Roland A. - PVD for microelectronics (779636), страница 18
Текст из файла (страница 18)
Patent No. 3,616,450 (1971).3.11. D. W. Hoffman, "Design and capabilities of a novel cylindrical post magnetron sputteringsource," Thin Solid Films 96:217 (1982).3.12. W. D. Sproul, "New routes in the preparation of mechanically hard films," Science 273:889-892 (1996).3.13. J.M. Schneider, W. D. Sproul, M. S. Wong, and A. Matthews, "Scalable process for pulsed DCmagnetron sputtering of nonconducting oxides, submitted (1996) to Surface and Coatings Tech.3.14.
S. Schiller, U. Heisig, Chr. Korndorfer, G. Beister, J. Reschke, K. Steinfelder, and J. Strumpfel,"Reactive DC high rate sputtering as a production technology," Surface and Coatings Tech. 33:405-423 (1987) also 61:331 (1993).3.15. S. Schiller, V. Kirchhoff, K. Goedicke, and N. Schiller, "Pulsed plasma deposition creates newera for PVD," Solid State Tech., S12-S14 (Dec. 1996).3.16. B. Windows and N. Savvides, "Charged particle fluxes from planar magnetron sputteringsources," J. Vac. Sci. & Tech.
A4:196-202 (1986). N. Savvides and B. Windows, "Unbalancedmagnetron ion-assisted deposition and property modification of thin films," J. Vac. Sci & Tech.A4:504-508 (1986). B. Window and N. Savvides, "Unbalanced DC magnetrons as sources ofhigh fluxes," J. Vac. Sci. & Tech. A4:453-456 (1986).3.17. C. M. Horwotz, "Radio frequency sputtering, the significance of power input," J. Vac. Sci.
&Tech. AI: 1795-1800 (1983).3.18. S. M. Rossnagel and H. R. Kaufman, "Langmuir probe characterization of magnetron operation," J. Vac. Sci. & Tech. A4:1822-1825 (1986).3.19. T. E. Sheridan, M. J. Goeckner, and J. Goree, "Electron and ion transport in magnetron plasmas," J. Vac.
Sci. & Tech. AS: 1623-1626 (1990).3.20. M. Dickson, F. Qian, and J. Hopwood, "Quenching of electron temperature and electron density in ionized physical vapor deposition," J. Vac. Sci. & Tech. AIS: 340-344 (1997).3.21. S. M. Rossnagel and K. L. Saenger, "Optical emission in magnetrons: nonlinear aspects," J.Vac. Sci. & Tech. AT: 968-971 (1989).3.22. L.
McCaig and R. Sacks, "Current sensitivity of plasma voltage and emission line intensitieson a planar magnetron glow discharge device," Appl. Spectr. 4 6 : 1 8 - 2 4 (1992).3.23. A. Okamota and T. Serikawa, "Reactive sputtering characteristics of Si in an argon-nitrogenmixture," Thin Solid Films 137:143 (1986).3.24.
W. D. Sproul, P. J. Rudnick, C. A. Gogol, and R. A. Mueller, "Advances in partial pressure control applied to reactive sputtering," Surface and Coatings Tech. 39/40:499-506 (1989).Chapter 4 The Planar Magnetron4.1 The DC MagnetronThe introduction of commercial DC magnetrons in the 1970s provided theIC industry with a production-worthy alternative to both evaporation anddiode sputtering. Magnetron-based sputter tools deposited thin films atmuch higher rates than diodes and operated at lower pressure, where gasphase scattering and gas-phase impurities were minimal.
As discussed indetail in Chapter 3, a DC magnetron is basically a magnetically enhanceddiode in which the spatial relationship of electric (E) and magnetic (B)fields is engineered to confine secondary electrons produced by Ar + bombardment of the target. Restricting these electrons to remain close to thetarget surface increases their probability of ionizing the Ar working gas,which in turn results in a more intense plasma discharge that can be sustained at a lower pressure. Since the Ar ions are much heavier than electrons (mass ratio ~ 7 x 104), they are not affected to first order by the confining magnetic fields, and sputter bombardment proceeds much as in adiode.The underlying physics of electron confinement is based on the Lorentzforce F given by the vector cross product of the two fields (F = eE x B).Since E is perpendicular to the target surface, application of a B field tangential to the surface gives the electron a component of velocity parallel tothe target.
Keeping the electrons confined is then achieved by creating asituation where the locus of these tangential B fields form a closed path. Inthis way, an electron whose initial velocity would have launched it awayfrom the target surface will remain in "low orbit" above it, undergoing acycloidal hopping motion along the orbit path. Each electron can then ionize many neutral gas atoms before being collisionally scattered out of theplasma region. A similar approach, but without the closed-path confinement, has been used to increase the sensitivity of high vacuum pressuregauges based on electron impact ionization of residual gas molecules.
Inthis case, thermionically emitted electrons undergo helical motion in thepresence of an applied magnetic field, which increases their path length inthe vicinity of a biased collector grid.As described in Section 3.7, the basic physics of the crossed-field DCmagnetron has been implemented in many different hardware designs overthe past 20 years, with the cathode shape and the spatial arrangement ofmagnetic fields limited only by the creativity of the designer (the reader isreferred to refs. 4 . 1 - 4 . 7 for excellent reviews of magnetron sputter sourceR. POWELL AND S.
