Hoffman D.M., Singh B., Thomas J.H. (Eds). Handbook of Vacuum Science and Technology

Подождите немного. Документ загружается.

724

Chapter

5.5:

Plasma-Enhanced Chemical Vapor Deposition

increased. If all dimensions, power density, and flow rates of a small plasma reac-

tor are multiplied by the same number, the outcome of the deposition process will

be different. The low-risk path to scaleup is therefore "modular" scaleup, where

one plasma environment (the plasma module) is fully developed to the right size.

The required process throughput is then achieved by duplicating the module,

placing them in series or parallel. Examples are drum cross-flow reactors [60] and

in-line coaters [61] shown in Figure 7.

Fig.

Examples

modular production reactors.

-T^n.

(a)

(a) Drum reactor with gas inlet and exhaust plenums (parallel modules) and (b) an in-line roll

coater (serial modules).

5.5.3 Process Scaleup 725

Modular scaleup minimizes the risk of meeting throughput demands, but the

basic difficulties of scaling PECVD to the modular level remain. We will briefly

discuss how the primary process variables of power, flow rate, pressure, and tem-

perature change with size.

The electrical power scales nearly linearly with electrode size for frequencies

<1 MHz or with the plasma volume for high frequencies. However, long be-

fore the free-space wavelength approaches the dimensions of the substrate, high-

frequency processes pose a difficulty in scaleup. At 13.56 MHz, it is difficult to

achieve uniformity when electrode sizes become >1 m. This problem can be

solved by decreasing the frequency <1 MHz or using multiple, smaller-size

synchronized electrodes. Pressure and substrate temperature usually remain un-

changed in scaleup, and the greatest challenge in scaling a PECVD process is in

properly engineering the gas flow pattern. Gas inlet systems have to be designed

correctly to achieve the required uniformity, maximizing both deposition rates

and gas utilization while minimizing particulate formation. Depending on the ap-

plication, hollow electrode effects can be engineered into a distributed gas system

to advantage [62] but can also be avoided by proper design. As a first rule, flow

rates should be much larger than the desorption and leak rates of a system. The

combined desorption and leak rate can be calculated from a pressure rate-of-rise

measurement. Flow rates can be further scaled by requiring that the system

doesn't operate in the gas-depleted mode; i.e., an increase in the total flow doesn't

affect the deposition rate proportionally. Unless graded films are desired, the av-

erage gas residency time (process pressure X system volume/gas flow rate) must

be much smaller than the total deposition time. Moreover, in a materials-efficient

process, the molecular diffusion length during the residency time must be much

longer than the interelectrode distance.

The hydrodynamic (= flow) part of the process is amenable to mathematical

modeling. Such models have been developed for a variety of reactor configura-

tions [63]. Some aspects of these models have been experimentally verified for

cross-flow [5] and longitudinal cylindrical reactor configurations. In practice, flow

rates are adjusted to ensure the uniformity of film thickness and materials proper-

ties of the film over the substrate area. Film thickness nonuniformities in the flow

direction generally result from depletion of precursor gases. Film nonuniformity

from gas depletion may be overcome by increasing the flow rate. At high flow

rates,

the gas composition of the exhaust gas will approach the composition of the

precursor gas. This approach to solving the nonuniformity problem obviously re-

duces the process efficiency, defined as the ratio between the film weight and the

weight of the condensable gas used to deposit the film. In practice it is often more

cost effective to increase the process efficiency by moving the substrates with re-

spect to the plasma to compensate for gas depletion effects. Pulsed plasmas [40]

offer another way to engineer around difficult flow problems by redistributing the

gas during the off cycle; fresh gas replaces the partially depleted gas. At the end

726 Chapter 5.5: Plasma-Enhanced Chemical Vapor Deposition

of the off cycle, the gas composition is spatially uniform along the substrate.

When the on cycle is short enough that the gas doesn't significantly deplete in the

flow direction, film thickness uniformity is assured, albeit at the cost of materials

efficiency.

