¡Descarga archivos ejecutados para programacion basica y más Ejercicios en PDF de Informática solo en Docsity! Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 1 SOLUTIONS: ECE 305 Homework: Week 3 Mark Lundstrom Purdue University (9/17/14) 1) Assume Silicon (bandgap 1.12 eV) at room temperature (300 K) with the Fermi level located exactly in the middle of the bandgap. Answer the following questions. a) What is the probability that a state located at the bottom of the conduction band is filled? b) What is the probability that a state located at the top of the valence band is empty? Solution: 1a) Begin with the Fermi function: f E( ) = 1 1+ e E−EF( ) kBT For our problem: E = EC f EC( ) = 1 1+ e EC−EF( ) kBT ≈ e− EC−EF( ) kBT The approximation is extremely good, because the bottom of the conduction band is very far above the Fermi level. The semiconductor is said to be nondegenerate. EF = EC + EV 2 (The Fermi level is exactly in the middle of the bandgap.) EC − EF = EC − EC + EV 2 = EC − EV 2 = EG 2 so the probability is f EC( ) ≈ e− EC−EF( ) kBT = e−EG 2kBT EG = 1.12 eV kBT = 0.026 eV EG 2kBT = 21.5 f EC( ) = e−EG 2kBT = e−21.5 = 4.43×10−10 What if we had not made the non-‐degenerate approximation? Then f EC( ) = 1 1+ e EC−EF( ) kBT = 1 1+ eEG 2kBT = 1 1+ 2.26×109 = 4.43×10−10 Same answer as with the non-‐degenerate assumption. Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 2 HW3 Solutions (continued): 1b) Begin again with the Fermi function: f E( ) = 1 1+ e E−EF( ) kBT For this part: E = EV f EV( ) = 1 1+ e EV −EF( ) kBT The Fermi function gives the probability that a state is full. We want the probability that the state is empty, which is 1− f EV( ) = 1− 1 1+ e EV −EF( ) kBT = e EV −EF( ) kBT 1+ e EV −EF( ) kBT = 1 e− EV −EF( ) kBT +1 1− f EV( ) = 1 1+ e EF −EV( ) kBT ≈ e− EF −EV( ) kBT The approximation is extremely good, because the Fermi level is very far above the top of the valence band. The nondegenerate approximation can be used for the valence band too. EF = EC + EV 2 (The Fermi level is exactly in the middle of the bandgap.) EF − EV = EC + EV 2 − EV = EC − EV 2 = EG 2 so the probability is 1− f EV( ) = e− EF −EV( ) kBT = e−EG kBT 1− f EV( ) = e−EG 2kBT = 4.43×10−10 2) For Si at room temperature, calculate the following quantities. (Numerical answers required, and don’t forget to include units with your answer). a) The density of states in the conduction band, gC E( ) , at an energy 26 meV above EC . b) The density of states in the valence band, gV E( ) , at an energy 26 meV below EV . c) The effective density of conduction band states. NC . d) The effective density of valence band states, NV . e) Compute the ratio of the effective density of conduction band states to the atomic density of Si. Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 5 HW3 Solutions (continued): gV E( ) = 1.05×1056 0.81( )3/2 EV − E = 7.65×1055 EV − E J-m3( )−1 Note that the DOS in the valence band is a little smaller than in the conduction band because the hole effective mass is smaller than the electron effective mass. We are asked for the density of states 26 meV (milli-‐electron volts) below the top of the valence band. In Joules EV − E = 0.026×1.60×10−19 = 4.16×10−21 J gC EV − 0.026 eV( ) = 7.65×1055 4.16×10−21 1 J-m3 = 4.94×1045 1 J-m3 The answer is correct, but the units are inconvenient. Let express the answer per eV per cubic cm, which is the more typical way of doing things for semiconductors. gV EV − 0.026 eV( ) = 4.94×1045 1 J × 1.6×10−19 J eV 1 m3 × 10−2 m cm ⎛ ⎝⎜ ⎞ ⎠⎟ 3 gV EV − 0.026 eV( ) = 7.90×1020 1 eV- cm3 Note: The total number of states in a range of energy of kBT = 0.026 eV is approximately gV EV − 0.026 eV( )× 0.026 = 2.06×1019 1 cm3 which is a reasonable number, as we shall see below. 