Physics · Book 4 · Bachelor Year 2

University Physics — Year 2

University Physics — Year 2 · Bachelor Year 2

31Potential Wells, Barriers and Tunnelling

A ball in a bowl rolls back and forth with whatever energy it was given; an electron in an atom, a nucleon in a nucleus, an electron in a nanometre-sized crystal can only have certain energies, and the lowest of them is not zero. A ball cannot cross a hill higher than its energy allows; an α\alpha particle leaves a nucleus through a Coulomb wall it could never climb, the needle of a scanning tunnelling microscope draws a current across a gap of vacuum, and the nitrogen atom of ammonia swings through the plane of its three hydrogens twenty-four billion times a second. This chapter solves the Schrödinger equation in the simplest potentials — wells with infinite and finite walls, a step, a barrier, a double well — and finds in them the two great quantum facts that classical mechanics cannot produce: the quantisation of bound energies, and tunnelling through classically forbidden regions, with its exponential sensitivity to the width and height of the barrier, which is why it can serve as a ruler of picometres and why radioactive lifetimes span thirty orders of magnitude.

A scanning tunnelling microscope: a metal tip held a few tenths of a nanometre above a surface; the tunnelling current, which changes tenfold per tenth of a nanometre, maps the surface atom by atom.
A scanning tunnelling microscope: a metal tip held a few tenths of a nanometre above a surface; the tunnelling current, which changes tenfold per tenth of a nanometre, maps the surface atom by atom.

31.1 The infinite well

Proposition 31.1 (Infinite square well)

A particle confined to 0<x<L0 < x < L by impenetrable walls (V=0V = 0 inside, V=V = \infty outside, so φ=0\varphi = 0 at the walls) has the stationary states and energies

φn(x)=2LsinnπxL,En=n2h28mL2=n2E1,n=1,2,3,\varphi_n(x) = \sqrt{\frac{2}{L}}\sin\frac{n\pi x}{L}, \qquad E_n = \frac{n^2h^2}{8mL^2} = n^2E_1, \qquad n = 1, 2, 3, \dots

The energies are quantised, grow as n2n^2, and the lowest is not zero: the confinement energy E1=h2/8mL2E_1 = h^2/8mL^2 is the price of localisation (ΔxL\Delta x \sim L forces Δph/2L\Delta p \sim h/2L). The φn\varphi_n are orthogonal, φnφm ⁣dx=δnm\int\varphi_n\varphi_m\dd x = \delta_{nm}, and any state of the well is a superposition of them.

Proof. Inside, φ=k2φ\varphi'' = -k^2\varphi with k2=2mE/2k^2 = 2mE/\hbar^2: φ=Asinkx+Bcoskx\varphi = A\sin kx + B\cos kx; φ(0)=0\varphi(0) = 0 kills BB, φ(L)=0\varphi(L) = 0 requires kL=nπkL = n\pi — the standing waves of a string, with E=2k2/2mE = \hbar^2k^2/2m. Normalisation gives A=2/LA = \sqrt{2/L}; orthogonality is that of the sines.

The infinite well: the first three levels, E_n n2, with their wave functions (left) and probability densities (right) — n - 1 nodes, and a ground state that is spread over the well and has a non-zero energy.
The infinite well: the first three levels, Enn2E_n \propto n^2, with their wave functions (left) and probability densities (right) — n1n - 1 nodes, and a ground state that is spread over the well and has a non-zero energy.

Example 31.2 (Orders of magnitude, and colours)

E1=h2/8mL2E_1 = h^2/8mL^2: an electron in 1nm1\,\mathrm{nm}, 0.38eV0.38\,\mathrm{eV}; in 0.1nm0.1\,\mathrm{nm} (an atom), 38eV38\,\mathrm{eV} — the scale of atomic energies; a nucleon in 5fm5\,\mathrm{fm}, 8MeV8\,\mathrm{MeV} — the scale of nuclear energies; a marble in a box, 1063J10^{-63}\,\mathrm{J} — no one will notice. A semiconductor nanocrystal a few nanometres across (a quantum dot) confines its electrons: the confinement energy adds to the crystal’s gap, so the smaller the dot, the bluer the light it emits — the same material glows red at 6nm6\,\mathrm{nm} and green at 3nm3\,\mathrm{nm}, tuned by size alone.

31.2 The finite well

Proposition 31.3 (Finite square well)

For V=0V = 0 in x<a|x| < a and V=V0V = V_0 outside, a state with 0<E<V00 < E < V_0 oscillates inside (k=2mE/k = \sqrt{2mE}/\hbar) and decays outside (κ=2m(V0E)/\kappa = \sqrt{2m(V_0 - E)}/\hbar): φeκx\varphi \propto \eu^{-\kappa|x|} beyond the walls. Matching φ\varphi and φ\varphi' at x=±ax = \pm a gives, for the even and odd states,

ktanka=κ(even),kcotka=κ(odd),with(ka)2+(κa)2=2mV0a22R2.k\tan ka = \kappa \quad\text{(even)}, \qquad -k\cot ka = \kappa \quad\text{(odd)}, \qquad\text{with}\quad (ka)^2 + (\kappa a)^2 = \frac{2mV_0a^2}{\hbar^2} \equiv R^2 .

