EE 143 Microfabrication Technology Sring 010 Prof Clark T-C Nguyen Det of Electrical Engineering & Comuter Sciences University of California at Berkeley Berkeley, CA 9470 LecM 5 C Nguyen /14/10 1 Semiconductor Doing LecM 5 C Nguyen /14/10
Semiconductors are not intrinsically conductive To make them conductive, relace silicon atoms in the lattice with doant atoms that have valence bands with fewer or more e - s than the 4 of If more e - s, then the doant is a donor P, As The extra e - is effectively released from the bonded atoms to join a cloud of free e - s, free to move like e - s in a metal Doing of Semiconductors Extra free e - P Doe P The larger the # of donor atoms, the larger the # of free e - s the higher the conductivity LecM 5 C Nguyen /14/10 3 Doing of Semiconductors (cont) Conductivity Equation conductivity σ = qμ n n + qμ electron mobility electron density charge magnitude on an electron If fewer e - s, then the doant is an accetor B B Doe Lack of an e - = hole = h + When e - s move into h + s, the h + s effectively move in the oosite direction a h + is a mobile (+) charge carrier LecM 5 C Nguyen /14/10 4 hole mobility hole density B hole
Ion Imlantation LecM 5 C Nguyen /14/10 5 Ion Imlantation Method by which doants can be introduced in silicon to make the silicon conductive, and for transistor devices, to form, eg, n-junctions, source/drain junctions, The basic rocess Control current & time to control the dose B+ B+ B+ B+ B+ Charged doant accelerated to high energy by an E-Field (eg, 100 kev) Masking material (could be PR, could be oxide, etc) Result of I/I B-B-B-Bx Deth determined by energy & tye of doant LecM 5 C Nguyen /14/10 6
Ion Imlantation (cont) Result of I/I Ion collides with atoms and interacts with e - s in the lattice all of which slow it down and eventually sto it B B Damage layer at to becomes amorhous B not in the lattice, so it s not electrically active High Temerature Anneal (also, usually do a drive-in diffusion) (800-100 C) Now B in the lattice & electrically active! (serves as doant) This is a statistical rocess imlanted imurity rofile can be aroximated by a Gaussian distribution LecM 5 C Nguyen /14/10 7 Statistical Modeling of I/I Imurity concentration One std dev away 061N N(x) N Unlucky ions Avg ions Lucky ions std dev away 0,14N 3 std dev away 011N ΔR ΔR R ΔR ΔR Distance into material, x R ΔR Δ Δ Projected range = avg distance on ion trends before stoing Straggle = std deviation characterizing the sread of the distribution LecM 5 C Nguyen /14/10 8
Mathematically Area under the imurity distribution curve Analytical Modeling for I/I N( x) N ex ( x R ) ( ) ΔR = Imlanted Dose = Q = N ( x) dx [ ions / cm ] For an imlant comletely contained within the Q = π N ΔR Assuming the eak is in the silicon (utting it in one-sided diffusion form) So we can track the doant front during a D I = Q subsequent diffusion ste D ( x R ) ( I ΔR ) N( x) = ex, where ( Dt) eff = π ( Dt) ( ΔR ) eff LecM 5 C Nguyen /14/10 9 0 I/I Range Grahs, R Roughly roortional to ion energy R α ion energy (some nonlinearties) Figure 61 R is a function of the energy of the ion and atomic number of the ion and target material Lindhand, Scharff and Schiott (LSS) Theory Assumes imlantation into amorhous material, ie, atoms of the target material are randomly ositioned Yields the curves of Fig 61 and 6 For a given energy, lighter elements strike with higher velocity and enetrate more deely LecM 5 C Nguyen /14/10 10
I/I Straggle Grahs Results for and O surfaces are virtually identical so we can use these curves for both Figure 6 LecM 5 C Nguyen /14/10 11