Roland A. - PVD for microelectronics (779636), страница 48
Текст из файла (страница 48)
It may also lead torather high levels of ionization for the metal that leaves the source, assputtered atoms that are not ionized and simply pass through the discharge are deposited and recycled on the opposite side of the cathode.8.2 Plasma AspectsThe general operating process of an I-PVD system is as follows. (1) Metalatoms are sputtered from a conventional magnetron cathode and ejected asneutrals. (2) A second plasma is formed from the background or workingIMa++++9H o l l___wo| r -q~2~///Z:~D|+III n e t r o nSubstrate--~*9+ 9+99999+X..--+99999.9*9*99MetalFIG. 8.6Cathode999*99*Plasma9999+§99B99999~H o l l o w c a t h o d e m a g n e t r o n I - P V D s y s t e m 18.101.99 9]9IONIZED MAGNETRON SPUTTER DEPOSETION: I-PVD25 1gas in the region between the cathode and the sample by the presence ofthe inductively coupled RF antenna.
(3) Some of the metal atoms thatenter this second plasma can be ionized under the right circumstances. (4)The metal ions (along with inert gas ions) can be accelerated from the second plasma to the sample or wafer surface by the presence of a bias on thesample.The plasma characteristics in this type of tool can be explored with theuse of a Langmuir Probe [X. 1 1 1, which is typically just a fine wire tip exposed to the plasma. By biasing the wire both positively and negatively upto a few tens of volts. it is possible to measure the plasma density as weilas the electron temperature at the probe's location.
However. this technique does not give any indication about the composition of the plasma,i.e., it cannot differentiate between metal and inert gas ions. So its use isprimarily to probe t h e electron dynamics and then to imply what might begoing on with the various ion species.For the purpose of examining the level of metal ionization, experimentshave been done where a small. gridded energy analyzer was located at thesample position 18.5, 8.61.
The cathode of the energy analyzer was configured to be a quartz crystal microhalance that could measure both the massof the depositing film and the net depositing ion current. Grids were configured above the energy analyzer cathode surface to suppress secondaryelectrons as well as to-admit or repel positive ions. This type of analyzer isappropriate for low-to-moderate plasma densities where the Dehye lengthof the plasma is still larger than the grid opening. Once the plasma becomes too dense (in this case at about 0.5 X 10" cm-'). the grids no longerfunction and the analyzer becomes shorted. An optical emission monitor(monochromator) was also configured to examine the emission of metalion and neutral lines at higher discharge powers.Several parameters can have a significant impact on the level of metalionization as well as the subsequent deposition of metal ions and neutrals.The most obvious is the density and electron temperature of the secondplasma.
Figure 8.7 shows the effect on relative metal ionization (of the depositing species) as a function of applied RF power to the RF coil electrode18.61. At zero RF applied power, the ionization level of the metal is quitelow. This is consistent with conventional magnetron sputter deposition. Asthe RF power is increased, the relative ionization increases and then saturates. There is a slight advantage in relative ionization with the use of Neinstead or Ar, presumably due to the higher electron temperature expectedwith Ne.The operating gas pressure of the I-PVD system is also quite critical dueto two primary effects. First, the plasma density and temperature are aR.
POWELL AND S. M. ROSSNAGEL2521.0~-0.8u_0.6O'1. . . . . .I . . . . ."A --"--- A ' ~ ~ A"~0.4cMa~na~on: 2 kWPressure: 36 mTorr(A) Neon0.20.0I'( ~ Argon0I.11002OO3OORF Induction Power (W)400FIG. 8. 7 Relative ionization of depositing metal species as a function of RF-inductive power for afixed magnetron sputtering rate [8.6].function of the gas density. Therefore, the level of metal ionization wouldbe expected to be a strong function of pressure. Second, the backgroundgas atoms tend to scatter the emitted, sputtered atoms due to collisions.This has been studied at great length for conventional sputter depositionprocesses.
