Opto-electronic Devices: Solar Cells, Photodetectors and LEDs
CSIR NET electronics: absorption and emission in semiconductors, photodiodes, PIN and APD detectors, solar cell I-V and efficiency, LEDs, with practice.
- 16 min
On this page · 14 sections
Prerequisites
- Semiconductor Devices: Junctions, Diodes, BJTs, FETs and Heterojunctions: the diode equation and the depletion region, which every device here reuses.
- Photon energy and the photoelectric relation \( E = h\nu \), from Wave–Particle Duality and the Schrödinger Equation.
- Total internal reflection and Fresnel reflection at an interface, from electromagnetic waves at boundaries.
The Hook: The Same Junction, Run Both Ways
Photons and the Band Gap
A photon is absorbed across the gap only if \( h\nu \geq E_g \). The conversion everyone should do in their head:
$$\boxed{\lambda\,(\mu\text{m}) = \frac{1.24}{E\,(\text{eV})}} \qquad (hc = 1240\ \text{eV nm})$$| Material | \( E_g \) (eV, 300 K) | Cut-off / emission \( \lambda \) | Gap |
|---|---|---|---|
| Ge | 0.66 | 1.88 µm | indirect |
| Si | 1.12 | 1.11 µm | indirect |
| GaAs | 1.42 | 0.87 µm | direct |
| GaP | 2.26 | 0.55 µm | indirect |
| GaN | 3.4 | 0.36 µm | direct |
Direct and indirect gaps
A photon carries negligible momentum compared with an electron at the zone boundary. So:
- In a direct-gap semiconductor (GaAs, InP, GaN) the conduction-band minimum and valence-band maximum are at the same \( \mathbf k \). Absorption and emission need only a photon. Radiative recombination is fast; these materials make LEDs and lasers.
- In an indirect-gap semiconductor (Si, Ge, GaP, AlAs) the extrema are at different \( \mathbf k \). A phonon must supply the momentum. Absorption near the edge is weak and radiative recombination slow, so non-radiative paths win. These materials still make excellent detectors and solar cells (absorption is weak near the edge but strong well above it), but poor emitters.
Absorption and penetration depth
Light intensity falls as \( I(x) = I_0e^{-\alpha x} \). The penetration depth \( 1/\alpha \) sets how thick the active region must be. Near the edge, \( \alpha \) of a direct-gap material rises steeply (roughly \( \propto \sqrt{h\nu - E_g} \)); that of an indirect-gap material rises slowly. This is why a GaAs solar cell needs only a few micrometres of material while a crystalline silicon cell is far thicker.
Photodetectors
Figures of merit
If optical power \( P \) at wavelength \( \lambda \) produces photocurrent \( I_{ph} \):
$$\text{photon flux} = \frac{P}{h\nu}, \qquad \text{quantum efficiency } \eta = \frac{I_{ph}/q}{P/h\nu}$$$$\boxed{\mathcal R = \frac{I_{ph}}{P} = \frac{\eta q\lambda}{hc} = \eta\,\frac{\lambda\,(\mu\text{m})}{1.24}\ \text{A/W}}$$The responsivity \( \mathcal R \) rises linearly with \( \lambda \) (each photon carries less energy but still gives one electron) up to the cut-off \( \lambda_c = hc/E_g \), then drops to zero. Other figures: response time (transit and RC limited), dark current, and noise-equivalent power or detectivity.
