메뉴 건너뛰기




Volumn 104, Issue 6, 2008, Pages

Boosting the on-current of a n -channel nanowire tunnel field-effect transistor by source material optimization

Author keywords

[No Author keywords available]

Indexed keywords


EID: 54749153664     PISSN: 00218979     EISSN: None     Source Type: Journal    
DOI: 10.1063/1.2981088     Document Type: Article
Times cited : (134)

References (27)
  • 8
  • 12
    • 54749106716 scopus 로고    scopus 로고
    • An analytical model exists for a simplified TFET with a non-conventional gate design (Ref.) as well as for a one-dimensional TFET (Ref.).
    • An analytical model exists for a simplified TFET with a non-conventional gate design (Ref.) as well as for a one-dimensional TFET (Ref.).
  • 14
    • 54749099722 scopus 로고    scopus 로고
    • Proceedings of the 63rd DRC, (unpublished), Vol.,.
    • J. Knoch and J. Appenzeller, Proceedings of the 63rd DRC, 2005 (unpublished), Vol. 1, p. 153.
    • (2005) , vol.1 , pp. 153
    • Knoch, J.1    Appenzeller, J.2
  • 15
    • 54749143705 scopus 로고    scopus 로고
    • note
    • MEDICI, Synopsys, Version 2007.12. MEDICI checks for the amount of band bending and only allows tunneling when there are valence band states at one end and conduction band states at the other end of the tunnel path. A limitation of MEDICI is that only Kane's model is available, which is a band- to-band-tunneling model for direct-bandgap materials. Even though the model parameters have been tuned for silicon, this implies that there may be quantitative deviations with experimental results. A limitation of MEDICI's implementation of Kane's band-to-band-tunneling model is that the E-k relation is not as shown in Fig., but instead a less accurate implementation is used whereby the value of the k -vector increases with the square-root of the energy-barrier which still has to be bridged (Ref.). This not only results in inaccuracies for band-to-band tunneling in one material, but for tunneling across heterojunctions, the inaccuracies can be quite significant because only the energy barrier which has to be bridged is considered (the energy barrier is given by the integral of the local electric field along the tunnel path) and therefore there is no dependence on the actual bandgap at the start and end point of the tunnel path. A second limitation of the band-to-band-tunneling model implementation is that only parameters which apply to the full tunnel path can be specified (model flags A.BTBT and B.BTBT), although the tunnel probability is dependent on local material parameters, like the reduced effective mass. Technical action is being taken by Synopsys and the above limitations may be removed in future product releases. As stated in the Release Notes of version 2007.03, previous versions of the program had a problem with non-local band-to-band tunneling across a heterojunction, because electrons were generated at the wrong place.
  • 16
    • 33748417557 scopus 로고    scopus 로고
    • 0163-1829 10.1103/PhysRevB.74.121302, (R).
    • F. Glas, Phys. Rev. B 0163-1829 10.1103/PhysRevB.74.121302 74, 121302 (R) (2006).
    • (2006) Phys. Rev. B , vol.74 , pp. 121302
    • Glas, F.1
  • 17
    • 54749155677 scopus 로고    scopus 로고
    • ITRS Roadmap
    • ITRS Roadmap, http://www.itrs.net
  • 18
    • 54749114139 scopus 로고    scopus 로고
    • The on-current is twice the actual ITRS on-current spec, to compensate for the fact that an effective oxide thickness of half the ITRS spec is used. The latter choice has been made because of the much stronger dependence of the TFET on-current on the gate dielectric thickness than the MOSFET (Ref.).
    • The on-current is twice the actual ITRS on-current spec, to compensate for the fact that an effective oxide thickness of half the ITRS spec is used. The latter choice has been made because of the much stronger dependence of the TFET on-current on the gate dielectric thickness than the MOSFET (Ref.).
  • 19
    • 54749100526 scopus 로고    scopus 로고
    • Ph.D. thesis, TU München.
    • P. -F. Wang, Ph.D. thesis, TU München, 2003.
    • (2003)
    • Wang, P.-F.1
  • 20
    • 50549156338 scopus 로고
    • 0022-3697 10.1016/0022-3697(60)90035-4.
    • E. O. Kane, J. Phys. Chem. Solids 0022-3697 10.1016/0022-3697(60)90035-4 12, 181 (1960).
    • (1960) J. Phys. Chem. Solids , vol.12 , pp. 181
    • Kane, E.O.1
  • 21
    • 0002930518 scopus 로고
    • 0021-8979 10.1063/1.1735965.
    • E. O. Kane, J. Appl. Phys. 0021-8979 10.1063/1.1735965 32, 83 (1961).
    • (1961) J. Appl. Phys. , vol.32 , pp. 83
    • Kane, E.O.1
  • 23
    • 54749139397 scopus 로고    scopus 로고
    • The device simulator only has a model available for band-to-band tunneling in direct-bandgamaterials (Ref.). It therefore seem most appropriate to us to also use a direct-bandgaapproximation for the E-k diagrams of silicon and germanium (see Fig.) as these diagrams are used to complement the simulation results.
    • The device simulator only has a model available for band-to-band tunneling in direct-bandgap materials (Ref.). It therefore seem most appropriate to us to also use a direct-bandgap approximation for the E-k diagrams of silicon and germanium (see Fig.) as these diagrams are used to complement the simulation results.
  • 25
  • 26
    • 54749137090 scopus 로고    scopus 로고
    • It is assumed that the electron occupation probability of the valence band remains close to 1 at the location of the shortest tunnel barrier, which is typically the case in our simulations. This may however no longer be true for extremely high source doping and/or when considering materials with extremely low effective density of states in the valence band.
    • It is assumed that the electron occupation probability of the valence band remains close to 1 at the location of the shortest tunnel barrier, which is typically the case in our simulations. This may however no longer be true for extremely high source doping and/or when considering materials with extremely low effective density of states in the valence band.


* 이 정보는 Elsevier사의 SCOPUS DB에서 KISTI가 분석하여 추출한 것입니다.