A new NEG coating setup with travelling thin solenoids for the SLS 2.0 complex vacuum chambers

The 288 m long Storage Ring of Swiss Light Source 2.0 (SLS 2.0) consists of several vacuum chambers with unique geometries. Complicated features, with many changes in the cross sections, are essential to provide the best impedance matching, and to allow synchrotron light extraction under the tight geometrical constraints. In order to speed up the commissioning time, it was decided to coat most of the vacuum chambers with non-evaporable getter (NEG) material. A new magnetron sputtering setup has been developed in Paul Scherrer Institut (PSI), where the plasma length, defined by thin solenoids, is relatively small. The solenoids are continuously travelling over the entire vacuum chambers more than hundred times per coating process with an average speed of 4 mm/s to assure the best possible thickness uniformity and to limit the cathode heat up. Flexibility provided by this solution allows to coat various vacuum vessels in one assembly. This paper will describe this NEG coating setup and show results on SLS 2.0 vacuum chambers.


SLS2.0 Vacuum System overview
Multi Bend Achromat (MBA) lattices are under implementation in many Synchrotron Light Sources worldwide.Those machines are characterized by extremely low emittance, but pose certain challenges for all the auxiliary systems [1].The vacuum chambers must fit inside the dense magnet layout, which results in small conductance limiting the effects of discrete pumping.Pressure rise caused by dynamic effects, mainly Photon Stimulated Desorption (PSD) affects the commissioning of a new storage ring by limiting the beam current increase.Since over two decades, the NEG coating technology has proven to reduce PSD and improve the ultimate pressure [2,3].
The SLS 2.0 lattice consists of 12 arcs and straight sections, resulting in over 15 different vacuum chamber geometries.The chambers for bending magnets will be manufactured internally at PSI.Their main characteristics are: Oxygen-Free Copper (OFE) as base material, minimal wall thickness of 1 mm, octagonal cross section of 18 mm and antechamber with slits as small as 3 mm for light extraction.The requirements of the NEG coating for those chambers have been chosen as follows: thickness: 500 nm ± 40 % over the whole electron channel, good adhesion (no signs of peel-offs or delayering), uniform composition of Ti/V/Zr (10-50 %), columnar structure.To achieve the uniform coating across the variable cross-section of the chambers (with examples shown in the figure 1), the adaptive coating conditions were needed.The magnetron sputtering setup developed at PSI consists of two thin (85 mm each) solenoids that can move along the 2.50 m setup.The idea of changing the position of the magnetic field while coating is already used by several institution, for instance by SAES Getters [4].At PSI, we have chosen to move the plasma along the geometry relatively fast (more than a hundred times over the entire chamber length).The advantage of this solution is that i) there is less local heat up of the cathode and ii) the short plasma length (around 20 cm) adapts better to the changing geometry as it moves along the chamber.As a consequence, a better thickness uniformity and transverse coverage can be achieved.Moreover, the coating parameters such as magnetic field can be adjusted during the coating process for each cross-section individually.

The Coating Setup
The Coating setup has been schematically presented in the figure 2. Its main components and their function will be briefly described.(2) Moving Solenoid Unit with two magnets of ø in =340 mm t=85 mm each, traveling along a linear spindle with a constant speed.Movement is executed by a servomotor that regulates the velocity and keeps it constant during the whole process.Maximal range of motion spread between the two limit switches is 2.50 m.
(3) Two cathodes inside a vacuum chamber -3 wires of Zr, V and Ti are twisted together to form a 3 mm bundle.It is held inside the top cross by an electrical insulator (macor).At the bottom, they are pulled downwards by the two mass loads.This helps to keep them tensioned, which is important to compensate the thermal dilatation during the coating process.The cathodes are centered along the curvature with small spacers installed length-wise that assure the desired cathode position, presented in the figure 3. (4) Bottom double cross containing the mass load and connected to the pumping station (Turbopump HiPace 300 l/s).The separation of inlet and outlet assures the continuous flow of Krypton and desorbed contaminants removal despite the spacers that limit the conductance.Figure 3.A Spacer at the inlet of the election chamber guiding the Ti-Zr-V cathode through its middle

