Unit 4 - Practice Quiz

PVD is a family of processes that use physical means, such as heating (evaporation) or ion bombardment (sputtering), to turn a source material into a vapor, which then condenses on a substrate to form a thin film.

In CVD, 'precursors' are the volatile chemical compounds that are introduced into the reaction chamber. They react or decompose on the substrate surface to create the desired solid thin film.

The 'target' in sputtering is a solid block or plate made of the material you want to deposit. It is bombarded by energetic ions, which physically eject or 'sputter' atoms from its surface.

Sputtering works by creating a plasma (usually from an inert gas like Argon). Ions from this plasma are accelerated by an electric field and collide with the target, knocking atoms loose for deposition.

MBE requires an ultra-high vacuum to ensure the purity of the deposited film and to allow the beams of evaporated atoms to travel in a straight line from the source to the substrate without colliding with other gas molecules.

The plasma provides the energy needed to break down the precursor gas molecules into reactive species. This means the substrate does not need to be heated to high temperatures, making it suitable for depositing films on temperature-sensitive devices.

ALD works by exposing the substrate to different chemical precursors one at a time. Each step is a self-limiting reaction, meaning it stops once all available surface sites have reacted. This cycle-based approach allows for atomic-level thickness control.

Spin-coating is a common solution-based method where a liquid precursor solution is applied to the center of a substrate, which is then spun at high speed to spread the liquid and evaporate the solvent, leaving a thin film.

Argon is a noble gas, so it is chemically inert and will not react with the target or the growing film. It is also relatively heavy, making it effective at dislodging atoms from the target upon impact.

Epitaxy refers to the growth of a crystalline film on top of a crystalline substrate, where the film's crystal lattice aligns with the substrate's lattice. This results in a high-quality single-crystal film.

Conformality is the ability to deposit a film of uniform thickness on all surfaces, even inside deep trenches or complex shapes. The self-limiting nature of ALD provides nearly perfect conformality, which is crucial for modern 3D device structures.

The energy from the plasma breaks chemical bonds in the stable precursor gases, creating highly reactive species, including ions and neutral fragments with unpaired electrons, known as radicals. These radicals readily react on the substrate surface to form the film.

Thermal evaporation is a PVD technique where a source material is heated in a high vacuum chamber, causing it to evaporate. The vaporized atoms then travel and condense on a cooler substrate.

Most CVD processes require the substrate to be heated. This thermal energy provides the activation energy needed for the precursor gases to react or decompose on the surface to form the solid film.

In DC sputtering, a direct current voltage is applied to the target. This requires the target to be electrically conductive to maintain the flow of current. For insulating targets, a charge would build up and stop the process.

Since each ALD cycle deposits a precise, repeatable amount of material (typically a fraction of a monolayer), the final film thickness is determined simply by the number of cycles performed. This allows for unparalleled thickness control.

Silicon nitride is a vital dielectric and passivation layer in semiconductor manufacturing. PECVD is the preferred method for depositing it because it can be done at low temperatures, preventing damage to underlying device structures like aluminum metallization.

The source material is heated in an effusion cell until it evaporates. In the ultra-high vacuum, the atoms or molecules travel in a straight line, like a beam, directly to the substrate without any collisions.

Films made from solutions can contain residual solvents or other impurities from the liquid phase. The process is also more exposed to ambient contamination compared to the highly controlled environment of a vacuum chamber, often resulting in lower purity films.

The fundamental difference lies in the source. PVD starts with a solid (or sometimes liquid) source that is physically vaporized. CVD starts with reactive gases (precursors) that are introduced into the chamber to create a solid film through a chemical reaction.

In DC sputtering, a negative voltage is applied to the target to attract positive ions. If the target is an insulator, positive charge from the ions accumulates on its surface, creating a positive potential that repels further ions and effectively stops the sputtering process. This is why RF sputtering is used for insulating targets.

