International Business Machines Corporation

Jody John Van Horn

Anne Elizabeth Gattiker

Marc Douglas Knox

David A Grosch

Franco Motika

Paul Steven Zuchowski

David Lewis Gardell

Phil Nigh

Kerry Bernstein

George Frederick Walker

Charles Hampton Perry

James Marc Leas

Robert William Koss

Edward J Nowak

Alvin Wayne Strong

Ernest Y Wu

Ronald Jay Bolam

Phillip J Nigh

Dennis R Conti

Ronald J Prilik

Steve Leo Dingle

Norman Jay Rohrer

Donald Lawrence Wheater

James M Crafts

Harold Wayne Chase

Brian C Noble

Leo A Noel

Raymond J Bulaga

John L Harris

Ann L Swift

Roger Gamache

Joel Thomas

William J Thibault

Andrew T Holle

Adrian Patrascu

Gregory V Miller

A method, apparatus and computer program product for testing semiconductor products that combines multiple techniques. Depending on the requirements, different ones of the techniques are emphasized over the other techniques. The testing applies a technique to achieve a higher single defect acceleration parameter at the expense of a second parameter, thus enabling acceleration of defects that require higher voltage or higher temperature than a traditional 'Burn In' can achieve, which defects would otherwise go unaccelerated. The method manages the adaptation of the different techniques, e.g., how it decides to favor one technique over the other, and how it carries out the favoring of one or more particular techniques in a given test situation. Thus, acceleration to defectivity (defect type and quantity) may be tailored in real time by uniquely leveraging the duration spent in a given section of a process flow based on the prevalence of unique defect types.

Diversified exerciser and accelerator

Test equipment provides interrupt capability to automatic testing as a means of actively controlling temperature of the device under test. A processor coupled to memory is responsive to computer-executable instructions contained in the memory. A test socket is coupled to a device under test and coupled to the processor. The processor is configured to interrupt an application pattern running on the device under test. In response to interrupting the application pattern, the processor is configured to cause a control pattern to run on the device under test and then cause the application pattern to restart running from the point of interruption on the device under test.

Programmable active thermal control

Methods for testing a semiconductor circuit () including testing the circuit and modifying a well bias () of the circuit during testing. The methods improve the resolution of voltage-based and IDDQ testing and diagnosis by modifying well bias during testing. In addition, the methods provide more efficient stresses during stress testing. The methods apply to ICs where the semiconductor well (wells and/or substrates) are wired separately from the chip VDD and GND, allowing for external control () of the well potentials during test. In general, the methods rely on using the well bias to change transistor threshold voltages.

Integrated circuit testing methods using well bias modification

A method for testing a semiconductor circuit () including testing the circuit and modifying a well bias () of the circuit during testing. The method improves the resolution of IDDQ testing and diagnosis by modifying well bias during testing. The method applies to ICs where the semiconductor well (wells and/or substrates) are wired separately from the chip VDD and GND, allowing for external control () of the well potentials during test. In general, the method relies on using the well bias to change transistor threshold voltages.

Integrated circuit testing method using well bias modification

A method and system for predicting gate reliability. The method comprises the steps of stressing a gate dielectric test site to obtain gate dielectric test site data and using the test site data to predict gate reliability. Preferably, the test structure and the product structure are integrated in such a manner that a test site occupies some of the product area and the product itself occupies the remainder of the product area. A preferred methodology, more specifically, is as follows: (1) Test structures at start both in parallel stress mode and in ring oscillator or 'product' mode; (2) Analyze the breakdown data as per the present state of the art for each of the areas based on the parallel stress mode; (3) Combine the above breakdown distributions using the area scaling to improve the confidence bounds of the Weibull slope of the cumulative distribution function; (4) Test the ring oscillators in the product mode to determine how many of the stress fails are also product fails as defined by an operational degradation; (5) Subdivide the failures to determine the relationship between the first fail, and the second fail, and the nth fail; (6) Investigate which stress fail, if not the first stress fail, is more likely to cause a product fail as defined by operational degradation; and (7) Based on the subdivision in step 5 and the results in step 6, make projection based on that fail which is most likely to cause fail. The methodology as outlined above bridges between dielectric stress fails and product degradation both in the case of each stress fail causing a product degradation, as well as in the case where more than one stress fail occurs before any product degradation occurs. And this relationship can be quantified.

