1. Sputter deposition of Thin Films
Course Tutor
Dr R E Hurley
Northern Ireland Semiconductor Research Centre
School of Electrical & Electronic Engineering
The Queen’s University of Belfast

2. Course content Applications Definitions Advantages of sputtering
Physical principles
The glow discharge
Types of sputtering system
Magnetron sputtering
Process parameters (effects on thin film properties)
Stress in films
Microstructure
Uniformity considerations
Etching
Compounds and reactive sputtering
Examples of systems

3. Application examples of sputtering
Developed in last 30yrs into a sophisticated coating tool
Optical-interference filters and protective coatings for
lenses, mirrors, transparent conductive coatings (ITO),
for displays, heat filters for architectural glass
Mechanical -hard coatings for tools, low friction coatings
for bearing surfaces, anti-corrosion coatings in aircraft industry
Electronics - especially semiconducting industry, metallisation,
barrier layers, display circuitry, discrete components
Data storage - CDs, DVDs, RW heads

4. Sputtering for thin films
Q. What is sputtering?
A. The removal of material (normally as atoms), from the
surface of a target by energetic ion (or neutral) bombardment.
The source of ions may be an electrical plasma discharge
surrounding the target, or alternatively a separate ion-source.
The incident ions cause cascades of collisions within the target,
where these intersect the surface, sputtering occurs.
A receiver, (substrate), collects the sputtered atom flux as a thin film.

5. Advantages of sputtering
Versatility
Wide range of materials including compounds,
alloys, and high purity targets available and interchangeable
Control
Particularly in the case of compounds and alloys when problems
of fractionation do not occur.
Good stability and control over deposition rate
Mass flow and pressure controllers work well. (Even without these,
good repeatability is possible once a system has achieved equilibrium)
Scalability
Substrates can be very small or very large, discrete or continuous as a roll

6. Sputtering - effect of incident ion

7. Particles in sputtering

8. Sputter yield from Cu target

9. D.C. Sputtering System

10. Glow discharge loss processes

11. Ion pair production in dark-space

12. Charge exchange in the dark-space

13. Substrate effects - argon sputtering

14. Types of sputtering systems
Diode (1960s, slow, rather high pressure [~0.1mb],
secondary electrons at substrate, tendency to arc, conducting targets only)
R.F. (60s,70s, good for insulators, slow,
lower pressure than diode, [~0.005 mb], secondary electrons at substrate)
Magnetron (DC and RF) (late 70s, 80s)
(the modern system, fast [minimising contamination], low pressure
[minimal pump throttling needed hence improved pumping of contaminants],
no secondaries to heat substrate)
Ion beam (for ultra-clean work, research, slow,)
Hybrids (all types of combination system,
may include evaporation, microwave sources etc.)

15. Commercial d.c. sputtering system

16. Problems with DC sputtering
Relatively high pressure overloads pumps, means
contamination, lack of control over process.
Bombardment of substrate by energetic neutrals and -ve ions (contamination)
Bombardment of substrate by energetic secondary electrons (heating)
Cannot be used with insulating targets; arcing by surface charging
Low deposition rate (increases contamination)

17. Alternatives to DC Sputtering
Reduce working pressure (increase plasma density
for a given working pressure), by additional excitation:
R.F. (13.56MHz)
2. External magnetic coils.
3. Use an ion beam system in UHV (very slow)
4. Magnetron sputtering - has become the industry norm.

18. Magnetron sputtering High deposition rates
(good for production, minimises effect of contaminants
Low pressure operation
(minimal throttling of pump, means higher pump speed
and less contaminants.
Elimination of secondary electrons on substrates
(Plastics and heat-sensitive substrates can be coated,
the substrate temperature can be regulated as required.
Good control over reactive sputtering
(high pumping speed = stability and control)

19. Magnetron sputtering - schematic

20. Magnetron sputtering
Unbalanced
Balanced

21. Sputtering parameters
Deposition rate (Å/sec, micron/min)
Substrate temperature
Gas pressure/plasma type (DC, RF, microwave, hot filament)
Target/substrate geometry, relative motion (uniformity)

22. Sputtering – thin film properties affected
Stress
Crystallinity/amorphousness
Gas and other impurity incorporation
Density and structure
Stoichiometry or composition
Effect of all above on optical, electrical, magnetic or mechanical properties.