Graphically, the solutions are the intersections of the curves η=ξtanξ\eta = \xi\tan\xi, η=ξcotξ\eta = -\xi\cot\xi with the circle ξ2+η2=R2\xi^2 + \eta^2 = R^2 (ξ=ka\xi = ka, η=κa\eta = \kappa a): there is always at least one bound state (the even ground state), and 1+2R/π1 + \lfloor 2R/\pi\rfloor in all. The particle penetrates the forbidden region over the depth 1/κ1/\kappa, and the levels lie below those of the infinite well of the same width.

Proof. The even solution is AcoskxA\cos kx inside and BeκxB\eu^{-\kappa|x|} outside; the continuity of φ/φ\varphi'/\varphi at x=ax = a gives ktanka=κ-k\tan ka = -\kappa. Same for the odd one with sin\sin. The circle is the definition of kk and κ\kappa. Each branch of ξtanξ\xi\tan\xi starting at ξ=nπ\xi = n\pi and of ξcotξ-\xi\cot\xi starting at (n+12)π(n + \tfrac12)\pi meets the circle once if it starts inside it: hence the count.

Graphical solution of the finite well: the bound states are the intersections of the circle of radius R = a√2mV_0/ with the branches  (even states) and - (odd). Here R = 4: three bound states; the dashed small circle (R = 1.2) still cuts the first branch — a well always binds at least one state.
Graphical solution of the finite well: the bound states are the intersections of the circle of radius R=a2mV0/R = a\sqrt{2mV_0}/\hbar with the branches ξtanξ\xi\tan\xi (even states) and ξcotξ-\xi\cot\xi (odd). Here R=4R = 4: three bound states; the dashed small circle (R=1.2R = 1.2) still cuts the first branch — a well always binds at least one state.

Example 31.4 (An electron in a nanometre well)

V0=1eVV_0 = 1\,\mathrm{eV}, width 2a=1nm2a = 1\,\mathrm{nm}: R=a2mV0/=0.5×109×5.1×109=2.6R = a\sqrt{2mV_0}/\hbar = 0.5 \times 10^{-9} \times 5.1 \times 10^9 = 2.6, so 1+2R/π=21 + \lfloor 2R/\pi\rfloor = 2 bound states; the ground state lies near 0.26eV0.26\,\mathrm{eV} instead of the infinite well’s 0.38eV0.38\,\mathrm{eV}, and leaks 0.2nm0.2\,\mathrm{nm} into the walls (1/κ=/2m(V0E)1/\kappa = \hbar/\sqrt{2m(V_0 - E)}). Such wells, grown as layers of semiconductors a few nanometres thick, are the heart of the diode lasers of Chapter 23.

31.3 Step and barrier: tunnelling

Proposition 31.5 (Potential step)

A particle of energy EE coming from x<0x < 0 onto the step V=V0V = V_0 for x>0x > 0:

  • E>V0E > V_0: φ=eik1x+reik1x\varphi = \eu^{\iu k_1x} + r\eu^{-\iu k_1x} on the left, teik2xt\eu^{\iu k_2x} on the right, with k1,2=2m(EV0,left/right)/k_{1,2} = \sqrt{2m(E - V_{0,\text{left/right}})}/ \hbar; continuity of φ\varphi and φ\varphi' gives r=(k1k2)/(k1+k2)r = (k_1 - k_2)/(k_1 + k_2) and the reflection probability (ratio of currents) R=r2R = r^2, T=1R=4k1k2/(k1+k2)2T = 1 - R = 4k_1k_2/(k_1 + k_2)^2 — a particle can be reflected by a step it has the energy to climb, exactly as a wave on a string by a change of impedance (Chapter 15);
  • E<V0E < V_0: on the right φ=teκx\varphi = t\eu^{-\kappa x}, κ=2m(V0E)/\kappa = \sqrt{2m(V_0 - E)}/\hbar — an evanescent wave; r=1|r| = 1 (total reflection, with a phase shift), no current flows into the step, but the particle is found there with a probability decaying over 1/κ1/\kappa.

Proof. Write the continuity equations at x=0x = 0 and solve; for the currents use j=(A2B2)k/mj = (|A|^2 - |B|^2)\hbar k/m (Chapter 30). For E<V0E < V_0, r=(kiκ)/(k+iκ)r = (k - \iu\kappa)/(k + \iu\kappa), of modulus one.

Theorem 31.6 (Tunnelling through a barrier)

A particle of energy E<V0E < V_0 meeting a rectangular barrier of height V0V_0 and width aa is transmitted with the probability

T=[1+V02sinh2(κa)4E(V0E)]1  16EV0(1EV0)e2κa(κa1),κ=2m(V0E).T = \Big[1 + \frac{V_0^2\sinh^2(\kappa a)}{4E(V_0 - E)}\Big]^{-1} \ \approx\ 16\,\frac{E}{V_0}\Big(1 - \frac{E}{V_0}\Big)\eu^{-2\kappa a} \quad (\kappa a \gg 1), \qquad \kappa = \frac{\sqrt{2m(V_0 - E)}}{\hbar} .

