Roland A. - PVD for microelectronics (779636), страница 3
Текст из файла (страница 3)
It also should be noted that the views expressed inthis book are those of the authors themselves and do not necessarily represent those of their employers past or present.Finally, we would like to express our appreciation to the many individuals, suppliers, and institutions who contributed information or figures that appear in the book. Of special note, we wish to acknowledgexiiPREFACEChuck Wickersham (Tosoh, SMD) for providing useful background material for Chapter 11 and Daniel Lee (Novellus Systems) for the use ofmaterial contained in the PVD training manual he has developed.Ronald A.
PowellPalo Alto, CAStephen RossnagelYorktown Heights, NYOctober 1998Useful Conversion Factors and ConstantsLength1111Angstrom (,&,)= lO .8 cmn a n o m e t e r (nm) = 10 Amicron (pm) = 10,000 A = lO .4 cminch = 2.54 cmMass and Force11111atomic mass unit (AMU) = 1.66 x 10 .2` gmp o u n d (Ib) = 454 gmdyne = 1 g m / c m - s e c 2Newton ( N ) = 1 k g m / m - s e c 2N e w t o n - 10 s d y n e sEnergy1111electron volt (eV) = 1.6 x 10 '2 ergeV/particle = 23.06 kcal/molejoule (J) = 107 ergwatt ( W ) = 1 J/secPressure and Vacuum-Related111111111111standard a t m o s p h e r e - 1.013 x 108 d y n e s / c m 2standard a t m o s p h e r e = 14.7 Ib/in. z - 14.7 psistandard atmosp_here - 760 TorrTorr = 1.33 x 103 d y n e s / c m 2Pascal (Pa) = 1 N/m2= 7.5 mTorrTorr = 133.3 Pamicron = 10 .3 Torr - 1 milliTorr (mTorr)standard cubic centimeter (std.
cc) = 0.76 Torr-literstd. cc per minute (sccm) - 12.7 mTorr-I/secliter/sec = 2.12 ft3/min~ / m i n ( C F M ) - 0.47 liter/secLangmuir = 10 ~ Torr-secMiscellaneous1 Tesla = 104 g a u s smass of electron (me) = 9.11 x 10 .28 gmcharge of electron (e) = 1.6 x 10 '~ C o u l o m bA v o g a d r o ' s n u m b e r (No) = 6.02 x 1023 molecules per moleB o l t z m a n n ' s constant (k) = 1.38 x 10 16 erg/~S t e f a n - B o l t z m a n n constant (a) = 5.67 x 10 s W/m2-K"permittivity of v a c u u m ( 4 ) = 8.85 x 10 12 C2/N-m 2permeability of vacuum (~o) - 4~ x 10 .7 N/A 2xiiiThis Page Intentionally Left BlankChapter 1 IntroductionThe phenomenon originally described as "cathodic disintegration" by SirWilliam Robert Grove in 1852 was renamed "spluttering" by Sir JohnThompson in 19? 1.
Spluttering refers to the rapid ejection of small particles,as in "frying bacon will splutter fat." In a scientific paper two years later,Thompson dropped the "1" from spluttering in favor of a less common variation, and it's been "sputtering" ever since.1.1 The Role of PVD in MicroelectronicsThe physical process that we now call sputtering was first reported in 1852by Sir William Robert Grove [1.1], who described the effect as "cathodicdisintegration." Grove's apparatus (shown in Fig. 1.1) utilized a cathodemade of silver-coated copper, but his manually pumped vacuum was sufficiently poor ( ~ 10 Torr) that the world's first sputter-deposited film wasprobably not silver but silver oxide.
Moreover, it was possible to "disintegrate" the as-deposited film by reversing the electrical leads to the cathodeand anode, in effect creating both the first sputter deposition system andthe first sputter etching system. Subsequent scientific investigations byother workers in the late 19th and early 20th centuries led to an understanding of the basic physics of the sputtering process and resulted in a variety of industrial coating applications such as the deposition of metalfilms for mirrors (c.
1875) and the deposition of gold films on wax phonograph masters (c. 1930). By the time the first microelectronic device m thesolid state t r a n s i s t o r - was demonstrated publicly in 1948, sputter deposition was nearly 100 years old. Since that time thin film deposition bys p u t t e r i n g - i.e., by physical vapor deposition ( P V D ) - has become anestablished and essential part of integrated circuit (IC) fabrication technology and has given rise to a multibillion dollar, global PVD equipmentmarket.In the early years of semiconductor electronics, thin films of metalswere typically deposited by electron-beam (e-beam) or hot filament evaporation.
However, with the introduction of production-worthy DC magnetron sources in the 1970s, sputtering began to displace evaporation. DCmagnetrons were capable of depositing high quality aluminum alloys ofA1-Cu and A1-Cu-Si at deposition rates and cost per wafer comparable toevaporation. In addition, the improved step coverage and better control ofalloy composition provided by PVD made it attractive for the productionR.