2c) This is a problem about the “effective density of states” which is different from the density-‐of-‐states. The effective density of states, is roughly the total number of states in an energy range of kBT near the bottom of the conduction band or top of the valence band. It is derived by integration as discussed in the text. The result is given in SDF on p. 51 NC = 2.510×1019 mn * m0 ⎛ ⎝⎜ ⎞ ⎠⎟ 3/2 1 cm3 Using mn * = 1.18m0 , we find Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 6 HW3 Solutions (continued): NC = 2.510×1019 1.18( )3/2 cm−3 = 3.22×1019 cm−3 NC = 3.22×1019 cm−3 Note that this is close to the value we estimated in part 2a): gC 0.026 eV + EC( )× 0.026 = 3.62×1019 1 cm3 2d) Repeat 2c) but for the valence band. NV = 2.510×1019 mp * m0 ⎛ ⎝ ⎜ ⎞ ⎠ ⎟ 3/2 1 cm3 Using mn * = 0.81m0 , we find NV = 2.510×1019 0.81( )3/2 cm−3 = 1.83×1019 cm−3 NV = 1.83×1019 cm−3 Note that this is close to the value we estimated in part 2b): gV EV − 0.026 eV( )× 0.026 = 2.06×1019 1 cm3 2e) Recall the atomic density of Si is NSi = 5.00×1022 cm-3 so the ratio NC NSi = 3.22×1019 5.00×1022 = 6.44×10−4 NC NSi = 0.06 % The number of states near the bottom of the conduction band that can be occupied is a small fraction of the density of silicon atoms. Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 7 HW3 Solutions (continued): 3) Consider a region of Si at room temperature. For each of the following cases, calculate the equilibrium electron and hole concentrations (n and p). Assume that the dopants are fully ionized. a) Intrinsic material ( N D = N A = 0 ) b) N D = 1.00×1013 cm-3 N A = 0 c) N D = 1.00×1017 cm-3 N A = 0 d) N D = 0 N A = 1.00×1017 cm-3 e) N D = 1.00×1017 cm-3 N A = 3.00×1017 cm-3 Solution: 3a) For intrinsic material, we know that n = p = ni . According to SDF, p. 54, ni = 1.00×1010 cm-3 so n = p = ni = 1.00×1010 cm-3 3b) First, let’s ask what to expect. The donor density is 1000 times the intrinsic density. All of the donor electrons will go in the conduction band, overwhelming the number of intrinsic carriers that were there. So we expect n ≈ N D = 1.00×1013 cm-3 We would determine the hole density from np = ni 2 , so p = ni 2 n ≈ ni 2 N D = 1020 1013 = 1.00×107 cm-3 Note that doping the semiconductor n-‐type means that we now have more electrons in the conduction band, and we have fewer holes in the valence band than for the intrinsic semiconductor. How would we do this problem more accurately? Begin by assuming that the semiconductor is neutral: p − n+ N D + − N A − = 0 Assume that the dopants are fully ionized and use p = ni 2 n ni 2 n − n+ N D − N A = 0 Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 10 HW3 Solutions (continued): p = N A − N D 2 + N A − N D 2 ⎛ ⎝⎜ ⎞ ⎠⎟ 2 + ni 2 Plug in numbers: p = 1017 − 0 2 + 1017 2 ⎛ ⎝⎜ ⎞ ⎠⎟ 2 +1020 = 0.5×1017 + 0.25×1034 +1020 = 0.5×1017 + 0.25×1034 + 0 p = 0.50×1017 + 0.50×1017 = 1.00×1017 p = 1.00×1013 cm-3 Just as expected. We find the electron density from n = ni 2 p ≈ ni 2 N A = 1020 1017 = 1.00×103 cm-3 n = 1.00×103 cm-3 3e) Since the net p-‐type doping is N A − N D = 3×1017 −1×1017 = 2×1017 is much greater than the intrinsic