Advanced Design and Epitaxy Tutorial Questions

Alloys

Question

1. What is meant by alloying in the context of a semiconductor? Why is this useful?

Answer

Combining two or more semiconductor materials to form a ternary (e.g. $\text{ce}A{l}_{x}G{a}_{(1-x)}As)$ or quaternary (e.g. $\text{ce}I{n}_{x}G{a}_{(1-x)}A{s}_y{P}_{(1-y)}$) alloy. By varying composition, you can continuously tune the band gap, lattice constant, and refractive index. This enables band gap engineering — designing heterostructures with specific emission wavelengths, carrier confinement, and optical waveguiding properties.

Question

2. What is the physical cause of band gap “bowing” in semiconductor alloys?

Answer

The band gap of an alloy doesn’t vary linearly with composition — it deviates (bows) due to differences in electronegativity and atomic size between the constituent atoms. The disorder in the alloy creates local potential fluctuations that lower the average band gap below the linear interpolation (Vegard’s law). This is characterised by a bowing parameter: Eg(x) = (1-x)Eg₁ + xEg₂ - bx(1-x), where b is the bowing parameter.

Question

3. AlxGa1-xAs is a useful semiconductor alloy for a number of device applications. At T=300K, the conduction band edge energies (also called valleys) for $\text{ce}A{l}_{x}G{a}_{(1-x)}As$ vary with Al composition, x as follows: $E_{\Gamma }(eV)=1.424+1.155x+0.37x^2$ $E_X(eV)=1.9+0.124x+0.144x^2$ $E_L(eV)=1.71+0.69x$ where the energies are all relative to the energy of the valence band maximum (at k=0). Note that $\Gamma$ is the direct conduction band edge at k=0 (at the Brillouin Zone Centre) while X and L are indirect ($k\ne 0$). Using this information, what is the useful composition range to make a bulk LED based on the $\text{ce}A{l}_{x}G{a}_{(1-x)}As$ alloy. Explain why. What corresponding range of operating wavelengths are therefore achievable with $\text{ce}A{l}_{x}G{a}_{(1-x)}As$?

Answer

The key constraint: LEDs need efficient radiative recombination, which requires a direct bandgap. That means the lowest conduction band minimum must be at Γ (k=0). If X or L becomes the lowest valley, the gap is indirect and the LED is useless — recombination requires a phonon and is overwhelmingly non-radiative.

Step 1 — Find where the crossover happens:

The material is direct as long as $E_{\Gamma }$ is the lowest of the three valleys. So find the value of $x$ where $E_{\Gamma }$ equals each of the other two.

Γ = X crossover:

$1.424+1.155x+0.37x^2=1.9+0.124x+0.144x^2$

$0.226x^2+1.031x-0.476=0$

Quadratic formula:

$x=\frac{-1.031+\sqrt{1.031^2+4(0.226)(0.476)}}{2\times 0.226}=\frac{-1.031+\sqrt{1.063+0.430}}{0.452}$

$x=\frac{-1.031+1.222}{0.452}=\frac{0.191}{0.452}\approx 0.42$

Γ = L crossover:

$1.424+1.155x+0.37x^2=1.71+0.69x$

$0.37x^2+0.465x-0.286=0$

$x=\frac{-0.465+\sqrt{0.465^2+4(0.37)(0.286)}}{2\times 0.37}=\frac{-0.465+\sqrt{0.216+0.423}}{0.74}$

$x=\frac{-0.465+0.800}{0.74}\approx 0.45$

Step 2 — Determine the binding crossover:

The Γ valley first crosses the X valley at $x\approx 0.42$. Above this, X is lower than Γ and the bandgap becomes indirect. (The L crossover at 0.45 doesn’t matter because X has already taken over by then.)

So the useful composition range is $0\le x\le 0.42$.

Step 3 — Wavelength range:

At $x=0$: $E_{\Gamma }=1.424$ eV → $\lambda =hc/E=1240/1.424=871$ nm

At $x=0.42$: $E_{\Gamma }=1.424+1.155(0.42)+0.37(0.42{)}^2=1.424+0.485+0.065=1.974$ eV → $\lambda =1240/1.974=628$ nm

Operating wavelength range: ~628 nm to ~871 nm (red to near-IR).

