Semiconductor Materials Engineering - Nanoscale Structures
Very Clean Materials
- Impurities less than 10^{-9}.
- Comparison:
- Polycrystalline metals and dielectric crystals: Impurities ~0.001.
- Amorphous polymers: Impurities ~0.01.
- Good mechanical parameters:
- Hardness can exceed that of steels.
- Allows for significant reduction in device dimensions.
- Increases production yield.
- High thermal conductivity:
- Devices can be placed more densely without concern for heating due to energy dissipation.
Unique Electrical Properties
- Electrical Conductivity:
- Semiconductors fall between metals and dielectrics in terms of electrical conductivity.
- Conductivity depends on the product of charge carrier concentration and their mobility.
- Electron mobility in semiconductors is several orders of magnitude higher than in metals or dielectrics.
- Even a small change in charge carrier density significantly alters the material's electrical conductivity.
- This principle underlies the operation of many devices.
Optical Properties
- Dependence on Material Purity:
- Optical properties are heavily influenced by the purity of the material.
- Semiconductors are unique in their purity, enabling the observation of charge carrier recombination.
- Recombination is due to direct electron transitions between energy levels of the crystal's constituent atoms, not impurity or defect centers.
- This characteristic is widely used in semiconductor light sources, making them more efficient than other types of devices.
Desired Attributes of Semiconductor Devices
- High-Speed Electronics:
- Large Frequency: Measured in GHz or even THz.
- Applications: Mobile phones, satellite TV, computers.
- Optoelectronics:
- Current ↔ Light conversion.
- Applications: Lasers, light-emitting diodes (LEDs), fiber optic communication.
- Solar cells, photodetectors.
- Miniaturization: Devices should be as small as possible.
- Speed: Devices should be as fast as possible.
- Cost-Effectiveness:
- Low service costs.
- Low energy consumption.
- Inexpensive.
US Patent 2402662 A
- Light-sensitive electric device
- Publication date: June 25, 1946
- Official filing date: May 27, 1941
- Patent registration date: May 27, 1941
- Also published as US2443542
- Inventors: Russell S. Ohl
- Original assignee: Bell Telephone Laboratories Inc.
- Russel Ohl - First semiconductor device.
Liquid Phase Epitaxy (LPE)
- Growth occurs from a solution.
- Suitable for heterostructures and doping.
- Offers moderate uniformity and thickness control.
- "Workhorse" in optoelectronics (LEDs, PDs, etc.).
Gas Phase Epitaxy
- Metal-Organic Chemical Vapor Deposition (MOCVD) (typically III/V).
- Growth occurs from the gas phase.
- Dominates in optoelectronics and Si-based electronics (LPCVD).
Molecular Beam Epitaxy (MBE)
- Evaporation in a vacuum.
- Dominates in III/V semiconductor-based electronics and optoelectronics.
- Allows for abrupt junctions and versatile selection of chemical elements.
- Popular research tool.
Liquid Phase Epitaxy (LPE) Details
- First demonstrated in 1963.
- Thin film deposition from a solution onto a substrate with crystalline orientation.
- Solvent can be In, Ga, or other low-melting-point metal.
- Solution, e.g., As
Liquid Phase Epitaxy (LPE) Growth Methods
- Step-cooling: Thickness \propto \sqrt{t}
- Equilibrium cooling: Thickness \propto t
- Supercooling: Thickness \propto t
- K is a constant dependent on the diffusion of each dissolved substance and the molar fraction of the soluble substance in the solution at the growth temperature.
- t is the growth time.
- R is the cooling rate.
Advantages of Liquid Phase Epitaxy (LPE)
- Inexpensive and simple setup.
- No high growth temperatures required (350-900^{\circ}C).
- Relatively high growth rate.
- Low concentration of point defects.
Disadvantages of Liquid Phase Epitaxy (LPE)
- Compositional non-uniformity.
- Insufficiently good morphology.
- Group III: Metal-organic precursors.
- Group V: Commonly hydrides.
