CER

N-A

TS-2

010-

177

30/0

8/20

10

EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH CERN – ACCELERATORS AND TECHNOLOGY SECTOR

Geneva, Switzerland, August 2010

CERN-ATS-2010-177

First Thoughts on a Higher-Energy LHC

Ralph Assmann, Roger Bailey, Oliver Brüning, Octavio Dominguez, Gijs de Rijk, José Miguel Jimenez, Steve Myers, Lucio Rossi, Laurent Tavian,

Ezio Todesco, Frank Zimmermann

Abstract

We report preliminary considerations for a higher-energy LHC (“HE-LHC”) with about 16.5 TeV beam energy and 20-T dipole magnets. In particular we sketch the proposed principal parameters, luminosity optimization schemes, the new HE-LHC injector, the magnets required, cryogenics system, collimation issues, and requirements from the vacuum system.

2

Table of Contents:

1. Parameters 2. Luminosity optimization 3. Injector 4. Magnets 5. Cryogenics studies 6. Vacuum system 7. Collimation issues 8. Conclusions and outlook

1. Parameters

Increasing the energy beyond the nominal energy of the present LHC [BRU04] can be obtained by raising the mean field in the dipole magnets. The main parameter of HE-LHC, 16.5 TeV beam energy, is based on the hypothesis of substituting all the present dipoles with new, more powerful ones, capable of operating at 20 T with beam. Magnets in the 20-25 T range for a higher-energy hadron collider in the LHC tunnel were first proposed by McIntyre [McI05]. Given the degree of approximation of this report, it is assumed that the LHC main bend filing factor of 2/3 remains unchanged. Intermediate energy can be obtained for a lower dipole field or if a reduced number of dipoles (e.g. 1 per cell or 2 per cell) are changed). If this second case turns out to be possible, one may think of a staged HE-LHC.

For a beam energy of 16.5 TeV a target luminosity value of 2x1034 cm-2s-1 has been defined, which should yield about the same radiation levels in the interaction regions as for the high luminosity upgrade, “HL-LHC,” at 7 TeV with 5x1034 cm-2s-1, for which IR magnets are presently being developed. We consider 1404 bunches with 50-ns bunch spacing and of close to nominal bunch intensity in order to limit the heat load on the beam screen from synchrotron radiation and image currents, to ease the demands on the cryogenic system, and to keep the stored beam energy comparable to that of the nominal LHC, in order to facilitate machine protection. This number of bunches per beam – 1404 – turns out to be sufficient to achieve the target luminosity value. Similar luminosity performance could also be achieved by operating with 2808 bunches of half the bunch charge and half the transverse emittance – a scenario which may, however, be more challenging for collimation and machine protection. A further side benefit of the 50-ns bunch spacing is that it also reduces the severity of electron-cloud effects, the suppression of which by other, additional means (coatings and clearing electrodes) will also be considered.

We consider both “flat” and “round” colliding beams. In the case of flat beams the horizontal beta function has been set to 1 m, about twice the nominal LHC beta function, which should be within reach at the higher energy, and the normalized horizontal emittance to the nominal value of 3.75 µm. The vertical beta function, the vertical emittance, plus the bunch charge have next been chosen so as to obtain a total beam-beam tune shift of 0.01, from two IPs, and the aforementioned target value for the luminosity. The vertical emittance and beta function come out to be about half the horizontal values. A smaller vertical emittance is possible thanks to the strong radiation damping. The small vertical emittance together with a beam crossing in the horizontal plane will allow for a vertical beta function between 0.4 and 0.5 m. Indeed, for flat beams it is assumed that the beams cross in the horizontal plane at two collision points. The crossing angle is chosen to provide a separation of 12 σ at the parasitic encounters, which is higher than the 9.5-σ separation of the nominal LHC, and ensures that long-range beam-beam effects are negligible. A set of key parameters is compiled in Table 1, which also shows that it is possible to achieve an identical luminosity performance with the same bunch intensity and the same maximum beam-beam tune shift by

3

colliding round beams, using a horizontal and vertical beta function of 0.6 m and a normalized emittance of 2.59 µm in both transverse planes.

Table 1: Preliminary parameters of HE-LHC at 33-TeV c.m. energy with flat and round beam collisions

nominal LHC HE-LHC beam energy [TeV] 7 16.5 dipole field [T] 8.33 20 dipole coil aperture [mm] 56 40 beam half aperture [cm] 2.2 (x), 1.8 (y) 1.3 injection energy [TeV] 0.45 >1.0 #bunches 2808 1404 bunch population [1011] 1.15 1.29 1.30 initial transverse normalized emittance [µm]

