Patent Application
20180301386
This invention relates to semiconductor burn-in methods to achieve increased reliability of high power III-N devices.
To date, most transistors used in power electronic applications have typically been fabricated with silicon (Si) semiconductor materials. Common transistor devices for power applications include Si CoolMOS, Si Power MOSFETs, and Si Insulated Gate Bipolar Transistors (IGBTs). While Si power devices are inexpensive, they suffer from a number of disadvantages, including relatively low switching speeds and high levels of electrical noise. More recently, silicon carbide (SiC) power devices have been considered due to their superior properties. III-N semiconductor devices, such as gallium nitride (GaN) devices, are now emerging as attractive candidates to carry large currents, support high voltages and to provide very low on-resistance and fast switching times. However, reliable fabrication and manufacturing of high-voltage III-N devices has proven to be very difficult
In one aspect, a plurality of III-N electronic devices are tested, each comprising a source, a gate and a drain. Each of the III-N devices is held at a first temperature less than or equal to 30° C. for a first period of time while applying a first gate-source voltage lower than a threshold voltage of the device an a first drain-source voltage greater than 0.2 times a breakdown voltage of the device. Each of the III-N devices is held at a second temperature greater than the first temperature for a second period of time while applying a second gate-source voltage lower than a threshold voltage of the device and a second drain-source voltage greater than 0.2 times a breakdown voltage of the device. After holding the devices at the first and second temperature, the electrical performance of one or more parameters of each of the devices is tested. Based on the results obtained from testing, at least on device of the plurality of devices is rejected.
In a second aspect, the quality and reliability of a III-N device comprising a package, a gate, a source, and a drain is assured. For a first period of time, a first gate-source voltage below the threshold voltage of the device is applied and a positive drain-source voltage is applied. During the first period of time, the package is held at a first temperature which is below 50% of the maximum operating temperature of the device. After the first period of time, the electrical performance of one or more parameters is tested and based on the results, the device is discarded if it fails to meet one or more of the parameters.
In a third aspect, a wafer of III-N electronic devices, each comprising a source, a gate, and a drain, is tested. The wafer is held at a first temperature less than or equal to 30° C. for a first period of time while simultaneously applying to a plurality of devices a first gate-source voltage lower than a threshold voltage of the devices, and a first drain-source voltage greater than 0.2 times a breakdown voltage of the devices. The wafer is held at a second temperature greater than the first temperature for a second period of time while simultaneously applying to a plurality of devices a second gate-source voltage lower than a threshold voltage of the devices, and a second drain-source voltage greater than 0.2 times a breakdown voltage of the devices. The electrical performance of one or more parameters of each of the devices is tested, and based on the results, at least on device on the wafer is rejected.
As used herein, the term “high temperature” refers to a temperature above ambient room temperature, and in some cases substantially above room temperature. Examples of high temperature can be 50° C. or higher, 85° C. or higher, 150° C. or higher, or in a range of 50° C. to 200° C., 85° C. to 200° C., or 150° C. to 200° C. The high temperature can be created by artificially heating a device, such as on a hot plate or in an oven. The term “low temperature” refers to a temperature which is either at ambient room temperature or at a temperature that is lower than ambient room temperature (e.g., less than the high temperature). Examples of low temperature can be 25° C. or lower, 0° C. or lower, −25° C. or lower, −55° C. or lower, or in a range of −85° C. to 25° C., −85° C. to 0° C., or −85° C. to −25° C. Low temperature can be created by artificial cooling the device, such as in a freezer. In addition, a low temperature can be created by using a hot plate in order to constantly control the temperature to be at or near room temperature.
As used herein, the terms III-Nitride or III-N materials, layers, devices, etc., refer to a material or device comprised of a compound semiconductor material according to the stoichiometric formula BwAlxInyGazN, where w+x+y+z is about 1 with 0≤w≤1, 0≤x≤1, 0≤y≤1, and 0≤z≤1. III-N materials, layers, or devices, can be formed or prepared by either directly growing on a suitable substrate (e.g., by metal organic chemical vapor deposition), or growing on a suitable substrate, detaching from the originally substrate, and bonding to other substrates.