This is the tunnel effect: classically impossible, quantum-mechanically exponentially small — and exponentially sensitive to the width aa, the height V0EV_0 - E and the mass mm. For an electron 1eV1\,\mathrm{eV} below the top, κ=5.1nm1\kappa = 5.1\,\mathrm{nm}^{-1}: T102T \sim 10^{-2} for a=0.5nma = 0.5\,\mathrm{nm}, 10410^{-4} for 1nm1\,\mathrm{nm}, 10910^{-9} for 2nm2\,\mathrm{nm}; for a proton, κ\kappa is 4343 times larger and nothing passes at these widths.

Proof. φ=eikx+reikx\varphi = \eu^{\iu kx} + r\eu^{-\iu kx} for x<0x < 0, Aeκx+BeκxA\eu^{\kappa x} + B\eu^{-\kappa x} inside, teikxt\eu^{\iu kx} for x>ax > a; four continuity conditions (of φ\varphi and φ\varphi' at 00 and aa) for four unknowns; eliminating AA, BB, rr gives 1/t2=1+(k2+κ2)2sinh2(κa)/4k2κ21/|t|^2 = 1 + (k^2 + \kappa^2)^2\sinh^2(\kappa a)/4k^2\kappa^2, which is the formula with k2=2mE/2k^2 = 2mE/\hbar^2, κ2=2m(V0E)/2\kappa^2 = 2m(V_0 - E)/\hbar^2. For κa1\kappa a \gg 1, sinhκaeκa/2\sinh\kappa a \approx \eu^{\kappa a}/2. For a barrier of arbitrary shape the exponent becomes 2κ(x) ⁣dx2\int\kappa(x)\dd x across the forbidden region (the Gamow factor, admitted).

Tunnelling through a barrier: the wave oscillates on the left, decays exponentially inside the classically forbidden region, and emerges on the right with a reduced amplitude — the same wavelength, a small probability.
Tunnelling through a barrier: the wave oscillates on the left, decays exponentially inside the classically forbidden region, and emerges on the right with a reduced amplitude — the same wavelength, a small probability.

Example 31.7 (The scanning tunnelling microscope)

A sharp metal tip is brought within d0.5nmd \approx 0.5\,\mathrm{nm} of a conducting surface and a small voltage applied: electrons tunnel across the vacuum gap, whose barrier height is the work function Φ4.5eV\Phi \approx 4.5\,\mathrm{eV}, so κ=2mΦ/=1.1×1010m1\kappa = \sqrt{2m\Phi}/\hbar = 1.1 \times 10^{10}\,\mathrm{m}^{-1} and the current Ie2κdI \propto \eu^{-2\kappa d} changes by a factor e2.210\eu^{2.2} \approx 10 for every 0.1nm0.1\,\mathrm{nm}. A feedback loop moves the tip to keep II constant, and the tip’s height, recorded as it scans, draws the surface with a vertical resolution of a picometre — and laterally atom by atom, because the last atom of the tip, being nearest, carries most of the current (Binnig and Rohrer, 1981).

Example 31.8 (Alpha decay)

An α\alpha particle of 5MeV5\,\mathrm{MeV} inside a heavy nucleus faces the Coulomb barrier of the remaining charge ZeZe, tens of MeV high at the nuclear surface and extending to the radius b=2Ze2/4πε0E50fmb = 2Ze^2/4\pi\varepsilon_0E \approx 50\,\mathrm{fm} where V=EV = E. The Gamow factor 2κ ⁣dr2\int\kappa\,\dd r is of order 8080: Te801035T \sim \eu^{-80} \sim 10^{-35}. The particle hits the wall some 102110^{21} times per second, so it escapes at a rate 1014s1\sim 10^{-14} \,\mathrm{s}^{-1}: a lifetime of a million years. Because the exponent varies as Z/EZ/\sqrt E, a 2MeV2\,\mathrm{MeV} change of EE shifts the lifetime by twenty orders of magnitude — the Geiger–Nuttall law, from microseconds to the age of the universe (Gamow, 1928: the first application of tunnelling).

Atomic resolution on a gold surface, imaged by a scanning tunnelling microscope: the rows of atoms of the reconstructed (100) face, a few tenths of a nanometre apart.
Atomic resolution on a gold surface, imaged by a scanning tunnelling microscope: the rows of atoms of the reconstructed (100) face, a few tenths of a nanometre apart.

31.4 The double well

Proposition 31.9 (Tunnel splitting)

Two identical wells separated by a barrier: if the barrier were impenetrable each well would have the same ground level E0E_0, twice. Tunnelling couples them and the two lowest states of the double well are the symmetric and antisymmetric combinations, φ±(φL±φR)/2\varphi_\pm \approx (\varphi_{\text{L}} \pm \varphi_{\text{R}})/\sqrt2, with energies E0Δ/2E_0 \mp \Delta/2: the level is split by an amount Δeκa\Delta \propto \eu^{-\kappa a} (the tunnelling amplitude, not its square). A particle placed in the left well, ψ=(φ++φ)/2\psi = (\varphi_+ + \varphi_-)/\sqrt2, is not stationary: it oscillates between the wells at the frequency ν=Δ/h\nu = \Delta/htunnelling oscillations, the two-level beats of Chapter 30.

Proof. The symmetric combination has no node under the barrier and a lower curvature, hence a lower energy; the antisymmetric one has a node and lies higher; the splitting is proportional to the overlap of the localised functions under the barrier, eκa\eu^{-\kappa a}. The beating is the two-level superposition already computed.