3.75 3.75 (x), 1.84 (y) 2.59 (x & y)

initial longitudinal emittance [eVs] 2.5 4.0 number of IPs contributing to tune shift 3 2 initial total beam-beam tune shift 0.01 0.01 (x & y) maximum total beam-beam tune shift 0.01 0.01 beam circulating current [A] 0.584 0.328 RF voltage [MV] 16 32 rms bunch length [cm] 7.55 6.5 rms momentum spread [10-4] 1.13 0.9 IP beta function [m] 0.55 1 (x), 0.43 (y) 0.6 (x & y) initial rms IP spot size [µm] 16.7 14.6 (x), 6.3 (y) 9.4 (x & y) full crossing angle [µrad] 285 (9.5 σx,y) 175 (12 σx0) 188.1 (12 σx,y0) Piwinski angle 0.65 0.39 0.65 geometric luminosity loss from crossing 0.84 0.93 0.84 stored beam energy [MJ] 362 478.5 480.7 SR power per ring [kW] 3.6 65.7 66.0 arc SR heat load dW/ds [W/m/aperture] 0.17 2.8 2.8 energy loss per turn [keV] 6.7 201.3 critical photon energy [eV] 44 575

photon flux [1017/m/s] 1.0 1.3 longitudinal SR emittance damping time [h] 12.9 0.98 horizontal SR emittance damping time [h] 25.8 1.97 initial longitudinal IBS emittance rise time [h] 61 64 ~68 initial horizontal IBS emittance rise time [h] 80 ~80 ~60 initial vertical IBS emittance rise time [h]

~400 ~400 ~300

events per crossing 19 76 initial luminosity [1034 cm-2s-1] 1.0 2.0 peak luminosity [1034 cm-2s-1] 1.0 2.0 beam lifetime due to p consumption [h] 46 12.6 optimum run time tr [h] 15.2 10.4 integrated luminosity after tr [fb-1] 0.41 0.50 0.51 opt. av. int. luminosity per day [fb-1] 0.47 0.78 0.79

4

The coil full aperture of the dipole coil is provisionally taken to be 40 mm. Leaving margin, as in the present LHC for the vacuum tube and the beam screen, the beam half aperture is 1.3 cm, which is about 30% lower than now. The maximum arc aperture is needed at injection. Since the injection energy will need to be considerably higher than the present 450 GeV, at least equal to 1 TeV, the reduced aperture should be adequate.

The IBS rise times are much longer than the radiation damping times. This has the consequence that, if no countermeasures are taken, the transverse and longitudinal emittances quickly shrink and the beam-beam tune shift increases to potentially unacceptable values. This and the possible loss of longitudinal Landau damping can be counteracted by injecting transverse and longitudinal noise for 3-D emittance control (see also [ZIM01]).

The stored beam energy of about 480 MJ is some 32% higher than for the nominal LHC. The synchrotron radiation is increased by about a factor of 18, which brings the total heat load close to, or beyond, the cooling capacity of the existing LHC cryoplants.

2. Luminosity optimization

The time evolution of beam parameters and luminosity during a physics store can be simulated starting from some initial set of beam parameters. As possible constraints a maximum total tune shift (e.g. equal to 0.01), both vertically and horizontally, can be demanded, which is realized through controlled horizontal and vertical emittance “blow up”. Also taken into account are radiation damping and intra-beam scattering (IBS). For the latter, horizontal and longitudinal growth rates are computed with MAD-X (using extended Conte-Martini formulae [CON85,ZIM06]). The vertical IBS growth rate is taken to be a certain fraction, e.g. 20%, of the horizontal rate. Proton burn off is another important ingredient. Further options include keeping the longitudinal emittance constant (through controlled longitudinal blow up) as well as leveling with the crossing angle. A dedicated simulation program produces a list of beam & machine parameters (including the optimum run time) and the time evolution of beam current, transverse emittances, longitudinal emittance, bunch length, momentum spread, luminosity, integrated luminosity, tune shifts, crossing angle etc.

Initial values for the vertical beta function and the vertical emittance plus the bunch charge are chosen so as to obtain a beam-beam tune shift of 0.01 from two IPs and an initial luminosity of 2x1034 cm-2s-1. At each time step the transverse emittances, subject to strong radiation damping, can be assumed to be blown up by noise injection in order for the total horizontal and vertical beam-beam tune shifts to stay at the limiting value of 0.01. Various cases have been considered, e.g. ones with and without a constant longitudinal emittance and ones with either constant or decreasing crossing angle. The principal results are almost the same for all these different situations.

Figure 1 compares the time evolution of the transverse emittances with a constant longitudinal emittance and constant crossing angle for the two cases that (1) the transverse emittances are blown up by noise injection so that the tune shifts in both planes stay at the postulated beam-beam limit and (2) that the emittances are shrinking under the sole influence of radiation damping and the weak IBS. Inspecting these curves reveals that controlled transverse emittance blow up is necessary throughout the entire store. Including this continuous controlled blow up, the horizontal normalized emittance decreases roughly from 3.75 micron to 0.5 micron over 20 hours, and the vertical emittance from 1.8 micron to, equally, 0.5 micron at the end. The luminosity time evolution is displayed in Fig. 2. Figure 3 demonstrates that the integrated luminosity is 0.5 fb-1 after 10 h and 0.7 fb-1 after 20 h. The optimum run time ranges from 10 to 14 hours, for a turnaround time of 5 to 10 hours.