- Examples:
- Arsenic, Phosphorus (As, P)
- Hydrogen, Carbon (H, C)
- Metal atom (Ga, In, Al)
Gas Phase Epitaxy System Components
- Hydrides
- Metal-organic sources
- Ventilation manifold
- H2 or N2 carrier gas
- Quartz tube reactor with heated graphite susceptor
- Throttle valve for total pressure control
- System pump
- Exhaust gas system
Advantages of Gas Phase Epitaxy
- Most flexible.
- Suitable for large-scale production.
- Abrupt junctions.
- Simple reactor, no ultra-high vacuum required.
- Economically viable.
Disadvantages of Gas Phase Epitaxy
- Low vapor pressure of many materials makes them difficult to transport via gas.
- Potential for carbon contamination and unintentional hydrogen introduction.
- Most parameters require precise control.
- Harmful precursors.
Molecular Beam Epitaxy (MBE)
- Atomic layer by atomic layer growth.
- High-resolution transmission electron micrograph (TEM) showing a GaAs/AlAs superlattice.
Advantages of Molecular Beam Epitaxy (MBE)
- Simple process.
- High uniformity.
- Abrupt junctions.
- Extensive in-situ monitoring capabilities.
Disadvantages of Molecular Beam Epitaxy (MBE)
- Problems with the P source.
- Expensive operation.
Semiconductor Usage
- Si (excluding solar cells): 400 ha.
- Applications: Microprocessors, memory, CCDs, etc.
- Other semiconductors (mainly GaAs and other III-V compounds): 10 ha.
- Applications: Lasers, LEDs, microwave devices, etc.
Direct vs. Indirect Bandgap
- Direct Bandgap:
- The conduction band minimum and valence band maximum are at the same point in the Brillouin zone (e.g., GaAs).
- Indirect Bandgap:
- The conduction band minimum and valence band maximum are at different points in the Brillouin zone (e.g., Silicon).
Infrared (IR) Spectral Regions
- Near-Infrared (NIR): 800 nm to 1.5 μm (sometimes up to 3 μm).
- Mid-Infrared (MIR): 1.5 μm (or 3 μm) to ~6 μm (or 30 μm).
- Far-Infrared (FIR): Remaining portion up to 1 mm.
- Terahertz Frequencies: 0.1 THz to 10 THz (30 μm – 300 μm).
- Creating alloys of two or more semiconductors.
- Using heterostructures to create quantum-confined structures.
- Epitaxial growth on mismatched substrates.
- These methods, alone or in combination, are widely used in the design and fabrication of various electronic and optoelectronic devices.
Forbidden Energy Gaps
- Elemental Semiconductors:
- Si: 1.12 eV (amorphous Si: 1.71 eV)
- Ge: 0.661 eV
- C (diamond): 5.46 eV
- Te: 0.33 eV
- Binary Semiconductors:
- GaAs: 1.41 eV (0.87 µm)
- GaP: 2.26 eV
- GaSb: 0.66 eV
- GaN: 3.4 eV (0.36 µm)
- InAs: 0.36 eV (3.4 µm)
- InSb: 0.17 eV (7 µm)
- InN: 0.65 eV (1.9 µm)
- InP: 1.34 eV (0.9 µm)
- SiGe: 1.12-0.66 eV
- CdTe: 1.5 eV (0.8 µm)
- ZnO: 3.3 eV (0.37 µm)
- The easiest way to change the properties of a material is to make an alloy of it with another material that differs in its characteristics.
- In the case of semiconductors, materials are alloyed to achieve two goals:
- To obtain the required forbidden energy gap.
- To obtain a material whose lattice constant matches the lattice constant of the substrate used in epitaxy.
Ternary Compounds
- Many parameters of ternary compounds depend on the parameters of the binary compounds that make them up, and vary linearly as the composition x changes.
- Vegard's Law
- Example: AlxGa{1-x}As lattice constant:
- a{AlGaAs} = xa{AlAs} + (1-x)a_{GaAs}
- Example: AlxGa{1-x}As bandgap:
- Eg^{AlGaAs} = xEg^{AlAs} + (1-x)E_g^{GaAs} + cx^2
- The compound becomes indirect.