In practice, scaleup is more complicated than making a few relatively simple

adjustments of the primary process variables. A change of reactor geometry po-

tentially affects any number of important intrinsic parameters. For instance, the

plasma potential is influenced by electrode geometry and details of power cou-

pling [51]. Plasma potentials have, through ion bombardment of the film and the

coater walls, a profound effect on the film properties [64]. Electric field patterns,

especially at the edges of substrates and electrodes, are nonuniform, causing an

effect known as "picture framing." The wall temperature of scaled-up reactors is

generally different from smaller prototypes. Gas retention or "memory" effects

are likely to be different, particularly if the reactor walls are covered by thick

deposits.

The ease with which plasma reactors can be cleaned should receive serious

consideration in the design phase of production systems. Deposits accumulating

around the deposition zone have to be removed regularly to avoid flaking and ex-

cessive outgassing. Flaking increases with the thickness of the deposit and tem-

perature cycling. The problem of cosmetic film damage is particularly relevant to

applications that require thick films, and to continuous processes that produce

thick films, as in roll coaters. Special constructions such as mesh-covered walls

minimize flaking. If the substrate is arranged between vertical electrodes, most of

the particulates generated during the process will settle on the reactor bottom

rather than on the substrates.

When the complexity of reactor construction does not allow the electrodes to

be easily exchanged, film buildup can be removed by plasma etching. Fluorinated

gases such as CF4 and

NF3

are commonly used for in situ chamber cleaning of sil-

icon, silicon oxide, and silicon nitride deposits. These gases, however, have long

retention times, and their use may well require several "sweep runs" before a re-

actor is restored to a controlled condition. The required purity level of a system,

as for instance measured by a pressure rate-of-rise measurement, depends on the

application and the nature of the contaminant. Group 3 and 5 dopants affect the

electrical behavior of silicon at ppm levels [65], whereas nitrogen, oxygen, and

hydrocarbons have a much smaller effect at comparable concentrations. Non-

electronic applications are generally more forgiving of impurities than electronic

devices.

Chemical memory results from long-term outgassing of the inside of a coater,

often exacerbated by the buildup of deposits. This carryover of small amounts

of gases from previous runs results in cross-contamination. This phenomenon

potentially limits the performance of multilayered devices such as thin-film tran-

sistors, photovoltaic structures, and superlattices. Some gases are worse than oth-

5.5.4 Conclusion 727

ers.

Diborane, fluorinated and chlorinated gases, and generally most high-vapor-

pressure liquids such as ammonia and metalloorganics, are especially bad actors.

In addition to procedural measures such as sweep runs, the problem can be mini-

mized by using multichamber reactors and gas lines dedicated to one gas.

But even multichamber reactors where isolated sections are dedicated to the

deposition of a single material, have their limitations. At the interface between

two films, the material composition is determined by startup transients that result

from the system responding to pressure changes due to plasma decomposition,

plasma-induced outgassing, and temperature increases. With gas flows stabilized,

10 seconds or more may pass before the process reaches a new stationary state af-

ter the discharge has first been turned on [66]. The material deposited in this time

is likely to be different from the bulk material. Startup transients especially limit

the performance of interfacially sensitive devices such as photosensors and thin-

film transistors [67,12].

5.5.4

CONCLUSION

The key advantage of PECVD is that semiconducting and dielectric films can

be deposited at relatively low temperatures. Ion bombardment can be substituted

for deposition temperature to achieve the required film density. Low-temperature

deposition is important for several different technologies. The continuing de-

crease of microelectronic device dimensions is synonymous with a decreasing

tolerance for high processing temperatures. On the other side of the device spec-

trum, PECVD enables the fabrication of electronic devices over very large areas,

often on low-temperature substrates. PECVD on plastics is rapidly emerging as

an important technology to enhance the barrier properties or the scratch resis-

tance of the plastic substrate. Significant advances have been made in enhancing

the biocompatibility of polymers with films deposited by PECVD techniques.

Process performance and equipment demands will continue to be dictated by

the needs of the single-wafer processing technologies for the manufacture of

high-density integrated circuitry. Process engineers have learned how to deal with

the inherent weaknesses and shortcomings of the process. Particularly, the forma-

tion of particles can be minimized by proper choice of process parameters, reac-

tor geometries and materials. It is well understood how to control and manipulate

the film stress. Multiple-frequency technology has become important to decouple

the chemical activation process from ion bombardment, allowing process win-

dows to be widened while maintaining tight control over the stress.