5

Figure 1: Evolution of the three HE-LHC emittances, for both flat and round beams, during a physics store with controlled blow up and constant longitudinal emittance of 4 eVs plus constant crossing angle (the thicker lines on the top), and the natural transverse emittance evolution due to radiation damping and IBS only (the thinner lines at the bottom) – still for constant longitudinal emittance and constant crossing angle –, which would lead to an excessive tune shift. In the following only the former cases are considered.

Figure 2: Time evolution of the HE-LHC luminosity, for both flat and round beams, including emittance variation with controlled blow up and proton burn off. The curves with constant or varying crossing angle lie on top of each other if the beam-beam tune shift is kept constant as assumed here.

6

Figure 3: Time evolution of the HE-LHC integrated luminosity, for both flat and round beams, during a physics store including emittance variation with controlled blow up and proton burn off.

From the preliminary studies reported above, we can conclude that when requiring ∆Qx,y= 0.01 the SR-IBS equilibrium is never reached, and transverse noise injection is needed throughout. The luminosity has proven almost independent of a variation in the crossing angle and of a variation in the longitudinal emittance. Both can, therefore, be kept constant, the latter by controlled longitudinal “pink” noise injection. High values for integrated luminosity are expected, close to 1 fb-1 per day. And the optimum running time before dumping the beam is of order 10-14 h with 5-10 hours turnaround time.

If the tune shift is not constrained to a maximum value of 0.01 by controlled blow up, but the transverse emittances are allowed to shrink according to the radiation damping, the tune shift acquires maximum values between 0.020 and 0.033 as is illustrated in Fig. 4. The maximum tune shift and also the peak luminosity are higher in the case of flat beams. Figure 5 shows the integrated luminosity for the same two cases, i.e. flat and round beams without controlled blow up and the pure emittance shrinkage from radiation damping (plus the weak IBS growth). After 10 hours the integrated luminosity reaches a value of 0.7-0.8 fb-1 at a maximum tune shift of 0.033 or 0.02, for flat and round beams, respectively, to be compared with a 10-h integrated luminosity of 0.5 fb-1 if the tune shift is limited to 0.01 (Fig. 3).

7

Figure 4: Time evolution of the HE-LHC tune shifts, for flat (left) and round beams (right), during a physics store including SR emittance shrinkage without controlled transverse blow up, and including proton burn off.

Figure 5: Time evolution of the HE-LHC instantaneous (left) and integrated luminosity (right), for both flat and round beams, including SR emittance shrinkage without controlled transverse blow up and including proton burn off.

3. New injector

To relax the aperture requirements at injection and to reduce the energy swing to a value similar to the present LHC, the injection energy into the HE-LHC should be raised above 1 TeV.

Several options are considered. The most attractive one may be to replace, or complement, the SPS with a new machine featuring fast ramping SC magnets that could reach a field of 4.5 T instead of 2 T for conventional magnets. The SC magnets could be of a type similar to the magnets developed for the FAIR SIS300 synchrotron, or, in case the machine is used only for filling the HE-LHC, the magnets could also be of the much slower ramping HERA type assuming power converters with higher performance than at HERA. The slim Tevatron magnets could perhaps be used in the transfer lines connecting the new “SPS+” and the HE-LHC.

Alternatively, the LHC itself could be used as an injector at a beam energy around 5 TeV, or, yet another possibility, a new ring based on small and inexpensive transmission-line type magnets could be installed in the LHC tunnel (see the earlier “LER” study).

8

4. Magnets

The success of the HE-LHC study depends critically on the success of the magnet R&D to reach dipole fields around 20 T in a useful bore. The history of HEP superconducting hadron colliders [HAN83, WOL95, STR91, ANE03, PER98] is reported in Table 2, while in Fig. 6 the various magnet cross sections are reported. While LHC has been the logical summit of 30 years of R&D based on Nb-Ti technology [ROS02], the HE-LHC must rely on novel and more advanced technology, capable to go well beyond 15 T and even to reach 20 T. This value seems today the maximum attainable on paper when we consider basic design limits and logistics restraints coming from the LHC tunnel. Although a basic zero-order design is presented, the main limitations coming from transportation and installation are included.