Aluminum has a low melting point (~660°C) and can be damaged at the high temperatures required for thermal CVD (700-900°C). PECVD uses plasma energy to dissociate precursor gases, allowing for deposition at much lower temperatures (200-400°C), making it compatible with temperature-sensitive underlying layers like aluminum.

FinFETs have complex, three-dimensional gate structures. ALD's layer-by-layer growth mechanism, based on self-limiting surface reactions, ensures the deposited film is perfectly uniform and conformal over these structures, which is critical for device performance. It also offers atomic-level thickness control, which is essential for ultra-thin gate dielectrics.

RHEED intensity oscillations are a classic indicator of 2D, layer-by-layer growth in MBE. The intensity is maximum when a layer is complete and smooth, and minimum when the layer is half-complete and at its roughest. Each oscillation period corresponds to the deposition of a single atomic monolayer.

While a certain pressure is needed to sustain plasma, excessively high pressure increases the density of gas atoms between the target and substrate. Sputtered atoms are more likely to collide with these gas atoms, losing energy and being scattered away from the substrate. This reduces the deposition rate and can also affect film density.

An insufficient purge step fails to remove all gas-phase TMA before the H₂O pulse is introduced. This allows TMA and H₂O to react in the gas phase (a CVD reaction) in addition to the desired surface reaction. This CVD component disrupts the self-limiting nature of ALD, leading to higher, uncontrolled growth rates and poor film quality.

Spin-coating relies on centrifugal force to spread a liquid precursor. This force, combined with surface tension effects, makes it very difficult for the liquid to uniformly fill and coat the vertical sidewalls of deep, narrow features. The result is a non-conformal film that is much thicker on the top surface than inside the trench.

The primary role of plasma in PECVD is to use electrical energy to break down stable precursor molecules into highly reactive ions and free radicals. These reactive species can then form a solid film on the substrate surface at much lower temperatures than would be required for the equivalent thermal reaction, effectively lowering the overall activation energy of the deposition process.

MBE relies on a direct, line-of-sight path for atoms or molecules from the source (effusion cell) to the substrate. A UHV environment ensures a very long mean free path, meaning the depositing species travels without colliding with background gas molecules. This preserves their chemical purity and directional nature, which is essential for high-quality epitaxial growth.

In an ideal, self-limiting ALD reaction, the growth per cycle is independent of precursor dose once saturation is reached. The GPC is a constant determined by the density of reactive sites on the surface and the physical size (steric hindrance) of the precursor molecules, which dictates how many can attach to the surface in a single monolayer.

Different elements in an alloy have different sputtering yields. Sputtering from a single alloy target can lead to a film composition that is different from the target's composition. Co-sputtering from separate, pure targets (e.g., one Iron, one Cobalt) allows the deposition rate of each material to be controlled independently (by adjusting power), enabling precise, real-time control over the final film's stoichiometry.

After the sol-gel is applied (e.g., by spin-coating), it forms a porous, amorphous film containing residual solvents and organic groups. The high-temperature annealing step provides the thermal energy needed to drive off these residuals, decompose organic precursors, and promote the densification and crystallization of the material into the desired final oxide phase (e.g., anatase or rutile TiO₂).

In mixed-frequency PECVD, low-frequency (LF) power increases ion bombardment energy, leading to film densification and higher compressive stress. High-frequency (HF) power primarily generates reactive radicals with less ion bombardment. By increasing the HF/LF power ratio, ion bombardment is reduced, which typically shifts the film stress from compressive towards tensile.

A magnetron uses strong magnets behind the sputtering target to create a magnetic field. This field traps electrons in a spiral path close to the target surface. The trapped electrons have a much longer path length, greatly increasing their probability of colliding with and ionizing neutral gas atoms (like Argon). This creates a much denser plasma near the target, leading to a higher sputtering rate at lower pressures.

Due to its self-limiting, layer-by-layer growth mechanism, ALD is renowned for producing highly conformal, dense, and pinhole-free films, even at very low thicknesses. PECVD, being a more continuous process driven by gas-phase reactions and ion bombardment, struggles to achieve the same level of perfection in conformality and pinhole density, especially on complex topographies.