density, we expect: p = 2×1017 and n = ni 2 p = 0.5×103 cm-3 . This is correct, as we can see by solving the problem properly. p = N A − N D 2 + N A − N D 2 ⎛ ⎝⎜ ⎞ ⎠⎟ 2 + ni 2 = 3×1017 −1×1017 2 + 2×1017 2 ⎛ ⎝⎜ ⎞ ⎠⎟ 2 +1020 p = 1×1017 + 2×1034 +1020 = 2×1017 p = 2×1017 cm-3 n = ni 2 p ≈ ni 2 2×1017 = 1020 2×1017 = 5.00×102 cm-3 n = 5.00×102 cm-3 Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 11 HW3 Solutions (continued): One might conclude from these examples, that there is no need to use space-‐charge neutrality and the quadratic equation. That would be wrong, as seen in the next problem. It only happened in this problem because in each case (except for case a)), the net doping was much larger than the intrinsic carrier density. 4) Assuming silicon with completely ionized dopants, compute n and p for the following case. N D = 5.00×1016 cm-3 N A = 0 T = 700 K Solution: We expect the intrinsic carrier density to be MUCH larger at higher temperatures. Using the information on p. 57 of SDF, we have ni 700 K( ) = 2.865×1016 cm-3 This is close to the doping density, so we need to use the quadratic equation. n = N D − N A 2 + N D − N A 2 ⎛ ⎝⎜ ⎞ ⎠⎟ 2 + ni 2 n = 2.5×1016 + 2.5×1016( )2 + 2.865×1016( )2 n = 2.5×1016 + 6.25×1032 +8.21×1032 = 6.3×1016 n = 6.3×1016 cm-3 p = ni 2 n = 2.865×1016( )2 6.3×1016 = 8.21×1032 6.3×1016 = 1.30×1016 cm-3 p = 1.30×1016 cm-3 Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 12 HW3 Solutions (continued): 5) For each of the cases (a-‐e) in Prob. 3, calculate the Fermi level position, with respect to the intrinsic level EF − Ei( ) . Note that you need to consider sign. Solution: In problem 3), we computed n and p. For this problem, we need an expression that relates n and p to the Fermi level and intrinsic level. The expressions are on p. 53 of SDF: n = nie EF −Ei( ) kBT p = nie Ei−EF( ) kBT Solving for EF − Ei( ) from the first equation, we have EF − Ei( ) = kBT ln n ni ⎛ ⎝⎜ ⎞ ⎠⎟ or in electron volts EF − Ei( ) q = kBT q ln n ni ⎛ ⎝⎜ ⎞ ⎠⎟ = 0.026ln n 1010 ⎛ ⎝⎜ ⎞ ⎠⎟ 3a) n = p = ni = 1.00×1010 cm-3 EF − Ei( ) q = 0.026ln 1( ) = 0 EF − Ei( ) q = 0 or we could say: EF − Ei( ) = 0 eV 3b) n = 1.00×1013 cm-3 EF − Ei( ) q = 0.026ln 1013 1010 ⎛ ⎝⎜ ⎞ ⎠⎟ = 0.180 EF − Ei( ) q = 0.180 Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 15 HW3 Solutions (continued): Case 3b) n = 1.00×1013 cm-3 EF − EC( ) q = 0.026ln 1013 3.22×1019 ⎛ ⎝⎜ ⎞ ⎠⎟ = −0.390 EF − EC( ) q = −0.390 Case 3c) n = 1.00×1017 cm-3 EF − EC( ) q = 0.026ln 1017 3.22×1019 ⎛ ⎝⎜ ⎞ ⎠⎟ = −0.150 EF − EC( ) q = −0.150 Note: The larger the n-‐type doping, the closer the Fermi level is to the conduction band. 3d) n = 1.00×103 cm-3 EF − EC( ) q = 0.026ln 103 3.22×1019 ⎛ ⎝⎜ ⎞ ⎠⎟ = −0.988 EF − EC( ) q = −0.988 Note: This value is getting close to the bandgap of Si. The Fermi level is getting close to the top of the valence band. 3e) n = 5.00×102 cm-3 EF − EC( ) q = 0.026ln 5×102 3.22×1019 ⎛ ⎝⎜ ⎞ ⎠⎟ = −1.01 EF − EC( ) q = −1.01 Also a p-‐type semiconductor, so the Fermi level is close to the top of the valence band. 7) For the case of problem 4), calculate the Fermi level position, with respect to the intrinsic level EF − Ei( ) . Note that you need to consider sign. Solution: The electron density was n = 6.3×1016 cm-3 and the intrinsic carrier concentration was ni 700 K( ) = 2.865×1016 cm-3 EF − Ei( ) q = kBT q ln n ni ⎛ ⎝⎜ ⎞ ⎠⎟ Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 16 HW3 Solutions (continued): at T = 700 K: kBT q = 1.38×10−23 × 700 1.6×10−19 = 0.0604 eV EF − Ei( ) q = 0.060ln 6.3×1016 2.865×1016 ⎛ ⎝⎜ ⎞ ⎠⎟ = 0.048 EF − Ei( ) q = 0.048 The Fermi level is almost in the middle of the gap – very close to the intrinsic level. This happens because n = 6.3×1016 cm-3 is only a little bigger than ni 700 K( ) . 