The principle of sputtering
For a new technology platform, anew material, or “disruptive” device, the tension can be exacerbated between technology validation&acceptance and the need to break into existing supply chains or market arenas. This talk will explore some of the encounters experienced by start-ups within t he nanotech space. Although a traditional stage-gate developmnet framework can be effective as a means to manage the development process, certain modifications and adjustments can facilitate greater success. Early integration of strategy and tactics can help to make the journey easier.
SVC Oct 2011 newsletter
Sputtering is often called a billiard game on atomic scale. Energetic ions hit the surface of a solid (the target) and cause the ejection of atoms by momentum transfer in a collision cascade. The ions may be generated by an ion source (ion beam sputtering or IBS) or in a low pressure plasma.
The essential quantity describing the sputter process is the sputter yield Y, which is the number of emitted target atoms divided by the number of incident ions. The first model for the sputtering process was published by Sigmund in 1969. Y depends on:
The ion energy Eg
the ion mass m1
the mass of the target atom m2
the surface binding energy of the target material Uo
the angle of ion incidence Ǿ
the target temperature T
Y ~ 1/Uo according to Sigmund’s theory, so materials with a high surface binding (or sublimation) energy like C (Uo = 7.36 eV) are difficult to sputter.
The invention o the planar magnetron cathode by Chapin (patent issued in 1979) marked the beginning of a new era in vacuum deposition technology. The main advantages of magnetron sputtering are as follows:
- low plasma impedance and thus high discharge currents fro 1 A to 100 A (depending on cathode length) at typical voltages around 500 V
- deposition rates in the range from 1 nm/s to 10nm/s
- low thermal load to the substrate
- coating uniformity in the range of a few percent even for several meters long cathodes
- easy to scale up
- dense and well adherent coatings
- large variety of film materials available (nearly all metals and compounds)
- broadly tunable film properties
During the first ten years the industrial users of magnetron sputtering have suffered from several severe limitations. The non homogeneous ion current distribution across the target surface caused by the trapping of electrons in the magnetic ield results in poor target material utilization (often less than 30%), if fixed standard magnet array;;s are used. Reactive processes have been developed to deposit various compound films like oxides or nitrides. The compound is formed by sputtering the pure metal and adding the reactive gas (O2, N2) to the working gas argon. Oxides or nitrides are then formed on all inner surfaces of the deposition chamber including the non-eroding target areas. A hysteresis occurs, and complex process stabilization methods had to be developed. If the compound is insulating in addition substantial arcing may occur, and process parameters may drift due to the so called disappearing anode. During the past 30 years many of these problems have been solved through various milestones. The R+D efforts focused on the following goals:
- improvement of target materials utilization
- stabilization of the reactive process in the transition regime
- long term stable high rate deposition of dielectric films
- improved film properties
- higher deposition rates
Target material utilization
Around 1985 the development of circular magnetrons was driven by the need for fast and cost effective metallization of a new entertainment medium, the Compact Disc. In such a configuration a race track shaped target is located between the poles of the magnet system. A very flat magnetic field is realized by soft magnetic pole pieces positioned on top of permanent magnets. The broadening of the erosion zone resulted in a target utilization of around 60%. Self-sputtering was observed for Cu at high power densities.
Further approaches to improve the life time of the target were oscillating permanent magnet arrays or modified static magnet fields. Such magnetrons permitted a materials utilization of 50-60% . finally, in the late eighties the rotatable magnetron or C.MAG was introduced to the market. With a material utilization up to 90% and other advantages, this tool is now state of the art in many industrial coaters.
Pulsed sputter processes
patents teaching that it might be advantageous to pulse a plasma range back to the late sixties. However, substantial development work started around 1990, driven by the needs of large area coating for architectural glass and flat displays. The optimum solution turned out to be a double on arrangement powered by mid frequency in the range of 10 kHz to 100 kHz. At any time, one o the magnetrons is on negative potential and acts as sputter cathode, while the second one acts as an anode. The first systems based on this concept were introduced to industry by the Fraunhofer FEP as well as the companies Leybold Systems and Von Ardenne Anlagentechnik. Mid frequency sputtering using rectangular or sine wave shaped pulses nowadays is standard for many applications in the coating industry. In an optimum solution, two rotatable magnetrons are used.
During the R&D work on mid frequency sputtering, it turned out soon that related films exhibited higher hardness, higher density, higher refractive index and a smoother surface than those deposited by conventional DC plasmas. First measurements of ion current and energy during mid frequency sputtering were reported by Jager et al in 1996. Enhanced ion bombardment of the growing film modifies film structure and properties.