The Golden Parameters
In order to efficiently coat hundreds of the chambers, a set of standard coating parameters needed to be established.Firstly, a round 19 mm tube was coated with one solenoid at fixed position.Samples cut off at different positions have been measured with Focused Ion Beam (FIB) and Scanning Electron Microscopy (SEM) techniques.The morphology of the NEG coating has been con-firmed to be columnar, with the grain sizes of 5-20 nm, which was a satisfactory result.The thickness of each sample has been measured (as shown in the figure 4).The values were used as reference values for calibration of our X-Ray Fluorescent (XRF) measuring system.This technique was then used for the further assessment of the thickness and composition of the following NEG coatings as a fast and efficient alternative to FIB/SEM.Furthermore, the measured samples thicknesses have been plotted against the solenoid position.It has been confirmed, that the thickness follows a Gaussian distribution and the total volume of deposited coating was around 15 mm 3 in the condition as specified in the figure 5.This input helped to assess total time needed to coat the full length of a chamber.Following tests on the round pipe with moving solenoid helped to establish the optimal coating speed.Subsequently, two half shells were used to imitate the actual geometry of the chamber, as previously done in MAX IV [5].Multiple coating iterations were performed.After each, the coating quality was visually assessed and the main parameters (thickness and composition) were measured.Root cause of the imperfections, such as delayering or poor uniformity, were assessed and corrected in the following trials.Eventually, the Golden Parameters for the coating were established as a standard for the SLS2.0 chambers: 0.03 mbar Kr; 20 mA ; 400 G ; 200-300 V ; 16 hours ; 160 passes for 1.5 m long chamber.
The final test of the half shells, has provided a full coverage and perfect adhesion.As showed in the figure 6, the thickness was not uniform radially, with the minimum of 100 nm in the slit.However, the measurement inside the electron chamber was distributed around 400 nm which is within the SLS 2.0 specification.As final step, FIB measurement were repeated to ensure the good calibration of XRF system and to calculate the final deposition rate that for the double solenoid was 0.1 µm per hour of coating (at the peak of the magnetic field).

First coating of the SLS 2.0 Vacuum Chambers
With those parameters, a first real vacuum chambers for the future SLS 2.0 storage ring has been coated.The adhesion and thickness were measured from a witness sample installed nearby the chamber inlet.The thickness of 400 nm and V/Zr/Ti composition of 25/30/45% was achieved.The chamber was visually tested with an endoscope.No signs of peel-offs or uncoated areas were found.The witness sample was also measured with X-ray Photo-electron Spectroscopy (XPS) method to assess the activation of the coating.As shown in the figure 7, the oxygen concentration has been reduced with the increased temperature, reaching about 45 % at 200 °C.Finally, the full chamber was activated to verify the pumping capabilities of the coating in practice.The NEG-coated vacuum chamber was baked-out for over 48 h with 150 °C followed by 24 h activation at 200 °C, as presented in the figure 8.At the end of the cool down ramp, the copper chamber was isolated from the system and the pressures were measured at the both end of the chambers with the penning gauges.
A pressure drop below 1.0×10 −10 mbar has been observed.Those values have not changed for a week after the isolation proving the successful activation of the NEG layer.The downstream side is equipped with a steel cube where in the real arc an Ion Pump, a NEG Pump and a crotch absorber will be mounted.This relatively large uncoated area served as an outgassing source.In consequence, pressure p 2 according to the figure 9 was one order of magnitude higher (1.0×10 −10 mbar) than at the opposite extremity, p 1 (4.0×10 −11 mbar).The chamber was vented with N 2 and exposed to the air for the two cycles, each time producing the ultimate pressures of similar magnitudes.

Future Perspectives
Until the end of the year 2024, over a hundred of those complex bending magnet chambers need to be NEG coated using the described setup.The main focus is now to develop an efficient handling procedure.The exact activation procedure that will be executed externally to the ring tunnel prior to the installation is also under assessment.It is crucial to assure the best conditions of pumping speed, low PSD yield and capacity from the very beginning, as the activation cannot be repeated in-situ after the installation [1].The strategy of exposing the sectors to the atmosphere during upgrades or repair works is also under assessment.
Moreover, the insertion devices and the injection line for the SLS2.0 are still under development.Few of those components, such as 2 m long, 9 mm thin, aluminum Apple-X 14th International Particle Accelerator Conference Journal of Physics: Conference Series 2687 (2024) 082028 undulator vacuum chamber is foreseen to be NEG coated using the same setup.The parameters for each of the chambers will be investigated using similar iterative process as for the bending magnet chambers.

Conclusions
We have built a NEG coating setup with two thin moving solenoids.Through a set of test on tubes and dummy chambers, we have gained the experience and found the Golden Parameters that assure good coating uniformity across the several rapidly changing crosssections.Subsequently, we have proved the maturity of the setup by coating a real vacuum chamber of a complex geometry.The coating demonstrated the pumping capabilities after activation with 200 °C.

Figure 1 .
Figure 1.Various cross-sections of one, 1.4m long Normal Bend Chamber (from the left: inlet, middle and outlet)

Figure 4 .
Figure 4. First coated NEG sample thickness measured with FIB/SEM

Figure 5 .
Figure 5.The NEG thickness along one solenoid after coating for 11 h with 200 G and 15.5 mA at 0.1 mbar.

Figure 6 .
Figure 6.Two half shells mimicking the chamber geometry and the NEG thickness distribution measured with XRF, the results of the last iteration of coating.

Figure 7 .
Figure 7. XPS measurement of the witness sample from the first NEG coated SLS 2.0 sample.