow. The reflow was performed using a well-controlled profile set on a SIKAMA reflow oven, SIKAMA INTERNATIONAL, INC., 118 E Gutierrez Street, Santa Barbara, Calif. 93101 USA. The Kapton film fixture was carefully removed after the reflow process, and the sample was rinsed thoroughly with isopropyl alcohol and blow-dried. FIG. 7A shows a photograph of the Kapton fixture and a pair of die after attachment. FIG. 7B also shows blank CNC machined bottom connectors.

(55) A dry film solder mask, Dynamask 5000 Series from Rohm and Haas Electronic Materials LLC, 455 Forest Street, Marlborough, Mass. 01752 was used for the solder masking process, a photo-sensitive material akin to dry film photoresist. A common desktop laminator was used for laminating a thin layer of dry film solder mask on the surface of the flip-chip device.

(56) The next step in the process was photolithography. A mask containing the pattern of the solder ball array was mounted on a transparent glass plate and aligned to the bond pads of the sample on a Karl Suss optical aligner, SUSS MicroTec Inc., 220 Klug Circle, Corona, Calif. 92880-5409 USA. FIG. 8 shows the optimization step performed to ascertain a proper exposure time. The openings for the solder balls were designed to be 10 mils in diameter. The solder balls themselves were 12 mils in diameter. Solder ball bond pads are undersized by 20% to help the solder ball retain a spherical shape after reflow.

(57) After exposure, the sample was stored in a dark enclosure for 60 minutes. The development process was performed in a bath of 10% sodium carbonate solution for 6 minutes. The development process was aided with mechanical scrubbing of the solder mask surface using a soft paintbrush. Once the features were clearly discernible, the solder mask was cured under continuous UV light for 90 minutes.

(58) In the solder ball attachment step, solder spheres of 12 mil diameter were attached to the solder masked flip-chip MOSFET. In order to secure the solder balls in position, a low temperature solder was used. The composition of the solder spheres was 63Pb36Sn2Ag with a reflow temperature of 220 C. The composition of the ball attachment solder was Sn42Bi58 with a reflow temperature of 165 C. The solder paste was stenciled on to the open pads on the MOSFET surface and the solder balls were placed on the wet paste. FIG. 9 shows a photograph of a solder-ball bumped sample.

(59) A flip-chip bonder was used for flip-chip attachment. The sample was mounted on the top arm of the flip-chip bonder. A substrate was patterned using the same process which was used to pattern the flip-chip MOSFET. The substrate material was FR-4 and a finished sample can be seen in the photograph presented in FIG. 10A. The substrate was designed such that it accommodated two flip-chip MOSFETs in a half-bridge configuration. Two flip-chip MOSFETs were flip-chip bonded to the substrate and reflowed in situ on the heated stage of the flip-chip bonder. No-clean tacky flux was used on the substrate surface to help secure the flip-chip MOSFETs in position during the placement and reflow process. The flip-chip devices were underfilled to help bolster the mechanical strength of the solder balls and to provide the necessary voltage isolation under operating conditions. The underfill was purchased from HENKEL CORPORATION, 200 Elm Street, Stamford, Conn. 06902-3800 and the part number was ECCOBOND FP4549. A photograph of the module with busbars and gate connection headers is shown in FIG. 10b.

(60) Electrical Testing

(61) Three samples were fabricated to study the beneficial effects of the low parasitic inductance offered by the proposed flip-chip bonded power MOSFETs. The first sample was a half bridge constructed using conventional discrete ROHM MOSFETs in a TO-247 package. The second module was constructed using bare die ROHM MOSFETs. The wire bonded electrical signal paths were kept to a minimum to realize the shortest possible lateral power loop. This approach is preferred over using discrete devices since it gives the designer more control over minimizing the parasitic inductances of the critical signal loops. It is also less volume intensive in case devices need to be connected in parallel to increase the current handling capability. The third half-bridge was constructed using the flip-chip bonded power devices. It must be noted that all the modules employed the same bare die MOSFET. This was intentionally planned to ensure that any observed differences in performance would be a result of the packaging alone. FIG. 11 shows a photograph showing the three samples compared to the size of a United States quarter for comparison. It was apparent that the half bridge module could be accommodated in much less FR4 real estate.

