Dust-induced beam losses in the cryogenic arcs of the CERN Large Hadron Collider

The interaction of dust particles with the LHC proton beams accounts for a major fraction of irregular beam loss events observed in LHC physics operation. The events cease after a few beam revolutions because of the expulsion of dust particles from the beam once they become ionized in the transverse beam tails. Despite the transient nature of these events, the resulting beam losses can trigger beam aborts or provoke quenches of superconducting magnets. In this paper, we study the characteristics of beam-dust particle interactions in the cryogenic arcs by reconstructing key observables like nuclear collision rates, loss durations and integral losses per event. The study is based on events recorded during 6.5 TeV operation with stored beam intensities of up to $\sim 3\times {10}^{14}$ protons per beam. We show that inelastic collision rates can reach almost ${10}^{12}$ collisions per second, resulting in a loss of up to $\sim 1.6\times {10}^8$ protons per event. We demonstrate that the experimental distributions and their dependence on beam parameters can be described quantitatively by a previously developed simulation model if dust particles are assumed to be attracted by the beam. The latter finding is consistent with recent time profile studies and yields further evidence that dust particles carry a negative charge when entering the beam. We also develop different hypotheses regarding the absence of higher-loss events in the measurements, although such events are theoretically not excluded by the simulation model. The results provide grounds for predicting dust-induced beam losses in the presence of higher-intensity beams in future runs of the High-Luminosity LHC.

Figure 1

Illustration of the LHC layout, showing the different insertion regions (IRs—yellow), dispersion suppressors (DS—dark gray) and arcs (gray). Each of the eight arcs consists of 23 cells. A cell is made up of two 53.45 m long half-cells (see top), which are composed of three bending dipoles (MBs), one quadrupole (MQ) and corrector magnets (not shown). The two counterrotating beams are represented by the blue line (beam 1) and the red line (beam 2), respectively. The beams are crossing each other in four interaction points.

Figure 2

Illustration of BLM positions in a standard LHC arc half-cell. The two layout plots at the bottom show a top view of the different BLM positions in run I and run II, respectively. Gray boxes represent magnets (bending dipoles: MB, quadrupoles: MQ), yellow boxes represent monitors. Monitors around the quadrupole are located on the horizontal plane (figure top right), while monitors between dipoles (run II only) are located on top of the dipole interconnects (figure top left). The blue and red arrows indicate the beam direction of beam 1 and beam 2, respectively.

Figure 3

Geometry model of a LHC arc cell used in the particle shower simulations. The picture shows the vicinity of a quadrupole. The cryostat and dipole next to the quadrupole are cut open to show the interior. The beam loss monitors of the quadrupole (yellow cylinders) are mounted on the outside of the cryostat.

Figure 4

Comparison of simulated BLM dose patterns for beam collisions with dust particles of different composition. The proton beam energy was 6.5 TeV. The dose is given per proton-nucleus collision. The assumed loss location is indicated by a cross on the $s$ axis. The beam direction is from the left to the right. The statistical error of the simulations is less than 3% for the highest signal, but can be as large as few 100% for the smallest signals. The data points are connected by lines to guide the eye.

Figure 5

Comparison of measured and simulated BLM dose patterns along LHC arc cells for dust-induced loss events in run II physics operation at 6.5 TeV. In four out of the six cases a dipole quench occurred (labels of dipoles which quenched due to showers are in bold red). The simulated patterns were scaled according to Eq. (1) to obtain the best match with the measurements. The scaling factors, which represent estimates of the inelastic proton-dust particle collisions, are indicated in the labels. The data points are connected by lines to guide the eye. The estimated $s$ location of the dust particles is identified by “x” labels.

