This site is an independent educational reference on thin-film deposition and metrology. It preserves technical material originally developed by the Philips Advanced Metrology Systems group in Natick, Massachusetts, whose work included non-contact optical measurement methods for semiconductor manufacturing.
The preserved archive focuses on the distinction between sputtering and evaporation as physical vapor deposition techniques. Sputtering uses energetic ions to eject atoms from a target, while evaporation relies on thermal heating to vaporize a source material. The archived documents also reference metrology approaches used to characterize deposited films, such as copper and low-k interconnect layers on 200mm and 300mm wafers.
No current company, product availability, or customer relationships are implied. The content is presented solely for historical and educational purposes. For current technical guidance, consult contemporary sources and qualified professionals.
Physical vapor deposition (PVD) is a workhorse family of techniques for depositing thin films in semiconductor, optics, and hard-coating applications. Among the many PVD variants, sputtering and thermal evaporation are the two most commonly contrasted. While both produce films atom-by-atom from a solid source, they differ fundamentally in the energy of the arriving species, the resulting film microstructure, and the practical constraints on composition and geometry. This article compares these two routes specifically for industrial semiconductor metrology and process integration, focusing on four key axes: arriving atom energy, adhesion, alloy retention, and step coverage.
The single most important difference between sputtering and thermal evaporation is the kinetic energy of the depositing species at the moment it reaches the substrate. In thermal evaporation, the source material is heated—either resistively, by electron beam, or by induction—until its vapor pressure is high enough to produce a useful deposition rate. The evaporated atoms or molecules leave the source with thermal energies, typically on the order of a few tenths of an electron volt (eV). They travel in a line-of-sight path from the source to the substrate, with little or no gas-phase scattering if the chamber pressure is kept low (often below roughly 10^-5 torr). Because their energy is low, they arrive at the substrate with insufficient momentum to disrupt the surface or to penetrate into the growing film’s subsurface layers.
In contrast, sputtering is a momentum-transfer process. A plasma is struck in an inert gas, usually argon, and positive argon ions are accelerated toward a target (the source material) by a negative bias. The ions strike the target with energies typically in the range of hundreds of eV to a few keV. This bombardment ejects atoms from the target via a collision cascade. The sputtered atoms leave the target with a distribution of energies that is much broader and higher than thermal evaporation—commonly in the range of 1 to 10 eV, with a high-energy tail extending to tens of eV. Furthermore, in magnetron sputtering, the sputtered atoms may pass through the plasma region and can be partially ionized or can collide with background gas atoms, but the average energy at the substrate remains substantially higher than in evaporation.
This energy difference has profound consequences. Higher-energy adatoms have greater surface mobility, meaning they can diffuse across the substrate surface before being buried by subsequent layers. This mobility promotes denser film growth, fewer voids, and a more columnar but tightly packed microstructure. Evaporated films, by contrast, often grow with a more open, columnar structure with significant void fraction, especially at low substrate temperatures. The energy also affects the intrinsic stress of the film: sputtered films often exhibit compressive stress due to atomic peening (the impact of energetic species), while evaporated films tend to have tensile stress due to the attractive forces between columns as they coalesce.
Adhesion is a direct consequence of the interface formed between the film and the substrate. For a film to adhere well, there must be strong chemical bonding across the interface, and ideally some degree of intermixing or interdiffusion. Thermal evaporation, with its low-energy atoms, produces a sharp interface. The arriving atoms have little ability to disrupt the native oxide or contaminant layer on the substrate surface, and they do not penetrate into the substrate lattice. As a result, evaporated films often adhere poorly to smooth, clean surfaces unless the substrate is heated to high temperatures (which promotes diffusion) or an intermediate adhesion layer (such as a thin chromium or titanium film) is deposited first.
Sputtering, on the other hand, offers a distinct advantage. The energetic sputtered atoms, and also the secondary electrons and ions from the plasma, can clean the substrate surface in situ. The initial stages of sputter deposition often involve a brief period of substrate bombardment that removes adsorbed contaminants and breaks up surface oxides. The arriving atoms then have enough energy to create a mixed interface layer, sometimes called a "pseudo-diffusion" zone, where the film and substrate atoms are intermixed over a few atomic layers. This mechanical and chemical interlocking dramatically improves adhesion for most material systems, particularly for metals on oxides, nitrides, and polymers. In practice, sputtering is the preferred route when adhesion is critical, such as for barrier layers, contact metallization, and reflective coatings on glass or plastic.