POWELL AND S. M. ROSSNAGELFIG. 1.1 Sputtering was first observed in 1852 using this simple apparatus, making it the world'sfirst PVD system 11.1]. The vacuum in the glass bell jar was manually produced using a hand-operated pump, and the working gas was introduced from a gas-filled bladder through a stopcock.of advanced large scale integrated (LSI) devices such as the 16K DRAM.As a result, PVD quickly displaced e-beam evaporation for leading-edgeapplications.There are a number of reasons why PVD has been so successful for microelectronic applications. First of all, sputtering can be used to deposit allof the conducting films currently used in interconnect metallizationschemes, including low-melting-point metals such as A1 (Tmelt ~ 660~Also, sputtering of imand refractory metals such as Ti (Tmelt ~ 1670~portant multicomponent alloys such as A1-Si-Cu and Ti-W can be depositedfrom a single alloy sputter target with the deposited film retaining the stoichiometry of the target.
This was problematic with evaporation since thedeposition rate of the respective alloy constituents depended on their individual vapor pressures. The PVD deposition rate is also well matched to thethroughput needs of wafer fabrication ( ~ 40 wafers per hour), being about1 /xm/min for thick films (e.g., an 8000/~ A1 alloy interconnect) and about1000 ]k/min for thin films (e.g., 500 ~, Ti/TiN barrier/liner combination).INTRODUCTION3Critical film a t t r i b u t e s - such as purity and microstructure, which affectelectrical conductivity; surface roughness, which affects lithography; andfilm adhesion to o x i d e s - have all proven acceptable for microelectronicapplications. Because sputtering is done from an extended area target andnot from a point source as in evaporation, shadowing is minimized and resulting step coverage is generally good ( > 50%) over features with relatively low aspect ratio (AR < 0.5"1).Another reason for the success of PVD is that the global film uniformityfrom a properly designed magnetron source has kept pace with wafer sizeincreases in the IC industry.
Advanced PVD sources can deposit films with3 0 " ( " t h r e e - s i g m a " ) n o n u n i f o r m i t y of 3 - 5 % over 2 0 0 - m m - d i a m e t e r(8-inch) Si wafers in a production environment. Being primarily a physical deposition process whose underlying physics is well understood, sputtering lends itself to first-principles type of modeling or Monte Carlo simulation. Since PVD utilizes nontoxic targets and low pressures of inert gas( ~ 1-10 mTorr of Ar), it is also in sync with increasing environmentalconcerns about the use and disposal of hazardous materials. PVD is alsocompatible with the established trend toward automated single-wafer,vacuum-integrated processing. Finally, PVD has demonstrated an acceptably low cost-of-ownership (Co0) consistent with the economic demandsof production-line wafer fabrication.Figure 1.2 shows the market for semiconductors and semiconductorprocess equipment since 1965 on a semilog scale.
While equipment salesfigures for a given year depend on the source (e.g., Dataquest reported capital equipment sales of ~ $30B in 1995 while VLSI Research reported$20B), it is the trend that is most important. A significant PVD equipment market emerged in the late 1970s as sputtering began to replacee-beam deposition for mainstream IC metallization. During the periodfrom 1980 through 1997, the PVD equipment market grew from $100M to$1.5B ~ a compound annual growth rate of ~ 23%.
This is reflected bythe total sales of semiconductor processing equipment (lithography, doping, etching, deposition, annealing, etc.), which increased from about $2Bto $30B over the same period, and by sales of semiconductors, which increased from about $20B to $150B" approximately 90% of today's marketis for ICs as opposed to discrete devices. These semiconductors in turn areincorporated in personal computers and in other consumer electronic products valued at about $800B.
Finally, these electronic products sustainlarger global industries such as automobiles and aviation. In fact, it hasbeen estimated that all those industries ultimately dependent on semiconductors or electronics represent annual sales of approximately $1,500B, or$15T.R. POWELLAND S. M. ROSSNAGELFIG. 1.2 Global sales of semiconductor process equipment and ICs that are fabricated using thisequipment have continued to rise steadily, albeit with fluctuations, for more than 30 years.The economics of the supplier-user interaction has made PVD technology part of a "food chain" within which levels are separated by about anorder of magnitude. Using data for 1997, we see that the total semiconductor equipment market was about 15 times larger than that for PVDspecific equipment alone.
This equipment in turn was used to produce ICswith sales 5 times larger, leading to electronic products valued about 5times greater still. Ultimately it is the end user that drives all of these sales,so that when the market for such things as personal computers slumps, oris predicted to slump by economic forecasters, there is a ripple effect backdown the chain.Strictly speaking, the term physical vapor deposition (PVD) can also be usedto describe methods such as electron-beam evaporation, thermal filamentevaporation, or molecular beam epitaxy (MBE) in which heated crucibles areused to produce vapors that condense at the wafer surface. In this volume, unless stated otherwise, PVD refers only to deposition by sputtering.Since PVD emerged in the 1970s as a production-worthy technology formicroelectronic fabrication, its major application has continued to be met-INTRODUCTION5allization and interconnection m i.e., the deposition of electrically connected, multiple levels of metal films.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