23. Illustration of stress in films
Film ‘wants’ to be ‘larger’
Film ‘wants’ to be ‘smaller’
Stress in thin films results from differences in thermal expansion (thermal stress) or from the microstructure of the deposited film (intrinsic stress). At substrate temperatures less than 20% of the melting point, intrinsic stress due to incomplete structural ordering dominates.
Thermal stress occurs because film depositions are usually made above room temperature. Upon cooling from the deposition temperature to room temperature, the difference in the thermal expansion coefficients of the substrate and the film cause thermal stress.
Intrinsic stress results from the microstructure created in the film as atoms are deposited on the substrate. Tensile stress results from microvoids in the thin film, because of the attractive interaction of atoms across the voids.
Figure 1. Tensile stress, conceptual diagram. The film wants to be "smaller" than the substrate because it was "stretched" to fit.
Compressive stress results when heavy ions or energetic particles strike the film during deposition. The impacts are like hitting the film with a hammer, packing the atoms more tightly.
Figure 2.Compressive stress, conceptual diagram. The film wants to be "larger" than the substrate, because it was "compressed" to fit.

24. Causes of stress in thin films
1. Thermal Stress
differences in expansion coefficient between film and substrate
2. Intrinsic Stress
incorporation of gas atoms
mismatch during epitaxy
recrystallisation
phase transformations
microscopic voids, porosity, and dislocation effects.
atomic peening
Tensile stress associated with voids and porosity
Compressive stress associated with gas atom incorporation
and atomic peening.

25. Stress in thin films Mechanism for compressive stress in sputtering
(Thornton and Hoffman, 1985)

26. Stress in thin films - pressure effect
K. O'Donnell, J. Kostetsky, and R.S. Post of NEXX Systems LLC
The grain size does not appear to change with increasing pressure but the grain morphology changes dramatically. The surface roughness increased from 2.91 nanometers (rms) at 2 mTorr to 5.24 nanometers (rms) at 20mTorr.
The surface morphology indicates a dense microstructure in the TiW sputter deposited at 2 mTorr which is reflected in the large compressive stress measured for this film ( MPa). The microstructure changes from a dense to a voided microstructure in the TiW film deposited at 20 mTorr which is reflected in the tensile stress measured for this film (+700 MPa). The increase in resistivity is consistent with the development of such a voided structure.
The energetic particles bombarding the substrate are the sputtered atoms of TiW and working gas ions that are neutralized and reflected at the cathode. No argon was detected using EDXS analysis of TiW samples, which is consistent with sputtered atoms having the greatest effect. Wehner and Anderson showed that sputtering with argon and krypton yielded similar stress variations as a function of process pressure and concluded that sputtered atoms dominate [2].
TiW (heavy mass, atomic peening effect at low pressures)
Hoffman and Thornton, 1979)

27. Effect on input power on stress, DC,AlN,sputtered
Kusaka, vac2000,59,812

28. Effect of negative bias on stress,, Cr, NiV (RF sputtered)
K. O'Donnell, J. Kostetsky, and R.S. Post of NEXX Systems LLC, 2002
Use of bias to reduce stress to near zero (more dense microstructure)
The resistivity of Cr (4000Å) does not vary significantly with increasing substrate bias.The NiV resistivity also does not vary with substrate bias. Energetic ions bombarding the growing film at increased substrate bias results in a high density of defect clusters, which modifies the film stress but not the resistivity.
The increase in resistivity seen above for TiW films deposited at increased process pressure is associated with a voided microstructure and increased tensile stress. The 'atomic peening' mechanism evident in TiW is not significant in the case of Cr and NiV due to the lower atomic mass of these elements. Reduction of tensile stress in Cr and NiV films is accompanied by reduced surface roughness and more uniform grain size.
CONCLUSIONS
Stress in TiW is controlled by varying gas pressure during deposition. TiW sputter deposited at pressures ranging from 2 to 20 mTorr shows a structure change from a dense fibrous microstructure to a voided structure of isolated columns which is associated with the stress reversal from compressive to tensile.
Reduction of tensile stress in Cr and NiV is achieved by applying substrate bias during deposition. Reduced surface roughness of films deposited with substrate bias together with a more uniform grain size distribution are consistent with a more dense microstructure in films with reduced tensile stress.

29. Microstructure of thin films
Affected by energy input into substrate during deposition
Increasing energy promotes grain growth and crystallisation
Gas pressure affects energetic neutral component.
Substrate temperature
Substrate bias affects energy of bombarding ions

30. Magnetron sputtering - microstructure diagram
Microstructure as a function of argon pressure and substrate temperature
Movchan and Demchishin, 1969

31. Use of AFM for roughness measurement
Sputtered AlN. AFM showing roughness v. power
Kusaka et al. Vac 2000,59, 810

32. Film uniformity in sputtering
Non-uniformity is a potential problem in magnetron sputtering
(caused by shape of the magnetic field lines which depend on design used)
In semiconductor processing of 300 mm wafers for ULSI the problem is acute.