Figure 6

Rate of inelastic nuclear collisions ${\dot{n}}_i(t)$ between 6.5 TeV protons and a dust particle entering the LHC beam. The two events were recorded in the LHC arcs in run II. In both cases, the losses lead to a magnet quench. The loss rate was reconstructed from BLM measurements using particle shower simulations. The time resolution of the measurements is $40\mu \mathrm{s}$. The labels in the figure indicate the integral number of collisions per event ($N_i$), the maximum collision rate ($R_i$) and the loss duration $\tau$. The different variables are discussed in more detail in Sec. 3a.

Figure 7

Measured time profile of a dust event recorded in LHC operation (top) and Gaussian profile used for estimating the loss duration $\tau$ (bottom). The standard deviation of the Gaussian distribution was calculated from the relative number of collisions $N_i^{80\mu \mathrm{s}}/N_i$ contained in the central $80\mu \mathrm{s}$ time window. The relative number of collisions in this interval was estimated from the measurements (ratio of red area and total area in top graph).

Figure 8

Reconstructed experimental distributions of dust events as a function of collisions per event (top), maximum collision rate (center), and loss duration (bottom). The figure considers about $N=5500$ loss events, which were recorded in the LHC arcs during 6.5 TeV proton operation in run II. In addition, one dust-induced quench at 6.39 TeV and one quench in the dispersion suppressor are included (at 6.5 TeV). The two bottom figures show different subsets of events exceeding a minimum number of collisions indicated by the vertical lines in the top figure. Events leading to a quench are represented by the dashed histograms.

Figure 9

Scatter plots of randomly sampled dust events based upon the simulation model described in Ref. [22]. The graphs show the number of inelastic proton-dust particle collisions per event versus the maximum collision rate. The color coding indicates the loss duration (left), the charge-to-mass ratio (center) and the radius (right). Only events with $\ge 5\times {10}^5$ collisions within $640\mu \mathrm{s}$ are shown. See the text for more details about the simulation parameters. The horizontal and vertical lines indicate the maximum $N_i$ and $R_i$ values observed in the experimental distributions in Fig. 8.

Figure 10

Measured and simulated distributions of dust-induced loss events as a function of collisions per event (top), maximum collision rate (center), and loss duration (bottom). The measurements are the same as in Fig. 9, including about $N=5500$ events. The simulation results correspond to different intervals of $|Q/m|$ and different maximum radii (see top of the figure).

Figure 11

Number of inelastic proton-nucleus collisions (top), maximum collision rate (center) and loss duration (bottom) of dust particle events as a function of the beam intensity. Every dot represents a loss event reconstructed from BLM measurements in run II. All considered events occurred at 6.5 TeV, except one event at 6.39 TeV. Only events with more than $5\times {10}^5$ collisions within $640\mu \mathrm{s}$ were considered. The crosses indicate events, which resulted in a magnet quench. The average and median values are indicated by the solid and dashed lines, respectively (beam 1 in blue and beam 2 in red). The yellow lines represent the simulation results (simulation C in Fig. 10).

Figure 12

Number of inelastic proton-nucleus collisions (top), maximum collision rate (center) and loss duration (bottom) of dust particle events as a function of the normalized transverse beam emittance. The measurements are a subset of the data shown in Fig. 11. The average and median values of the measurements are indicated by the solid and dashed lines, respectively (beam 1 in blue and beam 2 in red). Only events from 2016–2018 at beam intensities $I\ge 1.5\times {10}^{14}$ protons were considered. Simulations (yellow symbols) were performed for discrete emittance values only (average values are given by circles, median values are given by crosses).

Figure 13

Simulation predictions of the relative number of dust events exceeding a given number of collisions $N_i$. The absolute number of events was assumed to be independent of the intensity $I$.

Figure 14

Simulated distributions of $N_i$, $R_i$ and $\tau$ for different beam intensity intervals; $N$ indicates the number of events. The different $R_i$ and $\tau$ distributions correspond to different minimum $N_i$ thresholds as indicated in the figures ($N_i\ge {10}^6$: solid histograms, $N_i\ge {10}^7$: dashed histograms).