Epitaxial (single-crystal) growth requires that arriving atoms (adsorbed atoms or 'adatoms') have enough energy to migrate across the surface and find their correct position in the crystal lattice. This surface mobility is provided by thermal energy from heating the substrate. If the temperature is too low, atoms stick where they land, resulting in an amorphous or polycrystalline film.

While cost-effective, sol-gel processes rely on chemical precursors that can leave behind impurities like carbon from organic groups or hydroxyl (-OH) groups from water and solvents. Even after annealing, these residuals and a typically lower film density can disrupt the crystal lattice and scatter charge carriers, resulting in lower electrical conductivity compared to the purer, denser films produced by sputtering.

Amorphous silicon has a disordered atomic structure, which results in many unsatisfied or 'dangling' bonds. These dangling bonds act as traps for charge carriers (electrons and holes), severely degrading the material's electronic performance. Hydrogen atoms incorporated during PECVD readily bond with these silicon atoms, 'passivating' the dangling bonds and dramatically improving the material's quality for applications like solar cells and thin-film transistors.

An ALD process consists of discrete, sequential steps: precursor A pulse, purge, precursor B pulse, purge. Each of these steps takes time. To deposit a film of, for example, 10 nm with a growth-per-cycle of 0.1 nm, this four-step sequence must be repeated 100 times. This makes the overall process much slower than CVD, where precursors are introduced simultaneously for continuous growth.

During the negative part of the RF cycle, the insulating target is sputtered by positive ions, causing a buildup of positive charge. During the brief, less intense positive part of the cycle, the target attracts a flood of highly mobile electrons from the plasma. These electrons neutralize the positive charge accumulated during the sputtering phase, effectively 'resetting' the target surface and allowing the sputtering process to continue in the next negative cycle.

The 'disappearing anode' effect occurs when conductive surfaces that act as the anode (like the chamber walls) become coated with an insulating material. In this case, sputtered aluminum reacts with residual oxygen or water vapor to form insulating AlO. This coating prevents the return current of electrons from the plasma from reaching the anode, effectively reducing the anode area. To maintain the constant current, the system must increase the discharge voltage, which often leads to an unstable plasma and a lower sputtering rate.

The transition region in reactive sputtering is inherently unstable when controlled by gas flow, as a small change in flow can cause a large jump to either a metallic or poisoned target state. The intensity of a metallic emission line (e.g., from a Ti atom) is a direct indicator of the metallic fraction of the target surface. By using this signal in a fast feedback loop to control the reactive gas inlet, the process can be stabilized within this narrow, high-rate transition region, which is often desirable for producing stoichiometric films.

In a balanced magnetron, the magnetic field lines are closed loops near the target, effectively trapping the plasma. In an unbalanced magnetron, the stronger outer pole allows some field lines to extend towards the substrate. This creates a path for electrons to escape, which in turn drag ions along with them to maintain charge neutrality. The result is an increased flux of energetic ions bombarding the growing film, which can be used to densify the film, modify its stress, and improve its properties, a process known as ion-assisted deposition.

RHEED oscillations are a hallmark of layer-by-layer (Frank-van der Merwe) growth. Each oscillation corresponds to the completion of a single monolayer. The GaAs lattice is a zincblende structure. A single monolayer in the <001> direction has a thickness of a/2, where 'a' is the lattice constant. Given 15 oscillations (15 monolayers) for a 42.45 Å film, the thickness of one monolayer is 42.45 Å / 15 = 2.83 Å. Therefore, the lattice constant 'a' is 2 * 2.83 Å = 5.66 Å, which is very close to the known lattice constant of GaAs (~5.653 Å).

In GaN MBE, a slightly Ga-rich condition is often intentionally used. The excess Ga forms a thin liquid adlayer on the surface. This layer enhances the surface mobility of Ga and N adatoms, preventing the formation of 3D islands and promoting a smooth, layer-by-layer growth mode. This 'surfactant' effect is crucial for achieving high-quality epitaxial films. Extreme Ga-rich conditions, however, can lead to the formation of undesirable Ga droplets.