The most important applications of pulsed magnetron sputter processes are:
- antireflective, low emissivity and solar control coatings on glass and plastic web
- coatings for flat panel displays
- transparent conductive films for solar cells
- tribological coatings for components and tools
- coatings for sensors and precision optics
As an example the design of a wide band antireflective coating for archtectural glass
high power impulse magnetron sputtering HIPIMS came up about 10 years ago, after Kouznetsov et al reported on Cu sputter deposition using very short pulses at a target power density of up to 2800 W/cm2. In such HIPIMS processes 50-90% of the sputtered atoms are ionized, thus forming layers with increased density. Various results of the past years have shown the high potential of HIPIMS as sputter technique for the future. The following application fields are presently under evaluation:
- substrate pretreatment – etching
- substrate bias
- coating of 3D structures f-filling of trenches
- hard coatings for tribological applications
- coatings for precision optics
- transparent conductive coatings for thin film photovoltaics, heating and anti-fog
- insulating films
magnetron sputtering will be the key process in the future of films for precision optics. The main challenges are ultra precise control of film thickness and refractive index, reduction of defects and denser films. These tasks will be solved by enhanced methods for in situ film characterization, sputter up with rotatable magnetrons and HIPIMS processes. A notch filter for separation of extremely small wavelength bands contains 4410 individual films and has a total thickness of 122 μm.
Each new sputter technique comes up with applications that were not possible so far. HIPIMS was the basis for the development of a “coat and bend” process developed at Fraunhofer IST. By deposition of ITO on flat glass at room temperature and subsequent bending at 650 – 750 C a thoughened Low-E glass with an extremely hard coating is obtained. The coating exhibits an emissivity e< 20% and a light transmission Tvis> 80%. Such coatings are capable for use on the outer surface of a glazing. An ice-free windshield for car is now possible.
Towards higher deposition rates
There were many attempts to avoid the hysteresis in reactive sputter processes, and related work is still going on. In 1990 OCLI came up with the Meta Mode process where substrates are mounted on a rotating drum, they are alternating between a metallic sputtering and an oxygen rich zone. In each pass, a very thin metallic layer is deposited and subsequently oxidized. The targets thus can be operated at high deposition rates.
Belkind et al suggested a sputtering system consisting of two rotatable and one planar magnetron. For TiO2 deposition, the planar magnetron was operated in metallic mode while the two rotatables were operated in reactive mode. This patent publication together with work on atom assisted sputter yield amplification has inspi9red recent work resulting in th e so called serial co-sputtering or C2 technology. Planar and rotatable cathodes can be operated in separate gas atmospheres. The equipment offers useful modifications of reactive sputtering. In one emobdiment the main sources are equipped with TiO2 and the secondary sources are equipped with Bi targets. Using the sputter yield amplification effect, for TiO2:Bi films the deposition rate increases by a factor of 1.3 compared to the deposition of pure TiO2.
Among parameters influencing the sputter yield Y the angle of ion incidence and the target temperature T are of further interest. Y increases at oblique ion incidence up to a factor of 2.5 at 60 degrees, for very high ion energies in the range of 100-800 keV an increase by a factor of 10 and more has been found at grazing incidence. This amplificaiton effect is used in ion beam sputtering, but does not work for ions generated in a plasma discharge, since the ion incidence is always perpendicular tot he target surface.
The majority more than 80%, of the electric power delivered to a sputter target is transferred to heat, thus intensive colling is necessary to avoid melting. If targets are operated close to the melting point, the yield may increase drastically due to the formation of thermal spikes. Hot target sputtering processes have been investigated in the past and are also subject of present research at the Fraunhofer IST. They may be a promising approach for applications of PVD coatings on metal strips which usually are processed continuously at speeds in the range of 100 m/min.
The industrial implementation of magnetron sputtering was accompanied by various milestones. The 1980s may be described as the decade of reactive DC sputtering, the 1990s as the decade of pulsed sputtering and higher target utilization and the 2000s as decade of high ionization. Higher efficiency is the challenging task for the future, keeping in mind that the efficiency of sputtering is much less than 5%. the complexity of reactive magnetron sputter processes and their control has to be reduced, since the use of ceramic targets often is not advantageous. The importance of plasma simulation is steadily increasing. Finally, there is still a lot of work in exploring the potential of HIPIMS.
Source: SVC Fall newsletter, (Michael Vergohl et al.)