8) Your textbook defines a non-‐degenerate semiconductor as one for which the Fermi level is at least 3kBT below the conduction band edge and at least 3kBT above the valence band edge. What is the largest electron density for non-‐degenerate Si? Solution: Require: EC − EF( ) ≥ 3kBT n = NCe EF −EC( ) kBT n ≤ NCe −3kBT( ) kBT = NCe−3 n ≤ 3.22×1019( )× e−3 = 1.60×1018 n = 1.60×1018 cm-3 is the largest electron density in Si at room temperature for which we can use non-‐generate expressions. Note: Your book is fairly conservative on this. Many people would use non-‐ degenerate equations when n ≤ NC = 3.22×1019 cm-3 Some might even use the assumption for higher doping densities, but they understand what they are doing and what the errors are. Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 17 HW3 Solutions (continued): 9) For each case below, explain physically (i.e. in words in no equations) whether the intrinsic level is above or below the middle of the gap, then compute the location of the intrinsic level relative to the middle of the bandgap, Emid = EC + EV( ) 2 . a) Silicon at T = 300K with mn * = 1.18m0 and mp * = 0.81m0 . b) GaAs at T = 300K with mn * = 0.067m0 and mp * = 0.524m0 . Solution: a) Silicon: The carrier density is basically the number of states times the probability that the states are occupied. The density of states varies as effective mass to the 3/2 power, so the conduction band density of states is larger than the valance band density of states for Si. To get equal numbers of electrons and holes, the Fermi level (intrinsic level) will need to be a little below the middle of the bandgap. This will make the probability that conduction band statesare occupied a little smaller than the probability that valence band states are empty, and compensate for the larger DOS in the conduction band. To compute Ei − Emid we use eqn. (2.36) in p. 62 in SDF and find Ei − Emid q = 3 4 kBT q ln mp * mn * ⎛ ⎝ ⎜ ⎞ ⎠ ⎟ = 0.75× 0.026× ln 0.81 1.18 ⎛ ⎝⎜ ⎞ ⎠⎟ = −0.007 eV Ei − Emid q = −0.007 eV The intrinsic level is 7 meV below the middle of the gap in Si. b) GaAs: For GaAs, the conduction band density of states is smaller than the valance band density of states (just the opposite of Si). Using the same reasoning as in 9a), to get equal numbers of electrons and holes, the Fermi level (intrinsic level) will need to be a little above the middle of the bandgap. This will make the probability that conduction band states are occupied a little largerthan the probability that valence band states are empty, and compensate for the smaller DOS in the conduction band, Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 20 HW3 Solutions (continued): ρ = 1 nqµn + pqµ p = 1 niq µn + µ p( ) Ω-cm ρ = 1 1010 ×1.6×10−19 × 1400+ 460( ) = 3.4×105 Ω-cm ρ = 3.4×105 Ω-cm (silicon) This is a very large resistivity – not as large as an insulator, but very large for a semiconductor. For Ge: ρ = 1 niq µn + µ p( ) Ω-cm From Fig. 3.5, p. 80 of SDF for Ge: µn ≈ 4000 cm2 V-s µ p ≈ 2000 cm2 V-s From Fig. 2.20, p. 54 of SDF for Ge: ni 300 K( ) = 2×1013 cm-3 ρ = 1 2×1013 ×1.6×10−19 × 4000+ 2000( ) = 5.2×101 Ω-cm ρ = 5.2×101 Ω-cm (Ge) Intrinsic