As used herein, two or more contacts or other items such as conductive channels or components are said to be “electrically connected” if they are connected by a material which is sufficiently conducting to ensure that the electric potential at each of the contacts or other items is intended to be the same, e.g., is about the same, at all times under any bias conditions.
As used herein, “blocking a voltage” refers to the ability of a transistor, device, or component to prevent significant current, such as current that is greater than 0.001 times the operating current during regular conduction, from flowing through the transistor, device, or component when a voltage is applied across the transistor, device, or component. In other words, while a transistor, device, or component is blocking a voltage that is applied across it, the total current passing through the transistor, device, or component will not be greater than 0.001 times the operating current during regular conduction. Devices with off-state currents which are larger than this value exhibit high loss and low efficiency, and are typically not suitable for many applications. As used herein, the “device breakdown voltage” refers to the maximum voltage that can be sustained by the device while the device is biased in the off state such that the device off-state current does not exceed 0.001 times the operating current during regular conduction.
As used herein, a “high-voltage device”, e.g., a high-voltage switching transistor, is an electronic device which is optimized for high-voltage switching applications. That is, when the transistor is off, it is capable of blocking high voltages, such as about 300V or higher, about 600V or higher, or about 1200V or higher, and when the transistor is on, it has a sufficiently low on-resistance (RON) for the application in which it is used, e.g., it experiences sufficiently low conduction loss when a substantial current passes through the device. A high-voltage device can at least be capable of blocking a voltage equal to the high-voltage supply or the maximum voltage in the circuit for which it is used. A high-voltage device may be capable of blocking 300V, 600V, 1200V, or other suitable blocking voltage required by the application. In other words, a high-voltage device can block all voltages between 0V and at least Vmax, where Vmax is the maximum voltage that can be supplied by the circuit or power supply, and Vmax can for example be 300V, 600V, 1200V, or other suitable blocking voltage required by the application.
As used herein, a “III-Nitride” or “III-N device” is a device based on III-N materials. The III-N device can be designed to operate as an enhancement-mode (E-mode) transistor device, such that the threshold voltage of the device (i.e., the minimum voltage that must be applied to the gate relative to the source in order to turn the device on) is positive. Alternatively, the III-N device can be a depletion-mode (D-mode) device, having a negative threshold voltage. The III-N device can be a high-voltage device suitable for high voltage applications. In such a high-voltage device, when the device is biased off (e.g., the voltage on the gate relative to the source is less than the device threshold voltage), it is at least capable of supporting all source-drain voltages less than or equal to the high-voltage in the application in which the device is used, which for example may be 100V, 300V, 600V, 1200V, 1700V, or higher. When the high voltage device is biased on (e.g., the voltage on the gate relative to the source is greater than the device threshold voltage), it is able to conduct substantial current with a low on-voltage. The maximum allowable on-voltage is the maximum voltage that can be sustained in the application in which the device is used.
As used herein, the “operating temperature range” of a device is the range between the minimum and the maximum specified operating temperature of the device. In some applications, the minimum specified operating temperature of a device is −50° C. and the maximum specified operating temperature of the device is 150° C. The median operating temperature range is the mid-point of the operating temperature range.
The details of one or more disclosed implementations of the subject matter described in this specification are set forth in the accompanying drawings and the description below. Other features, aspects, and advantages will become apparent from the description, the drawings and the claims.
The reliability of III-N based devices has been the subject of much investigation over the last decade. Substantial progress has been made in demonstrating the viability of this technology. Yet, especially in the power market, the technology is still gaining general acceptance. Power devices, while similar to RF III-N devices, are subjected to much higher voltages in the off-state during circuit operation. These high voltages lead to different stability issues in the devices, thereby requiring a need to develop a robust device screening procedure which can accurately remove all weak devices. Such a procedure preferably screens devices which could exhibit early lifetime failure, or “infant mortality”, before the devices enter commercial applications. Electronic components often fail due to flaws resulting from the fabrication and manufacturing process. These components are commonly subjected to probe testing after completion to detect devices which do not electrically match the design criteria. However, probe testing typically will not detect devices with “latent” defects and some devices which pass probe testing can still result in early-life failure or infant mortality of the device.