With the increasing conmiercial availability of novel, high-vapor-pressure or-

ganometallic precursors at affordable prices, the applications of PECVD continue

728 Chapter 5.5: Plasma-Enhanced Chemical Vapor Deposition

to grow. Organosilanes are increasingly used as a lower cost and safer source of

silicon instead of

silanc.

Optical multilayer devices are made possible by the com-

mercial availability of aluminum-, titanium-, and tin-based organic compounds.

Fluorinated compounds of every description continue to fascinate materials sci-

entists, particularly for the modification of surface properties. New applications,

including diamond deposition, diffusion barriers, optical filters, abrasion-resistant

coatings on

polymers,

powder

coating,

fiber coating, and biomaterials have pushed

PECVD technology far beyond the early electronic applications.

Major gaps continue to exist in our detailed understanding of the deposition

process and the way in which the process affects the properties of materials and

devices. More guided by intuition than understanding, new reactor designs some-

times cleverly exploit our limited knowledge, often building on experience in

dif-

ferent areas of glow discharge technology such as plasma etching. Afterglow re-

actors, hollow cathode discharges, helical resonators, ECR, and other microwave

discharges have been extensively used in plasma etching but their application to

plasma deposition is relatively recent. Afterglow deposition schemes now pro-

vide additional control over, and understanding of, the plasma chemistry. As the

demand for high-density, large-area electronics and high-performance plastics

continues to rise, the unique advantages of PECVD as a low-temperature deposi-

tion technology guarantee its increasing role as a thin-film manufacturing tech-

nology. Industrial demand for new processing strategies will fuel fundamental re-

search at universities and improve our understanding of the critical processes in a

field that is ripe with opportunities for technological and commercial innovation.

REFERENCES

L G. Glockler and S. C. Lind,

The

Electrochemistry of Gases and Other Dielectrics (Wiley, New

York, 1939).

F. Kaufman, in Chemical Reactions in Electrical Discharges, Vol. 80, edited by R. F. Gould

(American Chemical Society, Washington, DC, 1969), p. 29.

3 A. T. Bell, in

Techniques

am/Applications of

Plasma

Chemistry, edited by J. R. Hollohan and

A. T. Bell (Wiley, New York, 1974), pp.

-56.

F. Jansen, in Plasma Deposited Thin Films, edited by J. Mort and F Janscn (CRC Press, Boca

Raton, FL, 1986), chap. 1.

I. Chen and F. Janscn, J. Non-Cfyst. Solids, 59, 60 (1983) 695.

6. W. Luft and Y. S. Tsuo, Hydrogenated

Amorphous

Silicon Alloy Deposition Processes (Marcel

Dekker, New York, 1993).

D. Pribat et al.. Rev. Tech. Thomson-CSF, 26 (1994) 73.

8. Properties of Amorphous Silicon, 2nd ed. (INSPEC, The Institution of Electrical Engineers,

London, 1989).

9. A. Lcttington and J. W. Steeds, Thin Film Diamond (Chapman and Hall, London, 1994).

10.

K. E. Spear and L Dismukes, Synthetic Diamond: Emerging CVD Science and Technology

(Wiley, New York, 1994).

References 729

11.

Third symposium on silicon nitride and silicon oxide thin films, Electrnchem. Soc. Proceedings,

94-16(1994).

12.

S. V. Nguyen, D. Dobuzinsky, D. Dopp, R. Gleason, M. Gibson, and S. Fridmann, Thin Solid

F;7;»5,

193/194(1990)595.

13.

P. K. Shufflebotham, H. C. Card, and A. Thanaiklakis, / Non-Oyst. Solids, 92 (1987) 183.

14.

M. Kuhr, S. Reinke, and W. Kulisch, Diamond and Related Materials, 4 (1995) 375.

15.

S. V. Nguyen, T. Nguyen, H. Treichel, and O. Spindler. J. Electrocheni. Soc, 141 (1994) 1633.

16.

