(Last year's presentation slides, links to talks: www.cpmt.org/scv/meetings/cpmt1110w.html).
Thursday, October 25, 2012
Registration is now CLOSED for the live on-site Workshop, but you can still attend the Webinar, or get on our DList to hear when the individual talks are available as WebEx lectures from our website.
|
The live presentation at Cisco is completely filled. You may register yourself, plus others from your company/institution, for the live Webinar over the Internet. Since registration is now closed (Oct. 23), please email Paul Wesling, p.wesling@ieee.org to get the Webinar login information. This is a free event -- there is no registration charge. |
Location:
On the campus of Cisco Systems:
Speaker / Company | Title and Abstract of Talk |
---|---|
Nelson Tam, Marvell |
An Investigation of Effects of Alpha Source Uniformity on Estimation of Soft Error Rates
(come back for speaker handouts and WebEx presentation ...) In accelerated measurements of alpha SER FIT rate in memory test chips, non-uniformity of alpha flux from large area Am241 source has been found to have significant effects on testing results. In this work, the alpha flux uniformity of two AM241 alpha sources are compared using a 28nm FF memory test chip by exposing different parts of the source to induce errors. Using this method, the observed FIT rates have significantly different values (up to a factor of 5) depending on the areas of the sources used. In fact the center of the source is found to have the lowest flux. If acceleration factor for soft errors is calculated using the activity of the entire source, the SER will certainly be under-estimated. Additionally, it is found that our SER hardened FFs are much more robust when the Am241 source is at a distance greater than 1cm from the DUT. The results suggest that normal incident alpha particles do not have high enough LET to cause the hardened FFs to fail. |
Robin Gardiner, Matheson Gas |
A Materials Perspective on B10 Molecules for Advanced Si Technology Chips
(come back for speaker handouts and WebEx presentation ...) The most significant materials-specific source of cosmic radiation generated Soft-Error Rates (SER) is the fission products produced by the absorbtion of thermal neutrons in 10B nuclei. Natural boron is about 20% 10B and 80% 11B, with 10B having the highest neutron cross section of any common silicon device material. Through purification and optimized synthetic routes isotopically enriched 11B materials can be available for immediate use. Couple the materials solution with the fact that Tri-Gate structures provide a significant reduction in SER sensitivity; and the problem can effectively be removed as a concern. However the integrated cost of ownership for enriched materials presents some key decision points across multiple parties within the silicon device supply chain and end user groups. A brief background to the problem and materials solution will be covered. Integrated cost of ownership is discussed and definition of the decision requirements required to take advantage of the immediate access to the solution will be explored. |
Bharat Bhuva, Vanderbilt University |
Combined SER of Bulk 28nm Technology from Logic and Flip Flops
(come back for speaker handouts and WebEx presentation ...) With the ever decreasing charge requirements for generating a single-event transient (SET) pulse, combinational logic circuits are fast becoming a major contributor to overall soft-error rate (SER) of a sequential circuit. This talk will focus on the contribution of combinational logic circuits and flip-flops to the overall SER of a circuit. Experimental results for logic SER and flip-flop SER at a bulk CMOS 28 nm technology node will be presented. Test circuits were designed to obtain individual contribution of logic and FF SER. Results show that at GHz operating frequencies, the Logic SER exceeds that of FF SER for 5 MeV alpha particle exposure. Since most designers focus their efforts to FF hardening, future designs may not show any overall SER improvements even when extremely hardened FF designs are used due to dominant logic SER contribution. This will affect the whole design flow and hardening techniques being developed for circuits fabricated at advanced technology nodes. |
Jeff Wilkinson, Medtronic |
Measurement of Alpha Emissivity for Low- and Ultra-Low Alpha Materials
(come back for speaker handouts and WebEx presentation ...) Measurement of alpha emissivity for low- and ultra-low alpha materials requires specialized equipment and careful technique for accurate results. A previous multicenter study (Wilkinson et al, IRPS 2011) revealed a 2.3X range of values for a low alpha material that could not be attributed to known sources of uncertainty. A second study was subsequently designed to investigate the hypothesized sources of variation. Data collection for this 9 center trial has been completed and analysis of the results has continued to demonstrate variability. This talk will summarize the work to date and the results of subsequent modeling that may provide insight into the errors. |
Brendan McNally, XIA |
Preliminary Results from a Field Assessment of the UltraLo-1800 Alpha Particle Counter
(come back for speaker handouts and WebEx presentation ...) Alpha emissivity measurements are used by the semiconductor industry to assess the suitability of materials for use in production processes. Recent work has been published which shows a wide variation in alpha emissivity results from measurements of the same sample conducted at several alpha counting centers, almost all of whom used gas proportional counters. A next-generation counter developed at XIA (the UltraLo-1800) operates in a fundamentally different manner from the gas proportional counters, and was recently made available to a handful of early customers. In this work we briefly review how the UltraLo-1800 operates, and examine the preliminary results of a round-robin field assessment conducted with the initial customers. We will explore how much variation is observed between participants in the measurement of a set of standard samples, and attempt to determine if observed variations are systematic by participant. |
Mike Gordon, IBM |
Alpha Testing - How Long is Long Enough?