- Virtual Crystal Approximation.
Semiconductor Technology Philosophy
- Make structures smaller and smaller.
- If needed materials don't exist, create them.
- Change electronic and optical properties using quantum confinement.
- At the nanometer scale, many new things happen.
- Requires searching for new fundamental physics, new applications, and new devices.
- Current advanced semiconductor technology relies on miniaturization and layer-by-layer growth.
Heterostructures
- Type I vs. Type II Heterostructures.
- In real junctions, charge carrier redistribution occurs, creating spatial charge regions and band edges distort.
AlGaAs/GaAs Heterostructure
- "Historical" AlGaAs/GaAs heterostructure.
- AlGaAs: depleted layer.
- GaAs: enriched layer.
- A 100-150 Å triangular potential well forms on the GaAs side, filled with two-dimensional electron gas.
Various Heterostructure Structures
- Quantum well (QW).
- Multiple quantum wells (MQW).
- Superlattice.
- Barrier.
- Double barrier with a quantum well.
Quantum Well (QW)
- Localized electron levels.
- Localized hole levels.
- New, larger forbidden energy gap.
Quantum Well Width and Energy
- From R. Dingle Festkoerperprobleme 15,21 (1975)
- Well Width
- 19.8 nm: 0.25 meV
- 12.2 nm: 0.4 meV
- 8.3 nm: 1.0 meV
- 5.1 nm: ~ 5 meV
- F. Pulizzi et al., Magnet Lab Nijmegen, 2001
Other Heterostructures
- Tunnel Barrier:
- Used in hot electron transistors and varactors.
- Resonant Tunneling Diode
Resonant Tunneling Diode (RTD)
- The current-voltage characteristic (IVACH) is similar to that of a tunnel diode (Esaki diode).
- The parasitic capacitance of the diode is greatly reduced, which is why it is suitable for higher frequencies than the tunnel diode.
- Frequency band above 700 GHz.
Superlattices
- When quantum wells are close to each other, their levels overlap and the degeneracy is removed, splitting the levels into minibands.
- When the wells are far from each other, they act independently—MQW (multiple quantum wells).
Quantum Cascade Lasers
- Traditional semiconductor laser: electron transitions occur between conduction and valence band levels.
- Quantum cascade laser: electrons fall from a higher quantum well subband to a lower one.
- The same injected electron can emit multiple photons.
- The wavelength depends on the layer thickness, not the material.
Quantum Wires and Quantum Dots
- Quantum wells are already used in many devices; why not try confining electrons in the remaining two directions.
Quantum Wires: Top-Down Technology
- Selective chemical etching to obtain 1D structures.
- Confinement is only achieved when the wires are about 10 nm thick.
- It turns out that it is too difficult to produce nanometer-sized 1D (wires) and 0D (dots) using chemical etching.
Self-Organization for Quantum Wires
- Growth on grooved substrates.
- Thicker GaAs (wire) occurs in the AlGaAs trench bottom due to differences in growth rates on various planes and surface diffusion of adsorbed atoms.
Strained Layers
- Crystalline layers grown on a crystalline substrate.
- Substrate and layer lattice constants match - lattice-matched layers.
- The lattice constant in the layer is larger than in the substrate - compressed layers.
- The lattice constant in the layer is smaller than in the substrate - tensile strained layers.
Influence of Strain on Band Structure
- Unstrained
- Compressed
- Strained
Quantum Dots
- When mismatch is too large, layers turn into dots.
InAs Quantum Dots
- Pyramidal QD energy gap dependence on dot size.
- Dots form spontaneously by rearranging on GaAs (311)B substrates upon sequential growth of AlGaAs and strained epitaxial InGaAs layers.
- Typical quantum dot size is about 30-150 nm.
Quantum Dots in Solar Cells
- Theoretically achievable efficiency - 65%.
QDIP - Quantum Dot Infrared Photodetectors
- Quantum Dot Infrared Photodetectors (QDIP) are other rapidly progressing QD optoelectronic devices.
- They have good applications in the near and mid-wavelength IR ranges (1 - 20 mm).