H. J. Frenck, W. Kulisch, M. Kuhr, and R. Kassing, Thin Solid Films, 201 (1991) 327.

17.

Y. Zhao and H. Suhr, Appl. Pliys., A55 (1992) 176.

18.

S. O. Kim, J. S. Byun, and H. J. Kim, Thin Solid Films, 206 (1991) 102.

19.

A. Hamerich, L. Lottermoser, and J. Muller, Smf. Coat.

Techno!.,

59 (1993) 212.

20.

E. T. Eisenbraun, B. Zheng, C. R Dundon, and R J. Ding, Appl. Phys. Lett., 60 (1992) 3126.

21.

R. A. Street, M. J. Thompson, and N. M. Johnson, Phil. Mag., B51 (1985) 1.

22.

R. S. Rosier, Solid State Technology, (June 1991) 67.

23.

W. E. Spear and R G. LeComber, Solid State Commun., 17 (1975) 1193.

24.

D. E. Carlson and C. R. Wronski, Appl. Phys. Lett., 28 (1976) 671.

25.

P. H. Singer, Semicnmluctor International, (March 1992) 44.

26.

A. Nagy and J. Helbert, Solid State Technology, (January 1991) 53.

27.

A. Nagy and J. Helbert, Solid State Technology, (March 1991) 77.

28.

G. W. Hills, A. S. Harms, and M. J. Thoma, Solid State Technology, (April 1990) 127.

29.

R. Williams, Modern Gas Processing (Artech House, Boston, 1991).

30.

G. J. Valco, V. J. Kapoor, M. D. Biedenbcndcr, and W. D. Williams, J. Electrocheni Soc, 136

(1989)175.

31.

D. E. Carlson, R. R. Arya, M. Bennett, and L. F. Chen, AIP

Conf.

Proc, Vol. 353 (1995).

32.

J. Mort and

Jansen, Plasma Deposited Thin Films (CRC Press, Boca Raton, FL, 1986), p. 187.

33.

Proc. Electrocheni. Soc, 94-35, (Electrochemical Society, Pennington, NY, 1995).

34.

H. Chatham,

Suif.

Coat.

TechnoL,

78 (1996) 1.

35.

S. Menichella, C. Misiano, E. Simonetti, L. D. Carlo, and M. Carrabino, Proceedings of the 37th

Conference of the Society of

Vacuum

Coaters, (1994) 37.

36.

C. Wohlrab and M. Hofer, Optical Engineering, 34 (1995) 2712.

37.

L. Shizhi, Z. Cheng, X. Xiang, S. Yulong, Y. Hongshun, X. Yan, and H. Wu, Suif Coat. Tech-

noL,

43/44(1990) 1007.

38.

S. Lim, J. H. Ryu, J. F. Wager, and T. K. Plant, Thin Solid Films, 245 (1994) 141.

39.

D. C. Bradley, Chem. Rev., 89 (1989) 1317.

40.

F. Jansen and S. Krommenhoek, Thin Solid Films, 252 (1994) 32.

41.

R. Monson, master's thesis, Department of Chemical Engineering, Brigham Young University,

Salt Lake City, Utah, 1991.

42.

C. Steinbruchel, Phys. Thin Films, 18 (1994) 289.

43.

M. Shiratani, H. Kawasaki, T. Fukuzawa, T. Yoshioka, Y. Ueda, S. Singh, and Y. Watanabe,

J. Appl. Phys., 79 (1996) 104.

44.

G. S. Selwyn, / Vac Sci. TechnoL, B9 (1991) 3487.

45.

G. S. Selwyn, J. Vac ScL TechnoL, AlO (1992) 1053.

46.

H. R W. Hey, B. G. Sluijk, and D. G. Hemmes, Solid State Technology (April 1990) 139.

47.

K. Kohler, D. E. Home, and J. W. Coburn,./. Appl. Phys., 58 (1985) 3350.

48.

R. J. M. M. Snijkers, M. J. M. v. Sambeek, M. B. Hoppenbrouwers, G. M. W. Kroesen, and

J. D. Hoog, J. AppL Phys., 79 (1996) 8982.

49.

