Phase Transformations in Microelectronics

1
Phase Transformations in Microelectronics
Rewriteable data storage Universal
non-volatile memory
Non-equilibrium phases at silicon-dielectric
interfaces
2
(No Transcript)
3
Writing Erasing Data
4
Time-Temperature-Transformation (TTT) Diagrams

• We learned about the concept underlying TTT
  diagrams within the context of steels
• But this is a universal concept
• Short, high-intensity laser pulse, followed by
  cooling abruptly results in amorphous material
  (if we miss the nose of the C-shaped curve)
• Long, low-intensity laser pulse will result in
  crystalline material

5
Present Day Technologies Using Phase Change
Materials
6
(No Transcript)
7
Discovery of phase change materials
Chalcogens Alloys containing Te, Se, S
8
Structure of Ge2Sb2Te5 or Ge1Sb2Te4

• Rock-salt (or sodium chloride) structure
• Te atoms occupy one sub-lattice of crystal
• Ge, Sb and vacancies occupy other sub-lattice
• Vacancies can be formed easily, and enables
  formation of amorphous phase

9
Mechanism of writing erasing
10
Phase Change Alloy Design

• Octahedral-like atomic arrangement
• Average valency of about 5

11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
Electronics Industry Moores Law
Size of devices halves every 18 months
15
Alternative (high-k) gate dielectrics
Thickness d
A

• Miniaturization ? shrinkage of A
• But, capacitance kA/d, and needs to be
  maintained at a constant value (k is dielectric
  constant)
• Industry has been using SiO2 dielectric for the
  last 45 years, and has accomplished
  miniaturization by decreasing d
• So far, the industry has been decreasing d, but
  we will soon reach the atomic limit
• Todays chips have d 20 Angstroms, and in a few
  years, d will shrink to less than 10 Angstroms
• The other possibility is to increase k, which
  means need to identify new material
• But there are serious problems with any
  replacement for SiO2

16
The Excitement Surrounding high-k Dielectrics

• Moores Law The International Technology
  Roadmap for Semiconductors requires continued
  shrinkage of electronic devices
• Miniaturization ? Decrease L for constant
  capacitance kL2/d
• Can be accomplished by decreasing d, or
• by replacing SiO2 by other dielectrics (e.g.,
  HfO2, Hf silicates, etc.) with larger dielectric
  constant (high-k)

(Craig R. Barrett, MRS bulletin 2006)
17
Potential High-k Dielectrics

• Six important criteria
• HfO2 meets the majority of these, and is hence
  considered as the most promising replacement for
  SiO2

18
Potential High-k Dielectrics
Unfortunately, the Band Gap of a dielectric bears
an inverse relationship with dielectric constant.
So, there is no miracle dielectric which will
have a large band gap and high K value. Since
neither of these values can be low, a good
dielectric candidate must be a compromise of the
two considerations.
19
High-k Metal Oxide DielectricsIssues
Challenges!

• HfO2 on Si is thermodynamically stable i.e., the
  following reactions are endothermic, based on
  bulk thermodynamics
• HfO2 Si ? Hf SiO2 (silica formation)
• HfO2 2Si ? HfSi SiO2 (silicide formation)
• 2HfO2 Si ? Hf HfSiO4 (silicate formation)
• However, all the above form when a thin layer of
  HfO2 is deposited on Si and annealed
• which is detrimental as interfacial layer
  impacts band offsets, threshold voltages, traps
  electrons, decreases dielectric constant, etc.

20
O Interstitials At InterfaceThermodynamics
Kinetics
Si
HfO2
Hf
Interfacial segregation Thermodynamically favored
O
Kinetic driving force further favors interfacial
segregation, and prevents O penetration into Si
O interstitials could lead to the formation of
SiOx
C. Tang R. Ramprasad, Phys. Rev. B 75, 241302
(2007)
21
Microelectronics The Near Future

• HfO2 based dielectrics grown on Si with well
  controlled stoichiometry
• HfO2 based dielectrics with modifications (e.g.,
  dopants)
• Dopants such as Si, Y, La, etc., are known to
  control defect chemistries, and stabilize the
  amorphous phase (which is preferred as grain
  boundaries are eliminated)
• The metal-HfO2 interface is still poorly
  understood

both phase change media next-gen transistors
require a detailed understanding of
thermodynamics, kinetics and phase
transformations in materials, and materials
design