At pressures on the order of 1 mTorr, the amount of gas scattering is low and most of the sputtered atoms travel in a line-of-sight, ballistic mode from the target to the film surface. At pressures of several tens ofmTorr, the path length for the sputtered atoms is much smaller than thecathode-to-film distance, and virtually all of the sputtered atoms lose theirinitial energy and direction due to gas atoms collisions.The effect of pressure on the plasma or electron temperature is shown inFig. 8.8 [8.11]. This figure shows a rapid fall in the electron temperatureas the gas pressure goes from 1 to 10 mTorr, and then a much slower fallas the pressure is increased. Conversely, the electron density (not shown)would show a roughly inverse dependence to the temperature, with a gradual rise in density as the pressure is increased.
This is a common dependence of a plasma powered at a constant power as a function of pressure:Low pressure correlates with fewer, hotter electrons, and higher pressureschange the plasma into a higher density but cooler plasma.IONIZED MAGNETRONSPUTTER DEPOSITION" I-PVD2535.00o4.0o%o"~,,."-..~3.0""-- A2.01." ---zx 13-cm cylinder3-~t'1.00.0Vv0v I]!v 1 J'i10''v9,!1 I]l]ITIq20il--l--V-[I'III301 V]Iv|vll40iiv v n l ] t50, l v l viv60Argon Pressure (mTorr)FIG. 8.8The backgroundpositing flux (Fig.ionization is quiteionization and thefound asElectrontemperature as a function of chamber pressure [8.11].gas pressure also alters the relative ionization of the de8.9) [8.6].
At low pressures (a few mTorr), the relativelow. This is due to the relatively small cross section forshort path length. The ionization mean free path can bevL =n~sve~>(8.1)where s is the average ionization cross section, v is the metal atom velocity, v is the electron velocity, and n is the electron density. For a plasmadensity of 1012/cm.3 ~ which is quite dense for a processing plasma ~ themean free path is on the order of 30 cm, which exceeds the throw distanceby 2 to 3 times.
However, if the background gas density is increased, twothings occur. First, the inert gas RF plasma becomes slightly denser, although with a lower electron temperature. Second, the sputtered atom canbe slowed or even stopped by collisions with gas atoms or other sputteredmetal atoms. This increases the net time that the metal ion spends in thesecond plasma, and as a result increases its chance of having an ionizingcollision.
In general, though, I-PVD will tend to be used at moderate system pressures in the tens of reTort. The choice of operating pressure willdepend on the required level of ionization, the system geometry, and theR. POWELL AND S. M. ROSSNAGEL1 .oII1I250 W RF Induction-0,8( e ) Argon (0)Neon/0°t0.-0mt0E20.4a0.20.00FiG. 8.90111020130Pressure (mTorr)I4050Relative ionization of the depositing species as a function c)f chamber pressure 18.61necessary deposition rate.
This final parameter falls rapidly as the operating pressure is increased.A third parameter that can strongly alter the level of metal ionization isthe number of metal atoms actually present in the discharge. The initialcase is with no metal atorns present. Here the plasma is composed only ofinert gas ions. Since the ionization potential of Ar is 15.7 eV and the electron temperature is typically I to 3 eV, only a small fraction of the Ar willbe ionized.
However, as metal atoms are sputtered into this inert gasplasma, their much lower ionization potential (5.9 eV for At. for example)will mean that a much larger number of electrons have sufficient energy toionize the metal atom and its ionization probability (compared to an Aratom) will be much higher. Thus, with sufficient RF power coupled intothe inert gas plasma.
the relative ionization of the metal atoms can exceed80%. even though the relative ionization of the Ar might be two orders ofmagnitude lower.This situation is complicated. though, as large numbers of metal atomsare sputtered into the plasma and the R F power to the plasma is held constant [8.6]. An example of this is shown in Fig. 8.10, where the relativeionization of the metal species (as deposited into the energy analyzer) ismeasured a s a function of RF power for three different magnetron powerlevels. As the magnetron power is increased, it corresponds roughly with a255IONIZED MAGNETRON SPUTTER DEPOSITION: I-PVD1.00.8AiF::,! /./'"~"Pressure- 36 mTorrO.2V /l(o) 3 kW Magnetron/0.0 ~ "0100200300400RF Induction Power (W)FIG.
8.10 Aluminum ion fraction as a function of RF induction power for three different magnetronpowers [8.61.linear increase in the number of metal atoms sputtered from the cathode.The figure shows a clear reduction in the relative ionization at all RF powers for the increased metal fluxes.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
