1. ECLOUD2018 WORKSHOP ELBA, ITALY
FIRST STUDY OF CONDUCTING
NEG (TiZrVHfCu)
ADRIAN HANNAH
STFC
University Of Liverpool
Reza Valizadeth
Vin Dhanak
Ruta Sirvinskaite
Nasser Sedghi
Taaj Sian
James Gibbon
Oleg Malyshev
Leanne Jones
Karl Dawson
Good afternoon, and thank you to the organising committee for allowing me to present our latest results on the study of thin films of NEG with copper
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

2. ECLOUD2018 WORKSHOP ELBA, ITALY
Non Evaporable Getter (NEG) FILM
Bulk
0.5 to several m thick film of transition metals or alloys with high oxygen solubility (Zr,Ti) and diffusivity(V).
It is firstly a barrier to stop hydrogen to defuse into Vacuum from the vessel bulk metal.
Once the surface is activated it will work as an active surface for pumping H2O, CO, CO2, and N2 by chemical reaction.
Hydrogen dissolves in the bulk of NEG and forms solid solution.
It has a very low outgassing rate since it is deposited in XHV condition and reactive molecules after chemical reaction forms stable carbides, oxides, and nitrides.
NEG
Interface
Our NEG films are typically deposited in the range of half to several microns thick. Their primary function is to provide a barrier to prevent the outgassing, usually of Hydrogen, from the bulk of the vacuum chamber into the vacuum system, The metals chosen tend to favour Zr and Ti due to their high oxygen solubility and V due to its high diffuse-zivity. However metals of a similar transition can also be used. The secondary function of the film becomes apparent after the film has been activated and the metallic layer on the surface, becomes an active pump for water, CO, CO2 and Nitrogen. As the film is deposited in XHV conditions, and the chemical reactions form stable carbides, oxides and nitrides, there is very little outgassing from the film
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

3. ECLOUD2018 WORKSHOP ELBA, ITALY
NEG ACTIVATION
To activate NEG film, the vessel needs to be heated under vacuum.
The activation takes place by diffusion of oxygen into the bulk.
The diffusion depends exponentially with the temperature and on the square root of time.
Either higher temperature and less time or low temperature and long activation time.
Diffusion predominantly take place through grain boundaries.
To lower the activation temperature, the NEG film should be synthesis under a condition that it will grow with grain size in nm range.
To activate the NEG, the system needs to be heated whilst under vacuum. This allows the surface oxygen to diffuse into the bulk of the film via the grain boundaries. This diffusion depends exponentially with the temperature and on the square root of the time. This means either a short time at high temperature or a long activation time at a low temperature. The smaller the grain size of the film, hence a greater grain boundary density, the lower the activation temperature. To decrease the grain size is predominately the reason for adding additional metals to the NEG mixiture.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

4. ECLOUD2018 WORKSHOP ELBA, ITALY
NEG PUMPING ACTION
where  is the sticking probability ( ≤ 1), v is the mean molecular velocity and
A is the geometrical surface area
The pressure P inside a system is given by the gas load Q divided by the pumping speed S. For a NEG pump, the speed is given by this formulae here. The geometric area and velocity are fixed for a particular system, Hence to increase the pumping speed, we need to increase the sticking probability alpha. One way we can do this is to grow the film so that it has an open, high aspect ratio, columnar structure.
Hence, to increase the pumping speed of NEG it is necessary:
Increase the sticking probability
( this can be achieved by growing the film in an open large aspect ratio columnar structure)
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

5. ECLOUD2018 WORKSHOP ELBA, ITALY
OTHER IMPORTANT PROPERTY
In particle accelerator beam interacts with the vacuum vessel wall.
eg: Wall resistive impedance depends on surface roughness and beam material.
In general the beam pipe can be made of Cu, Al and SS.
In case of Cu and Al the Wakefield impedance can be influenced by the NEG coating which has significantly higher surface resistance than the Cu or Al wall.
It should be remembered that in an accelerator, the particle beam interacts with the vacuum vessel wall and that wall resistive impedance effects can depend on properties such as the roughness of the vacuum vessel wall and its material properties. Although beampipes can be made of Copper, Aluminium or stainless steel, a NEG coating on Copper or Aluminium can significantly increase the surface resistance resulting in changes to the Wakefield impedance
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

6. ECLOUD2018 WORKSHOP ELBA, ITALY
MATERIAL RESISTANCE
Defects in metals results in electron–defect collision, they lead to a reduction in mean free path.
Crystal lattice vibrations: phonons which are temperature dependent.
Impurities: can be inclusion of foreign atoms, lattice defects, dislocations which are not temperature dependent.
Grain boundaries: These are also not temperature dependent.
Dense or porous structure.
As can be seen, the material resistance is dependent on the Electron mean free path within the material.
A good NEG film will typically consist of different elements, causing lattice defects and dislocations. It will require a high grain density and a porous or columnar structure. These requirements means a good NEG film is unable to have a low material resistance.
Hence there is mismatch between having high pumping, low activation NEG and having a good conducting NEG even if the material had high conductivity.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