Table 2: Main SC magnets for HEP colliders

Project (Lab) Energy (TeV/beam)

Operating dipole field (T)

Operating temperature (K)

Tunnel length (km)

Status

CBA♣ (BNL) 0.4 5 4.4 3.8 Cancelled in 1983 Tevatron† (FNAL) 0.98 4.3 4.4 6.3 Operated in 1987 HERA (DESY)* 0.92 5.3 4.4 6.3 Operated in 1989 SSC (Dallas) 20 6.8 4.4 87 Cancelled in 1993 RHIC (BNL) 0.1/nucleon⊗ 3.4 4.4 3.8 Operated in 1999 LHC (CERN)** 7

3.5 8.3 4.2

1.9

27 Operation in 2013 LHC from 2010

♣ last name of the early “Isabelle”.

† operation of superconducting magnets began in 1983 at lower field.

* after upgrading in 1998 from 0.82 TeV, 4.7 T

⊗ heavy ion accelerator

** Operation of LHC started at 0.45 TeV in September 2008, interrupted by the incident, and then resumed in 2009.

9

Figure 6: Cross section of the dipoles of the main superconducting circular colliders that have been built (or proposed, like SSC).

4.1 Operating temperature for the superconducting magnets

It is better to right away rule out any possible idea of working at a temperature higher than LHe: the best HTS (High Temperature Superconductor) has critical currents that are not useful above 15 K for very high field. The necessity of good cooling inside the coil, with efficient transverse and longitudinal removal of the heat, rules out the use of purely conduction-cooled magnets (cryogen-free magnets). Past experience has shown that from the discovery of a new superconducting material, it takes more than 20 years to develop a useful conductor and a practical magnet design. Therefore, even if new discoveries in the HTS domain may change the scenario, the choice of LHe (4.4 K) or HEII (2 K) bath as operating mode is straightforward for HE-LHC. Despite the fact that 4.4 K may be an attractive option that may be considered in future studies, for this report we concentrate on a solution based on 2 K cooling, since it is more favorable for the maximum field level. Because coils are impregnated and because of weak dependence of the main superconductors (Nb3Sn and HTS) on temperature, we do not need a great uniformity in temperature like in the LHC, which offers a considerable advantage for cryogenics.

4.2 Superconductor

Despite the variety of SC materials and the continuous new discoveries, the practical superconductors (i.e. with good physics characteristics, good workability and suitability to cabling) are limited. In Fig. 7 the superconductor critical current density (i.e. the intrinsic SC characteristics) is reported as a function of applied magnetic field [LEE].

Figure 7: Practical superconductor current density (i.e. critical current over the Sc cross section). For Nb-Ti the curve represents an industrial production, for Nb3Sn the best industrial production and for HTS it is the best result on short samples or very small production.

Considering the very preliminary state of YBCO material, the candidate materials are the classical Nb-Ti and Nb3Sn (Internal Sn) and round wire 2212 (or Bi-2212). It should be noticed that today Bi-2212 data are rather optimistic (by a factor 3-4), for magnet conductors. However, considering that reported data have been obtained in industry, we are confident that reasonable improvements will enable using the Bi 2212 curve of Fig. 7.

10

The main considerations in selecting the materials for the magnet design are:

a) Below 7-8 T, the moderate field region, considering its low cost and its robustness, Nb-Ti is the natural choice. Nb-Ti can attain also 10-10.5 T, if operated at 1.8 K, however in this region it suffers training and a very low stability margin.

b) Above 9 T and up to 15 T, the high field region, Nb3Sn is certainly the best choice. Its maturity is by far better than HTS and, as mentioned above, the actual cross over between Nb3Sn and HTS for real conductor today is about 18-19 T. Both Nb3Sn and HTS are brittle materials with many difficult consequences in terms of magnet design and manufacturing. An alternative to Nb3Sn could be offered by Nb3Al, for its better mechanical properties: however for the moment this material is lagging behind Nb3Sn in terms of performance and maturity of the process. Today Nb3Sn is a well developed material used for:

i. 10-20 T solenoids of various types for NMR spectroscopy and for high field laboratory facilities. About 15-20 tons of material is produced every year.

ii. ITER toroidal coils and central solenoid. The quantity produced in the past was however less than 1 ton per year on average. With the start of the project in 2006 now the production must increase to 100 tons/year for 5 years. This figure is comparable to the LHC Nb-Ti production, which was 250 tons/year over 5 years.

iii. Accelerator R&D, mainly in USA, driven by the DOE program started in 1998. The product is of much higher Jc than the material for ITER and of much better quality than the standard Nb3Sn production for solenoids. It took about 10 years for the program to increase performance from 1200 A/mm2 at 12 T (ITER) to 2700-3000 A/mm2 of today's HEP material. The production today is around 2 tons per year. The HL-LHC project will increase the production at 4-5 tons/year. For the HE-LHC the production must increase to 100 tons/year.

c) Above 15 T, in the very high field region, the HTS

materials are in principle superior to classical superconductors. Of course they need much more development, especially in the 16-18 T range, and the choice (vs. Nb3Sn) in that range will depend a lot on the evolution of material science and conductor industrial processes in the next decade. Today two types of superconductor are available for this range, the bismuth copper oxide BSCCO, in the form of Bi-2212, and yttrium copper oxide YBCO, in the form Y-123 (in industry sometimes called 2G, i.e., second generation). As mentioned, today Bi-2212 is more suitable for compact cable needed for accelerator magnets. However, YBCO is under development for many other applications, raising big hope in performance and cost improvement. Today the production of Bi-2212 is virtually zero, i.e., it is confined to R&D. However, recently the DOE has launched a 6 M$ three-year-program aimed at developing Bi-2212 and the EuCARD collaboration in Europe is also pursuing, on more modest scale, HTS development for very high field dipoles.