A method to reduce the occurrence of infant mortality is to subject the devices to accelerated life-time testing. A method of accelerated life-time testing, or stress test, is known as “burn-in.” Burn-in testing can be a critical process to assure high product reliability of new technologies, especially during their introduction phase. A type of burn-in can be increasing the temperature and voltage of a device for a set period of time beyond the typical operating temperature and voltage of the device. This will accelerate the failure of devices with latent defects that were not identified during initial probe testing and allow for their removal from known good parts before the devices enter their intended field of use.
Unfortunately, this type of burn-in is a lengthy process and can require significant capital expenditures to implement. III-N devices, in particular, could require hundreds of hours of burn-in to accurately detect all latent defects. It is therefore a desire to minimize the time required for burning-in devices. One method to reduce the required burn-in time is to increase the temperature or voltage of the burn-in test, however III-N devices cannot withstand severe burn-in conditions and can begin to incur different failure mechanisms, such as an increase of dynamic on-resistance beyond the specification.
Silicon based semiconductor devices typically result in latent defects from the manufacturing of the device but do not suffer from defects in the bulk silicon substrate. These silicon device defects can be accelerated by increasing the temperature of the burn-in test to a temperature that is higher than the typical operating temperature of the device. Burn-in can be performed on silicon devices at temperatures between 90° C. to 150° C. for anywhere between 24 to 168 hours. These conditions, however, may not be suitable for burn-in testing for III-N devices. III-N devices can suffer from defects found in the bulk III-N material layers of the device (EPI defects) as well as from the manufacturing of the electrical metal routing layers and passivation layers (e.g., device layers) over the III-N material layers. These different types of defects can exhibit different activation energy and may not respond to an accelerated life-time testing in the same manner. In order to reduce the burn-in time and successfully screen latent defects in III-N devices it may be required to perform burn-in testing at multiple temperatures, voltages and times.
At 101, each diced non-inked III-N device is packaged. Here, each III-N device that electrically measures within the critical parameters of the on-wafer testing is assembled into a final package. This involves mounting the device into a metal, plastic or ceramic casing. The package holds the contact leads which are used to connect the device to external circuits. It also acts to dissipate heat and protect the device from the external environment. The package can be a discrete electronic package or an electronic module containing multiple semiconductor devices.
At 102, each packaged III-N device undergoes initial testing. The initial testing can include a variety of electrical tests. These typically include measuring on-resistance (RON), off-state drain leakage (IDS), and threshold voltage (VTH). These test can be performed at room temperature or at an elevated temperature, for a period of time of one second or less. The main purpose of testing the packaged III-N device at this step is to screen out any defects or mechanical failures caused by the device packaging performed at 101 (e.g., broken wire bond or poor solder interface). The short duration of these tests are insufficient to detect devices with latent defects.
After the initial testing performed at 102, each of the performance parameters of the packaged III-N devices are compared, at 103, against the device specification. If the device fails to meet a minimum specification, the device is rejected at 109N. Passing devices are sent through, at 109Y, to an accelerated lifetime test, known as a stress test, or burn-in test.
During the burn-in test, the packaged III-N device can be put into a socket mounted on a printed circuit board (PCB), or directly soldered to a PCB. The PCB can be configured to test a plurality of devices simultaneously. The PCB can be installed into a testing chamber that has a temperature controller. The temperature controller can be capable of controlling the testing chamber temperature between −50° C. and 200° C. The packaged device is then stressed under a positive voltage bias from drain to source while in the off-state (i.e., while the gate is biased relative to the source below the device threshold voltage) for a specified period of time. The drain-source voltage of the device may be biased at a voltage which is equal to or greater than the typical operating voltage of the device. The device may be biased at a voltage which is greater than the specification of the device as listed on the data sheet. Since the III-N devices described herein are high-voltage devices, the burn-in step may be performed at a voltage greater than 400V, greater than 600V, greater than 900V or greater than 1200V. In some implementations, the drain-source voltage of the device during burn-in testing is at least 0.2 times, at least 0.4 times, at least 0.5 times, at least 0.6 times, at least 0.7 times, or at least 0.8 times the device breakdown voltage. Furthermore, the drain-source voltage of the device may be stepped between multiple voltages during the burn-in process.
