Lecture 15 OUTLINE MOSFET structure & operation (qualitative) Review of electrostatics The (N)MOS capacitor Electrostatics t ti Charge vs. voltage characteristic Reading: Chapter 6.1 6.2.1 EE105 Fall 2007 Lecture 15, Slide 1 Prof. Liu, UC Berkeley
The MOSFET Metal-Oide-Semiconductor Field-Effect Transistor: GATE LENGTH, L g Gate OXIDE THICKNESS, T o Source Drain Substrate JUNCTION DEPTH, X j M. Bohr, Intel Developer Forum, September 2004 Current flowing through the channel between the source and drainis controlled by the gate voltage. N-channel & P-channel MOSFETs operate in a complementary manner CMOS Complementary MOS CURRENT V TH GATE VOLTAGE EE105 Fall 2007 Lecture 15, Slide 2 Prof. Liu, UC Berkeley
N Channel MOSFET Structure Circuit symbol The conventional gate material is heavily doped polycrystalline silicon (referred to as polysilicon or poly Si or poly ) Note that the gate is usually doped the same type as the source/drain, i.e. the gate and the substrate are of opposite types. The conventional gate insulator material is SiO 2. To minimizecurrent flow between the substrate (or body ) and the source/drain regions, the p type substrate is grounded. EE105 Fall 2007 Lecture 15, Slide 3 Prof. Liu, UC Berkeley
Review: Charge in a Semiconductor Negative charges: Conduction electrons (density n) Ionized acceptor atoms (density N A ) Positive charges: Holes (density p) Ionized donor atoms (density N D ) The net charge density [C/cm 3 ] in a semiconductor is ρ q ( p n + ) N D N A Note that p, n, N D, and N A each can vary with position. The mobile carrier concentrations (n and p) in the channel of a MOSFET can be modulated by an electric field via V G. EE105 Fall 2007 Lecture 15, Slide 4 Prof. Liu, UC Berkeley
Channel Formation (Qualitative) As the gate voltage (V G ) is increased, holes are repelled away from the substrate surface. The surface is depleted of mobile carriers. The charge density within the depletion region is determined by the dopant ion density. As V G increases above the threshold voltage V TH, a layer of conduction electrons forms at the substrate surface. For V G > V TH, n > N A at the surface. The surface region is inverted to be n type. V G < V TH V G V TH The electron inversion layer serves as a resistive path (channel) for current to flow between the heavily doped (i.e. e highly conductive) source and drain regions. EE105 Fall 2007 Lecture 15, Slide 5 Prof. Liu, UC Berkeley
Voltage Dependent Resistor In the ON state, the MOSFET channel can be viewed as a resistor. Since the mobile charge density within the channel depends on the gate voltage, the channel resistance is voltage dependent. EE105 Fall 2007 Lecture 15, Slide 6 Prof. Liu, UC Berkeley
Channel Length & Width Dependence Shorter channel length and wider channel width each yield lower channel resistance, hence larger drain current. Increasing W also increases the gate capacitance, however, which limits circuit operating speed (frequency). EE105 Fall 2007 Lecture 15, Slide 7 Prof. Liu, UC Berkeley
Comparison: BJT vs. MOSFET In a BJT, current (I C ) is limited by diffusion of carriers from the emitter to the collector. I C increases eponentially with input voltage (V BE ), because the carrier concentration gradient in the base is proportional to e V BE / V T In a MOSFET, current (I D ) is limited by drift of carriers from the source to the drain. I D increases ~linearly with input voltage (V G ), because the carrier concentration in the channel is proportional to (V G V TH ) In order to understand how MOSFET design parameters affect MOSFET performance, we first need to understand how a MOS capacitor works... EE105 Fall 2007 Lecture 15, Slide 8 Prof. Liu, UC Berkeley
MOS Capacitor A metal oide semiconductor structure can be considered as a parallel plate capacitor, with the top plate being the positive plate, the gate insulator being the dielectric, and the p type semiconductor substrate being the negative plate. The negative charges in the semiconductor (for V G > 0) are comprised of conduction electrons and/or acceptor ions. In order to understand dhow the potential ti and charge distributions ib ti within the Si depend on V G, we need to be familiar with electrostatics... EE105 Fall 2007 Lecture 15, Slide 9 Prof. Liu, UC Berkeley
Gauss Law E ρ ρ is the net charge density is the dielectric permittivity If the magnitude of electric field changes, there must be charge! In a charge free region, the electric field must be constant. Gauss Law equivalently says that if there is a net electric field leaving a region, there must be positive charge in that region: ρ E dv dv V V ρ Q dv E dv E ds V S V Q E ds charge within the enclosed volume The integral of the electric field over a closed surface is proportional to the EE105 Fall 2007 Lecture 15, Slide 10 Prof. Liu, UC Berkeley
Gauss Law in 1 D E( ) E de de d ρ d 0 ρ ρ( ') E( 0) + d' Consider a pulse charge distribution: ρ() 0 qn A X d E() 0 X d EE105 Fall 2007 Lecture 15, Slide 11 Prof. Liu, UC Berkeley