Increasing the RF frequency in a CCP reactor reduces the ion transit time across the sheath relative to the RF period. Ions can no longer follow the instantaneous electric field, resulting in a lower time-averaged sheath voltage (). Since ion energy is proportional to sheath voltage, ion bombardment energy decreases. Concurrently, higher frequency leads to more efficient power absorption by electrons and reduced electron loss to the walls, resulting in a higher plasma density (). The combined effect is a higher deposition rate (due to higher and thus more radical generation) and reduced film stress and damage (due to lower ion energy).

Low RF power minimizes the multi-step dissociation of SiH, favoring the creation of the primary radical SiH. High power tends to create more highly dissociated and reactive species like SiH, SiH, and Si. A high H dilution serves two purposes: first, it reduces the partial pressure of SiH, further limiting unwanted gas-phase polymerization. Second, atomic H produced from the H plasma can scavenge weakly bonded Si atoms from the surface and selectively etch away strained Si-Si bonds, promoting a more ordered network with fewer defects. This combination favors the dominance of SiH as the growth precursor, leading to higher quality films.

In a CCP reactor, the RF power used to generate the plasma also determines the DC self-bias and thus the ion energy. Plasma density and ion energy are coupled. In an ICP reactor, the plasma is primarily sustained by the inductive coil (source power), which controls the plasma density. A separate RF bias can be applied to the substrate holder to independently control the sheath potential and thus the energy of ions bombarding the substrate. This decoupling allows for the creation of a high-density plasma (for high deposition rates) while maintaining low ion energy (to minimize substrate damage).

The ALD window is the temperature range where the growth is self-limiting. Below the window, precursors may condense, leading to thick, uncontrolled growth. Above the window, metalorganic precursors like TMA can thermally decompose (e.g., TMA -> Al + 3CH). This decomposition is not a self-limiting surface reaction. It deposits aluminum continuously as long as the precursor is present, breaking the layer-by-layer nature of ALD. This results in a CVD-like growth mode, a GPC greater than expected, poor thickness control, and potentially poor conformality.

In high-aspect-ratio (HAR) structures, precursor transport to the bottom is limited by Knudsen diffusion, which is slow. The conformality depends on achieving a sufficient 'exposure' (pressure × time) throughout the feature. If the pulse time is too short, the precursor will fully react at the top of the feature before it has time to diffuse to the bottom, resulting in a film that is thick at the top and thin or non-existent at the bottom. Therefore, pulse and purge times must be made much longer than for a planar surface to allow for complete saturation and purging of the entire HAR structure.

Low-temperature deposition techniques like PEALD cannot provide enough thermal energy for surface species to rearrange into a dense, stable network and for byproduct desorption to be complete. PEALD of SiN often uses precursors like bis(tert-butylamino)silane and an N/H plasma. Incomplete reactions and the incorporation of H from both the precursor and plasma result in a significant amount of Si-H and N-H bonds in the film. This high hydrogen content, combined with the low deposition temperature, leads to a less dense, amorphous film compared to the dense, low-hydrogen, stoichiometric films produced by high-temperature LPCVD.

Stoichiometry in ALD of ternary compounds is controlled by the number of cycles for each component, not directly by their individual GPCs. The goal for SrTiO is to deposit one layer of SrO for every one layer of TiO to build the perovskite crystal structure. The ALD process is self-limiting, so within one SrO cycle, a 'monolayer' of SrO precursor deposits, and similarly for TiO. To achieve the 1:1 atomic ratio of Sr to Ti required for SrTiO, a 1:1 cycle ratio (e.g., [1x SrO cycle + 1x TiO cycle] repeated) is needed. The different GPCs simply mean that the 'monolayers' have different thicknesses.