Ge is fairly conductive! This happens because the mobilities are higher, but mostly because the bandgap is much lower, so the intrinsic carrier concentration is much larger. For GaAs: ρ = 1 niq µn + µ p( ) Ω-cm From Fig. 3.5, p. 80 of SDF for GaAs: Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 21 µn ≈ 8500 cm2 V-s µ p ≈ 430 cm2 V-s HW3 Solutions (continued): From Fig. 2.20, p. 54 of SDF for Ge: ni 300 K( ) = 2.25×106 cm-3 ρ = 1 2.25×106 ×1.6×10−19 × 8500+ 430( ) = 3.1×108 Ω-cm ρ = 3.1×108 Ω-cm (GaAs) Intrinsic GaAs has a resistivity that is orders of magnitude larger than Si! This happens even though the electron mobility is much larger than Si because the bandgap is much larger, so the intrinsic carrier concentration is orders of magnitude smaller. 11b) Since we are heavily doped n-‐type, we can ignore the contribution from holes. ρ = 1 nqµn Ω-cm For Si: From Fig. 3.5, p. 80 of SDF for Si: µn ≈110 cm2 V-s ρ = 1 1×1019 ×1.6×10−19 ×110 = 5.7 ×10−3 Ω-cm ρ = 5.7 ×10−3 Ω-cm (Si) This is about eight orders of magnitude lower than the intrinsic resistivity of Si. The ability to control the resistivity over orders of magnitude is what makes semiconductors so useful. For Ge: From Fig. 3.5, p. 80 of SDF for Si: Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 22 µn ≈ 900 cm2 V-s HW3 Solutions (continued): ρ = 1 1×1019 ×1.6×10−19 × 900 = 6.9×10−4 Ω-cm ρ = 6.9×10−4 Ω-cm (Ge) Even lower than Si because the mobility is higher. For GaAs: From Fig. 3.5, p. 80 of SDF for GaAs: µn ≈ 3200 cm2 V-s ρ = 1 1×1019 ×1.6×10−19 × 3200 = 2.0×10−4 Ω-cm ρ = 2.0×10−4 Ω-cm (GaAs) Even lower than Ge because the mobility is higher. Comments: For Si, we found that the range of possible resistivities is 3.0×105 Ω-cm < ρ < 6×10−3 Ω-cm about eight orders of magnitude. In practice, we could never have pure, intrinsic Si, there will always be some unintentional dopants no matter how pure we try to make Si, so the resistivities would not be as high as indicted here, but it’s common to find resistivities in the 1000’s. Also, one can dope Si to 1020 cm-3 , so resistivities 10 times lower than indicated above can be obtained. How do these limits compare to insulators and metals? A Google search shows Resistivity of copper: ρCu = 1.68×10−8 Ω-cm Resistivity of diamond: ρdiamond = 1013 −1020 Ω-cm Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 25 N D − N A = 4.66×1015 cm-3 HW3 Solutions (continued): 13e) Where is the intrinsic level located with respect to the conduction band? (Note: You can simply write down the answer without any work, if you explain your reasoning. Since NC = NV putting the Fermi level exactly at the middle of the bandgap will produce equal numbers of electrons and holes. Exactly at mid-‐gap. 14) For the energy band diagram sketched below, answer the following questions. 14a) Sketch the electrostatic potential, V x( ) , vs. position, x. 14b) Sketch the electric field, E x( ) , vs. position, x. 14c) Sketch the electron density, n x( ) vs. position, x. 14d) Sketch the hole density, p x( ) , vs. position, x. Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 26 HW3 Solutions (continued): Solution: 14a) Sketch the electrostatic potential, V x( ) , vs. position, x. 14b) Sketch the electric field, E x( ) , vs. position, x. Mark Lundstrom 09/11/2014 ECE-‐305 Fall 2014 27 HW3 Solutions (continued): 14c) Sketch the electron density, n x( ) vs. position, x. 14d) Sketch the hole density, p x( ) , vs. position, x.