Electrostatic Potential The electric field (force) is related to the potential (energy): E dv d 2 d V ( ) 2 d ρ ( ) Note that an electron ( q q charge) drifts in the direction of increasing potential: F e qe q dv d ρ() 0 X d E() ( ) V () 0 X d 0 X qn A 0 X d EE105 Fall 2007 Lecture 15, Slide 12 Prof. Liu, UC Berkeley
Boundary Conditions Electrostatic potential must be a continuous function. Otherwise, the electric field (force) would be infinite. Electric field does not have to be continuous, however. Consider an interface between two materials: E 1 ( 1 ) E ds 1 E1S + 2E2S Qinside E 2 ( 2 ) S If Q inside 0, 0 E E 1 E1S + 2E2S 1 2 2 1 then Discontinuity in electric displacement E charge density at interface! EE105 Fall 2007 Lecture 15, Slide 13 Prof. Liu, UC Berkeley 0
MOS Capacitor Electrostatics Gate electrode: Since E() ( ) 0 in a metallic material, V() ( ) is constant. Gate electrode/gate insulator interface: The gate charge is located at this interface. E() ) changes to a non zero value inside id the gate insulator. Gate insulator: Ideally, there are no charges within the gate insulator. E() is constant, and V() is linear. Gate insulator/semiconductor interface: Since the dielectric i permittivity of SiO 2 is lower than that t of Si, E() is larger in the gate insulator than in the Si. Semiconductor: If ρ() is constant (non zero), then V() is quadratic. EE105 Fall 2007 Lecture 15, Slide 14 Prof. Liu, UC Berkeley
MOS Capacitor: V GB 0 If the gate and substrate materials are not the same (typically the case), there is a built in potential (~1V across the gate insulator). Positive charge is located at the gate interface, and negative charge in the Si. The substrate surface region is depleted of holes, down to a depth X do ρ(( ) 0 V () X do V So S,o Q dep -t o 0 X do EE105 Fall 2007 Lecture 15, Slide 15 Prof. Liu, UC Berkeley
Flatband Voltage, V FB The built in potential can be cancelled out by applying a gate voltage that is equal in magnitude (but of the opposite polarity) as the built in potential. This gate voltage is called the flatband voltage because the resulting potential profile is flat. ρ(ρ () -t t o0 V () -t o 0 There is no net charge (i.e. ρ()0) in the semiconductor under for V GB V FB. EE105 Fall 2007 Lecture 15, Slide 16 Prof. Liu, UC Berkeley
Voltage Drops across a MOS Capacitor V V V + V GB FB o S V () -t 0 X t o X d If we know the total charge within the semiconductor (Q S), we can find the electric field within the gate insulator (E o ) and hence the voltage drop across the gate insulator (V o ): Q Q Q S S Q S E ds E o A V o Eoto to A o o Co where Q S is the areal charge density in the semiconductor [C/cm 2 ] C and is the areal gate capacitance [F/cm 2 ] o o t o EE105 Fall 2007 Lecture 15, Slide 17 Prof. Liu, UC Berkeley
V GB < V FB (Accumulation) If a gate voltage more negative than V FB is applied, then holes willaccumulateat at the gate insulator/semiconductor interface. ρ() -t o 0 V () -t o Areal gate charge density [C/cm 2 ]: 0 Q G C o ( V V ) GB FB EE105 Fall 2007 Lecture 15, Slide 18 Prof. Liu, UC Berkeley
V FB < V GB < V TH (Depletion) If the applied gate voltage is greater than V FB, then the semiconductor surface will be depleted of holes. If the applied gate voltage is less than V TH, the concentration of conduction electrons at the surface is smaller than N A ρ() qn A () ρ(( ) -t o0 V () X d -t o 0 X d Areal depletion charge density [C/cm 2 ]: Q qn X dep AX d V GB V FB V + V 2 2 C ( V V ) o S qn AX C o d + qn AX 2 Si o GB FB X d 1+ 1 Co q SiN A EE105 Fall 2007 Lecture 15, Slide 19 Prof. Liu, UC Berkeley Si 2 d
V GB > V TH (Inversion) If the applied gate voltage is greater than V TH, then n > N A at the semiconductor surface. At V GB V TH, the total potential dropped in the Si is 2φ B where φ B N V T ln n i A ρ() X d,ma -t o V () -t o 0 X d,ma V V + 2 φ + TH FB B 2q N Si C o A (2φ ) B EE105 Fall 2007 Lecture 15, Slide 20 Prof. Liu, UC Berkeley
Maimum Depletion Depth, X d,ma As V GB is increased above V TH, V S and hence the depth of the depletion region (X d ) increases very slowly. This is because n increases eponentially with V S, whereas X d increases with the square root of V S. Thus, most of the incrementalnegative negative charge in the semiconductor comes from additional conduction electrons rather than additional ionized acceptor atoms, when n eceeds N A. X d can be reasonably approimated to reach a maimum value (X d,ma ) for V GB V TH. Q dep thus reaches a maimum of Q dep,ma at V GB V TH. If we assume that only the inversion layer charge increases with increasing V GB above V TH, then inv o ( VGB VTH ) and so QG ( VGB) Co( VGB VTH ) Qdep, ma Q C + EE105 Fall 2007 Lecture 15, Slide 21 Prof. Liu, UC Berkeley
Q V Curve for MOS Capacitor Q G X d,ma 2 Si (2φ B ) qn A Q inv C o ( V V ) GB TH Q dep,ma V GB [ V ] V FB V TH Q qn X 2qN (2φ ) dep, ma A d,ma A Si B EE105 Fall 2007 Lecture 15, Slide 22 Prof. Liu, UC Berkeley
Eample EE105 Fall 2007 Lecture 15, Slide 23 Prof. Liu, UC Berkeley