The pH acts as a catalyst for the hydrolysis and condensation reactions. Under acidic conditions (H catalyzed), the hydrolysis rate is fast, but the condensation rate is slow and favors reactions at the end of chains, leading to linear or weakly-branched polymers. This forms a fine, microporous gel network. Under basic conditions (OH catalyzed), the condensation rate is much faster than hydrolysis, especially between more highly condensed species. This favors the growth of discrete, highly-crosslinked, dense colloidal particles (like spheres), which then aggregate to form a gel with larger pores (mesoporous) between the particles.

The final thickness is directly proportional to the initial thickness , which is proportional to . Therefore, . We can write this as a ratio: . We are given nm at rpm, and we want nm. Plugging in the values: . So, . Squaring both sides gives . Solving for yields rpm.

The coffee-ring effect is primarily caused by an outward capillary flow. The droplet edge is 'pinned', and evaporation is fastest at the edge, so liquid from the center flows outward to replenish the edge, carrying the suspended particles with it. While several strategies exist, a highly effective one is to induce an inward Marangoni flow that counteracts the outward capillary flow. This can be achieved by using a solvent mixture (e.g., water and ethylene glycol). The more volatile solvent (water) evaporates faster from the edge, increasing the concentration of the less volatile, higher-surface-tension solvent (ethylene glycol) there. This creates a surface tension gradient that pulls liquid (and particles) from the edge back toward the center, promoting a uniform deposit.

Preferential sputtering initially removes the lighter/more weakly-bound element (oxygen) at a higher rate. This enriches the target's surface layer with the other element (silicon). This altered surface layer continues to evolve until a steady state is reached. In this steady state, the sputtering yields of the components, multiplied by their concentrations in the altered surface layer, result in a sputtered flux that has the same stoichiometry as the bulk target. Therefore, after a brief initial period where the film may be non-stoichiometric, the process self-corrects, and the deposited film will have the same stoichiometry as the bulk SiO target.

In a batch LPCVD furnace, wafers downstream will see a slightly lower concentration of the precursor (SiH) due to depletion from reacting with upstream wafers. In the mass-transport-limited regime, the growth rate is directly proportional to this precursor concentration. Therefore, downstream wafers will have a thinner film, leading to poor uniformity. In contrast, in the surface-reaction-limited regime, the growth rate is controlled by the slow surface reaction kinetics, which depend strongly on temperature but are only weakly dependent on precursor concentration (as long as some precursor is available). Since temperature is well-controlled in the furnace, the growth rate is nearly identical on all wafers, even with slight precursor depletion, resulting in excellent wafer-to-wafer uniformity.

While shutter speed and purity are important, the ultimate physical limit to interface abruptness in many material systems (like InGaAs/GaAs) is surface segregation. This is a thermodynamic effect where one atomic species (e.g., In) has a lower surface free energy than the other (e.g., Ga). As the new layer (GaAs) is grown on top, it is energetically favorable for In atoms to swap positions with overlying Ga atoms and remain on the surface. This 'riding' of the growth front leads to a gradual, rather than abrupt, change in composition across the interface, smearing it over several atomic layers, even with instantaneous shutter action.

The reaction between SiH and O is extremely rapid at deposition temperatures. If they are premixed, homogeneous (gas-phase) nucleation will occur almost instantly, creating a 'snow' of SiO particles that contaminates the chamber and the wafers. To prevent this, APCVD reactors are designed to keep the reactant gases separate until the very last moment. They are injected through separate manifolds or a 'showerhead' that is designed so that the gases only mix in a thin boundary layer just above the hot substrate. This ensures that the heterogeneous reaction on the wafer surface is the dominant process, minimizing particle formation.

Steric hindrance refers to the spatial arrangement of atoms and ligands on a molecule that can impede chemical reactions. When a precursor molecule like TMA adsorbs, its methyl groups take up a certain amount of space. A bulkier precursor has much larger ligands, which occupy a significantly larger 'footprint' on the surface. These large ligands prevent other precursor molecules from adsorbing on nearby available reaction sites. This leads to a lower number of precursor molecules being able to chemisorb per unit area during the saturation step, which directly translates to a lower mass gain and thus a smaller growth-per-cycle (GPC).