The double well: the degenerate level of the two isolated wells splits into a symmetric state (lower) and an antisymmetric one (higher) by - a; a particle started on one side tunnels back and forth at the frequency /h.
The double well: the degenerate level of the two isolated wells splits into a symmetric state (lower) and an antisymmetric one (higher) by Δeκa\Delta \propto \eu^{-\kappa a}; a particle started on one side tunnels back and forth at the frequency Δ/h\Delta/h.

Example 31.10 (Ammonia)

In NH3_3 the nitrogen atom sits on one side of the plane of the three hydrogens or on the other — a double well with a barrier of about 0.25eV0.25\,\mathrm{eV}. The tunnel splitting of the ground level is Δ=1×104eV\Delta = 1 \times 10^{-4}\,\mathrm{eV}: ν=Δ/h=24GHz\nu = \Delta/h = 24\,\mathrm{GHz}, the inversion frequency, the transition of the first maser and of the first atomic clock (1949). Replace hydrogen by deuterium and the heavier molecule tunnels less: 1.6GHz1.6\,\mathrm{GHz}. In heavier pyramids (PH3_3, AsH3_3) the splitting collapses to kilohertz and below: the molecule stays on its side for years — which is why left- and right-handed molecules exist at all, and why sugar does not racemise on the shelf.

Charles Townes with the first maser (1954): a beam of ammonia molecules, sorted by an electric field into the upper inversion state, amplified the 24\, GHz transition in a cavity — stimulated emission’s first device, six years before the laser.
Charles Townes with the first maser (1954): a beam of ammonia molecules, sorted by an electric field into the upper inversion state, amplified the 24GHz24\,\mathrm{GHz} transition in a cavity — stimulated emission’s first device, six years before the laser.

Method 31.11 (Wells and barriers)

(1) Write φ\varphi region by region: e±ikx\eu^{\pm\iu kx} where E>VE > V, e±κx\eu^{\pm\kappa x} where E<VE < V, with k,κ=2mEV/k, \kappa = \sqrt{2m|E - V|}/\hbar. (2) Match φ\varphi and φ\varphi' at each boundary (at an infinite wall, φ=0\varphi = 0). (3) Bound states: the matching has solutions only for discrete EE — count them graphically. (4) Scattering: currents (A2B2)k/m(|A|^2 - |B|^2)\hbar k/m give RR and TT. (5) Tunnelling: Te2κaT \sim \eu^{-2\kappa a}; order of magnitude first, prefactor after. (6) Double well: splitting eκa\propto\eu^{-\kappa a}, beats at Δ/h\Delta/h.

31.5 Exercises

Exercise 31.1

Infinite well: electron in 1nm1\,\mathrm{nm} (E1E_1, E2E_2, E3E_3, wavelength of the 212 \to 1 photon); electron in 0.1nm0.1\,\mathrm{nm}; nucleon in 5fm5\,\mathrm{fm}; a 1g1\,\mathrm{g} marble in 10cm10\,\mathrm{cm} (E1E_1, and the quantum number for a speed of 1cm/s1\,\mathrm{cm}/\mathrm{s}).

Solution

Solution of Exercise 31.1.

1nm1\,\mathrm{nm}: 0.380.38\,, 1.501.50\,, 3.38eV3.38\,\mathrm{eV}; 212 \to 1: 1.13eV1.13\,\mathrm{eV}, 1.1µm1.1\,\text{µ}\mathrm{m}. 0.1nm0.1\,\mathrm{nm}: 38eV38\,\mathrm{eV}. Nucleon: 8MeV8\,\mathrm{MeV}. Marble: E1=5.5×1063JE_1 = 5.5 \times 10^{-63}\,\mathrm{J}; n=2mvL/h=3×1027n = 2mvL/h = 3 \times 10^{27}.

Exercise 31.2

Quantum dots: a nanocrystal of gap 1.74eV1.74\,\mathrm{eV} confines an electron–hole pair whose confinement energy is h2/8mL2h^2/8m^*L^2 with m=0.1mem^* = 0.1\,m_{\text{e}}. Emission wavelength for L=6L = 6, 44, 33, 2.5nm2.5\,\mathrm{nm}; which colours?

Solution

Solution of Exercise 31.2.

Confinement 3.76eVnm2/L23.76\,\mathrm{eV}\,\mathrm{nm}^{2}/L^2: 0.100.10, 0.240.24, 0.420.42, 0.60eV0.60\,\mathrm{eV}; total 1.841.84, 1.981.98, 2.162.16, 2.34eV2.34\,\mathrm{eV}: 673673\,, 628628\,, 574574\,, 530nm530\,\mathrm{nm} — red, orange, yellow-green, green.

Exercise 31.3

Tunnelling: electron with V0E=4.5eVV_0 - E = 4.5\,\mathrm{eV}: κ\kappa; e2κa\eu^{-2\kappa a} for a=0.3a = 0.3, 0.50.5, 11, 2nm2\,\mathrm{nm}; factor per 0.1nm0.1\,\mathrm{nm}; same for a proton at a=0.3nma = 0.3\,\mathrm{nm}.

Solution

Solution of Exercise 31.3.

