Pressure Effects in a PVD system on Thin Film Chemistry and Mechanical Properties Introduction Physical Vapor Deposition (PVD) is a vacuum deposition technique used to describe any of a variety of methods to deposit thin films by the condensation of a vaporized form of the material onto various surfaces. Techniques may include sputter, thermal evaporation, e-beam, and pulsed laser. Multilayer films produced by PVD are currently used in a variety of applications, eg: microelectronics (electrical barriers, diffusion barriers ); sensors (magnetic sensors, gas sensors ); optics (anti-reflection coatings); electrical conductors; corrosion resistance and protection. Films that were once microns in thickness are now measured in nanometers. Many of these applications require the use of an adhesion layer to help the primary film adhere to the substrate. vacuum conditions in a SMART PVD System intended to replicate the current conditions found in many labs today. The materials were chosen as representative of an adhesion layer and a conductor: an adhesion layer of Ti, followed by an electrical conducting layer of Ag. While there have been a number of scientific studies on the effect of gases on the properties of thin films, most of these are ignored in the everyday production and subsequent use of nanometer thin films. Equipment and process protocols have not kept up as thinner films have been put into use. Vacuum levels during deposition often vary a lot. Some multilayer films are still carried to a 2 nd system which is dedicated to a specific material for the deposition of the 2 nd film thus exposing the 1 st film to the air. Experimental Details Three nano-thin films (~100nm) of two different material layers were processed under 3 different Figure 1 The basic SMART PVD System. The two sputter sources were mounted at the bottom of the vacuum chamber. Page 1
A two layer film set was produced: first a 100 nm layer of Ti was DC sputter deposited on a Si wafer, and then a 100 nm layer of Ag was DC sputter deposited onto the Ti. Three different vacuum conditions were chosen: sample 1 - base pressure of 1.3E-7 torr for Ti; Ag followed immediately sample 2 - base pressure of 1.3E-6 torr for Ti; Ag followed immediately sample 3 - base pressure of 5E-7 torr for Ti; the chamber was then vented, followed by a pump down to 5E10-7 torr; Ag followed. The chemical composition of the multilayer film structure was measured using ESCA chemical depth profiling by sputtering with Zalar rotation in a Physical Electronics VersaProbe XPS Microprobe. The mechanical adhesion was measured with a Hysitron TI 950 TriboIndenter using ramp force scratching. The Si substrates had been broken in half before the depositions so as not to effect the mechanical measurements. ESCA: Results Each of the samples was sputter depth profiled by ESCA. The ESCA results are shown in the following 3 figures: A very thin TiO2 layer is seen at the interface with the Si substrate, followed by a thin partial Ti nitride. There is also a thin titanium partial oxide at the Ag/Ti interface which formed between the depositions of the metal films. Figure 3 ESCA depth profile through sample 2. Though the starting pressure was a factor of 10 larger, the titanium interface with the Si shows a more complex chemical structure. There is less oxygen but more nitrogen and carbon present, indicating the formation of a more complex Titanium oxide-nitride-carbide. Figure 2 ESCA depth profile through sample 1. Figure 4 ESCA depth profile through sample 3. Page 2
This sample had a starting base pressure half way between samples 1 and 2. However, the chemistry of the Ti/Si interface is much more similar to sample 1. The titanium oxide layer is about twice as thick as in sample 1, or about 45nm. Similar to sample 1, there is no nitride or carbide at the interface. In all cases, the original surface Si oxides were reduced as measured by the Si ESCA line shapes. None of the oxygen at the interface was bound to the Si. The Ti layers are the same thickness as the Ag layers, but they sputter more slowly. NanoScratch thin film adhesion testing: The other half of each of the samples was analyzed by ramp force scratching. A typical analysis is shown in figure 5. The critical loads for the Ti films on Samples 1 and 3 are shown in the following table. Sample Test # 1 3 Critical Load (μn) 1 1156.0 2 1128.6 3 1326.3 4 1065.6 5 1083.7 Ave: 1152.0 Std Dev: 103.8 1 845.9 2 882.0 3 801.6 4 845.9 5 863.9 Ave: 847.9 Std Dev: 29.9 The scratch tests on sample 2 did not show any observable delamination, indicating a greater adhesion than either sample 1 or 3. From these results, we see that the Ti film on sample 1 adheres much better to the Si substrate than it does on sample 3. This is seen more clearly in the following chart. Figure 5 Ramp Force Scratch: Normal force, normal displacement, lateral force, and lateral displacement vs time on sample 1. The red bar indicates the critical point. In all cases, the Ag films adhered well. There was no delamination, with the testing tip simply plowing through the film until it reached the Ti film. Each of the samples was analyzed five times. The thin film delaminates when the normal force during the scratch reaches the critical force, or critical load. Page 3
Each of the scratches was also examined by scanning probe microscopy (SPM). These are all 14μm x14μm scans made following the scratch using the same diamond tip. The micrographs are shown: Sample 1 Sample 3 Start of ramp force scratch End of ramp force scratch Sample 2 The remnants of the delaminated film are clearly seen for samples 1 and 3. Sample 2 showed no clear delamination of the Ti film, consistent with the scratch measurements. There is no visible delamination of the Ag films, also consistent with the quantitative scratch force measurements. Page 4
The overall results of the three films are summarized in the following table: Sample 1 2 3 Deposition Pressure for Ti Base Pressure of 1.3E-7 torr Base Pressure of 1.3E-6 torr Base Pressure of 5E-7 torr Ag / Ti Interface Thin Ti sub oxide Thin TiO Ti oxide formed after exposure to air Ag Adhesion No delamination detected No delamination detected No delamination detected Ti Film Chemistry Ti / Si Interface Chemistry Primarily Ti with some mixed complex Ti -O-N-C Thin TiO2 layer at Si Interface followed by nitride layer Mixed complex C Ti -O-N- Thin TiO layer at Si interface followed by mixed complex Ti -O-N Thin Ti layer with some mixed complex Ti -O-N-C 30 nm TiO2 layer at Si Interface followed by nitride layer TiO2 Adhesion Critical Load of 1150 μn No delamination detected Critical Load of 850 μn Conclusions Small changes in the pressure at the start of a deposition can substantially affect the film interface chemistry and the adhesion strength of the film. The interface of the Ti adhesion layer which was exposed to air had a thicker oxide layer on it than those which were not exposed. This film also had poorer adhesion. There is evidence that the oxide layer acted as the weak boundary layer and the delamination occurred within the oxide. The interface consisting of the complex Ti Oxy- Nitride- Carbides showed no gross delamination, implying better film adhesion. These interface layers can be a substantial fraction of a film which is 100nm or less. Analytical measurements are necessary to characterize the film changes brought on by small changes in deposition conditions. Precise control of the deposition conditions is critical to film reproducibility Page 5
Recommendations If you intend to work with nano-thin films (<100nm), you need an ideal PVD system. An ideal PVD system is one where you can control the total base pressure as well as individual gas pressures before and during the deposition. The ideal system will have UHV capabilities so you can start with the cleanest environment possible. The ideal system will have the ability to use a variety of deposition sources so you can tailor the films and the interfaces. The SMART NanoTool from SVTA can be configured as such an ideal PVD system System Overview Multi-Technique thermal sputter e-beam laser Versatile vary sample size and heating co-deposition capability Flexible multiple configurations custom design available UHV capable <5 x 10-10 Torr Load lock capability Field Upgradable Compact Cost Effective The author would like to thank Physical Electronics for providing the ESCA measurements and Hysitron for providing the nanomechanical measurements. Page 6