11

4.3 Magnet Design

Magnet design is going to be probably the most challenging issue of the whole project. The main issues to be solved to reach fields in the region of 20 T are:

1. Finding a coil design with a reasonable amount of superconductor. The cost of the whole project will be dominated by the cost of the SC, from five to ten times the cost of the LHC superconductors (about 300 MCHF). That is why the coil must be subdivided into three sections of different materials, Nb-Ti for the low and medium field region, Nb3Sn for the high field region and HTS for the very high field region.

2. Keeping the magnet size within reasonable limits to fit, once cryostated, inside the LHC tunnel while allowing transport of a nearby magnet. For the moment only “classical” solutions, based on an iron yoke for flux return, have been explored. In such a case the size of the magnet is determined by the aperture and the field level, beside the inter-beam distance. This is the main reason of fixing the magnet aperture to 40 mm. In the ill-fated SSC the magnet aperture was for long time 40 mm and then, around 1990, it was increased to 50 mm, a major driver of the cost increase that seriously affected the project. Also LHC increased the dipole aperture from 50 mm to 56 mm around 1992, a decision that led to a modest increase in the cost but certainly contributed to a decrease of the main field and it prolongated the R&D. However, at this preliminary stage and based also on the first positive observations of LHC behavior, it seems to us we should consider 40 mm as best choice compatible with the 20-T operating field.

3. Enlarging the beam-to-beam distance to 300 mm is a necessity to make room for coils and to maximize the field by reducing the cross talk. As for the aperture, this choice may be disputable and it will be thoroughly investigated in the next studies.

4. Controlling the forces and resulting stresses in presence of brittle materials (Nb3Sn and HTS). This point may be one of the hardest to face, because it can reduce significantly the conductor performance.

5. Having a field design of collider quality. The preliminary design shows all harmonics below 5 units (i.e. 5·10-4 of the main field), despite strong saturation for the iron. We have not yet examined in detail the effects of some features of these novel superconductors:

a. Conductor positioning accuracy: for Nb3Sn it is five times worse than for Nb-Ti (results from US magnet R&D). No estimate is available for HTS.

b. Persistent currents: filament size from 5 to 10 times larger than in Nb-Ti. c. Coupling currents of various types (absence of snap-back in Nb3Sn is one of the few

good news).

We have ideas, partly under development, how to cure b. and c., and more experience is needed to understand a.

The design we have devised is based on a coil thickness of about 75 mm, a scale-up from previous experience, see Fig. 8, by keeping the same overall Joverall ~ 350 A/mm2. Also, there is no grading inside each layer of superconducting material, to avoid further complication at this stage. Dealing with three materials, which is the most important grading, is for the moment difficult enough.

12

Figure 8: Relation of coil width (or thickness on horizontal axis) vs field. Triangles are real Nb-Ti based projects, large dots are single R&D magnets.

The proposed 2D design of the main HE-LHC dipole cold mass is shown in Fig. 9. The iron yoke is 800 mm in diameter, some 200 mm larger than for the LHC, an extra that may be reduced, in part, and may be accommodated in a different cryostat design. In Fig. 10 the coil structure and the field map is shown for a quadrant cross-section. In Fig. 11 the coil subdivision is shown, with peak field for each block, and a table with the material. The ratio between peak field and central field is 1.03, while the stray field at 200 mm distance is 50 mT.

Figure 9: Cross section of the cold mass of the 20-T dipole proposed for the HE-LHC. The outer diameter is 800 mm.

0

5

10

15

20

0 20 40 60 80

Ope

ratio

nal f

ield (

T)

Coil width (mm)

HE-LHC

LHCSSC

Hera

Tevatron

RHIC

D20 (max. reached)

HD2(max. reached)

13

Figure 10: Cross section of the coils in a quadrant with field levels in colour (dimensions in mm).

Figure 11: Coil partition into different materials and peak field in each block.

The design is for a 20 T operational field, working roughly at 80% of the load line limit, i.e., the magnet is designed with a 25 T short sample limit, see load lines in Fig. 12 (left). We believe than a 20% margin is needed to be realistic for a large production (experience from LHC). In any case we do not claim this to be a 25-T magnet, since the forces and stresses are just manageable at 20 T, see Fig. 12 (right). Of course there is ample room for optimization: from this first calculation it seems possible to keep stresses on brittle conductors below 180 MPa, a viable level according to recent results.