κ=1.09×1010m1\kappa = 1.09 \times 10^{10}\,\mathrm{m}^{-1}; e2κa\eu^{-2\kappa a}: 1.5×1031.5 \times 10^{-3}, 1.8×1051.8 \times 10^{-5}, 3×10103 \times 10^{-10}, 101910^{-19}; ×8.8\times 8.8 per 0.1nm0.1\,\mathrm{nm}; proton: κ\kappa is 4343 times larger, e280\eu^{-280} at 0.3nm0.3\,\mathrm{nm}.

Exercise 31.4

Step: electron of 2eV2\,\mathrm{eV} onto a step of 1eV1\,\mathrm{eV}: k1k_1, k2k_2, RR, TT. Same for a step down of 1eV1\,\mathrm{eV}. Why is this impossible classically, and what is the analogue for light?

Solution

Solution of Exercise 31.4.

k1=7.3×109m1k_1 = 7.3 \times 10^{9}\,\mathrm{m}^{-1}, k2=5.1×109m1k_2 = 5.1 \times 10^{9}\,\mathrm{m}^{-1}; R=0.03R = 0.03, T=0.97T = 0.97. Step down: k2=8.9×109m1k_2 = 8.9 \times 10^{9}\,\mathrm{m}^{-1}, R=0.01R = 0.01. A classical particle never turns back when it has the energy; a light wave is partly reflected at any change of index.

Exercise 31.5 ★★

Infinite well in detail. (a) Derive the levels and normalise. (b) Show the orthogonality. (c) For n=1n = 1: x\langle x\rangle, Δx=L1/121/2π2\Delta x = L\sqrt{1/12 - 1/2\pi^2}, p=0\langle p\rangle = 0, Δp=π/L\Delta p = \pi\hbar/L; check Heisenberg. (d) Sketch φn2|\varphi_n|^2 for large nn and compare with the classical probability of finding a bouncing particle.

Solution

Solution of Exercise 31.5.

(a), (b) Proposition 31.1. (c) L/2L/2; 0.18L0.18L; 00; π/L\pi\hbar/L; product 0.57>/20.57\hbar > \hbar/2. (d) Rapid oscillations about the mean 1/L1/L — the classical uniform density.

Exercise 31.6 ★★

Finite well. (a) Derive ktanka=κk\tan ka = \kappa for the even states. (b) Show graphically there is always a bound state and count them for R=1R = 1, 44, 1010. (c) Electron, V0=1eVV_0 = 1\,\mathrm{eV}, 2a=1nm2a = 1\,\mathrm{nm}: RR, number of states, penetration depth of the ground state (take E0.26eVE \approx 0.26\,\mathrm{eV}). (d) What happens to the number of bound states as V0V_0 \to \infty, and to their energies?

Solution

Solution of Exercise 31.6.

(a) Proposition 31.3. (b) 11, 33, 77. (c) R=2.6R = 2.6, two states; κ=4.4×109m1\kappa = 4.4 \times 10^{9}\,\mathrm{m}^{-1}, depth 0.23nm0.23\,\mathrm{nm}. (d) Infinitely many, tending to the infinite well’s from below.

Exercise 31.7 ★★

Step, E<V0E < V_0. (a) Write φ\varphi on both sides and find r=(kiκ)/(k+iκ)r = (k - \iu\kappa)/(k + \iu\kappa). (b) Show r=1|r| = 1 and compute the phase of rr. (c) Show the current vanishes for x>0x > 0 although φ20|\varphi|^2 \ne 0 there. (d) Electron 1eV1\,\mathrm{eV} below the top: penetration depth; compare with the evanescent wave of total internal reflection.

Solution

Solution of Exercise 31.7.

(a) 1+r=t1 + r = t, ik(1r)=κt\iu k(1 - r) = -\kappa t. (b) Numerator and denominator are conjugates; phase 2arctan(κ/k)-2\arctan(\kappa/k). (c) ψψ\psi^*\psi' is real for a real exponential. (d) 0.2nm0.2\,\mathrm{nm}; like the evanescent wave of total reflection: present, carrying no flux, and able to feed a second medium — that is tunnelling.

Exercise 31.8 ★★

The barrier. (a) Write the four continuity conditions and derive TT. (b) Evaluate exactly and with the approximation for E=V0/2E = V_0/2, κa=3\kappa a = 3; for κa=1\kappa a = 1. (c) For E>V0E > V_0 show T=[1+V02sin2(k2a)/4E(EV0)]1T = [1 + V_0^2\sin^2(k_2a)/4E(E - V_0)]^{-1} and find the energies of perfect transmission. (d) Interpret them (think of the anti-reflection layer of Chapter 15).

Solution

Solution of Exercise 31.8.

(a) Theorem 31.6. (b) κa=3\kappa a = 3: exact 1/(1+sinh23)=9.9×1031/(1 + \sinh^23) = 9.9 \times 10^{-3}, approximation 4e6=9.9×1034\eu^{-6} = 9.9 \times 10^{-3}; κa=1\kappa a = 1: 0.420.42 against 0.540.54. (c) κik2\kappa \to \iu k_2; T=1T = 1 for k2a=nπk_2a = n\pi. (d) The reflections from the two faces cancel when the barrier is a whole number of half-wavelengths — resonant transmission.