- Interband and intraband electron jumps are used for IR detection.
Laser Diode History
- Heterolasers (3D).
- Quantum wells (2D).
- Quantum dots (0D).
Advantages of Lasers with Quantum Dots
- First laser diodes with QDs were created in 1994.
- They allow overlapping a wide spectral range (from 0.87 to 1.9 μm).
- The lowest threshold current density (6 A/cm2, 2005) has been achieved in a continuous operation laser, as well as the widest tunable wavelength range (210 nm, 2000).
- Parameters better than other laser diodes: maximum power emitted in continuous mode from a single channel (12 W, 2002), relaxation oscillation frequency (7.5 GHz, 2005), optical radiation resistance, etc.
- Fujitsu. Laser, 25 Gb/s stream.
Disadvantages of Quantum Dot Lasers
- QD inhomogeneity: Inhomogeneous line broadening due to different QD sizes.
- Spatial discreteness of QD distribution: Carriers recombine in bursts, there is no diffusion to replenish QDs with new ones.
- Internal optical losses: Free carrier absorption in cladding.
Nanowires
- Planar 1D technology proved not very fruitful.
- Around 2008, nanowires emerged - 1D structures grown perpendicular to the substrate.
- They are obtained in two main ways: using templates and VLS (vapor-liquid-solid) method. The latter is based on self-formation.
Vapor-Liquid-Solid (VLS) Growth
- Discovered when growing Si "whiskers" (Wagner and Ellis, 1964)
- Au/Si eutectic droplets formed at the growth temperature.
- Droplets were supersaturated with Si, entering them from the surrounding vapors.
- Si in the droplet condensed.
- This process started at the liquid and crystal interface and continued in the same direction.
- Highly anisotropic growth
GaAs Nanowire Growth on (111)B planes
- [111] is energetically the most favorable direction r Pout Pin.
GaAs Nanowire Temperature Dependence
- GaAs nanowires: Important to choose the correct temperature.
Other III-V Nanowires
- GaAs, InAs, and InP nanowires (wurtzite - hexagonal).
- AlGaAs (zinc blende - cubic).
Core-Shell Nanowires
- Low T → axial growth (GaAs core, 450 ºC)
- High T → radial growth (AlGaAs shell, 650 ºC)
- GaAs core is passivated by the AlGaAs shell.
Axial Heterostructure Nanowires
- Axial segments are obtained by alternating molecule flows, e.g., turning off TMG and turning on TMI
- Growth is interrupted for 1 to 5 minutes to allow the Au droplet to empty from the previous Group III material.
- Thin segments → axial quantum wells.
- Diffusion of additional atoms from the substrate is possible, so it is difficult to predict the exact composition.
Catalyst-Free VLS Growth
- (111) Si plane.
- Covered with oxide in which windows of the desired diameter are opened using microimprint technology.
- Nanowire growth begins in these windows.
Nanowires as Part of Electronic and Photonic Devices
- Overcome fundamental limitation of lattice mismatch.
- Future integrated optoelectronic circuits made of nanowires.
Single Nanowire Laser
- NW self-catalyzed VLS, 610oC -80 nm.
- Thickened to 340 nm (480oC and higher Ga and As flows).
Nanowire Solar Cell
- p-i-n junction, 2015. 13,8%. Islamabad
Nanocrystals
- A nanocrystal is a particle of material with a diameter of less than 100 nm, composed of atoms with or without a crystalline lattice (amorphous).
- Properties differ greatly from the bulk crystal for diameters <5 nm.
- Properties are closer to those of bulk crystals for diameters >20-50 nm.
Nanocrystals in History
- Nanoparticles were known and used in Roman times.
- Colloidal metals were used to dye glassware and fabrics and as therapeutic aid for arthritis treatment.
Nanocrystals and Diameter
- When the diameter of nanocrystals is much larger than the exciton diameter ab, the absorption spectrum is similar to that of the bulk crystal.
- When the diameter of the nanocrystal approaches or becomes smaller than ab, the absorption maximum shifts to the blue.