(come back for speaker handouts and WebEx presentation ...) In this work we discuss the time required to make accurate alpha particle measurements of ultra-low emissivity (e< 2a/khr/cm2) samples. This time is related to the counter background, the level of radon adsorption and its decay, and counting (Poisson) statistics. We show examples of each using an early commercial version of the XIA UltraLo-1800 ionization counter. The lowest emissivity we have measured to date, on a variety of known ultra-low alpha-particle emissivity samples, is about 0.3 a/khr/cm2. We have measured the alpha-particle emissivity of the same sample at two locations within our research laboratory; both underground, and on the second floor, and we observed ~2X difference in the number of alpha particles detected. We will present modeling results which show that neutron-induced reactions on silicon samples and in the argon counter gas can produce a measurable alpha-particle emissivity (that needs to be accounted for). We will also present a review of Poisson statistics relevant for the time distribution between adjacent detected alpha-particle events. We show modeling results that demonstrate the variability of emissivity data one might obtain for repeated multi-day experiments, and use these simulated data to discuss the confidence intervals in the measurements. The effect of long or short measurement times on the confidence interval is discussed. |
Panel Discussion with all speakers |
(come back for speaker handouts and WebEx presentation ...) Our authors and industry experts will discuss critcal issues related to alpha emissivity measurments, device performance, and test methodologies for advanced Si nodes. |
Shah Jahinuzzaman, Intel |
Alpha-particle Induced Soft Error Rates in 22nm Bulk Tri-Gate Technologies
(come back for speaker handouts and WebEx presentation ...) In this paper, we present the measured alpha-particle soft error rate (SER) of latches and flip-flops implemented in 22nm high-k + metal gate bulk Tri-Gate technology. While the high energy neutron SER in 22nm Tri-Gate devices can be up to 4x lower compared to its 32nm planar counterparts, the observed alpha-particle induced SER reduction is in excess of 10x. This benefit can be attributed to a significant decrease in the charge collection volume with no or minimal decrease in the critical charge. |
Brett Clark, Honeywell |
The Distribution and Transport of Alpha Activity in Tin
(come back for speaker handouts and WebEx presentation ...) The presence of alpha emitters in packaging materials pose an increased risk to device reliability as device size decreases and flip chip and 3-D packaging technology becomes more widespread. Particularly troubling is the observation that alpha emission increases over time in some reported instances. Several experiments were conducted to examine alpha flux behavior in tin materials and determine the cause for the temporal alpha emissivity increase. The experiments determined the distribution of the alpha emitters within the material volume, and a mechanism based on microsegregation is presented to explain the non uniform distribution. Data describing differential isotope distribution within tin are also presented, providing confirmation of 210Po transport within a solid tin matrix. The effects of microsegregation and 210Po migration on alpha emissivity over time will be discussed. Potential impacts from these mechanisms on alpha emissivity in packaging applications will also be presented. |
Yi-Pin Fang, TSMC |
Cross-Technology Alpha SER and Thermal Neutron SER Trend Prediction For Memory and Logic Devices
(come back for speaker handouts and WebEx presentation ...) Single event upsets due to energetic alpha particles emitted from package materials now dominates SER for advanced technology nodes. Moreover, thermal neutron SER for memory and logic devices has become a threat from the 65nm technology and beyond since B2H6 carrier gas containing the natural B10 isotope is used during W plug processing. Here we describe a simple model to predict alpha and thermal neutron SER across-technology nodes and validate the model with experimental data. While alpha particle SER continues to be concerned, we show that thermal neutron SER decerases significantly with technology progression. |
Sang H. Baeg, Hanyang University |
High-Energy Alpha Particle SEU Measurements for Flip Chip Devices
(come back for speaker handouts and WebEx presentation ...) Alpha SEU measurement methodology is introduced for flip chip with high energy alpha particles. The flip chip has silicon substrate depth more than 200 um. In typical alpha SEU measurements, alpha particles below 10MeV are used. They do not have sufficient energy to penetrate the substrate and to cause the upsets in active elements. Instead of thinning the substrate or other apparatus for low energy alpha to penetrate the thick silicon, the high energy alpha is alternatively used to study alpha upset in silicon. In one of flip chip SRAM testing, about 23 MeV energy of alpha particle was needed to start cause SEU, which showed the validity of the proposed methodology. |
SCV Chapter
Home Page |
How to Join IEEE |
Contact our Chapter Chair |
CPMT Society
Home Page |
IEEE Home Page |
Email
to Webmaster |
Last updated on