(62) FIGS. 12A-12B illustrate an advantageous configuration enabled by this flip-chip device invention. In conventional SiC power modules and systems the power loop is always lateral. This is depicted pictorially in FIG. 12A. This restricts a module designer to a minimum loop area defined by critical tolerances in the circuit. Even the most optimized lateral power loop often has a parasitic inductance of 15-20 nH. The situation is understandably even worse in the case of discrete devices. Also, in conventional power modules, there needs to be a ground plane beneath the switching loops to minimize switching transients. Flip-chip devices enable a vertical loop design as described in FIG. 12B. The ground plane requirement is eliminated for a vertical loop due to return path cancellation. Such innovative circuit designs are not possible to implement with conventional power module technology. The topside of the power devices are unavailable for stacked 3D topologies due to the presence of bond wires. The proposed flip-chip topology solves that problem. The critical switching loops are outlined in FIGS. 12A-12B.

(63) Finite element simulations were conducted in ANSYS Q3D to estimate the critical loop inductances of the wire bonded and flip-chip approach. Frequencies from 100 kHz to 10 MHz were simulated and the results are shown plotted and compared in FIG. 13. It was observed that the vertical loop design was much superior to a conventional design in principle.

(64) The electrical test results confirmed the simulation results. A double pulse test was used to test the samples. FIG. 14A shows a schematic of a standard double pulse test setup, and FIG. 14B shows a photograph of the test setup used for this study. The double pulse test is a relatively safe method to measure the switching characteristics of a power module. The ringing frequency during the turn-off of the low-side MOSFET can also be used to estimate the loop inductance of the power loop. It must be noted that all waveforms were measured at a maximum DC bus voltage of 300 V due to the limitation of the available oscilloscope and probes. While the results are expected to be even better at higher voltages close to the device rating, the trends observed at 300V are clearly indicative of the major benefits of the invention.

(65) A double-pulse waveform was applied to the gate of the low side MOSFET of each module. The high-side device was forcibly held in the OFF state by electrically shorting the gate and source. Hence the high-side MOSFET essentially functioned as a diode. The turn-off and turn-on waveforms were measured across the drain and source terminals of the low side MOSFET. The current through the inductor was also measured. FIG. 15A shows the turn-off waveforms for the MOSFET. Compared to the discrete and wire bonded modules, the overshoot and ringing was reduced by 14.2% and 8.7% respectively using the flip-chip MOSFETs. The time period of the ringing in each waveform was measured using the oscilloscope cursors, and the estimated loop inductances are shown in Table 1.

(66) TABLE-US-00001 TABLE 1 The measured parasitic power loop inductance from the oscilloscope waveforms. Output Parasitic Loop capacitance Time period of Inductance of the ringing (s) (At) (nH) = Coss 10.sup.9 MOSFET Wire- Flip- Wire- Flip- (Coss) Discrete bonded chip Discrete bonded chip 110e-12 13e-9 8e-9 4.6e_9 38.96 15.4 4.88

(67) FIG. 15B shows a comparison between the simulated and measured power loop inductances for the wire-bonded and flip-chip module, and the error in the estimated inductance was found to be 4.96% and 3.84% respectively. FIG. 15C shows the turn-on waveforms for the MOSFET. The turn-on waveforms were nearly identical for all the modules. The turn-on time for all the modules was about 20 ns and the turn-on time was measured to be about 16 ns. An inductor current of 12 A was drawn during all the tests. FIG. 15D shows the graph of the measured inductor current plotted alongside the turn-off waveform for the flip-chip half bridge module.

(68) FIG. 16A shows a comparison between the estimated loop area and the measured power loop parasitic inductance of the respective modules. FIG. 16B shows the relationship between the percentage overshoot with the measured loop inductance. This reaffirms the fact that tightly constrained critical loops are essential for obtaining high fidelity textbook waveforms even under high switching frequencies.

(69) The gate signal of the low-side MOSFET of the modules was also measured during the double pulse test. The fidelity of the gate signal is critical under high-frequency switching conditions. An excessive overshoot in the gate signal often results in a false turn-on condition resulting in module failure. As with the power loop, the key to reducing the overshoot in the gate loop also lies in low inductance signal paths. The discrete module had the longest (and consequently the most inductive) gate-source path due to the absence of a dedicated Kelvin source connection. The wire bonded module was provided with a dedicated Kelvin source return, as was the flip-chip module. The MOSFET gate turn-off waveforms are shown plotted in FIG. 17A. There are overshoots observed at multiple locations during the gate turn-off in the discrete and wire bonded modules. A transient-free smooth waveform can be obtained using a flip-chip module with a vertical power loop. The gate turn-on waveforms are shown in FIG. 17B. As with the drain-source turn-on waveforms, there was nothing much to distinguish the three waveforms.