Nanoparticles Synthesis Methods
- Chemical synthesis by reduction/precipitation reactions (A).
- Physical nanoparticle synthesis (B).
- Biological synthesis, using microorganisms or plant extracts as reducing agents (C).
Chemical Methods for Nanocrystal Synthesis
- Colloid chemistry methods: Precipitation of insoluble salts from a mixture of soluble salt solutions.
- Requires a stabilizer to ensure that the points do not get too large.
- Metastasis reaction: Molecule-precursors that initiate synthesis.
- For example, CuInS quantum dots are produced this way.
- Metal-organic method: A precursor is injected with a syringe into a heated flask with TOPO (triocytlphospine oxide), which is vigorously stirred by blowing an inert gas.
- CdSe quantum dots are produced this way.
Colloidal Materials
- Colloidal NCs are most often composed of elements of group II-VI
- CdSe, CdTe, CdS
- HgSe, HgTe, HgS
- PbSe, PbS
- ZnS
- The most common are CdSe and CdTe.
Nanocrystal Applications
- PbSe in glass matrix: Used for Ho laser mode synchronization.
- PbSe: Relaxation time decreases monotonically from 25 to 1 ps with decreasing radius from 2.9 to 1.4 nm. Dependence ~ 1/r^3.
Fast Optical Switches
- CdTe crystals in a glass matrix.
- Excitation probing experiment.
- Excited by one or two beams with a delay of 1 and 2.5 ps.
Biological Marking
- Advantages compared to dye molecules:
- Protein-coated QDs are very stable (>2 years).
- Narrow spectral width (FWHM ~ 25nm).
- Wide excitation spectrum.
- High quantum yield (40-50%).
- Compared to rhodamine 6G:
- 20x brighter.
- 100-200x more stable.
CdSe Quantum Dots
- By changing the size of nanocrystals, a wide spectrum of colors can be realized by excitation from a single source.
Quantum Dot Light Sources
- What makes quantum dot lasers superior to others?
- Higher differential gain - wider modulation frequency band.
- Electrons are trapped by QDs and cannot recombine in impurities and defects - lower threshold current.
Light-Emitting Diodes with QD
- By changing the size of the nanoparticles, it is possible to change the wavelength of the emitted wave.
- It is possible to produce white-light-emitting diodes.
Allotropes of Carbon
- a) diamond.
- b) graphite (composed of individual graphene layers).
- c) lonsdaleite.
- d) - f) fullerenes (C60, C540, C70).
- g) amorphous carbon.
- h) carbon nanotubes.
Graphene Band Structure
- Graphene is considered to be a semiconductor with a zero band gap because its conduction and valence bands meet at Dirac points.
Superconductivity in Graphene
- In 2018, a unique study found how to turn graphene into a superconductor under completely natural conditions.
- MIT, together with Harvard University scientists, found that two layers of graphene, stacked and rotated at a "magic angle" (1.1 degrees), allow electrons to pass with zero resistance.
Fullerenes
- Fullerenes are similar in structure to graphene, which consists of layers with hexagonal rings, but may also have pentagonal rings, which prevent the sheets from remaining flat.
- They can be hollow spheres, ellipsoids, tubes, or flat shapes.
- Spherical fullerenes are called "buckyballs", cylindrical ones - nanotubes.
- Graphene is an example of planar fullerene.
- In a C60 molecule, the distance from core to core is ~0.7 nm.
Unique Properties of Carbon Nanotubes
- Significantly stronger and more resistant to deformation compared to other materials.
- Extraordinary electrical properties.
- Stable up to 2800 oC (observed in vacuum).
- Charge-carrying capacity 1000 times better than copper wires.
- Has twice the thermal conductivity compared to diamond.
- By compressing or stretching, it is possible to change electrical properties by changing the quantum states of electrons.
- Depending on the structure (determined only by diameter and lattice angle), they are conductors or semiconductors.
Single-Walled vs. Multi-Walled Nanotubes
- Multi-Walled (MWNT):
- Inner diameter 1.5 - 15 nm; Outer diameter 2.5 – 50 nm ~ 50 layers. Many structural defects.