M. ROSSNAGELdevelopment). While a variety of magnetron designs are still in active usefor IC production, the planar magnetron cathode with circular target is byfar the most widely deployed in VLSI and ULSI device manufacturing.4.2 The Planar MagnetronPlanar magnetrons have been designed for a variety of target shapese.g., square, rectangular, wedge, and circular.
However, the circular shapeis widely used in IC production because it matches the circular geometryof the wafer and lends itself to rotating the magnet arrays that are employed to improve target utilization and film uniformity. Figure 4.1schematically presents a DC planar magnetron with a circular target(Fig. 9.10 in ref. 4.7). In the design shown, the surface-parallel B fieldneeded to produce closed electron paths is provided by a ring of permanentbar magnets arranged around a central magnet, the entire array being located behind a nonmagnetic target backing plate.
The magnets are connected to each other by afield return plate made of a highly permeable material such as iron. This iron rail completes the magnetic circuit andprevents magnetic flux from spreading into regions other than the desiredone in front of the target surface. The field lines pass through the target andarch over the target surface, which for thick targets at start of life can be1 inch or more away from the backing plate. Therefore, to obtain the desired field strength at the target surface, strong permanent magnets havingan energy product on the order of 30 megagauss-oersted (MGO) are oftenused. The value of the energy product is determined from the point on anexperimental plot of B versus H, where the product of B X H is maximum - - analogous to determining the maximum power point on a loadlineplot of current versus voltage for an electrical device.
The square of magnetic field strength has units of energy density that would normally begiven as ergs/cm 3 in the CGS system or joules/m 3 in the MKS system.However, since the convention is to give B in gauss and H in oersted, thevalue of the energy product is given in mixed units of gauss-oersted.A variety of high-energy-product permanent magnets have been used inplaner magnetrons, including ferrites and rare earth alloys such as samariumcobalt (Sm-Co) and neodymium-boron-iron (Nd-B-Fe). Using dense arraysof compact bar magnets with high coercive force (e.g., ~ 30 MGO andvolume < 30 cm3), one can produce tangential fields at the target surfaceon the order of 500 G, or 0.05 Tesla (T) in SI units. A modern planer magnetron design might incorporate perhaps 30-50 such magnets.
Electromagnets have also been used in some modern designs and offer the abilityPlasma RingElectron Trajectory~Erosion,,," ',).'.-9.'.,.9-'.,, 4i.p ~;.,,,'i,.:::/'JIll#,-',, ~'_ ,";,,' ~ 1," ,,, z,, ,~f~9|/.i",-'.-.-''i'%-..Trenchet Surface" - .,. -".-'/~A~_~~o.,..t----ler~Wall~ ~ ~ - - - - - , ' ".',..-~.O/"-Magntetr~~I . . ..,.'"//A ~ / ' / ~ / / / / / / / / / / / / J ~///IM agnetro nCross Section7Insulating RingO-Ring Seal,XCu Backing PlateBar MagnetstDeionizedWatert, Fe Field Return Plate-500 VFIG. 4.1 Schematic of a planar magnetron with circular target and stationary magnet array. (Reproduced with permission of the McGraw-HillCompanies from D.
L. Smith, Thin Film Deposition: Principles & Practice, McGraw-Hill, NY, 1995.)90R. POWELL AND S. M. ROSSNAGELto electrically vary field strength (e.g., to jump-start the plasma at a lowpressure by temporarily using a high field) or to optimize field distributionfor improved film uniformity. However, permanent magnets offer a number of advantages over electromagnets. For example, there is no DC powerrequired; no additional heat to dissipate; no danger of magnet insulationfailure or field interruption from a power failure; and far less weight, volume, and complexity for the equivalent magnetic field. Figure 4.1 showshow the ring of magnets confines the electrons into an annular closed path,which in turn produces an annulus of intensified sputtering plasma and acorresponding "race-track" sputter erosion groove in the target surface.The narrowness of the groove results from radial compression of theplasma by a magnetic-mirror effect, which is common to many magneticplasma systems [4.5].
The boundaries of the track correspond to the location of the pole pieces of the magnetic array where the E and B fields arenearly parallel (E • B ~ 0). Since the crossed-field confinement effect isminimal here, electrons can escape from the cathode without causing anylocalized enhancement of plasma density.The planar magnetron shown in Fig. 4.1 with fixed permanent magnetelements suffers from poor utilization of the target material, which is preferentially etched in a narrow annular region.
If the target is too close to thewafer, the deposited thin film thickness profile will mirror the race trackand be highly nonuniform. By using a source-to-substrate distance that isvery large compared to the mean free path of the sputtered neutrals,isotropic gas-phase scattering can be used to smooth out this nonuniformity; however, the same scattering will reduce the deposition rate by directing atoms away from the wafer and toward the chamber walls. Also, asshown in Fig. 4.2, the angular distribution of atoms leaving the target willchange as the groove deepens since the local surface normal is no longerin the same direction as when the target was planar. This will affect boththe step coverage and thickness uniformity of thin film deposition as thetarget ages.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
