Commissioning of the ThomX Storage Ring

We will report on the ongoing ThomX ring commissioning, its status, its main challenges, our results and our planning. ThomX is a compact Compton-based X-ray source under commissioning at IJCLab in Orsay (France). This facility is composed of a 50-70 MeV linac, a transfer line and a storage ring whose circumference is 18 m. Compton scattering between the 50 MeV electron bunch of 1 nC and the 30 mJ laser pulses stacked in a Fabry-Perot cavity will result in the production of X-rays with energy ranging between 45 keV and 90 keV. We aim at a total flux of about 1013 X-rays per second. The injector commissioning started in the spring of 2021. The ongoing storage ring commissioning faces many challenges due to the rings low energy, its compactness, its non-linear beam dynamics, the time-limited beam storage and the need to achieve a very accurate and stable geometry of the collision region between the laser pulses and the electron bunch. The commissioning and operational experience is of great importance for the future Compton sources.


Introduction
Authorisation from French Nuclear Safety Authority for commissioning of the ThomX Transfer Line (TL) and the Storage Ring (SR) was obtained in August 2022 allowing us in this phase to operate the SR at 10 Hz injection repetition rate and electron bunch charge up to 100 pC.
The transfer line is used to transport the electron bunch from the linac to the SR.At injection electron beam encounters on its path (Fig. 1) an injection dipole (1), a septum (2) to use on-axis injection, and a kicker (3) to put the beam on the stored orbit.ThomX ring employs a double bend achromat optics with a two-fold symmetry [1].The main parameters are presented in Table 1.Ring diagnostics [2] is based on 12 BPMs, processed by Libera Brilliance+ electronics, which provide the beam position information for both single pass data and turn-by-turn data averaged over two turns.In addition, there are beam loss monitors and synchrotron radiation monitor.

Beam injection and first turns
Measured electron beam parameters at the injection are the following: 50 MeV energy, 100 pC charge and 4 mm mRad emittance.The measurements of the injected bunch length is in progress.Once transfer line optics parameters for e − beam transport were determined, we started to experiment with the e − beam injection in the ring by monitoring the signal at a BPM  located just after the septum.In order to facilitate the finding of the correct parameters for the first injections we installed additional LaneX screen near the beampipe in the vicinity of the injection dipole (DPI).In this way one could observe if the beam is lost due to too high or too low DPI current.To detect injection losses we installed a Fiber Beam Loss Monitor (FBLM) along the injection path, and scintillator at the septum exit.We manually scanned the currents of the DPI, septum and transfer line correctors to obtain injection.On September 9th, the first beam injection was successfully observed and detected at BPM1.Afterwards, we turned on the ring dipoles and set them to the current of the TL dipoles.This resulted in the signal read by the BMP2, BPM3.In such a way, the beam reached the Interaction Point (IP) or, in other words, made quarter of a turn, as it was predicted by the simulations.
With the goal of beam threading we turned on the ring quadrupoles to nominal values obtained from the model, the half turn was achieved.Manually adjusting the SR correctors one by one and maximizing the BPM signal, we established the beam transport untill septum, making the first turn on September 19th.Losses generated at the septum were detected with the scintillator detector, and showed two distinct signals spaced by 60 ns, which equals to one revolution period of the ThomX SR.
Although the orbit corrections were performed by applying the SVD to an open line model of the ring, the beam could not be stored for more than one turn.With the FBLM at the last arc of the ring (3/4) we detected important losses.The origin of this behavior was wrong polarity in one of the quadrupoles in this arc.Once quadrupole polarity was fixed and orbit correction was preformed, we achieved beam storage for more than 300 turns in the ring without power in the RF Cavity (3rd October 2022)2 .Beam storage was improved by scaling of the focusing and defocusing families of quadrupoles with small steps of ±0.1%.
Injector beam parameters were not stable and varied from shift to shift.To optimize and facilitate injection search we applied Robust Conjugate Direction Search (RCDS) method [3].Set of 7 parameters among which are two pairs of transfer line correctors (horizontal and vertical), DPI, septum and kicker currents is varied by RCDS in order to maximize objective function, which in our case the sum of turn by turn sum signals from all electrodes for the first and last BPM in the ring: (T BT 1 + T BT 12 ).In less than 15 minutes, with 500 iterations of the algorithm, we found systematically optimal injection parameters.Cases of non-converging algorithm signal some possible hardware issues.