R.H.BrucQ, J. Appl. P/i>'5., 52 (1981) 7064.

50.

A. Tsukune et al., in Extended Abstract: Properties uf silicon nitride prepared by dual frequency

RF plasma deposition (Electrochemical Society, San Diego, CA, 1986), p. 580.

51.

K. Kohler, J. W. Coburn, D. E. Home, and E. Kay, / AppL Phys., SI (1985) 59.

52.

J. Schlote, S. Hinrich, B. Kuck, and K. W Schroder, 5//;/ Coating

TechnoL,

59 (1993) 316.

730 Chapter 5.5: Plasma-Enhanced

Chemical

Vapor Deposition

53.

G. Lucovsky, D. V. Tsu, R. A. Rudder, and R. J. Markunas, in Thin Film Processes II, edited by

J. L. Vossen and W. Kern (Academic Press, New York, 1991), Section IV-2.

54.

G. Lucovsky, Y. Ma, T. Yasuda, C. Silvestre, and J. R. Hauser, Jpn. J. Appl Phys., Part I, 31

(1992)4387.

55.

S. Matsuo and M. Kiuchi, Jpn. J. Appl. Phys., 22 (1983) L210.

56.

K. L. Seaward, J. E. Turner, K. Nauka, and A. M. E. Nel, J.

Vac.

Sci. TechnoL, B13 (1995) 118.

57.

S. Dzioba, Mat Res. Soc. Symp. Proc, 165 (1990) 91.

58.

C. S. Pai, J. Electrochem. Soc, 139 (1992) 850.

59.

B. Fowler, D. Stark, J. Xie, R McDonald, R. Carpio, and S. Akbar, Proc. SPIE-Int. Soc. Opt.

£/ig., 2090 (1993) 203.

60.

J. Mort and F. Jansen, in Plasma Deposited Thin Films, edited by J. Mort and F. Jansen (CRC

Press,

Boca Raton, FL, 1986), chap. 7.

61.

H. Chatham, A. Smith, R. Smith, J. Rietzel, and C. Woolley, 39th Annual Technical Conference

Proceedings of the Society of

Vacuum

Coaters (Philadelphia 1996) 64.

62.

M. Chang, J. M. White, D. N. K. Wang, and D. Maydan (Applied Materials, European Patent

Application EP 303508, 1989).

63.

I. Chen, Thin Solid Films, 101 (1983) 41.

64.

J. L. Vossen, J. Electrochem. Soc., 126 (1979) 319.

65.

F Jansen and D. Kuhman, J.

Vac.

Sci. Technol, A6 (1988) 13.

66.

S. V. Nguyen and R H. Pan, Appl. Phys. Lett., 45 (1984) 134.

67.

Y. Nakayama, T. Ohtsuchi, M. Nakano, and T. Kawamura, / Non-Cryst. Solids, 77, 78 (1985)

757.

CHAPTER

5.6

Common Analytical

Methods for Surface

and Thin Film

John H.Thomas,III

Corporate Research Laboratory

5.6.1

INTRODUCTION

The surface of a material is generally considered the topmost atomic layers or,

on a crystalline surface, a monolayer (single atomic layer). To determine surface

structure or chemistry, the surface under investigation must be kept in a pristine

state such that the surface is not contaminated by atmospheric contaminates or

residual gas components resulting from poor vacuum quality. In earlier chap-

ters 4.3 and 1.8, it was shown that vacuum systems that were not properly pre-

cleaned could contain substantial amounts of contaminate residual gases from

outgassing of components and the chamber and/or backdiffusion of gas compo-

nents from the pumps; hydrogen, argon, methane, and water vapor are a few of

the more conmion residuals. These residual vacuum gas components can readily

react with ahnost any surface. For example, assume the surface is a single-crystal

surface of a very reactive metal (Ti (100)). To produce a very clean surface, pro-

cedures such as in situ fracture, and sputtering followed by in situ annealing must

be performed under very good vacuum conditions (<10~^^ torr). Measurements

* All rights of reproduction in any form reserved.