Exercise 31.9 ★★

Gamow. For the Coulomb barrier from the nuclear radius RnR_{\text{n}} to b=2Ze2/4πε0Eb = 2Ze^2/4\pi\varepsilon_0E, the exponent is G=2κ ⁣dr=(22mE/)b[arccosRn/b(Rn/b)(1Rn/b)]G = 2\int\kappa\,\dd r = (2\sqrt{2mE}/\hbar)\,b\,[\arccos\sqrt{R_{\text{n}}/b} - \sqrt{(R_{\text{n}}/b)(1 - R_{\text{n}}/b)}]. (a) Z=90Z = 90, E=5MeVE = 5\,\mathrm{MeV}, Rn=8fmR_{\text{n}} = 8\,\mathrm{fm}: bb, GG, TT. (b) Frequency of hits v/2Rnv/2R_{\text{n}} and the lifetime. (c) Repeat for E=4E = 4 and 6MeV6\,\mathrm{MeV}: ratio of lifetimes. (d) Why do nuclei emit α\alpha particles rather than protons or 12^{12}C?

Solution

Solution of Exercise 31.9.

(a) b=52fmb = 52\,\mathrm{fm}; G=82G = 82; T=e822×1036T = \eu^{-82} \approx 2 \times 10^{-36}. (b) v=1.55×107m/sv = 1.55 \times 10^{7}\,\mathrm{m}/\mathrm{s}, 102110^{21} hits per second: rate 2×1015s12 \times 10^{-15}\,\mathrm{s}^{-1}, lifetime 107\sim 10^7 years. (c) G=100G = 100 and 6969: e18108\eu^{18} \approx 10^8 longer, e13106\eu^{-13} \approx 10^{-6} shorter. (d) The α\alpha is tightly bound and leaves with positive energy; a proton would not; a 12^{12}C has three times the charge and a far larger exponent (cluster decay exists, at 101010^{-10}).

Exercise 31.10 ★★★

Ammonia. Splitting Δ=1×104eV\Delta = 1 \times 10^{-4}\,\mathrm{eV}. (a) Inversion frequency and the period of the tunnelling oscillation. (b) The N atom is prepared on one side: write ψ(t)\psi(t) and the probability of finding it on the other side. (c) ND3_3 tunnels at 1.6GHz1.6\,\mathrm{GHz}: deduce κa\kappa a from the ratio of the splittings if Δeκa\Delta \propto \eu^{-\kappa a} and κm\kappa \propto\sqrt m (mD=2mHm_{\text{D}} = 2m_{\text{H}}, barrier unchanged). (d) Why can a chiral molecule be left-handed for years?

Solution

Solution of Exercise 31.10.

(a) 24GHz24\,\mathrm{GHz}; 41ps41\,\mathrm{ps}. (b) ψ=(φ+eiE+t/+φeiEt/)/2\psi = (\varphi_+\eu^{-\iu E_+t/\hbar} + \varphi_-\eu^{-\iu E_-t/\hbar})/\sqrt2; P=sin2(πνt)P = \sin^2(\pi\nu t). (c) e(21)κa=15\eu^{(\sqrt2 - 1) \kappa a} = 15: κa=6.5\kappa a = 6.5. (d) Heavier groups and higher barriers make the tunnelling period longer than years.

Exercise 31.11 ★★★

The STM. Φ=4.5eV\Phi = 4.5\,\mathrm{eV}, d=0.5nmd = 0.5\,\mathrm{nm}, bias 0.1V0.1\,\mathrm{V}, I=1nAI = 1\,\mathrm{nA}. (a) κ\kappa, the factor of II per 0.1nm0.1\,\mathrm{nm}, the height resolution if II is measured to 1%1\%. (b) Electrons per second in the current. (c) The tip’s apex atom is 0.25nm0.25\,\mathrm{nm} closer than its neighbours (laterally 0.25nm0.25\,\mathrm{nm} away, so d=d2+0.252d' = \sqrt{d^2 + 0.25^2}): fraction of the current it carries. (d) A 1cm1\,\mathrm{cm} steel frame expands by αLΔT\alpha L\Delta T with α=1×105K1\alpha = 1 \times 10^{-5}\,\mathrm{K}^{-1}: what temperature stability does a picometre need, and how do designs get round it?

Solution

Solution of Exercise 31.11.

(a) 1.09×1010m11.09 \times 10^{10}\,\mathrm{m}^{-1}; ×8.8\times 8.8; 1%1\% of II is 0.5pm0.5\,\mathrm{pm}. (b) 6×1096 \times 10^9. (c) Each neighbour at d=0.56nmd' = 0.56\,\mathrm{nm} carries e2κ×0.06nm=0.28\eu^{-2\kappa \times 0.06\,\text{nm}} = 0.28 of the apex’s current; the apex carries about half. (d) 105K10^{-5}\,\mathrm{K} — impossible; symmetric designs in which tip and sample expand together, fast scans, low-expansion materials, and the feedback.

Exercise 31.12 ★★★

A quantum-well laser. A 10nm10\,\mathrm{nm} layer of GaAs (gap 1.42eV1.42\,\mathrm{eV}, me=0.067mem_{\text{e}}^* = 0.067\,m_{\text{e}}, mh=0.45mem_{\text{h}}^* = 0.45\,m_{\text{e}}) between barriers 0.3eV0.3\,\mathrm{eV} higher. (a) Ground confinement energies of electron and hole (infinite well). (b) Photon energy and wavelength of the laser. (c) Width for 780nm780\,\mathrm{nm}. (d) With the finite barrier, how many bound electron states does the 10nm10\,\mathrm{nm} well hold?