33. Solutions for non-uniformity
Use relative motion of target and substrate
For rotation over circular targets, +/- 8% is typical for many systems
For circular targets, sun and planets will reduce to < 1%
For large systems, lateral (shuttle), motion over rectangular targets is good
These systems are used for architectural glass and roll-coating of plastic sheet but also for batches of smaller components, e.g. silicon wafers

34. RF Mag. with sun and planets holders for ZnO films
for SAW filters, Ar/O2. (Yoshino et al Vac.2000,59, )

35. In-line reactive sputtering for ITO, dc and mf
Strumpfel and May, Vacuum, 59, 2000,

36. Industrial magnetrons for coating Cu, MF-pulsed for solar absorb.
1m wide at 1m/min.Cr-carbide, Milde etal. Vac.,2000,59,

37. Use of sputtering for etching substrates
Place substrate on target → controlled etching
Dedicated etchers are normally used with or without chemistry
Vertical side-walls are possible (anisotropic etching)
Reactive chemistry (Br, Cl, O2), increases etch-rate
but increases isotropy
A mask allows pattern delineation, but leads to tapering and trenching
A separate broad-area ion source may be used

38. Use of sputtering for etching substrates
Sputter etch
Wet chemical etch

39. Sputtering - erosion of mask

40. Sputter etching - trenching effect

41. Sputtering of compounds
Compounds of two or more components are possible
Methods
Several targets (co-sputtering)
Compound or alloy target
(Note: components sputter at different rates, but an equilibrium will be established
after a certain time which results in composition of original target material being
deposited as a thin film. The exception is when component is released as a gas
(O2, N2), and hence depleted from depositing film.)
Multi-component target (discrete pieces of second material)

42. ZrTi target/PbO pellets, rf diode PZT ferroelectrics
Kim et a. Vacuum, 59,

43. The main problem is control:
Reactive sputtering for oxides and nitrides
The main problem is control:
The critical reaction occurs at the target surface
A mixture of argon and reactive gas (O2, N2) is used.
Too much reactive gas, sputter-yield falls, gas build-up uncontrollably and deposited layer is gas-rich.
Too little reactive gas, sputter-yield increases uncontrollably and layer is metal-rich

44. Control problem in reactive sputtering
Hysteresis effect during reactive sputtering, (Sproul, 1984)
The partial pressure follows a hysteresis loop with gas flow supply.
The operating point for stoichiometric films is at a point of abrupt change of slope.

45. Reactive sputtering of oxides and nitrides
Methods of controlling the process:
Setting the flow/pressure at fixed values and using trial and error
Unsatisfactory, may work with a high-speed pumping system.
Plasma monitoring using Optical Emission Spectrometry
Works well and costs can be moderate with CCD system.
With fast enough pumping, manual ‘tweaking’ is OK.
A more sophisticated system will use feedback from spectral
lines to control flow ratio of argon/reactive gas, and a
pressure controller to control total pressure.

46. Optical emission spectrometry (OES)
Suitable spectral lines of M and O2 (or N2)
These are normally in the near UV and a quartz window is necessary.
The ratio of the two lines is held at a fixed value by adjustment of the Ar/reactive gas flows.
The ratio can be chosen for stoichiometry or metal or gas rich films as required.

47. Optical emission spectroscopy
Library file, Ti(I)

48. Optical emission spectroscopy
Reactive sputtering of ZrN showing effect of varying N2 flow-rate
(Berg et al., 1986) Ar N2 Zr

49. Biased directional sputtering
(Sato et al, Vacuum 59, 2000, )
Ti, Ta comps. Barrier layers

50. RF magnetron system for ferroelectric films
(Ichikawa et al. Vacuum,59, 2000, )
PbT and SrT effect of 02 partial pressure in Ar on
film growth.

51. ZnO/In2O3 magnetron sputtering for improved
ITO type . (Tominaga et al, Vac 2000, 59, )

52. Sputtering- effect of H2O on SrTiO3
Nakagawara et al vac. 59, 2000,742-47

53. DC magnetron sputtering of AlN, stress v. power
Kusaka et al, Vac. 59, 2000,

54. ECR-RF reactor, TiO2
Lee, IBM,VAC1998,51-4, 503-9

55. Sputter, unbalanced-closed-field for DLC
Teer et al Vac, 1999,52-1-2,

56. DLC-pulsed plasma, ion bombardment
Tochitsky et al., Vacuum, 2000,58,79-86