7. ECLOUD2018 WORKSHOP ELBA, ITALY
CONDUCTIVE NEG
In order to enhance the NEG conductivity with out compromising any vital NEG characteristic such as:
Low Activation Temperature
High Pumping speed (large surface area and high sticking probability).
In the first instance it was decided to reduce the resistivity by adding a more conducting metal such as Copper, Aluminium or Silver with different concentration.
In order to try and produce a NEG coating, which maintains the vital NEG characteristics, but has a higher conductivity, we decided to add a metal, such as copper, aluminium or silver to the NEG film. The hope was that by varying the concentration of this metal, we could produce a film with the required NEG characteristics, but with a suitable conductivity.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

8. ECLOUD2018 WORKSHOP ELBA, ITALY
NEG DEPOSITION METHOD
TARGET: Twisted 3 mm diameter Ti-Zr-V-HF with and without 2 mm and 1 mm copper wire.
Bp: 2x10-10 mbar
Wp: 2x10-2 mbar (Kr)
Power : 100 W, DC
Magnetic field: 250 Gauss
Deposition temperature : 110 C
Deposition time: 4 hours
Witness Sample: SS, Si and Glass
Cylindrical Magnetron
To test our theory, we used our standard tube deposition facility. The target material consisted of 4 3mm diameter wires of Ti, Zr, V and Hf, twisted together to produce a single target. These materials were chosen as we had already shown that they produce the lowest activation temperature. Around this target we then added a further target wire of either 1mm or 2mm diameter copper. We then deposited a film onto a half meter long 2 ¾” conflate tube, at a power of 100W DC for 4 hours with a working pressure of Krypton of 2 times 10 to the minus 2. these are the conditions we use to grow open columnar structures. Beneath the tube we also placed three witness samples of stainless steel, silicon and glass to assist our analysis of the coating.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

9. ECLOUD2018 WORKSHOP ELBA, ITALY
SET UP FOR NEG PUMPING SPEED EVALUATION
Once the tube had been coated, we transferred it to our NEG pumping speed evaluation facility as shown here, where we can activate the film using our bakeout and activation profiles shown here. This allows us to activate to a range of temperatures and measure the sticking probability for each activation. So the NEG is held at 80C whilst the main system is baked out at 250C for 24 hours. The main system is then reduced to 150C or the current activation temperature for the NEG, whichever is lower, whilst we degass all the gauges. After this the NEG components are raised to the same temperature and we allow the system to reach equilibrium. After this, the NEG section is raised to its activation temperature and after reaching this, the main non coated system is reduced to room temperature. After activation for 24 hours, the NEG section is reduced to room temperature as well. By injecting a known volume of a gas mixture into one end of the NEG tube to be measured, we can record the partial pressures of the components at both ends of the tube to determine the sticking probability for that species using Mollflow simulations.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

10. ECLOUD2018 WORKSHOP ELBA, ITALY
If we compare the results of 1mm copper with NEG shown here in Black, against Columnar NEG shown here in Red, we can see that the pumping for CO2 shown by triangles, is comparable or better than that for the columnar structure, whilst CO, shown as squares is slightly lower. hydrogen, shown as circles is about a decade lower. The onset of activation also occurs for the Cu-NEG at a slightly higher temperature. In all cases, the results are better than the green lines, which represent a NEG coating, but in a dense, non columnar coating.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

11. ECLOUD2018 WORKSHOP ELBA, ITALY
In terms of capacity, the columnar 1 mm Cu-Neg film is comparable to that of the dense NEG.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

12. ECLOUD2018 WORKSHOP ELBA, ITALY
XPS SPECTRA OF 1 mm Cu-NEG AS RECEIVED
XPS spectra of the as received 1mm diameter Copper wire and NEG shows an oxide state with a top layer of carbon, CO and CO2 at the surface. The dominate metal within the range of the XPS sample depth is Ti and no Hf is detected on this witness sample.
The surface is at an oxide state with a top layer of carbon with CO and CO2 trapped at the surface.
the dominate metal within the sample depth is Ti and no Hf has been detected.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

13. ECLOUD2018 WORKSHOP ELBA, ITALY
XPS SPECTRA OF 1 mm CU-NEG HEATED TO 150 C
After activation to 150C for 12 hours, we can see some activation has taken place. Copper and Zr are still in their oxide state, but Vanadium and Titanium have gone through partial activation. For vanadium, as well as surface activation, there has also been an increase in surface concentration. For Ti 458 to 454 for V 516 to 512eV
After heating for 12 hours at 150 C some activation has taken place.
Both Cu (CU auger peak shape) and Zr (No shift in the Binding energy) remained in the oxide state.
Both Vanadium and Titanium has gone through partial activation with V being more effected.
For Vanadium as well as surface activation there is also an increase in surface concentration
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

