First Observation of Luminosity-Driven Extraction Using Channeling with a Bent Crystal

Luminosity-driven channeling extraction has been observed for the first time from a 900 GeV circulating proton beam at the Fermilab Tevatron. The extraction efficiency was found to be of the order of 30%. A 150 kHz beam was obtained during luminosity-driven extraction with a tolerable background rate at the collider experiments, and a 900 kHz beam was obtained when background limits were doubled. This is the highest energy at which channeling has been observed. PACS numbers: 29.27.A~ Since the original suggestion of bent crystal channeling [l] there has been interest in exploiting the technique for accelerator extraction. While the planar channeling critical angle is small, 5.8 prad at 900 GeV for the Si( 111) plane compared to the Tevatron beam divergence of -10 p rad, this is less of a limitation than might be thought. Many unchanneled particles multiple scatter in the crystal and remain in the accelerator to channel on

One crystal was mounted in a goniometer with four degrees of freedom so that it could be translated and rotated with small step sizes.The crystal was cut so that the (111) atomic plane was parallel with the top optical surface of the crystal.
The beam side was optically flat.The 3 9 mm long, 3 mm high, 9 mm wide crystal was bent through a vertical angle of 6 4 2+5 prad with a four point bender (see Fig. 1).The angle of the (111) plane was prealigned to within 300 prad of the nominal beam angle.Several mechanisms were available to drive halo beam onto the crystal.A fast kicker magnet at El7 could provide transverse kicks of 0.5 mm at the crystal for an individual bunch.
Results of these studies have already been published [6].Noise sources such as beam-gas scattering, power supply modulation, and magnetic field non-linearities also produced beam growth, called natural diffusion.
Diffusion could be stimulated with an RF electrical horizontal damper.Most significantly, proton-antiproton collisions at the collider detectors created halo.In operation the crystal was gradually moved horizontally into the halo from the outside of the ring.Note that in contrast with the CERN experiment, the crystal moved into the beam in the horizontal plane, but bent the beam up, so that any lack of parallelism between the atomic planes and the top optical surface would not reduce the extraction efficiency.The simulation [7] predicts a oV of 21 to 24 prad compared to the 32 prad measured in Fig. 3.
We The maximum extraction rate achieved was 150 kHz.I n this mode the limitation was the impact of particles scattered from the crystal in creating backgrounds for the operating collider experiments.The DO "lost protons" monitor, which was l/6 of the ring downstream from the crystal, reached the limit set by that experiment at an extraction rate between 50 and 150 kHz.
The crystal edge was between 5 and 8 times o, from the beam center.
This limitation was removed during a special store with 36 proton bunches and 3 antiproton bunches during which DO was not taking data.There were 3 x 10 12 protons circulating, and an extraction rate of 900 kHz was achieved.
The DO lost proton monitor exceeded its upper limit by a factor of two.
During that same store, the extraction rate was also studied as a function of luminosity.
Only with the crystal when its vertical angle is not aligned to the beam with the number that interact when it is correctly aligned for maximum channeling.Fewer interactions are observed when the crystal is well aligned with the beam because the channeled protons do not come close to nuclei [8].
We call this the "channeling efficiency" and define it as the difference between the aligned and unaligned interaction monitor rate, divided by the unaligned rate.The "surface loss" mentioned above does not lower this efficiency, and the dechanneling losses contribute only partially (once a proton has dechanneled after channeling through part of the crystal, it has less than 8.8% probability of a nuclear interaction).Thus we expect this efficiency to be slightly higher than the extraction efficiency.In operation, the interaction counter rates were sensitive to fluctuations arising from such effects as small horizontal deviations of the circulating beam.
Some of these effects could change in an unpredictable way in the time it took to do a typical 0, scan.To mitigate this time dependence, the best measurements were obtained by moving the crystal quickly back and forth from an aligned to a very unaligned vertical angle.
An example of such data is shown in Fig. 3

(top).
These data were taken within minutes after the 0, scan shown in Fig. 3 (bottom).
The crystal was moved repeatedly to three different angles, one at the peak of the 0, scan, and one angle each in the left and right tail of the 0, scan.
In two stores in which the extraction was luminosity driven, the weighted average channeling efficiency was 3 2+4%.During the 84-bunch proton-only fill, the efficiency was 32*15%.The errors in these efficiences are derived from the rms scatter of the many data points about their average value [9].The simulation [7] predicted an extraction efficiency of 35% for a realistic crystal.