- Single-Walled (SWNT):
- Single atomic layer wall, diameter 1-1.5 nm. Unique mechanical properties.
Chirality
- The properties of nanotubes depend on the orientation of the hexagonal lattice with respect to the longitudinal axis of the nanotube.
- This property is called chirality.
- Chiral vector: \vec{Ch} = n\vec{a1} + m\vec{a_2} \equiv (n, m)
- Diameter: dt = \frac{Ch}{\pi}
- C_h = a\sqrt{n^2 + m^2 + nm}
- If 3n + m is a whole number - metal.
- If not a whole number - semiconductor.
- Electrical conductivity is measured by the four-probe method.
Semiconducting Nanotubes
- As a rule, they are p-type.
- Basic hypotheses explaining p-type:
- Influence of contact metal.
- Impurities that entered during synthesis.
- Influence of atmospheric oxygen.
- No one of these hypotheses has been confirmed or denied.
Empirical Guidelines
- The band gap is inversely proportional to the diameter of the tube.
- 1.6 nm diameter NV – 0.6 eV.
- The work function is F~4.5 eV, which is larger than most metals used for contacts (Au, Pt).
- Metallic NVs can transmit 10^9 A/cm^2
Nanotube Production Technologies
- Electric discharge with carbon electrodes
- Laser evaporation, laser sputtering
- Chemical vapor deposition
- High-pressure technologies
Carbon Arc Method
- Historically the first method.
- The electric arc evaporates the carbon anode with catalysts (Ni and Co).
- He: 500 Torr, Current: 100 A and 35 V.
- The camera is water-cooled.
- Nanotubes settle in the internal walls of the tube.
Laser Vaporization
- Target: 1 at.% Ni and Co well-mixed in graphite powder.
- Ar gas flow.
- In an oven, temperature of 1200oC.
- Nd:YAG pulsed laser operating at 60Hz.
- Cleanest, but very low yield (~0.4 g/h).
Chemical Vapor Deposition (CVD)
- Idea: to create a certain catalyst distribution on the substrate in advance and grow nanotubes on it by CVD.
- The most important step: put the catalyst exactly where it needs to be.
- Advantage: nanotubes are grown in the places where they need to be.
- Source: hydrocarbons.
CVD Process Steps
- Apply photoresist.
- Expose the photoresist.
- Apply catalyst.
- Etch the resist.
- Grow carbon nanotubes.
High-Pressure (HiPCO) Method
- Single-walled nanotubes in the gas phase (1200C, 10 atm).
- CO+CO C+ CO2 catalyst (25 mTorr).
- CO flows through the catalyst particles at high pressure and high temperature.
- It is possible to produce single-walled nanotubes in kilograms.
- No cleaning is needed because CO is used instead of hydrocarbons.
Nanotube Applications
- Transistors:
- Field-effect Transistor (FET).
- Single-electron Transistor (SET).
- Scanning microscope tips.
- Field emission displays.
FET Transistor
- Similar to MOSFET
- p-type channel
- By annealing or doping with K, an n-type can be obtained
- Using both p- and n-type nanotubes, a CMOS analog can be obtained.
IBM Transistor
- At 4 K temperature, it acts as a single-electron device.
Nanotube Weaving
- Using AFM needle.
- Van der Waals forces fixed the changed shape.
- Band structure parameters change at bends.
- By weaving as needed, it is possible to create energy barriers.
N-Type Nanotubes
- Usually p-type, but can be made n-type.
- Doping with K in vacuum (IBM).
- Vacuum annealing (IBM).
- Electrostatic doping (TU Delft).
- Returns to p-type in contact with air.
- Needs to be coated with SiO2 or PMMA (Poly(methyl methacrylate)).
CNTFET Inverter
- p-i-n FET can be made from the same tube.
Nanotube Needles
- Most were made of metals and often broke.
- Carbon nanotubes have a higher length-to-thickness ratio, are more resistant, and have a sharper tip.
- GaAs surface scanned by AFM with a nanotube needle and with a traditional SiN needle.
- Resolution is 100 nm.