Epitaxy

Question

4. Describe the basic operating principle of MBE for crystal growth.

Answer

MBE operates under ultra-high vacuum (~10⁻¹⁰ mbar). Source materials (e.g. Ga, Al, As) are heated in effusion cells (Knudsen cells) to produce molecular beams directed at a heated substrate. The mean free path exceeds the chamber dimensions, so molecules travel in straight lines without scattering (ballistic transport). Shutters control which beams reach the substrate, enabling abrupt composition changes. Growth rate is ~1 monolayer/s. In-situ monitoring via RHEED provides real-time feedback on surface crystallinity and growth rate.

Question

5. Describe the basic operating principle of MOCVD for crystal growth.

Answer

MOCVD (also called MOVPE) uses metal-organic precursors (e.g. trimethylgallium, trimethylaluminium) and hydride gases (e.g. arsine, phosphine) carried by hydrogen into a heated reactor chamber at moderate pressures (~10-760 mbar). Precursors decompose (pyrolyse) at the hot substrate surface, releasing the metal atoms which incorporate into the growing crystal. Gas flow rates control composition and growth rate. Higher throughput than MBE — multiple wafers can be processed simultaneously.

Question

6. Compare and contrast MBE and MOCVD.

Answer

MBE advantages: UHV gives extremely clean interfaces, RHEED provides in-situ monitoring, monolayer-level thickness control, excellent for research and quantum structures. MBE disadvantages: slow growth rate, expensive to operate (UHV), low throughput (typically one wafer), difficult to grow phosphorus-containing materials.

MOCVD advantages: higher growth rates, multi-wafer processing (higher throughput), lower cost per wafer, handles phosphide and nitride materials well, industry standard for mass production. MOCVD disadvantages: less precise interface control, toxic precursors (arsine, phosphine), carbon incorporation from organics, no in-situ surface monitoring equivalent to RHEED.

Question

7. How does lattice constant mismatch impact epilayer quality? When is strain beneficial or problematic for a semiconductor laser?

Answer

If the epilayer lattice constant differs from the substrate, the epilayer is strained (compressed or tensile) to match the substrate. Below a critical thickness, the layer grows pseudomorphically (strained but defect-free). Above the critical thickness, strain relaxes via misfit dislocations which act as non-radiative recombination centres, degrading device performance and reliability.

Beneficial: controlled strain in thin quantum wells modifies the valence band structure — compressive strain reduces the in-plane hole effective mass, lowering threshold current and improving differential gain. Tensile strain can switch the polarisation of emission (useful for semiconductor optical amplifiers).

Problematic: exceeding the critical thickness generates dislocations → increased non-radiative recombination → higher threshold current, reduced efficiency, and device degradation over time.

Light Emission

Question

8. Illustrate the main differences between spontaneous emission and stimulated emission.

Answer

Spontaneous emission: an electron in the conduction band recombines with a hole in the valence band randomly. The emitted photon has energy ≈ Eg but random direction, phase, and polarisation. Rate depends on carrier density.

Stimulated emission: an incoming photon with energy = Eg triggers an electron-hole recombination. The emitted photon is identical to the stimulating photon — same wavelength, direction, phase, and polarisation (coherent). This is the basis of optical amplification and lasing.

Diagram would show: spontaneous — single electron dropping, photon emitted in random direction. Stimulated — incoming photon passes near excited electron, two identical photons emerge together.

Question

9. How does stimulated emission give rise to optical gain in a semiconductor?

Answer

When population inversion exists (more electrons in the conduction band than the valence band at the lasing energy), an incoming photon is more likely to stimulate emission than be absorbed. Each stimulated emission event produces an additional coherent photon, so the optical intensity grows as light propagates through the material. The gain coefficient g describes the net amplification per unit length: I(z) = I₀·exp(gz). Above a threshold injection current, gain exceeds losses (absorption, scattering, mirror losses) and lasing occurs.

Question

10. What are the three essential requirements to make a semiconductor laser?

Answer

(1) Gain medium: a semiconductor with a direct band gap capable of efficient radiative recombination under population inversion. (2) Pumping mechanism: electrical injection (forward-biased pn junction) to achieve population inversion — more electrons in the conduction band than holes at the emission energy. (3) Optical feedback: a resonant cavity (e.g. cleaved facets forming a Fabry-Perot cavity, or DFB grating) to provide feedback so photons make multiple passes through the gain medium, enabling stimulated emission to dominate.