Device types
| Detector | Mechanism | Gain | Speed | Remarks |
|---|---|---|---|---|
| Photoconductor | light raises conductivity of a bulk sample | \( G = \tau/t_{tr} \), can be \( \gg 1 \) | slow (set by \( \tau \)) | simple; gain–bandwidth trade-off |
| p-n photodiode | carriers separated by the junction field | 1 | moderate | diffusion from outside the depletion region slows it |
| PIN photodiode | wide intrinsic absorption layer | 1 | fast | low capacitance, most light absorbed in the field region |
| Avalanche photodiode (APD) | high reverse bias, impact ionisation | \( M \) (tens to hundreds) | fast | extra noise from the random multiplication |
| Phototransistor | photocurrent acts as base current | \( \approx \beta \) | slow | large Miller-type capacitance |
Photoconductive gain. In a photoconductor of length \( L \) with bias \( V \), a carrier crosses in \( t_{tr} = L^2/(\mu V) \). If the photogenerated carrier lives for \( \tau \), the contacts keep replenishing it, so each absorbed photon pushes \( \tau/t_{tr} \) electrons round the circuit:
$$\boxed{G = \frac{\tau}{t_{tr}} = \frac{\tau\mu V}{L^2}}$$The response time is also of order \( \tau \), so gain is bought with speed.
Photodiode operating modes
The illuminated diode obeys
$$I = I_s\left(e^{V/\eta V_T} - 1\right) - I_{ph}$$which shifts the dark I–V curve down by \( I_{ph} \).
- Photoconductive mode (third quadrant, reverse bias): current \( \approx -(I_s + I_{ph}) \), linear in light power; reverse bias widens the depletion region, lowers capacitance and speeds response. Used for detection.
- Photovoltaic mode (fourth quadrant, \( V > 0 \), \( I < 0 \)): the device delivers power. This is the solar cell.
Solar Cells
Write the cell current as positive when delivered to the load:
$$\boxed{I = I_L - I_0\left(e^{V/\eta V_T} - 1\right)}$$with light-generated current \( I_L \) and dark saturation current \( I_0 \).
| Parameter | Definition | Result |
|---|---|---|
| Short-circuit current | \( I \) at \( V = 0 \) | \( I_{sc} = I_L \propto \) light intensity |
| Open-circuit voltage | \( V \) at \( I = 0 \) | \( V_{oc} = \eta V_T\ln\left(\dfrac{I_L}{I_0} + 1\right) \) |
| Fill factor | \( FF = \dfrac{V_{mp}I_{mp}}{V_{oc}I_{sc}} \) | ideal cells roughly 0.8–0.85 |
| Efficiency | \( \eta_{cell} = \dfrac{P_{\max}}{P_{in}} = \dfrac{FF\,V_{oc}I_{sc}}{P_{in}} \) |
Consequences that Part C uses:
- Intensity: \( I_{sc} \) scales linearly; \( V_{oc} \) rises only logarithmically, by \( \eta V_T\ln2 \approx 18 \) mV per doubling. So efficiency rises slightly under concentration (until heating and series resistance bite).
- Temperature: \( I_0 \propto n_i^2 \) rises steeply, so \( V_{oc} \) and efficiency fall with temperature.
- Band gap trade-off: a small gap absorbs more of the spectrum (more current) but gives low voltage; a large gap gives high voltage but wastes sub-gap photons; photons far above the gap lose their excess energy as heat (thermalisation). For a single junction under sunlight the detailed-balance (Shockley–Queisser) limit is about 33%, for gaps near 1.3–1.4 eV.
- Parasitic resistances: series resistance \( R_s \) lowers the slope near \( V_{oc} \) and reduces \( FF \) (and eventually \( I_{sc} \)); a low shunt resistance \( R_{sh} \) lowers the slope near \( I_{sc} \) and reduces \( FF \) and \( V_{oc} \).
Cells are rated at standard test conditions: 1000 W/m\( ^2 \), AM1.5 spectrum, 25 °C.
Light-Emitting Diodes
Forward bias injects electrons into the p side and holes into the n side. In a direct-gap material they recombine radiatively, emitting photons of energy close to \( E_g \).
$$\lambda_{peak} \approx \frac{1.24}{E_g}\ \mu\text{m}$$The spectrum is set by thermal spread of carriers in the bands: in the simple parabolic-band model the peak is near \( E_g + kT/2 \) and the full width at half maximum is about \( 1.8kT \) (about 0.047 eV at 300 K). In wavelength, \( \Delta\lambda = \lambda^2\Delta E/hc \): about 28 nm at 870 nm. This is far broader than a laser line; an LED is spontaneous emission, a laser diode stimulated emission above a threshold current.