The specified period of time used for a burn-in process can affect the production capacity, the cost, and the time to market of a finished product. As a result, the lowest duration burn-in time is desired which can still accurately capture all early lifetime failures. Burn-in practice for silicon based semiconductor devices involves increasing the temperature and voltage beyond the typical operating temperature and voltage of the device. This is done in order to stress the device at levels that are beyond the typical operating conditions the device would see during normal applications. In many cases, for burn-in testing of silicon based semiconductor devices, increasing the temperature, the bias voltage, or both reduces the time required for a burn-in process while still allowing devices which would have otherwise exhibited early failure to be screened out. This is due to the positive activation energy that latent defects typically exhibit in silicon based devices. Often times, for silicon based devices, the burn-in process may be conducted at a temperature or voltage beyond the specification rating of the device as found on the data sheet. This is done in order to reduce the required burn-in time even further. Examples of burn-in times can be as much as 24 hours to 168 hours.
However, III-N based semiconductor devices have been found to also exhibit different failure mechanisms in addition to those typically found in silicon based semiconductor devices. Exposing III-N devices to high temperature burn-in can inject charge into the device and relax certain types of defects with negative activation energy, actually making the device stronger. This would indicate that performing the burn-in test at elevated temperatures could increase the time required to screen devices containing certain types of latent defects.
It has been found that the burn-in process can be improved for III-N devices by performing burn-in at multiple temperatures. For example, two burn-in steps can be performed sequentially, each at their own temperature. The first step can be a low temperature burn-in step, and the second step can be a high-temperature burn-in step. In some implementations, more than two burn-in steps can be performed. For example, three, four, five, or more different burn-in steps, each performed sequentially at a different temperature, e.g., at sequentially higher temperatures, could be used to screen the device. In the example shown in
At 104, a low temperature burn-in of III-N packaged devices is performed. One example of a low temperature burn-in can be performed at ambient room temperature (e.g., no artificial increase of the packaged III-N device temperature). In some instances, a hot plate or oven may be used to control the temperature at or near room temperature. The indicated temperature typically refers to the device package case temperature. Since the burn-in process is performed while the device is biased in the off state (i.e., where the gate is biased below the device threshold voltage relative to the source), the device does not dissipate much power, and therefore device self-heating is minimal. Another example of low temperature burn-in can be performed at a temperature below ambient room temperature. In order to perform the burn-in process below room temperature, the packaged device can be artificially cooled (e.g., placed in a freezer). Artificially cooled devices can be maintained under constant nitrogen purge to reduce the humidity and prevent condensation from forming and shorting the device. During low temperature burn-in, examples of temperatures the device may be stressed at are 30° C. or lower, 0° C. or lower, −25° C. lower, −50° C. or lower or the lowest specification temperature listed on the device data sheet. Temperatures used for a low temperature burn-in may even be below the minimum specified operating temperature of the device listed on the data sheet. The temperature can be less than the median operating temperature range of the device. The time used for low temperature burn-in can be less than 96 hours, less than 48 hours, or less than 24 hours. The time used can be greater than 0.5 hours, greater than 1 hour, greater than 24 hours, greater than 48 hours, between 1 hour and 96 hours or between 24 hours and 96 hours.
At 110, a test can optionally be performed to screen out packaged III-N devices which do not meet specific critical parameters after the low temperature burn-in. Performing this screening can eliminate the failed III-N devices from utilizing further resources during the high temperature burn-in step. In some cases, it is preferable that this step be omitted, as it can increase testing time, thereby decreasing product cycle time.
At 105, a high temperature burn-in of III-N packaged devices is performed. High temperature burn-in can be performed by heating the device, typically by placing the device on a hot plate or in an oven. Examples of high temperature can be 50° C. or higher, 85° C. or higher, or 150° C. or higher. The temperature can be greater than the median operating temperature range of the III-N device. The temperature can be greater than the maximum specified operating temperature of the III-N device.
During the low temperature burn-in and the high temperature burn-in, the gate of the device is biased relative to the source below the threshold voltage of the device while a positive voltage is applied to the device drain relative to the source for a specified period of time. It is desirable to use the lowest specified period of time required to successfully screen out latent defects. The time used for high temperature burn-in can be less than 96 hours, less than 48 hours, or less than 24 hours. The time used can be greater than 0.5 hours, greater than 1 hour, greater than 24 hours, greater than 48 hours, between 1 hour and 96 hours or between 24 hours and 96 hours.