Circuitry and methodology to establish correlation between gate dielectric test site reliability and product gate reliability

According to the present invention, a method of controlling the burning in of at least one I/C device in a burn in tool is provided. For high power device, the tool has a heat sink positioned to contact each device being burned in, and has a socket for mounting each device to be burned in, and a power source to supply electrical current to burn in each device. The method includes the steps of continuously monitoring at least one process parameter selected from the group of current, voltage, power and temperature, and varying the voltage to maintain at least one of the parameters at or below a given value. Also, a technique for burning in low power devices without a heat sink is provided. The invention also contemplates a tool for performing the above method.

Device burn in utilizing voltage control

Disclosed is an integrated circuit device, comprising: a first power rail for supplying power to first latch and a circuit during a first clock phase; a second power rail for supplying power to a second latch during a second clock phase; and the circuit coupled between an output of the first latch and an input of the second latch.

Integrated circuit phase partitioned power distribution for stress power reduction

A method and system for predicting gate reliability. The method comprises the steps of stressing a gate dielectric test site to obtain gate dielectric test site data and using the test site data to predict gate reliability. Preferably, the test structure and the product structure are integrated in such a manner that a test site occupies some of the product area and the product itself occupies the remainder of the product area. A preferred methodology, more specifically, is as follows: (1) Test structures at start both in parallel stress mode and in ring oscillator or 'product' mode; (2) Analyze the breakdown data as per the present state of the art for each of the areas based on the parallel stress mode; (3) Combine the above breakdown distributions using the area scaling to improve the confidence bounds of the Weibull slope of the cumulative distribution function; (4) Test the ring oscillators in the product mode to determine how many of the stress fails are also product fails as defined by an operational degradation; (5) Subdivide the failures to determine the relationship between the first fail, and the second fail, and the nfail; (6) Investigate which stress fail, if not the first stress fail, is more likely to cause a product fail as defined by operational degradation; and (7) Based on the subdivision in step 5 and the results in step 6, make projection based on that fail which is most likely to cause fail. The methodology as outlined above bridges between dielectric stress fails and product degradation both in the case of each stress fail causing a product degradation, as well as in the case where more than one stress fail occurs before any product degradation occurs. And this relationship can be quantified.

A system for modifying the power up and diagnostic procedure of systems such that the system voltage is lowered to a predetermined voltage level that has been shown to detect delay faults. The system conducts the normal procedure of power up/diagnostic routines at the described VLV condition and then logs failures to this VLV condition. Upon completion of the VLV power up, the system is shut down normally and then subsequently powered up again at the normal voltage conditions. Discrepancies between the VLV power up/diagnostics are noted in the system log and communicated appropriately.

AC defect detection and failure avoidance power up and diagnostic system

A method and system are provided that minimize wafer or package level test time without adversely impacting yields in downstream manufacturing processes or degrading outgoing quality levels. The method provides optimization by determining, a priority, the most effective set of tests for a given lot or wafer. The invention implements a method using a processor-based system involving the integration of multiple sources of data that include: historical and realtime, product specific and lot specific, from wafer fabrication data (i.e., process measurements, defect inspections, and parametric testing), product qualification test results, physical failure analysis results and manufacturing functional test results. These various forms of data are used to determine an optimal set of tests to run using a test application sequence, on a given product to optimize test time with minimum risk to yield or product quality.

Method for test optimization using historical and actual fabrication test data

An apparatus and a method for simultaneously testing or burning in all the integrated circuit chips on a product wafer. The apparatus comprises a glass ceramic carrier having test chips and means for connection to pads of a large number of chips on a product wafer. Voltage regulators on the test chips provide an interface between a power supply and power pads on the product chips, at least one voltage regulator for each product chip. The voltage regulators provide a specified Vdd voltage to the product chips, whereby the Vdd voltage is substantially independent of current drawn by the product chips. The voltage regulators or other electronic means limit current to any product chip if it has a short. The voltage regulator circuit may be gated and variable and it may have sensor lines extending to the product chip. The test chips can also provide test functions such as test patterns and registers for storing test results.

Semiconductor wafer test and burn-in

Growing community of inventors

Underhill, VT, United States of America

Jody John Van Horn