HUP: Direct Ionization-Induced SEE Rate Calculation
Overview
The HUP module in CREME96 calculates the single-event effect (SEE) rate due to direct ionization. For most (but not all) devices, protons produce too little direct ionization to cause SEEs. Thus, upsets and other single-event effects due to direct-ionization are often colloquially referred to as "heavy-ion-induced". (Hence, this program module's name, "HUP".)
(Protons are more likely to cause SEEs through nuclear reactions, which produce highly-ionizing recoil nuclei. You should also run the PUP module to evaluate the rate of proton-induced SEEs.)
Inputs
Inputs to the HUP routine are:
Integral LET Spectrum
This is provided by an .let file created by the LETSPEC routine. This file must be created before you run HUP.
Device Label
In your inputs to the HUP and PUP routines, you must provide a label for each SEE rate calculation, such as "93L422" or "RR160-25", for record-keeping purposes. (The quotes are not necessarily part of the label.) The label may have a maximum of 12 characters.
In addition, you may also specify for each calculation two comments, each with a maximum of 40 characters. These comments may contain additional useful information, such as "Cross-section fit from Stapor data".
These device labels and comments will be displayed on the screen at the end of the calculation and copied to the HUP/PUP Report File.
RPP Dimensions
HUP calculates direct-ionization SEE rates using the integral rectangular parallelepiped (IRPP) method. You must supply the RPP dimensions (in microns). Optionally, you can also include funnels.
Number of Bits per Device
The CREME96 SEU rate calculation returns the SEU rate on both a per-bit and per-device basis. You must specify the number of bits per device. By default, the number of bits per device is initially set to one
NOTE: All SEU cross-section data must be input on a per bit (NOT per device!) basis. See HUP SEU Cross Section Data Inputs for further details.
SEE Cross Section Parameters
- A Weibull function can be used to specify the "heavy-ion" SEE cross-section for the IRPP calculation. The fit parameters must yield the cross-section (in square microns/bit) as a function of effective LET (in units of MeV-cm2/milligram; NOT /gram!)
- Alternatively, the cross-section may also be specified in a table of cross-section values with values in the same units as above.
- As an option, CREME96 also supports the Critical Charge Method from the old CREME(1986) code, in which the LET-dependent cross-section is modeled as a step function. The critical charge must be given in picoCoulombs (= 10-12 Coulombs).
- Heavy-ion cross-sections must be specified in units of square-microns/bit (not device!) vs effective LET in MeV-cm2/milligram (not gram!)
- The reason for this requirement is that CREME96 SEE-calculations are done in the context of the RPP model. Thus, a key ingredient of the calculations is the distribution of path lengths through the sensitive volume of the bit. Those distributions are radically different when the lateral dimensions are those of a bit or of the device as a whole. For example, suppose you have a device consisting of 10000 bits, each with a 1 x 1 x 1 microns sensitive volume. A single sensitive volume with dimensions of 100 x 100 x 1 microns is not equivalent, since it contains many, many chords through it much longer than 1.732 microns, which is the maximum chord length in the real sensitive volume. Treating the whole device as a single sensitive volume therefore leads to gross overestimates of the SEE-rate, by incorrectly including contributions of low-LET particles traveling along very long paths, which do not actually exist.
- RPP dimensions of the bit sensitive volume must be specified in microns.
- If you set x=y=0 in the RPP dimension input fields, HUP will automatically reset them to the square root of the limiting cross-section per bit.
- Alternatively, if you put zero in the limiting cross-section input field under the Weibull and Critical Charge options in HUP, the limiting cross-section will be automatically reset to "xy", the product of the lateral RPP dimensions.
- If you set x=y=0 and the limiting cross-section = 0, HUP will return an SEU rate of zero.
See also HUP SEU Cross Section Data Inputs
Output
In a single run of HUP (that is, with a given input LET file), you can do calculations for as many as 10 devices and/or different cross-section parameters. Results from HUP are written both to the screen and to an output file, which records both your inputs and the calculated results.
SEE Rate Report File
These files are produced by the HUP and PUP routines. They record the results of your SEE rate calculations, as well as the various input parameters.
Note that each click of the SUBMIT button on the HUP/PUP pages will produce a new report file. Since the HUP/PUP input pages allow you to specify inputs for up to 10 devices at once, the results of many calculations, all with the same input .let file (for HUP) and .tfx/.flx file (for PUP), can be included in one report file. This allows for easy comparison among devices in a given environment.
In order to locate your report files, the output file from HUP is called "something.hup", where "something" is the rootname you specify and the extension ".hup" is added automatically. Similarly, report files from the PUP program are named "something.pup".
At present, these files are simply ASCII text. You can review their contents on the screen. You can also download these files from the CREME server to your computer.