Efficiency chain
$$\eta_{ext} = \eta_{inj} \times \eta_{int} \times \eta_{extr}$$- Injection efficiency: fraction of current that injects carriers into the active region (heterojunctions push this near 1).
- Internal quantum efficiency: radiative recombination rate over total recombination rate, \( \eta_{int} = \dfrac{1/\tau_r}{1/\tau_r + 1/\tau_{nr}} \).
- Extraction efficiency: fraction of internally generated photons that escape. A semiconductor has a high refractive index \( n \), so only light inside the escape cone \( \sin\theta_c = 1/n \) gets out of a flat surface:
further multiplied by the Fresnel transmission \( 1 - \left(\frac{n - 1}{n + 1}\right)^2 \) at normal incidence. With \( n = 3.6 \), only about 1.3% escapes a flat face: hence domed encapsulation, textured surfaces and shaped chips.
Materials and colours. Infrared: GaAs. Red and amber: AlGaInP and GaAsP families. Blue and green: InGaN on GaN. Akasaki, Amano and Nakamura received the 2014 Nobel Prize in Physics for efficient blue LEDs. White LEDs commonly combine a blue InGaN chip with a yellow-emitting phosphor. GaP, though indirect, can emit through isoelectronic impurity (nitrogen) centres that relax the momentum rule.
Putting the Three Together
graph LR
A[Photon in, h nu above Eg] --> B[Electron-hole pair]
B --> C{Junction field?}
C -->|reverse bias, collect| D[Photodetector: current proportional to power]
C -->|no bias, deliver power| E[Solar cell: fourth quadrant]
F[Forward current] --> G[Injected carriers]
G --> H{Direct gap?}
H -->|yes| I[LED: photon near Eg]
H -->|no| J[Mostly heat]Memory Tricks
Don’t Confuse
| This | With this | The difference |
|---|---|---|
| Quantum efficiency | Responsivity | electrons per photon (dimensionless, \( \leq 1 \) without gain); amperes per watt (depends on \( \lambda \)) |
| Photoconductive mode | Photovoltaic mode | reverse-biased photodiode, third quadrant; unbiased, power-delivering, fourth quadrant |
| Photoconductor gain | APD gain | \( \tau/t_{tr} \), bulk recirculation, slow; impact ionisation in a high field, fast but noisy |
| \( I_{sc} \) | \( V_{oc} \) | linear in light; logarithmic in light |
| LED | Laser diode | spontaneous emission, broad (\( \sim 1.8kT \)) spectrum, no threshold; stimulated emission, narrow line, threshold current |
| Internal efficiency | External efficiency | radiative fraction inside; photons out per electron in, including extraction |
Common Mistakes
Practice
All questions below are original practice questions written for this page. They are not from any CSIR NET paper. Use \( hc = 1240 \) eV nm and \( V_T = 25.9 \) mV at 300 K.
Level 1: Foundation
- What is the cut-off wavelength of a silicon photodiode (\( E_g = 1.12 \) eV)?
- In which quadrant of the I–V plane does a solar cell deliver power?
- Name one direct-gap and one indirect-gap semiconductor.
Answers
- About 1.11 µm. \( 1240/1.12 = 1107 \) nm.
- The fourth quadrant (positive voltage, negative current in the diode sign convention).
- Direct: GaAs (or GaN, InP). Indirect: Si (or Ge, GaP).
Level 2: Part B Style
Practice Q1. A photodiode with quantum efficiency 0.8 is illuminated with 850 nm light. Its responsivity is closest to:
(a) 0.34 A/W (b) 0.55 A/W (c) 0.69 A/W (d) 1.17 A/W
Answer: (b) 0.55 A/W
\( \mathcal R = \eta\lambda(\mu\text{m})/1.24 = 0.8 \times 0.85/1.24 = 0.548 \) A/W. (c) is \( \lambda/1.24 \) without \( \eta \); (d) inverts the ratio.