Figure 12: Magnet load lines in the J-B space (left): square marker indicates the operative points for each material. Horizontal stress on the coils (right): peak stress is 220 MPa, and each colour covers a 25-MPa range.

0

20

40

60

80

0 20 40 60 80 100 120

y (m

m)

x (mm)

HTS

HTS

HTS

Nb3Sn

Nb3Sn

Nb3Sn

Nb3Sn

Nb-Ti

Nb-Ti

0

20

40

60

80

0 20 40 60 80 100 120

y (m

m)

x (mm)

16.9 T

20.6 T

19.9 T

12.7 T

12.5 T

12.6 T

11.4 T

8.2 T

6.3 T20 T

Turns %Nb-Ti 40 28%Nb3Sn 58 41%HTS 45 31%

0

500

1000

1500

2000

0 5 10 15 20 25

Cur

rent

den

sity

(A/m

m2)

Field (T)

Nb-Ti

Nb3Sn

Bi-2212

14

The operational current is 13 kA. The viability of such a relatively low current (basically the same as for the LHC) must be confirmed after detailed quench protection studies. Containment of the hot spot temperature following a quench may require to increase the operational current, so for the moment we assume a current in the range 13-18 kA for the main magnets.

There is a strong saturation of the iron yoke and the vacuum vessel will have to shield about 50 mT of stray field, which may require a special cryostat design.

At this stage, we have not yet considered the 3D part of the design. The rectangular block lay-out of the coils will favour a solution based on flat race-track coils with flared ends, such as in HD2 (see Fig. 3). However, the cosθ solution and its classical end design is not at all discarded.

4.4 Final magnet considerations and issues

The main drive in this design has been volume optimization and a 20% margin from the superconductor limit. No consideration on cost issues have been injected at this stage. This would certainly call for a reduction (and a strong one!) of the HTS cross section in favor of the Nb3Sn one. Furthermore, today's results favour the use of Nb3Sn up to 15 T, rather than 13 T, as in this design. However, we need to be optimistic on HTS because of the larger optimization margin in Jc that HTS has over Nb3Sn. The constraints in terms of space are severe, so we should try to profit of the better intrinsic Jc of Bi-2212 (not to mention YBCO for the future) in the 14-17 T region, even if today this material is not yet available in industrial wires; in addition, the much larger stability of HTS is certainly a big advantage, since it may allow working at 90% or more of the critical surface. Development of conductor, including cost reduction strategy, is the number one issue for a mass production of a 20 T operational field dipole.

From the design point of view, many issues are still pending even in a conceptual design stage. To reduce the size, while reducing the iron saturation as well as the stray field, a design based on anti-coils to compress or cancel the return flux must be explored. However, this has never been done so far for accelerator magnets and it will require a careful study of all implications.

A design with a larger aperture (50 mm) must also be explored and the penalties in terms of field level and stray field, as well as in terms of cost, must be determined.

As for field quality, experience has shown that so far Nb3Sn coils have five times higher geometric harmonics than Nb-Ti coils. This ratio can improve in the future. However, we may need to devise a strategy to correct these harmonics or their consequences on the beam, as well as the other superconductor effects. A possible solution is to couple the best geometric design with a sectioning of the coil powering. We will examine in the future a powering scheme of the magnets with two or three independent power supplies, a solution that may present multiple advantages:

a) Optimization of the cable size and current for best use of each superconductor. This is very important considering that superconductor properties will dominate the cost of the project.

b) Dynamic correction of field quality through different current ramps of the three sections. This is important also to release the request of small filament size, certainly an issue for the project.

c) The quench protection, which will become more complex, can be “trimmed” for each section. This will require developing special techniques, as those used in NMR magnets.

Of course a variant of these extreme solutions is a nested power supply solution, where all coils are in series for a basic current and then small currents of different values and ramps can be injected into some blocks.

15

The magnet length can be as long as for the LHC dipoles, 15 m. However, given the difficulties for conductor production in long lengths and for the coil reaction and handling, a solution based on shorter lengths could be envisaged, like 2x7.5 m long magnets in the same cold mass. The loss in field coverage and in energy (about 10%) could be worth the advantages.

A point that needs to be investigated soon, also considering the strong implication on other systems, is the performance vs. an increase of the operating temperature (easing beam screen and synchrotron radiation removal) and a use of Nb3Sn only, should the HTS prove to be not realistic.

4.5 Conclusions on the magnets

The magnet design is certainly the most controversial issue of the HE-LHC design, the performance of the machine being directly related to the attainable field level. A design has been presented, using solid parameters for classical superconductors and the best OBTAINED performance for HTS. The design shows that the 20-T operational dipole is in reach of the known superconductors and that there is no obstacle in principle to attain it, even if a lot of deep studies must be pursued still.