(70) The results from these experiments demonstrate that flip-chip power device embodiments of the disclosed invention provide the ease of use and assembly of discrete devices. They do away with the cost and complexity associated with manufacturing wire bonded power modules, while providing even better performance.

(71) The disclosed flip-chip methods were also used to construct 650V-rated Schottky diode packages. To evaluate these devices on a Keysight curve tracer, they were engineered to fit a TO-247 form factor. An image of the samples is shown in FIG. 18A. One of the modules encapsulated in a black glob-top is a wire bonded sample for comparison. FIG. 18B shows the IV curve of the samples. It was observed that the on-state resistance of the wire bondless flip-chip sample was 24% lower than the wire bonded sample, despite being in a TO-247 package. This implies a 24% reduction in the conduction losses of the Schottky diode merely as a result of doing away with wire bonds in favor of the disclosed flip-chip methods. The turn-off waveforms of the two modules were also measured and are contrasted in FIG. 18C. As with the MOSFET modules, the time period of the ringing was measured using the oscilloscope cursors and the inductances were calculated. The calculations and results are shown in Table 2.

(72) TABLE-US-00002 TABLE 2 Measured parasitic loop inductances of the two modules. Output capacitance Parasitic Loop of the Time period of Inductance (nH) = MOSFET ringing (s)(At) (992 10.sup.9 Coss (Coss) Wire-Bonded Flip-chip Wire-bonded Flip-chip 110e12 9.2e9 8.2e9 19.51 15.5

(73) The overall loop inductance 20.5% lower despite the TO-247 package. FIG. 18D shows the turn-on waveforms for the two modules. FIG. 18E shows the voltage across the diode and the inductor current measured at a DC bus voltage of 450 V and 15 A.

(74) Reference numerals used throughout the detailed description and the drawings correspond to the following elements: bare die power device 10 bottom electrical contact 101 top electrical contact 102 bottom pad connector 11 top surface second contact 103 the lower level pad 104 electrically conductive die-attach material 105 trench 106 top surface first contact 107 gate pad 108 source pads 109 solder mask 12 top pad solder spheres 13 bottom pad solder spheres 14 power devices 15 metallic connector 16 substrate 17 underfill 18 bare die converted to flip-chip power device 19 electroless plating 21 machining 22 die attachment to bottom connector 23 solder masking 24 solder balls attached 25 bonding to substrate 26 mount sample 31 clean 32 remove oxide 33 etch 34 apply first zincate 35 strip 36 apply second zincate 37 nickel plating 38 gold plating 39 mount die 41 rinse 40 argon clean 42a alkaline clean 42b rinse 42c etch 44a apply first zincate 44b strip 45 etch 46a apply second zincate 46b rinse 47 nickel plate 48

(75) From the foregoing, it will be seen that this invention well adapted to obtain all the ends and objects herein set forth, together with other advantages which are inherent to the structure. It will also be understood that certain features and subcombinations are of utility and may be employed without reference to other features and subcombinations. This is contemplated by and is within the scope of the claims. Many possible embodiments may be made of the invention without departing from the scope thereof. Therefore, it is to be understood that all matter herein set forth or shown in the accompanying drawings is to be interpreted as illustrative and not in a limiting sense.

(76) When interpreting the claims of this application, method claims may be recognized by the explicit use of the word method in the preamble of the claims and the use of the ing tense of the active word. Method claims should not be interpreted to have particular steps in a particular order unless the claim element specifically refers to a previous element, a previous action, or the result of a previous action. Apparatus claims may be recognized by the use of the word apparatus in the preamble of the claim and should not be interpreted to have means plus function language unless the word means is specifically used in the claim element. The words defining, having, or including should be interpreted as open ended claim language that allows additional elements or structures. Finally, where the claims recite a or a first element of the equivalent thereof, such claims should be understood to include incorporation of one or more such elements, neither requiring nor excluding two or more such elements.