
Even experts in the field of ESD design, testing, and failure analyses have difficulty in the accurate use of the words "always", "never", "all", "none" and "I know." Because of the nature of ESD energy, the ESD protection circuitry needs to be re-examined every time the semiconductor technology changes. The key characteristic of ESD energy (just as with lightning) is that it will follow "the path of least resistance". Therefore, in order to guarantee that an integrated circuit meets the "robust" ESD requirements of customers, extensive testing through all possible "pin-to-pin" combinations must be performed.
Also, it is important to understand that good ESD protection is achieved by making good "IC level design decisions", not just an "IO pad design decision." The "IC level design decisions" involve how a given padring is designed as well has how the core components of the IC design interconnect to the outside world through the padring. This has become particularly important as more analog circuitry has been brought on-chip and RF functions have been integrated into the single IC with the advent of SiGe technologies.
The key to good ESD design is to "predict" the paths that the ESD energy might flow and then either "manage" the current flow or provide an alternate path for the current to flow. Because of the high voltages, it is seldom (almost never) an option to "block" the ESD energy. The key behind the current flow management is power management. Because the ESD energy is stored on a known size capacitor at a known voltage, the total energy that is being discharged through the device is known. The sudden discharge of this energy causes significant voltage transients as well as concentrated power that creates "hot spots" within the integrated circuit. These two items (voltage and power) cause nearly all the failures identified in ESD testing. Therefore, prediction and management of the high current flow is key to good protection. High voltages cannot be allowed across the ultra-thin gate oxides and power must be "spread out" (distributed).
There are three basic types of failures:
As semiconductor technology advances into the "deep-submicron" arena, transistors continue to shrink in gate thickness, transistor length, transistor widths, dielectric insulator thickness, conductor thickness, and device/conductor spacings. In addition to shrinking, new conductors and materials are being introduced to provide additional capability. All of these changes can and do affect ESD performance (tolerance). (Dimensions are generic and not specific to any foundry).
|
Name of Technology |
Description Condutor Layers (minimum sizing) |
Conductor Dimensions (Microns) |
Gate Oxide |
||||
|---|---|---|---|---|---|---|---|
|
Metal 1 |
Poly |
||||||
|
Width |
Space |
Width |
Space |
Microns |
Angstroms |
||
|
SPDM_1.0 |
Single Poly Double Metal (1.0 Micron) |
1.7 |
1.3 |
1.0 |
1.5 |
0.19 |
180 to 200 |
|
SPDM_0.8 |
Single Poly Double Metal (0.8 Microns) |
1.2 |
1.2 |
0.8 |
1.2 |
0.17 |
160 to 180 |
|
SPDM_0.6 |
Single Poly Double Metal (0.6 Microns) |
1.0 |
0.8 |
0.6 |
0.9 |
0.15 |
135 to 165 |
|
SPTM_0.5 |
Single Poly Triple Metal (0.5 Microns) |
0.6 |
0.7 |
0.5 |
0.75 |
0.12 |
110 to 130 |
|
DPTM_0.35 |
Double Poly Triple Metal (0.35 Microns) |
0.5 |
0.5 |
0.35 |
0.65 |
0.0075 |
60 to 90 |
|
DPQM_0.25 |
Double Poly 4 to 6 Metal (0.25 Microns) |
0.35 |
0.35/0.6 |
0.24/0.25 |
0.5 |
0.005 |
40 to 60 |
|
TPQM_0.18 |
Triple Poly 4 to 6 Metal (0.18 Microns) |
0.3 |
0.3/0.5 |
0.18 |
0.4 |
0.0035 |
30 to 40 |
|
QPQM_0.13 |
Four Poly & "N"-- Metal (0.13 Microns) |
0.18 |
0.2/0.4 |
0.13 |
0.2 |
0.002 |
17 to 24 |
|
QPQM_0.09 |
Four Poly & "N"-- Metal (90 nanometers) |
0.15 |
0.16/0.3 |
0.1 |
0.14 |
0.0015 |
14 to 16 |
|
QPQM_.065 |
Four Poly & "N"-- Metal (65 nanometers) |
0.1 |
0.12-0.2 |
0.07 |
0.1 |
0.0013 |
12 to 14 |
** Shaded rows show technology with "salicided" diffusions and poly gates.