To give a guide for the time-schedule, we note that the first single bore 9 T model showing potential for LHC was tested in 1987 while the first twin 10 m long LHC dipole of almost final configuration was successfully tested to 9 T in 1994. The final prototype was successfully tested in 1999-2000; the last LHC dipole has been delivered in November 2006, for a start of the machine in September 2008. It is clear that 10 years of R&D is not a big lapse of time for such a project. Even counting on the necessary development that US-LARP and CERN are doing for the luminosity upgrade and on its final success (i.e. installation of 13 T Nb3Sn magnets in LHC by 2020), the leap toward 20 T is hard. For this reason a vigorous program should be pursued on the following points:

• Superconductor development, aimed at increasing Jc in the 15-20 T range and aiming at understanding all issues related to cabling (and strain) and to field quality, especially the dynamic effects. In this effort, starting from 2013-14 the accent must be on HTS; however it is not to be forgotten that this magnet mainly relies on the performance of the Nb3Sn part. If we accept a compromise in performance we can do a 15 T magnet only with Nb3Sn (and Nb-Ti), while we cannot do a magnet with HTS only, mainly for cost reasons (unless a big breakthrough, may be in the YBCO region). This conductor R&D must include qualification of the conductor in a SMC (Small Coil Model) test bed.

• Magnet design, modeling and short (1-2 m long) models to examine the different designs and select the most promising route.

• Energy extraction from the magnet system. Assuming that the stored electro-magnetic energy in the magnets is proportional to the magnetic field, one expects a total stored EM energy of ca. 25 GJ for the HE-LHC. The LHC has been sectorized into 8 parts in order to keep the EM energy per sector at the order of 1 GJ. For the HE-LHC this strategy would imply ca. 30 independent sectors.

16

5. Cryogenics studies

Similar to the LHC, the heat deposited in the HE-LHC will reach 3 different temperature levels:

- the thermal shield temperature level (TS) between 50 and 75 K, - the 5-K heat intercept (HI) and beam screen temperature level (BS) between 4.6 and

20 K (40-60 K or 85-100 K as an alternative compatible with vacuum specification), and - the cold mass temperature level (CM) at 2 to 4.5 K.

It is also assumed that specific cryogenic systems will be needed for insertion magnets and RF cavities. These insertions are not defined yet; consequently, in the following, only the continuous cryostats (CC) will be considered, i.e. arcs plus dispersion suppressors and associated current feed boxes.

5.1 Heat inleaks

In first approximation the thermal performance of the HE-LHC cryostat is assumed to be similar to the one of the LHC cryomagnet. In addition, it is assumed that the LHC cryoline (QRL) is used with its present thermal performance. The specific heat inleaks are given in Table 3.

Table 3: Specific heat inleaks on magnets and cryoline

Temperature level LHC HE-LHC

TS (50-75 K) [W/m] 7.7 7.7

HI (4.6 K) [W/m] 0.23 0.23

CM (2 K) [W/m] 0.21 0.21

5.2 Resistive heating in superconducting splices

The resistive heating in the magnet splices is proportional to the square of the magnet current, to the splice electrical resistance and to the number of splices. The corresponding heat load is deposited at the CM temperature level. Table 4 gives the main parameters related to the resistive heating. The increase of the magnet current and of the number of splices, both by 50%, translates into an increase of the resistive heating by a factor 3.4 with respect to the nominal LHC.

Table 4: Resistive heating in magnet splices

LHC nominal HE-LHC

Main magnet current [kA] [kA] 12 18

Splice resistance [nOhm]

Number of splice per arc [-]

[nOhm]

[-]

0.5

2500

0.5

3750

Resistive heating on CM [W/m] [W/m] 0.1 0.34

5.3 Current lead cooling

Concerning the cooling of the current leads (CL), it is assumed that HE-LHC is using the same type of HTS current lead as the LHC with the same cooling performance, i.e. a specific cooling rate per kA of 54 mg/s of helium between 20 and 300 K. In addition, as the optics of the HE-LHC is not yet fully

17

defined the number of individually powered magnets is not known; consequently, it is assumed that the total current entering or exiting is proportional to the main magnet current. In addition, as for the LHC, it is assumed that high-load sectors enter two times more current than low-load sectors. Table 5 lists the main parameters for the current lead cooling.