Fig. 2
Fig. 2 shows a vertical beam profile obtained with a finger counter scan.The beam width was crV = 0.25 mm after correcting for the height of the finger counter.The width expected, based on the critical angle and the beam optics, was oV = 0.23 mm.A tail is visible below the beam resulting from such factors as horizontal misalignment and dechanneling.The bottom of the tail was cut off by the Lambertson magnets.The number of particles in the visible tail is 20% of the peak, twice that expected on the basis of a simulation of the experiment [7].The crystal was aligned to the circulating beam by scanning the crystal through the vertical angle 0,.Fig. 3 (bottom) shows a two air gap scintillator coincidence distribution.The width of the 0, distribution for diffusion mode is due mostly to the effect of multiple scattering from crystal multiple passes convoluted with the angular distribution of the circulating beam and the critical angle.The simulation[7] predicts a oV of 21 to 24 prad compared to the 32 prad measured in Fig.3.We have measured extraction rates under three conditions: extraction driven by natural diffusion during proton-only stores, RF noise-driven diffusion during a proton-only store, and luminosity-driven extraction during proton-antiproton stores.In a typical proton-only store, 1 Ott protons were circulating in six bunches.The maximum extraction rate achieved was 200 kHz.Higher rates could have been achieved by moving the crystal even closer to the beam, but with only six bunches, a rate of 287 kHz corresponded to extracting on average one proton per bunch, and the counters could not count more than one particle per bunch.The crystal edge was between 4 and 6 times the beam width (0,) from the beam center.To mitigate this limitation, a special proton-only store was arranged with 1011 protons circulating in 84 bunches.Additional diffusion was induced by transverse RF horizontal noise using an electrical damper, creating an rms diffusion rate at the crystal of 0.023 pm per turn.The extraction rate achieved was greater than 450 kHz.In the luminosity-driven stores, typically 1012 protons were circulating in six bunches.The maximum extraction rate achieved was 150 kHz.I n this mode the limitation was the impact of particles scattered from the crystal in creating backgrounds for the operating collider experiments.The DO "lost protons" monitor, which was l/6 of the ring downstream from the crystal, reached the limit set by that experiment at an extraction rate between 50 and 150 kHz.The crystal edge was between 5 and 8 times o, from the beam center.This limitation was removed during a special store with 36 proton bunches and 3 antiproton bunches during which DO was not taking data.There were 3 x 10 12 protons circulating, and an extraction rate of 900 kHz protons are detected with a system of scintillators in two air gaps separated by 40 m.The inset shows the location of the crystal extraction system, the fast kicker, the RF damper, and the collider experiments at BO and DO. 2. Vertical profile of the extracted beam taken with a thin finger counter.Note the tail extending below the main peak.To better illustrate this tail, the values have been multiplied by 20 and replotted (diamonds see right ordinate).

3 .
The lower data set (right ordinate) is the counting rate in a coincidence between scintillators in the two air gaps as the vertical angle of the crystal was varied.The curve is a fit to a Gaussian plus a flat background.The upper data set (left ordinate) is the counting rate in the interaction monitor at three different vertical angles.The curve is a Gaussian with the same width and central value as the curve in the lower half of the figure.

5The University of New Mexico, Albuquerque, NM 87131 6 Petersburg Nuclear Physics Institute, Gatchina, Russia 7tnstitute for High Energy Physics, Serpukhov, Russia 8Superconducting Super Collider, Dallas, TX 75237 9The University of Texas, Austin, TX 78712 lOThe University of Texas Southwestern Medical Center at Dallas, Dallas, TX 75235 ' ' Vanderbilt University, Nashville, TN 37235 ' 2The University of Virginia, Charlottesville, VA 22901
critical angle is small, 5.8 prad at 900 GeV for the Si( 111) plane compared to the Tevatron beam divergence of -10 p rad, this is less of a limitation than might be thought.