Electricity: measuring and testing – Fault detecting in electric circuits and of electric components – Of individual circuit component or element
Reexamination Certificate
1999-07-29
2001-09-18
Metjahic, Safet (Department: 2858)
Electricity: measuring and testing
Fault detecting in electric circuits and of electric components
Of individual circuit component or element
C324S763010
Reexamination Certificate
active
06292009
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to a semiconductor wafer mode controlling assembly and, more particularly, to such an assembly in which modes of circuitry of dice (IC's) on the wafer are controlled through alternating signals applied to the dice through probe pads on the wafers. The invention also includes methods for constructing and operating such wafers and the assembly.
2. State of the Art
Typically, finished integrated circuit chip assemblies include a die or dice attached to a lead frame and encapsuled with an encapsulant. Numerous expensive and time consuming steps are involved in producing such chip assemblies. These steps may include the following: (1) forming dice on a wafer substrate, (2) testing the dice, (3) cutting dice from the wafer, (4) connecting a die or dice to a lead frame, (5) encapsulating the die or dice, lead frame, connecting wires, and any auxiliary circuitry, (6) performing burn-in and providing other stresses to the dice, and (7) testing the assembly.
Defects in a finished chip assembly can prevent it from operating as intended. In spite of painstaking attention to detail, defects may be introduced at various levels of production. For example, manufacturing defects in the die may cause a failure. It has been found, however, that some defects are manifest immediately, while other defects are manifest only after the die has been operated for some period of time.
Reliability curves are used to express a hazard rate or instantaneous failure rate h(t) over time t, and often have a “bath tub” shape. The reliability curves for many, if not all, IC's are generally like that shown in FIG.
1
. The reliability curve in
FIG. 1
may be divided into three regions: (1) an infant mortality region, (2) a random failures region, and (3) a wearout region.
The infant mortality region begins at time t
0
, which occurs upon completion of the manufacturing process and initial electrical test. Some IC's, of course, fail the initial electrical test. Inherent manufacturing defects are generally expected in a small percentage of IC's, even though the IC's are functional at time t
0
. Because of these inherent manufacturing defects (that may be caused by contamination and/or process variability), these IC's have shorter lifetimes than the remaining population. Typically known as IC's suffering “infant mortalities,” while the IC's may constitute a small fraction of the total population, they are the largest contributor to early-life failure rates.
Once IC's subject to infant mortality failure rates have been removed from the IC population, the remaining IC's have a very low and stable field failure rate. The relatively flat, bottom portion of the bathtub curve, referred to as the random failure region, represents stable field-failure rates which occur after the IC failures due to infant mortalities have been removed and before IC wearout occurs.
Eventually, as wearout occurs, the failure rate of the IC's begins to increase rapidly. However, the average lifetime of an IC is not clearly understood, because most lab tests simulate only a few years of normal IC operation.
“Burn-in” refers to the process of accelerating failures that occur during the infant mortality phase of component life in order to remove the inherently weaker IC's. The process has been regarded as critical for product reliability since the semiconductor industry began. There have been two basic types of burn-in. During the process known as “static” burn-in, temperatures are increased (or sometimes decreased) while only some of the pins on a test IC are biased. No data is written to the IC, nor is the IC exercised under stress during static burn-in. During “un-monitored dynamic” burn-in, temperatures are increased while the pins on the test IC are biased. The IC is cycled under stress, and data patterns are written to the IC but not read. Hence, there is no way of knowing whether the data written is retained by the cell.
In recent years, as memory systems have grown in complexity, the need for more and more reliable components has escalated. This need has been met in two ways. First, manufacturing process technology has reached a level of maturity and stability where inherent manufacturing defects, caused by contamination and process variation, have been reduced. As a result, latent failures have been significantly reduced, resulting in lower field failure rates. Secondly, more sophisticated methods of screening infant mortalities have been developed. As IC manufacturing practices have become more consistent, it has become clear that burn-in systems that simply provide stress stimuli in the form of high temperature and VCC (power) to the IC under test may be inadequate in two areas: (1) such burn-in systems cannot detect and screen infant mortality failure rates measured in small fractions of a percent; (2) such burn-in systems are unable to confirm random failure rates that are claimed to be significantly lower than 100 FITs (i.e., fewer than 100 failures per billion IC hours) at normal system operating conditions.
To address these issues, an “intelligent” burn-in approach can be utilized. The term “intelligent burn-in,” as used in this discussion, refers to the ability to combine functional, programmable testing with the traditional burn-in cycling of the IC under test in the same chamber. Advantages to this approach include:
(1) The ability to identify when a failure occurs and, thereby, compute infant mortality rates as a function of burn-in time. As a result, an optimal burn-in time for each product family can be established.
(2) The ability to correlate burn-in failure rates with life test data typically obtained by IC manufacturers to determine the field failure rates of their products.
(3) The ability to incorporate into the burn-in process certain tests traditionally performed using automatic test equipment (ATE) systems, thereby reducing costs.
Some IC's have internal test modes not accessible during normal operation. These test modes may be invoked on ATE by applying a high voltage to a single pin. The IC is then addressed in a manner so as to specify the operating mode of interest. Operating modes such as data compression, grounded substrate, and cell plate biasing can be enabled, thus allowing evaluation of IC sensitivities and help in isolating possible failure mechanisms.
The electrical characterization data gathered from these tests is used to identify which part of the circuit appears to be malfunctioning, the possible location(s) on the IC, and the probable type or nature of the defect. To facilitate discussion and reporting, failures are often classified according to their electrical characteristics, referred to as the failure mode. Typical classification of these modes includes the following: single cell defect, adjacent cell defect, row failure, column failure, address failure, open pin, supply leakage, pin leakage, standby current leakage, and entire array failure (all dead cells).
In anticipation that some IC's will have defects, many IC's are designed with redundancies. In such IC's, a defective section of the IC may be shut off and a redundant but properly operating section activated and used in place of the defective section. For example, typical integrated memory circuits include arrays of memory cells arranged in rows and columns. In many such integrated memory arrays, several redundant rows and columns are provided to be used as substitutes for defective rows or columns of memory. When a defective row or column is identified, rather than treating the entire array as defective, a redundant row or column is substituted for the defective row or column. This substitution is performed by assigning the address of the defective row or column to the redundant row or column such that, when an address signal corresponding to the defective row or column is received, the redundant row or column is addressed instead.
To make substitution of the redundant row or column su
Beffa Raymond J.
Cloud Eugene H.
Farnworth Warren M.
Nevill Leland R.
Kerveros J.
Metjahic Safet
Micro)n Technology, Inc.
TraskBritt PC
LandOfFree
Reduced terminal testing system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Reduced terminal testing system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Reduced terminal testing system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2496196