Question

11. What is the main difference between a semiconductor laser and an LED?

Answer

An LED operates below threshold on spontaneous emission — output is incoherent (broad spectrum, random phase/direction, linear L-I characteristic). A laser operates above threshold on stimulated emission — output is coherent (narrow linewidth, well-defined phase and direction, with a distinct threshold in the L-I curve). The laser requires an optical cavity for feedback; the LED does not. Lasers also have much faster modulation bandwidth than LEDs.

Semiconductor Lasers

Question

12. Describe the advantages of a double heterostructure laser over a homojunction laser.

Answer

Homojunction: single pn junction of one material. Carriers diffuse away from the junction over the diffusion length (~µm), giving a large active region and low carrier density. No optical confinement — light spreads throughout the structure. High threshold current.

Double heterostructure (DH): a thin narrow-gap active layer sandwiched between wider-gap cladding layers. Provides: (1) Carrier confinement: the band gap discontinuities at both interfaces form potential barriers, confining injected carriers to the thin active layer → high carrier density for population inversion at lower current. (2) Optical confinement: the wider-gap claddings have lower refractive index, forming a slab waveguide that confines the optical mode to overlap with the gain region. Result: dramatically reduced threshold current (orders of magnitude lower than homojunction).

Question

13. What is quantum confinement? Advantages of quantum well over bulk lasers?

Answer

Quantum confinement occurs when a semiconductor layer is thin enough (~<20 nm) that carrier wavefunctions are confined, quantising the energy levels. The continuous density of states becomes step-like.

QW laser advantages over bulk: (1) Modified density of states: step-function DOS concentrates carriers at specific energies → higher gain per carrier → lower threshold current. (2) Tunable emission: quantised energy levels depend on well width, allowing wavelength tuning independent of material composition alone. (3) Reduced temperature sensitivity: the step DOS means gain peak shifts less with temperature compared to bulk. (4) Strain engineering: thin QWs can be strained without generating dislocations (below critical thickness), enabling further band structure optimisation. (5) Lower threshold current density and higher differential gain → faster modulation.

Question

14. F-P cavity: n = 3.4, L = 1 mm, λ = 1550 nm. Mode spacing? Modes within 50 nm gain bandwidth?

Answer

Mode spacing: Δλ = λ²/(2nL) = (1550×10⁻⁹)²/(2 × 3.4 × 1×10⁻³) = 2.4025×10⁻¹²/6.8×10⁻³ = 3.533×10⁻¹⁰ m ≈ 0.353 nm

Number of modes in 50 nm bandwidth: N = 50/0.353 ≈ 142 modes.

This is why F-P lasers are inherently multi-mode — there are many longitudinal modes within the gain bandwidth.

Question

15. How small would the F-P cavity need to be for single mode? Is this practical? Two other approaches?

Answer

For single mode: need Δλ ≥ gain bandwidth, so L ≤ λ²/(2n·Δλ_gain) L ≤ (1550×10⁻⁹)²/(2 × 3.4 × 50×10⁻⁹) = 2.4025×10⁻¹²/3.4×10⁻⁷ = 7.07×10⁻⁶ m ≈ 7 µm

A 7 µm cavity is impractically short for an edge-emitting laser — the gain path length is too short to overcome mirror losses. VCSELs can have ~µm-scale cavities but use high-reflectivity DBR mirrors to compensate.

Two alternatives for single mode operation: (1) Distributed Feedback (DFB) laser: a Bragg grating integrated along the cavity length selects a single wavelength satisfying the Bragg condition (λ_B = 2nΛ). Only one mode has sufficient feedback. (2) Distributed Bragg Reflector (DBR) laser: Bragg gratings at one or both ends of the cavity act as wavelength-selective mirrors, reflecting only a narrow band and suppressing other modes.

Question

16. Describe two approaches to manufacturing single mode lasers and outline their operating principle.

Answer

(1) DFB laser: a periodic grating (period Λ) is etched into or near the active layer during fabrication (typically by e-beam or holographic lithography, followed by regrowth). The grating provides continuous distributed feedback via Bragg reflection. Only the mode satisfying λ_B = 2n_eff·Λ experiences constructive feedback and lases. A quarter-wave phase shift is often introduced at the centre to select a single mode from the two degenerate Bragg modes.

(2) VCSEL: a very short vertical cavity (~λ) between two DBR mirror stacks (alternating high/low-n layers, each λ/4 thick). The short cavity means mode spacing exceeds the gain bandwidth, naturally yielding single longitudinal mode operation. The DBR mirrors provide >99% reflectivity to compensate the short gain path. Light emits perpendicular to the wafer surface.