Solution

Solution of Exercise 31.12.

(a) 0.0560.056\, and 0.008eV0.008\,\mathrm{eV}. (b) 1.48eV1.48\,\mathrm{eV}, 835nm835\,\mathrm{nm}. (c) Confinement 0.17eV0.17\,\mathrm{eV}: L=6.1nmL = 6.1\,\mathrm{nm}. (d) R=3.6R = 3.6: three.

31.6 Problem: The STM, the alpha particle and the ammonia maser

Problem 31.1

Weekend problem — three tunnels

Data: =1.05×1034Js\hbar = 1.05 \times 10^{-34}\,\mathrm{J}\,\mathrm{s}, h=6.63×1034Jsh = 6.63 \times 10^{-34}\,\mathrm{J}\,\mathrm{s}, me=9.11×1031kgm_{\text{e}} = 9.11 \times 10^{-31}\,\mathrm{kg}, mα=6.64×1027kgm_\alpha = 6.64 \times 10^{-27}\,\mathrm{kg}, e2/4πε0=1.44MeVfme^2/4\pi\varepsilon_0 = 1.44\,\mathrm{MeV}\,\mathrm{fm}, 1eV1\,\mathrm{eV} =1.60×1019J= 1.60 \times 10^{-19}\,\mathrm{J}.

Part I — The scanning tunnelling microscope. Tip and sample are tungsten (Φ=4.5eV\Phi = 4.5\,\mathrm{eV}); gap dd; bias 50mV50\,\mathrm{mV}; the current is I=I0e2κdI = I_0\eu^{-2\kappa d}.

  1. κ\kappa for electrons at the Fermi level facing the vacuum barrier Φ\Phi.
  2. Factor of change of II for Δd=0.1nm\Delta d = 0.1\,\mathrm{nm}; for 1pm1\,\mathrm{pm}.
  3. The feedback holds II constant to 1%1\%: height noise.
  4. I=1nAI = 1\,\mathrm{nA}: electrons per second; average time between two electrons, compared with the tunnelling “time” /Φ\sim\hbar/\Phi.
  5. The apex atom and its neighbours 0.25nm0.25\,\mathrm{nm} away laterally, d=0.5nmd = 0.5\,\mathrm{nm}: share of the apex atom; why a single atom is enough to image atoms.
  6. An adsorbed atom 0.1nm0.1\,\mathrm{nm} high: change of II in constant-height mode; why constant-current mode is preferred.
  7. A 2cm2\,\mathrm{cm} metal frame, α=1×105K1\alpha = 1 \times 10^{-5}\,\mathrm{K}^{-1}: temperature change that moves the tip by 1pm1\,\mathrm{pm}; how do instruments cope (symmetry, speed, low-expansion materials)?
  8. With a 1V1\,\mathrm{V} bias the barrier is no longer rectangular: sketch it and say whether II grows faster or slower than linearly with VV.
  9. On an oxidised patch the work function is 3eV3\,\mathrm{eV} instead of 4.5eV4.5\,\mathrm{eV}: by how much does the tip retract in constant-current mode over a perfectly flat surface? What does the STM image, then?

Part II — Alpha decay of a nucleus Z=92Z = 92, A=238A = 238. The α\alpha (E=4.2MeVE = 4.2\,\mathrm{MeV}) moves in a nucleus of radius Rn=8fmR_{\text{n}} = 8\,\mathrm{fm}; outside, the Coulomb energy is V(r)=2(Z2)e2/4πε0rV(r) = 2(Z - 2)e^2/4\pi\varepsilon_0r.

  1. Height of the barrier at RnR_{\text{n}}; classical turning point bb.
  2. Speed of the α\alpha inside and its frequency of hits on the wall.
  3. Gamow exponent G=(22mE/)b[arccosRn/b(Rn/b)(1Rn/b)]G = (2\sqrt{2mE}/\hbar)\,b\,[\arccos\sqrt{R_{\text{n}}/b} - \sqrt{(R_{\text{n}}/b)(1 - R_{\text{n}}/b)}] and T=eGT = \eu^{-G}.
  4. Decay rate and half-life; compare with the measured 4.5×1094.5 \times 10^9 years.
  5. The same with E=5.2MeVE = 5.2\,\mathrm{MeV} (another isotope): half-life; the Geiger–Nuttall sensitivity.
  6. Taking Rn=9fmR_{\text{n}} = 9\,\mathrm{fm} instead of 8fm8\,\mathrm{fm}: effect on the half-life; comment on the precision of such estimates.
  7. Why is the tunnelling of a proton (charge 11, mass 1/41/4) not observed from the same nucleus, and that of a 12^{12}C (charge 66) either?
  8. The α\alpha emerges with the full 4.2MeV4.2\,\mathrm{MeV} although it “passed under” a 30MeV30\,\mathrm{MeV} wall: is energy conserved?

Part III — The ammonia maser. The N atom tunnels through the H3_3 plane; the ground level is split by Δ=9.8×105eV\Delta = 9.8 \times 10^{-5}\,\mathrm{eV}.