The specified period of time for the low temperature burn-in can be the same time as the high temperature burn-in or it can be substantially different than the time for the high temperature burn-in. In one example, the time of the low temperature burn-in can be between 24 hours and 96 hours, and the time for the high temperature burn-in can be between 24 hours and 96 hours. In another example, the time for both burn-in temperatures can be 24 hours or less.
In addition, the positive drain-source voltage applied during the low temperature burn-in can be the same drain-source voltage applied during the high temperature burn-in, or the voltages applied can be different. In one example, the drain-source voltage of the III-N device applied during the low temperature burn-in is at least 0.2 times the device breakdown voltage and the drain-source voltage applied during the high temperature burn-in is at least 0.2 times the device breakdown voltage. In another example, the drain-source voltage applied during both burn-in temperatures is 400V or higher. Furthermore, the drain-source voltage of the device may be stepped between multiple voltages during the high temperature or the low temperature burn-in process steps.
At 106, each packaged III-N device undergoes final testing. The final testing can include a variety of electrical tests. These typically include measuring on-resistance (RON), off-state drain leakage (IDS), and threshold voltage (VTH). These test parameters can be the same as those tested at 102. At 107, packaged devices which show critical test parameters out of specification or devices which show a statistical deviation out of family compared to the initial testing 102 results can be rejected as shown at 111N. Rejected devices are removed from the plurality of devices and are not considered for commercial use. For example, devices that are within 10% or greater of the maximum specified value of the test parameter can be rejected. In addition, devices which show a dynamic on-resistance increase of 25% or more from the initial testing can be rejected. Devices which show an off-state drain leakage greater than 1 μA can be rejected.
At 108, all devices which have completed the low temperature burn-in and the high temperature burn-in and still meet the desired device performance requirements can be placed into final inventory 108 and be released for commercial use.
Wafer fabrication at 201, can include multiple fabrication processes, each which can result in the occurrence of device defects. Each wafer typically contains a plurality of III-N semiconductor devices. Once wafer fabrication is completed, each III-N device is electrically tested to measure one or more critical parameters. These typically include measuring on-resistance (RON), off-state drain leakage (IDS), and threshold voltage (VTH). The initial on-wafer testing is shown at 202 of
At 203, a plurality of devices on the wafer can be simultaneously subjected to low temperature burn-in testing. This is done by employing a specially made testing apparatus, called a probe card, which has contact points that match the specific device layout on the wafer. The probe card can contain spring-like electrodes which extend away from the probe card. These spring like electrodes, when lowered towards the wafer surface, can contact each device electrode and apply the desired bias voltages required for burn-in to a plurality of devices simultaneously. Furthermore, low temperature burn in 203 and high temperature burn-in 204 can be completed sequentially. Low temperature burn-in 203 and high temperature burn-in 204 can have the same process requirements as low temperature burn-in 104 and high temperature burn-in 105, respectively, previously described with reference to process 100 in
As shown in
At 205, the wafer is placed into final on-wafer testing after completion of the burn-in processes. The final on-wafer testing can include a variety of electrical tests. These tests typically include measuring on-resistance (RON), off-state drain leakage (IDS), and threshold voltage (VTH). These test parameters can be the same as those tested in the on-wafer testing process step 202. The III-N devices which show critical test parameters out of specification or devices which show a statistical deviation out of family compared to the initial on-wafer testing 202 results can be rejected at 209N. Rejected devices are screened out and marked with an ink so as to avoid accidental assembly of these devices into final device packages. After final on-wafer testing is complete, the wafer is sawed (or diced) between scribe lines and each “good” die (i.e., non-inked III-N device) is subjected to further assembly and packaging. This packaging can be the same device packaging as described in 101 of
A number of implementations have been described. Nevertheless, it will be understood that various modifications may be made without departing from the spirit and scope of the techniques and devices described herein. Accordingly, other implementations are within the scope of the following claims.
This patent document claims the priority of U.S. Provisional Application No. 62,486,417, filed on Apr. 17, 2017. The above referenced application is incorporated by reference as part of the disclosure of this document.
| Number | Date | Country | |
|---|---|---|---|
| 62486417 | Apr 2017 | US |