Practice Q2. A 1 mW beam at 650 nm falls on a detector. The number of photons arriving per second is approximately:
(a) \( 3.3\times10^{15} \) (b) \( 6.2\times10^{15} \) (c) \( 3.3\times10^{18} \) (d) \( 1.9\times10^{15} \)
Answer: (a) 3.3 × 10¹⁵
\( h\nu = 1240/650 = 1.91 \) eV \( = 3.06\times10^{-19} \) J. Flux \( = 10^{-3}/3.06\times10^{-19} = 3.27\times10^{15} \) s\( ^{-1} \). (b) is \( 10^{-3}/q \), the number of electrons per second in 1 mA; (c) slips a factor of \( 10^3 \).
Practice Q3. A silicon photoconductor is 100 µm long, biased at 10 V, with electron mobility 1350 cm\( ^2 \)/V s and photocarrier lifetime 1 µs. Its photoconductive gain is about:
(a) 1 (b) 13.5 (c) 135 (d) 1350
Answer: (c) 135
\( t_{tr} = L^2/(\mu V) = (10^{-4})^2/(0.135 \times 10) = 7.4\times10^{-9} \) s. \( G = \tau/t_{tr} = 10^{-6}/7.4\times10^{-9} = 135 \). (b) and (d) are off by factors of 10 from slips in unit conversion (\( 1350 \) cm\( ^2 \)/V s \( = 0.135 \) m\( ^2 \)/V s; \( 100\ \mu\text{m} = 10^{-4} \) m). Keep SI throughout.
Practice Q4. Which semiconductor is the best choice for an efficient LED emitting near 870 nm?
(a) Si (b) Ge (c) GaAs (d) GaP
Answer: (c) GaAs
GaAs is direct-gap with \( E_g = 1.42 \) eV, so \( \lambda \approx 1240/1.42 = 873 \) nm. Si and Ge are indirect and emit at the wrong wavelengths anyway; GaP’s gap (2.26 eV) corresponds to green.
Level 3: Part C Style
Practice Q5. A solar cell of area 100 cm\( ^2 \) under 1000 W/m\( ^2 \) has \( I_{sc} = 3.0 \) A, dark saturation current \( I_0 = 10^{-9} \) A, ideality factor 1, and fill factor 0.78 at 300 K. Consider:
- \( V_{oc} \approx 0.56 \) V.
- The conversion efficiency is about 13%.
- If the light intensity is doubled (fill factor unchanged), \( V_{oc} \) rises by about 18 mV.
- If the light intensity is doubled (fill factor unchanged), the efficiency doubles.
Which are correct?
(a) 1 and 2 only (b) 1, 2 and 3 only (c) 1, 3 and 4 only (d) 1, 2, 3 and 4
Answer: (b) 1, 2 and 3 only
- \( V_{oc} = V_T\ln(I_{sc}/I_0 + 1) = 0.02585\ln(3\times10^9) = 0.02585 \times 21.82 = 0.564 \) V. True.
- \( P_{in} = 1000 \times 0.01 = 10 \) W. \( P_{\max} = 0.78 \times 0.564 \times 3.0 = 1.32 \) W. Efficiency 13.2%. True.
- \( \Delta V_{oc} = V_T\ln2 = 17.9 \) mV. True.
- New efficiency \( = 0.78 \times 0.582 \times 6.0/20 = 13.6\% \): a small rise, not a doubling, because \( P_{in} \) doubles too. False.
Practice Q6. An LED is made from a semiconductor with refractive index 3.6 and internal quantum efficiency 0.9. Light leaves only through the flat top face. Consider:
- The critical angle is about 16°.
- Ignoring Fresnel loss, about 2% of internally generated photons escape the top face.
- Including normal-incidence Fresnel loss, the external quantum efficiency is about 1.2%.