14. ECLOUD2018 WORKSHOP ELBA, ITALY
XPS SPECTRA OF 2 mm Cu-NEG AS RECEIVED
For the 2mm Copper Neg, we see a similar oxide state, with a top layer of Carbon with CO and CO2 trapped on the surface. Zr and hf can be seen at trace levels.
The surface is in an oxide state with a top layer of carbon with CO and CO2 trapped at the surface.
The dominate metal within the sample depth is Cu which is in Cu(II) state (presence of satellite).
Both Zr and Hf are at trace level of detection.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

15. ECLOUD2018 WORKSHOP ELBA, ITALY
XPS SPECTRA OF 2 mm CU-NEG HEATED TO 230 C
After heating to 230C for 20hours, partial activation has taken place. Hf and Zr remain predominately in their oxide states. Vanadium and Titatinium have gone through activation and become the dominant metals at the surface. Copper two has reduced to a mixture of copper one and copper zero, with a reduced concentration at the surface.
After heating for 20 hours at 230 C partial activation has taken place.
Hf and Zr predominantly remained in the oxide state.
Both Vanadium and Titanium has gone through activation and became the most dominant metal at surface.
Cu(II) has been reduced to mixture of Cu(I) and Cu(0) with much reduced concentration at the surface.
There is also presence of metal carbide.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

16. ECLOUD2018 WORKSHOP ELBA, ITALY
Planar SEM of Cu-NEG on Silicon and Glass
The film are grown in nm diameter perpendicular columns with 5-10nm separation.
Aspect ratio of 100 can be easily achieved to give high pumping speed.
The composition depends on the position of the sample with respect to elemental wire in the twisted wire.
SEM of the Copper Neg shows the film has grown in perpendicular columns of 20 to 40 nano meters diameters with a separation of about 5 to 10 nano meters. Aspect ratios of 100 are easily achieved which lends itself to high pumping speeds. As can be seen by comparing the XPS data from Silicon and glass, the surface film composition depends on the position of the sample relative to the elemental wires in the twisted wire target. Zr Ti V Hf Cu
4.6
39.2
49.4
2.2 Zr Ti V Hf Cu
2.6
53.5
33.3
4.5
4.6
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

17. ECLOUD2018 WORKSHOP ELBA, ITALY
X-SECTION SEM OF Cu-NEG on Si
The cross section SEM on Silicon shows the substrate is around 2 micron thick. Hafinium and copper being the most predominate elements through the thickness of the film. There is also a high concentration of Copper at the interface.
The film grown on Silicon substrate is 2 micron thick.
Hafnium and Copper are most predominate element in the film through the thickness.
There is a high copper concentration at the interface.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

18. ECLOUD2018 WORKSHOP ELBA, ITALY
SHEET RESISTIVITY OF CU-NEG VS NEG
Rs (Ω/□) for NEG
Rs (Ω/□) for NEG +1 mm Cu
Rs (Ω/□) for NEG +2mm Cu
Si/NEG film -
7.00
1.34
Glass/NEG film
15.27
7.07
2.41
Rs = X (V/I)
Four point probe measurements of the samples show a sheet resistance for the NEG material to be around 15 ohms per square. This drops to around 7 with the NEG and 1mm copper wire and to around 1.3 to 2.4 for the NEG with 2mm copper wire depending on Silicon or glass.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

19. ECLOUD2018 WORKSHOP ELBA, ITALY
Here we can see the SEY results, taken as received, and after heating to 150C. You can see that it has dropped from as received of about 1.7 to about 1.1 after activation
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

20. ECLOUD2018 WORKSHOP ELBA, ITALY
And similarly for the 2mm diameter copper wire. After activation to 150C, the SEY has dropped from 1.4 to slightly above 1.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY

21. ECLOUD2018 WORKSHOP ELBA, ITALY
CONCLUSION
It is possible to reduce NEG sheet resistance (impedance) by adding a more conducting material to the conventional ternary/quaternary NEG.
The higher the copper concentration the lower the sheet resistance.
The decrease in resistance was in expense of NEG pumping properties, whilst keeping the expected SEY of around 1 for activated NEG.
The pumping properties for the Cu-NEG (1mm copper wire) was comparable to conventional NEG for CO and CO2 but hydrogen sticking probability was reduced by order of magnitude.
The activation temperature is also increased from 160 C to 200 C.
The surface composition changes with thermal activation with more reactive metal (V and Ti) diffusing to the surface.
Both samples 1mm and 2mm Cu Neg have shown low SEY of around 1 after activation.
Although the film composition is uniform in depth but there seems to be large variation in lateral direction which means better target preparation should be employed.
More study is needed to optimise the conductance versus Pumping properties and direct measurement of impedance rather than sheet resistivity.
Thank you for your attention.
A.N Hannah
ECLOUD2018 WORKSHOP ELBA, ITALY