731

732 Chapter 5.6: Common Analytical Methods

for

Surface and Thin Film

Fig.

Excitation source:

x-rays (< 1500 eV)

UV (20-40 eV)

electrons (<10 KeV)

ions (2-15 KeV)

alpha particles/protons (2500 KeV)

Detected Particles:

electrons

ions

alpha particles

protons

The analytical process of surface analysis.

must then be made before a monolayer of contamination forms on the surface,

which generally takes only a short period of time, depending on the composition

of the background gas in the system.

Most surface-analytical techniques are particle based; that is, they involve the

detection of low- or high-energy particles (electrons and ions) that are produced

by X-rays, ultraviolet irradiation, or electron or ion bombardment. Figure 1 sum-

marizes the general methodology used to study surface structure and chemistry.

Chemical analysis of surfaces, the determination of elemental composition and

some details of the chemistry existing on the material surface, is performed by

two experimental procedures: the energy analysis of excited electrons, and the

energy or mass analysis of

ions.

Instruments used in energy analysis of particles

are based on electrostatics. Mass analysis is performed by one of three methods:

magnetic field deflection, E® mv (RF

fields)

and electrostatic (time of flight) mea-

surements. Table 1 summarizes the most commonly used analytical techniques

and some important properties of the method.

The techniques just listed require the use of ultra-high vacuum. The following

sections present an overview of these methods.

5.6.2

THE ELECTRON SPECTROSCOPIES

The electron spectroscopies are the most common surface analytical techniques

used in industry, especially. Auger electron spectroscopy (AES). Both AES and

X-ray photoelectron spectroscopy (XPS or ESCA) are used for submonolayer

qualitative and quantitative surface analysis, and with the addition of ion sputter-

ing and elemental imaging thin-film in-depth analysis. AES is most often used in

5.6.2

The

Electron Spectroscopies

733

Table

Common Analytical Techniques and Their Properties

Vnalytical

Technique

•C-ray photoelectron

;pectroscopy

XPS/ESCA)

\uger electron

ipectroscopy (AES)

secondary ion mass

ipectrometry

:dynamic SIMS)

Secondary ion

mass spectrometry

(surface TOP

SIMS)

Rutherford back-

scattering spec-

trometry (RBS)

Excitation

Probe

1.5 keV

photons

2-20 keV

electrons

5-20 keV ions

l-4MeV

a-particles

Emitted

Particles

Slow e~

from core

levels

Slowe"

from core

levels

Slow ions

a-particles

Range

Z> I

Z>2

All

Spatial

Resolution

10/im

imaging

500 A

imaging

2-10/xm

imaging

2-10/xm

imaging

2 mm

Detection

Limits

-0.1 at.

~0.1 atomic

<1 ppb

Surface

<1 ppm

-1%

Accuracy

(%)

<5%

-10%

Varies with Z

and matrix

Varies with Z

and matrix

<2%

the depth-profiling mode but can be used to obtain chemical information. XPS

can also be used in the depth-profiling mode, but as a result of the "X-ray beam

size,"

depth profiling is much slower, because of

the

larger sputtered area required.

5.6.2.1

Introduction

Electron spectroscopy is produced experimentally by bombarding a surface with

either electrons or photons (see Figure 1) and detecting the excited electrons in

an ultra-high vacuum. The two most conmion spectroscopies are Auger electron

spectroscopy (AES) and X-ray photoelectron spectroscopy (XPS) or ESCA (elec-

tron spectroscopy for chemical analysis), electron excitation, and X-ray excita-

tion of electrons. Electrons emitted into the analysis vacuum chamber are ulti-

mately collected and kinetic energy analyzed to determine the chemistry of a

surface of unknown composition. The kinetic energy of the emitted electron

(jmv^) is conmionly measured to an accuracy of 0.1 eV in 1000 eV (1 part in

10"^) using electrostatic energy designs

[1-4].

A typical instrument consists of

several components: an excitation source of either electrons or photons (X-rays),

a means of analyzing the kinetic energy of the emitted particles (electrons), a de-

tection and data acquisition system, and an ultra-high vacuum system. These

components are discussed later in this section.