Beam storage and first measurements
First turns and orbit corrections of an electron storage ring are a complex and time-consuming process being essential for ensuring the proper functioning of the machine.Orbit correction with SVD was performed to minimize closed orbit in the horizontal and the vertical planes independently (see Fig. 2).The maximum deviation is 2 mm and 2.2 mm for horizontal and vertical orbit respectively, with std measured from turn-by-turn data at different injections of about 0.08 mm.
The RF cavity was completely conditioned in late fall 2022 [4].For ThomX SR, radiation damping is negligible during storage time, and main role of the RF cavity is to preserve beam quality, namely the electron bunch length and longitudinal emittance.
With the RF cavity turned on, December 8th we achieved beam storage for more than 100 ms.During the RF frequency and phase search we found a significant offset of the resonance frequency, which turns out to be 500.4MHz, whereas the design value is 500.02MHz.Such offset has to be understood.
Synchrotron oscillations (see Fig. 3 ) are measured at different power stored in RF cavity and matched fairly well with the one estimated from the model of the SR.Precisely, at ∼10 kV stored in the RF cavity, we observe synchrotron oscillations with a period of 248 turns, which is close to one predicted by the model (272 turns); at ∼40 KV stored in the RF cavity, the measured period of the synchrotron oscillations is 151 turns and the one given by the model is 136 turns.
Measurement of tunes is crucial for circular machines.We do not observe significant betatron oscillations after injection, which makes tune measurement non reliable.Therefore we used extraction kicker to excite transverse beam oscillations in horizontal plane (see Fig. 4).Beam is excited after turn 524, and with harmonic analysis by FFT the fractional part is measured to be 0.18.One can also observe 3 oscillation along the ring orbit after such excitation, which is in total close to the model value of the horizontal tune of 3.17.Our current tune measurement is limited due to fast decoherence of the betatron oscillations.A simple model shows, that if this fast decay of the coherence of the betatron oscillation is due to radial wave number spread, this spread should have a standard deviation of about 0.01.

Summary and Outlook
The first part of the commissioning of the THOMX Ring, starting from injection, beam threading and first turns, beam storage and orbit corrections was successfully performed.During this commissioning multiple issues were diagnosed and resolved, such as quadrupole polarity inversion, overvoltage in steerers, wrong BPMs cabling, synchronization problems, injector instabilities, replacement of a broken klystron (on November 9th).These issues resulted in the commissioning delays.We operate at 10 Hz and stored the beam for more than 100 ms with the RF cavity switched on.Measured values of the horizontal tune integer and fractional parts are in good agreement with the design value, and allows us to conclude that there no big errors in the machine optics.A few observations still need to be understood, such as the decoherence of betatron oscillations, measurement of vertical tune, big difference in the operating frequency of the RF Cavity with respect to the nominal one.
We found very useful to have additional diagnostics for beam losses like a scintillator and a FBLM, which can be easily located in the places of interest in the ring, and can be directly monitored with an o-scope without passing by control and acquisition system.
Currently we entering to phase of precise and accurate measurements, leading to a better understanding of the behavior of the beam and the storage ring.Soon we will perform fine machine physics studies, BBA, chromaticity measurements, LOCO, and beam optimization at the interaction point where the electron beam meets with laser beam from FP cavity are foreseen.

Figure 1 .
Figure 1.ThomX ring layout.On axis injection in the ring is done by pulsed magnets: septum (2) and fast injection kicker (3).BMP1 located 30 cm downstream of septum is used in search of injection.

Figure 2 .
Figure 2. Top: measured horizontal and vertical closed orbit.Orbit correction is performed with the theoretical response matrix from the model.Sextupoles are off.Bottom: ThomX SR optics.

Figure 3 .
Figure 3. Top: synchrotron oscillations observed in turn-by-turn data at BPM3.At the RF cavity voltage of 40 kV (10 kV ), the oscillation period is 151 (245) turns compared to the values given by the model 136 (272) turns.Bottom: beam is stored with RF cavity switched on for more than 100000 turns and only for 300 turns (measured by the Libera electronics), when the cavity is not powered.

Figure 4 .
Figure 4. Top: beam oscillations observed in turn-by-turn data at BPM3.Low frequency radial oscillations are due to synchrotron oscillations.High frequency oscillations follow an excitation of coherent betatron oscillations with extraction kicker at turn 524.Bottom: comparison of the FFT performed on the of the model turn-by-turn data (100 turns) including the noise and averaged over 2 turns with the FFT performed on the measured data.Fractional part of the tune is measured to be 0.36/2 = 0.18, which is consistent with the theoretical value.