- Encapsulating the chip in a dome of higher refractive index than air increases the extraction efficiency.
Which are correct?
(a) 1 and 2 only (b) 1, 2 and 4 only (c) 1, 2, 3 and 4 (d) 3 and 4 only
Answer: (c) 1, 2, 3 and 4
- \( \theta_c = \sin^{-1}(1/3.6) = 16.1^\circ \). True.
- \( (1 - \cos\theta_c)/2 = (1 - 0.9607)/2 = 0.0197 \approx 2\% \) (the approximation \( 1/4n^2 = 0.0193 \) agrees). True.
- Fresnel transmission \( 1 - (2.6/4.6)^2 = 0.68 \). External efficiency \( = 0.9 \times 0.0197 \times 0.68 = 0.012 \), about 1.2%. True.
- A dome of index \( n_d > 1 \) enlarges the escape cone to \( \sin\theta_c = n_d/n \), and the dome’s curved surface lets the light leave it near normal incidence. True.
Practice Q7. Three detectors are offered for a 1.55 µm optical link: a silicon PIN diode, an InGaAs PIN diode (\( E_g \approx 0.75 \) eV) and a silicon APD. Consider:
- Only the InGaAs diode responds at 1.55 µm.
- At 1.55 µm, an ideal (\( \eta = 1 \)) detector has responsivity 1.25 A/W.
- The silicon APD’s internal gain lets it detect 1.55 µm light weakly.
- A PIN diode is faster than a simple p-n diode of the same material mainly because absorption takes place in the high-field region.
Which are correct?
(a) 1, 2 and 4 only (b) 1 and 3 only (c) 2, 3 and 4 only (d) 1, 2, 3 and 4
Answer: (a) 1, 2 and 4 only
- Silicon’s cut-off is 1.11 µm; InGaAs at 0.75 eV cuts off at \( 1240/0.75 = 1650 \) nm. True.
- \( \mathcal R = 1.55/1.24 = 1.25 \) A/W. True.
- Gain multiplies carriers that are generated; sub-gap photons generate none. False.
- In a PIN the intrinsic layer is depleted; carriers drift in the field instead of diffusing slowly, and the wide layer also lowers capacitance. True.
Quick Revision Table
| Idea | Formula or fact |
|---|---|
| Wavelength–energy | \( \lambda(\mu\text{m}) = 1.24/E(\text{eV}) \) |
| Direct gap | GaAs, InP, GaN: efficient emitters |
| Indirect gap | Si, Ge, GaP, AlAs: phonon needed, weak emission |
| Responsivity | \( \mathcal R = \eta q\lambda/hc = \eta\lambda(\mu\text{m})/1.24 \) A/W |
| Photoconductor gain | \( G = \tau/t_{tr} = \tau\mu V/L^2 \) |
| Illuminated diode | \( I = I_s(e^{V/\eta V_T} - 1) - I_{ph} \) |
| \( V_{oc} \) | \( \eta V_T\ln(I_L/I_0 + 1) \); +18 mV per doubling |
| Efficiency | \( FF\,V_{oc}I_{sc}/P_{in} \) |
| Temperature | \( V_{oc} \) and efficiency fall as \( T \) rises |
| SQ limit | about 33% near 1.3–1.4 eV, single junction |
| LED linewidth | \( \approx 1.8kT \); \( \Delta\lambda = \lambda^2\Delta E/hc \) |
| Escape cone | \( (1 - \cos\theta_c)/2 \approx 1/4n^2 \) |
Key Takeaways
What’s Next
- Operational Amplifiers: the transimpedance amplifier that turns photocurrent into voltage is an op-amp circuit.
- Semiconductor Devices: Junctions, Diodes, BJTs, FETs and Heterojunctions: heterostructures behind modern LEDs and laser diodes.
- Data Interpretation, Error Analysis and Least-Squares Fitting: photon-counting statistics and fitting I–V data.
- Electronics and Experimental Methods (Core) overview.