1. 1 Materials Beyond Silicon Materials Beyond Silicon By Uma Aghoram

2. TI Fellows Forum 2 Moore’s Law MSILSIVLSI Moore’s Law states that the number of transistors on a chip doubles about every two years. At each stage of scaling fundamental limits are being reached

3. TI Fellows Forum 3 SCALING LIMITS Lithographic Limits –XRay and E-beam in future Short channel effects –Vt roll off –DIBL Oxide thickness scaling reaching a few atomic layers of silicon –Large leakage currents –Higher power dissipation

4. TI Fellows Forum 4 Exponential Forever ? 130nm90nm60nm45nm30nm ?

5. TI Fellows Forum 5 KEEPING MOORE’S LAW ALIVE Strained Silicon Novel Device structures High K dielectrics Carbon Nanotubes, SET Alternate channel materials

6. TI Fellows Forum 6 REPLACING SILICON Cost effective Easy incorporation into existing technology Reliable oxide Performance enhancement

7. TI Fellows Forum 7Germanium ADVANTAGES Has high electron and hole mobility as compared to silicon – High speed high-K dielectrics Good compatibility with III-V materials DISADVANTAGES No reliable native oxide Small Bandgap may lead to larger tunneling currents Material nn hh Silicon1450500 Germanium39002270

8. TI Fellows Forum 8 Gallium Arsenide and other III-V materials ADVANTAGES Has 6X larger electron mobility as compared to silicon – High speed Have excellent Optoelectronic properties High resistance to radiation damage High flexibility to forming alloys Heat resistant Material nn hh Silicon1450500 GaAs9200400

9. TI Fellows Forum 9 DISADVANTAGES No reliable native oxide Composite nature leads to high defect density- Unstable Vt More expensive to manufacture pure GaAs to meet industry standard Lower thermal conductivity than Silicon More fragile More quantum effects Small recombination time results in poor performance of GaAs BJTs. High leakage current in narrow gap III-V materials Gallium Arsenide and other III-V materials

10. TI Fellows Forum 10 Where do we stand? Germanium Strained Si Mosfets with SiGe layers have been manufactured Dual channel heterostructure MOSFETS are being researched. Gallium Arsenide: Bell Labs and Freescale recently reported that they have grown high quality Ga 2 O 3 on GaAs with low interface states and achieved enhancement and depletion mode MOSFETs. HBT’s, HEMT, Power transistor

11. TI Fellows Forum 11 Toy Problem Position of charge centroid in a MOSCAP for three different materials namely: –Silicon –Germanium –Gallium Arsenide The position where the charge peaks is of relevance as it determines additional thickness of the dielectric and effectiveness of gate control of device

12. TI Fellows Forum 12Assumptions One dimensional problem One band effective mass Hamiltonian Voltage on gate directly applies to the channel 20A of oxide and varying well widths Wave functions are strongly excluded form the oxide Lattice spacing and dimension varies with material (100) surface Inversion charge density was a constant at 1e13/cm2 insulator channel source drain X Z Vg 0

13. TI Fellows Forum 13PROCEDURE 1-D Schrodinger Poisson Solver U(r) n(z) [H z +U]  m =  m  m

14. TI Fellows Forum 14Procedure From Schrödinger part of the solution we get the eigen values and eigen functions. We can then calculate the electron density n(z) Using the Poisson's equation we then solve for the self consistent potential, and use this new potential in the Schrödinger's equation and thus solve self consistently. For the Hamiltonian the mass to be used is the out of plane effective mass and for the density of states expression use the DOS mass.

15. TI Fellows Forum 15Results Si Ge GaAs InSb N inv =1e13/cm2

16. TI Fellows Forum 16CONCLUSIONS As the mass reduces the quantum confinement effects are more significant and peaking of the charge concentration takes place deeper in the substrate. This means that the effective gate capacitance is smaller in Ge and GaAs as compared to silicon Also there is better gate control of channel in silicon devices as compared to GaAs and Ge Higher voltages are required by III-V materials to reach the same inversion charge density when compared to silicon Lower transconductance in materials with lower mass.

17. TI Fellows Forum 17 Existing Results

18. TI Fellows Forum 18 Current Research Germanium The main supporters of Germanium are Sematech, IBM, Umicore and Soitech. A lot of research is also being carried out in this field by IMEC in partnership with Umicore and Soitech. GaAs Intel is one of the main supporters of GaAs Quintec researches in this area

19. TI Fellows Forum 19 References and Acknowledgement Quantum Transport – Atom to transistor Supriyo Dutta Cambridge university press Physics of Strain Effects in Semiconductors and MOSFETs Y. Sun, S. E. Thompson, and T. Nishida, awaiting publication Gallium Arsenide –GaAs as a semiconductor, its turbulent past, shaky present and promising but distant future Aseem Srivastava IEEE 1989 http://www.edn.com/article/CA6314526.html?ref=nbra http://www.investorrelations.umicore.com/en/pressReleases/2003/germanium_E.p df Fundamentals of modern Vlsi devices Taur and Ning http://mems.caltech.edu/courses/EE40%20Web%20Files/effective%20mass%20exp lanation.pdf Indium Phosphide and related materials, 2005 International conference on www.intel.com ftp://download.intel.com/research/silicon/Gordon_Moore_ISSCC_021003.pdf www.compoundsemiconductor.net/articles/news/10/1/25/1

20. 20 Thank you!! Questions???

21. TI Fellows Forum 21