When depositing an alloy or a compound with multiple components, the composition of the film must match the source material. Thermal evaporation faces a fundamental limitation here: different elements have different vapor pressures at a given temperature. In a multi-component source, the more volatile element (the one with the higher vapor pressure) evaporates preferentially. This leads to a phenomenon called "fractionation" or "congruent evaporation failure." For example, if you try to evaporate a nickel-chromium alloy, the chromium, which has a higher vapor pressure, will leave the source faster initially, producing a chromium-rich film early in the run and a nickel-rich film later. This makes it nearly impossible to deposit a stoichiometric alloy film of uniform composition from a single evaporation source without complex feedback control or the use of multiple independently controlled sources.
Sputtering, by contrast, is a non-thermal process. The sputter yield (the number of atoms ejected per incident ion) depends on the momentum transfer and the surface binding energy of each element, but the differences between elements are far smaller than the differences in vapor pressure. For many alloys, the sputter yield ratios are close enough that the film composition closely matches the target composition, especially after a brief "conditioning" period during which the target surface reaches a steady-state composition. This is why sputtering is the standard method for depositing alloy films such as titanium-tungsten, nickel-chromium, and aluminum-copper. The target itself is manufactured to the desired composition, and the film inherits that composition with only minor deviations that can be corrected by adjusting the target composition slightly. For compound targets (e.g., oxides, nitrides), reactive sputtering—where a reactive gas like oxygen or nitrogen is added to the argon plasma—allows precise control of stoichiometry, which is very difficult with evaporation.
Step coverage refers to the ability of a deposition process to coat non-planar surfaces, such as the sidewalls and bottoms of trenches, vias, and contact holes. This is a critical metric in semiconductor manufacturing, where topography is ubiquitous. Thermal evaporation is inherently a line-of-sight process. Atoms travel in straight lines from the source to the substrate. On a flat surface, this is fine. But on a step, the top surface receives a full flux, while the sidewall receives a flux proportional to the cosine of the angle between the sidewall normal and the source direction. The bottom of a deep trench may receive almost no flux at all if the aspect ratio is high. The result is poor step coverage: thick films on the top surface, thin or absent films on sidewalls, and a "keyhole" void at the bottom of narrow features. This is a severe limitation for any application requiring conformal coating.
Sputtering offers better, but not perfect, step coverage. The higher energy of the sputtered atoms gives them some ability to "bounce" or re-sputter from the surface, and the presence of a background gas at moderate pressure (in the millitorr range) causes some scattering, which randomizes the arrival angles. This scattering allows atoms to reach sidewalls and trench bottoms that would be shadowed in evaporation. However, sputtering is still not fully conformal. The step coverage is typically better than evaporation but worse than chemical vapor deposition (CVD) or atomic layer deposition (ALD). For high-aspect-ratio features, sputtering often requires substrate bias (to attract ions and enhance re-sputtering) or collimation (to filter out off-angle atoms). Even then, the bottom coverage is often thicker than the sidewall coverage, and the top corners can be eroded by preferential sputtering. In practice, for sub-micron features, neither evaporation nor sputtering is used for conformal films; CVD or ALD is chosen. But for moderate topography, sputtering with a collimator or a long-throw configuration is often acceptable.
Given these tradeoffs, the choice between sputtering and thermal evaporation is driven by the application requirements. Thermal evaporation is still used where its simplicity and low cost are advantageous, and where the film requirements are modest. Examples include depositing soft metals like aluminum or gold for non-critical contacts, making free-standing films for lift-off patterning (where the line-of-sight deposition is actually beneficial because it avoids coating the sidewalls of the photoresist), and producing optical coatings where the film stress is not a primary concern. Evaporation also excels at depositing materials with very high melting points, such as refractory metals, when using an electron beam source, because the source can be locally heated without heating the entire crucible.
Sputtering is the default choice for most semiconductor front-end-of-line (FEOL) and back-end-of-line (BEOL) metallization, barrier layers (e.g., tantalum, tantalum nitride, titanium nitride), and silicide-forming metals. It is also preferred for any alloy film, for films requiring good adhesion to oxides or polymers, and for applications where film density and low void fraction are critical. The tradeoff is that sputtering equipment is more complex, requires higher power, and has a lower deposition rate for some materials compared to evaporation. The plasma also introduces the risk of substrate damage from ion bombardment, which is why sputtering is often avoided for very sensitive device layers.
In summary, the energy of arriving atoms is the master variable. Evaporation delivers low-energy, line-of-sight atoms that produce porous, poorly adhering films with poor step coverage but excellent composition control for pure elements and simple lift-off processes. Sputtering delivers higher-energy atoms that produce dense, well-adhering films with good alloy retention and moderate step coverage, at the cost of increased complexity and potential substrate damage. For a process engineer, the decision is rarely about which is "better" in the abstract, but rather which set of tradeoffs is acceptable for the specific film, substrate, and topography at hand.
This independent educational reference summarizes general technical concepts. Verify current standards, dimensions, and manufacturer specifications before making a procurement or engineering decision.