  1. Frequency and wavelength of the inversion transition.
  2. Write the symmetric and antisymmetric states in terms of “N up” and “N down”; which is lower, and why?
  3. A molecule prepared “N up”: ψ(t)\psi(t), the probability of “down” at time tt, the period.
  4. Boltzmann ratio of the two levels at 300K300\,\mathrm{K}: can the thermal population give a maser? How is the inversion obtained (a state-selecting electric field)?
  5. Photon energy in joules; number of molecules that must emit per second for an output of 1×1010W1 \times 10^{-10}\,\mathrm{W}.
  6. ND3_3 tunnels at 1.6GHz1.6\,\mathrm{GHz}: deduce κa\kappa a for NH3_3 assuming Δeκa\Delta \propto \eu^{-\kappa a} with κm\kappa \propto \sqrt{m}.
  7. The maser was the first atomic clock: why is a tunnelling frequency a good clock and which effects shift it?
  8. Summarise: quantisation from confinement, tunnelling from evanescence, and what fixes the exponents.
Solution

Solution of Problem 31.1.

1. κ=2mΦ/=1.09×1010m1\kappa = \sqrt{2m\Phi}/\hbar = 1.09 \times 10^{10}\,\mathrm{m}^{-1}.

2. e2.2=8.8\eu^{2.2} = 8.8; e0.022\eu^{0.022}: 2.2%2.2\% per picometre.

3. 0.5pm0.5\,\mathrm{pm}.

4. 6×1096 \times 10^9 per second; 0.16ns0.16\,\mathrm{ns} apart, against /Φ1016s\hbar/\Phi \sim 10^{-16}\,\mathrm{s}: one at a time.

5. About half; the rest decays so fast with distance that the image is that of one atom.

6. ×8.8\times 8.8; constant current keeps the tip safe and turns the exponential into a linear height.

7. ΔT=1012/(105×0.02)=5×106K\Delta T = 10^{-12}/(10^{-5} \times 0.02) = 5 \times 10^{-6}\,\mathrm{K}; symmetric mounts, speed, low-expansion materials, and a feedback that follows the drift.

8. A trapezoid, lowered on the collecting side: the effective barrier thins and II grows faster than linearly.

9. κ=0.89×1010m1\kappa' = 0.89 \times 10^{10}\,\mathrm{m}^{-1}; at 0.5nm0.5\,\mathrm{nm} the exponent drops from 10.910.9 to 8.98.9, I×7I \times 7: the tip retracts by ln7/2κ0.1nm\ln 7/2\kappa' \approx 0.1\,\mathrm{nm} — a bump that is electronic, not geometric.

10. V(Rn)=2×90×1.44/8=32MeVV(R_{\text{n}}) = 2 \times 90 \times 1.44/8 = 32\,\mathrm{MeV}; b=62fmb = 62\,\mathrm{fm}.

11. v=1.4×107m/sv = 1.4 \times 10^{7}\,\mathrm{m}/\mathrm{s}; 9×10209 \times 10^{20} hits per second.

12. G=96G = 96; T=e961042T = \eu^{-96} \approx 10^{-42}.

13. Rate 1021s1\sim 10^{-21}\,\mathrm{s}^{-1}, half-life 1013\sim 10^{13} years — a thousand times the measured value: for an exponent of a hundred, a crude model does well to land within a few orders of magnitude.

14. G=79G = 79: e17107\eu^{17} \approx 10^7 times shorter, about 10610^6 years — one extra MeV, seven orders of magnitude.

15. G=92G = 92: e4.590\eu^{4.5} \approx 90 times shorter — a fermi of radius is two orders of magnitude.

16. A proton is not pre-formed with positive energy (it is bound by some 7MeV7\,\mathrm{MeV}); a 12^{12}C has three times the charge and an exponent three times larger.

17. Yes: the α\alpha always has 4.2MeV4.2\,\mathrm{MeV}; its wave function merely has an evanescent part under the barrier, where no measurement finds it with negative kinetic energy.

18. 23.7GHz23.7\,\mathrm{GHz}; 1.27cm1.27\,\mathrm{cm}.

19. (±)/2(|{\uparrow}\rangle \pm |{\downarrow}\rangle)/\sqrt2; the symmetric one is lower — no node, less curvature.

20. ψ=(φ+eiE+t/+φeiEt/)/2\psi = (\varphi_+\eu^{-\iu E_+t/\hbar} + \varphi_-\eu^{-\iu E_-t/\hbar})/\sqrt2; P=sin2(πνt)P_{\downarrow} = \sin^2(\pi\nu t); period 42ps42\,\mathrm{ps}.

21. eΔ/kBT=0.996\eu^{-\Delta/k_BT} = 0.996: almost equal, no inversion; an inhomogeneous electric field sorts the beam and sends the upper-state molecules into the cavity.

22. 1.6×1023J1.6 \times 10^{-23}\,\mathrm{J}; 6×10126 \times 10^{12} molecules per second.

23. κa=6.5\kappa a = 6.5.

24. It is fixed by the molecule alone, identical for every molecule, and insensitive to the outside; shifts come from the Doppler effect in the beam, stray electric fields, collisions and the cavity’s pulling.

25. Confinement quantises (Enn2/L2E_n \propto n^2/L^2); evanescence lets the wave through (Te2κaT \sim \eu^{-2\kappa a}); the exponents are set by 2m(V0E)/\sqrt{2m(V_0 - E)}/\hbar and the width — mass, height, distance.