Data processing: measuring – calibrating – or testing – Measurement system – Statistical measurement
Reexamination Certificate
2001-11-02
2004-02-03
Hoff, Marc S. (Department: 2857)
Data processing: measuring, calibrating, or testing
Measurement system
Statistical measurement
C710S058000
Reexamination Certificate
active
06687648
ABSTRACT:
FIELD OF THE INVENTION
Embodiments of the present invention relate to the manufacture, test and qualification of integrated circuits. More particularly, embodiments of the present invention provide a method and computer aided system for predicting the reliability of oxide-nitride-oxide non-volatile memory.
BACKGROUND ART
Silicon Nitride based Non-Volatile Memory has many advantage as compared to its floating gate and tunneling oxide based counterparts. Silicon-Oxide-Nitride-Oxide-Semiconductor (SONOS) is potentially very dense in terms of number of cells per unit area that can be used and it requires fewer process steps as compared to the floating gate memory. Moreover, it can be easily integrated with the standard SRAM technology. The other advantages of using SONOS devices include their suitability for applications requiring large temperature variations and radiation hardening. There has been much apprehension in using SONOS on a wide scale in the industry because the behavior of SONOS with respect to data retention and endurance has been found unpredictable especially with variations in temperature.
There is a need for an understanding of the physical mechanism and a working model to predict the behavior of SONOS in order to reap the benefits of this technology. Attempts have been made in the past to model the physical mechanism of data retention at high temperatures. Miller, McWhorter, Dellin and Zimmermann (Journal of Applied Physics 67(11), Jun. 1, 1990, pp 7115-7124) have studied in detail the performance of “Excess Electron” states and “Excess Hole” states and their differences. The data used for this study is taken from a transistor device. A significant difference is behavior has been shown between the “Excess Electron” and “Excess hole” threshold states when the programming temperature is different from the storage temperature and the storage temperature is varied.
Other studies have led to conflicting conclusions. A study by Ross, Goodman and Duffy (RCA rev. 31,467-1970) showed that the threshold voltage decay rate of Metal-Nitride-Oxide-Silicon (MNOS) transistors was insensitive to temperatures for temperatures up to 125 degrees C. A study on NMOS transistor cycling endurance with temperature variations by Neugebauer and Burgess (Journal of Applied Physics 47,3182-1976) suggested that while charge injection for both threshold states (Excess Electron and Excess Holes) was temperature dependent, the decay rate of only the excess hole state was increased by elevating the temperature. They also found that the programming temperature, relative to the storage temperature affected the subsequent decay rate. Williams and Beguwala (IEEE Transactions on Electron Devices ED-25, 1019-1978) have found that the decay rate of only the excess electron state of MNOS transistors was temperature dependent.
Much work has been done in the past to model the erase and program properties of the SONOS non-volatile memory. Many researchers including Minami and Kamigaki (“A Novel MONOS Nonvolatile Memory Device Ensuring 10-Year Data Retention after 10
7
Erase/ Write Cycles,” IEEE Transactions on Electron Devices, vol. 40. No.11, November 1993, pp. 2011-2017) have suggested that data is written by electrons injected from the semiconductor substrate by modified Fowler-Nordheim tunneling through the SiO
2
and part of Si
3
N
4
layer and are stored at traps in the Si
3
N
4
layer. On the other hand, the holes too are injected from the semiconductor by direct tunneling through the SiO
2
layer and are stored at the traps in the Si
3
N
4
layer.
There are theories about the physical location of the charge traps as well. R. Paulsen et al., (R. Paulson, R. Siergiej, M. French, M. White, “Observation of Near-Interface Oxide Traps with Charge -Pumping Technique,” IEEE Transactions on Electron Devices, Vol. 13, No. 12, December 1992, pp 627-629) suggest that the traps in silicon nitride are at a well-defined trapping distance corresponding to the tunnel oxide thickness.
While all the work done in the past explains some portion of the SONOS memory program and erase mechanism, there is a need for an explanation of the behavior of SONOS memory over the entire range of temperature (e.g., the industrial temperature range) and a lifetime of program-erase cycles. The advantage of having such a prediction of data retention is that reliable products could be produced at a cost advantage compared to other types of non-volatile memories.
SUMMARY OF THE INVENTION
Therefore, it would be advantageous to provide a method and computer implemented system providing for predicting the reliability of oxide-nitride-oxide based non-volatile memory. A further need exists for a method of predicting the yield of semiconductor devices containing oxide-nitride-oxide non-volatile memory at wafer sort. A still further need exists for a method of determining activation energies for data retention and threshold voltages in oxide-nitride-oxide non-volatile memory.
A purpose of this invention is to provide a tool and a complete process to predict the life of a SONOS Non Volatile Memory element with respect to data retention at various temperatures. This invention provides relationship which utilizes the bake temperature and initial threshold margin (difference between the program threshold and the erase threshold) as input and produces the predicted time taken to reach the final threshold margin as the output. A calculation of constants is required which involves a few data points on threshold voltage margin, time and temperature as input when the model is used for a new process. The model allows yield improvement and shortens test times. The model permits the estimation of data retention life in the field based on simple test data collected in well-known procedures.
A method for predicting the reliability of oxide-nitride-oxide (ONO) non-volatile memory is disclosed. ONO memory devices may be programmed. Margin voltages may be recorded initially, and during baking at 100 degrees C. and 300 degrees C. From this data, constants and activation energy may be determined through a first formula. Frenkel-Poole activation energy may be determined using well-known methods such as charge pumping. According to embodiments of the present invention, through the use of a second formula, decay time of the information stored in the ONO memory may be predicted from the activation energy. The first formula may also be used to predict the decay time. The two decay time predictions may be compared to establish confidence. In this manner, data retention of an ONO memory may be reliably predicted, which was not possible in the conventional art.
Another embodiment of the present invention may use predictions of data retention of ONO memory in sorting integrated circuit devices containing ONO memory.
This is a first time invention of a model to predict the retention lifetime for data stored in a SONOS structure. Previous calculations have all tried to determine the distribution of charge storage or trapping sites, which is difficult to predict in a chemical vapor deposition deposited nitride and hard to calculate. This method circumvents this problem by calculating the change in the height of the oxide potential barrier which is dependent upon the decayed charge. Previously attempted methods have a strong dependence upon the quality of nitride and estimation of the location of “emission frontier” and therefore do not provide good results over the entire range of temperatures and quality of nitride. There is no standardized method to do this calculation.
The products must pass qualification requirements which involves a minimum of ten years of life with respect to data retention. Due to the lack of a model it has not been possible so far to determine how long the required data retention bake time should be. This process utilizing the model mentioned above for the first time provides a tool to accomplish this task. This process will drastically reduce the test time required to find reliable dies (with respect to data retention) in production line.
REFER
Kumar Santosh
Russell Edmund L.
Cypress Semiconductor Corporation
Hoff Marc S.
Raymond Edward
Wagner , Murabito & Hao LLP
LandOfFree
Method of predicting reliabilty of oxide-nitride-oxide based... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of predicting reliabilty of oxide-nitride-oxide based..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of predicting reliabilty of oxide-nitride-oxide based... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3291024