Table 5: Current lead cooling

LHC nominal HE-LHC

Main magnet current [kA] 12 18

Total current in/out

Total current high load sector CC

Total current low load sector CC

Specific CL cooling flow

High-load sector CL cooling flow

Low-load sector CL cooling flow

[kA]

[kA]

[kA]

[mg/s per kA]

[g/s]

[g/s]

2750

460

230

54

25

12

4130

690

345

54

37

19

5.4 Beam-induced loads

The parameters impacting the beam-induced loads are the beam energy, the bunch population, the number of bunches, the bunch length and the beam-screen aperture. Table 6 gives the scaling laws to be applied for the different beam-induced loads. Table 7 lists the parameters and the beam-induced loads for the nominal LHC and the HE-LHC. Compared with the nominal LHC, all beam-induced loads on the beam screens increase for the HE-LHC. The biggest change concerns the synchrotron-radiation load, which increases by a factor 17.

Table 6: Scaling laws of beam induced heat loads

Beam parameter

Beam-induced load

Energy

E

Bunch Population

N

bunch

Bunch Number

n

bunch

Bunch Length

σ

z [rms]

Beam-screen

aperture b

Temp. level

Synchrotron radiation E4

Nbunch

nbunch

BS

Image current Nbunch

2 n

bunch σ

z

-3/2 b-1 BS

Photo-electron cloud Nbunch

3 n

bunch b-2 BS

Beam gas scattering Nbunch

nbunch

CM

18

Table 7: Parameters and specific beam-induced loads

LHC nominal HE-LHC

Beam energy [TeV] 7 16.5

Bunch population

Bunch number

Bunch length

Beam-screen aperture radius

Synchrotron radiation

Image current

Photo-electron cloud

Beam gas scattering

[1011 p]

[-]

[cm]

[cm]

[W/m]

[W/m]

[W/m]

[W/m]

1.15

2808

7.55

2

0.33

0.36

0.90

0.05

1.29

1404

6.55

1.3

5.71

0.44

1.50

0.03

5.5 Operating the beam screens at a higher temperature

Other possible temperature operating ranges for the beam screen cooling compatible with the beam vacuum specification are 40-60 K or 85-100 K. Increasing the operating temperature of the beam screen will have the following consequences:

- As the electrical resistivity of the copper on the beam-screen surface increases with the temperature, the image-current load will also increase proportionally. Measurements at 20 K, 50 K and 92.5 K on LHC beam-screen samples give a copper resistivity increase by factors 5.5 and 23. Consequently, the image current heat-load will increase from 0.44 to 2.4 and 9.8 W/m (see Table 8). A coating with HTS (like Bi-2223 or Y-123) may improve this figure dramatically. However, today this is a speculation.

- The temperature difference between the beam screen and the cold bore will increase, i.e. the heat inleaks on the cold mass will increase as well. Measurements on String 2 (see LHC PN 330 [TAV03]) indicate a heat inleak increase on the cold-mass of 0.17 and 0.71 W/m (see Table 8).

- The present design of the LHC beam screen cooling loop based on a unit length of 53 m and two 3.7-mm inner-diameter capillaries per aperture is locally limited to a heat extraction of 2.4 W/m per aperture, i.e. 4.8 W per meter of machine. Changing the operating conditions and the specific heat load has a direct impact on the cooling capillary diameter. Table 9 gives the operating conditions of the beam screen cooling loops and the corresponding required capillary diameter assuming the same cooling loop configuration as today. The operation of the beam screen at 20 bar and between 40 and 60 K minimizes the cooling capillary diameter.

19

Table 8: Cryogenic specific heat loads

Temperature level Heat load source LHC

nominal

HE-LHC

BS @ 4.6-20 K

BS @ 40-60 K

BS @ 85-100 K

TS Heat inleaks

Total TS

[W/m]

[W/m]

7.7

7.7

7.7

7.7

HI Heat inleaks

Total HI

[W/m]

[W/m]

0.23

0.23

0.23

0.23

BS

Heat inleaks [W/m] 0 0 -0.17 -0.71

Synchotron radiation [W/m] 0.33 5.71 5.71 5.71

Image current [W/m] 0.36 0.44 2.40 9.81

Photo-electron cloud [W/m] 0.90 1.50 1.50 1.50

Total BS [W/m] 1.82 7.65 9.45 16.3

Heat inleaks [W/m] 0.21 0.21 0.38 0.92

CM Resistive heating

Beam-gas scattering

[W/m]

[W/m]

0.10

0.05

0.34

0.03

0.34

0.03

0.34

0.03

Total CM [W/m] 0.36 0.58 0.74 1.29

Table 9: Beam screen cooling capillary diameter

BS @ 4.6-20 K BS @ 40-60 K BS @ 85-100 K Inlet temperature [K] Inlet pressure [bar] Outlet temperature [K] Outlet pressure [bar] Specific heat load [W/m] Loop length [m] Number of capillary per aperture Capillary inner diameter [mm]

4.6 3.0 20 1.3

7.65 50 2

4.4

40 20 60 18

9.45 50 2

3.8

85 20

100 18

16.3 50 2

6.0

5.6 Heat load summary

Table 8 resumes the specific cryogenic heat load for the different temperature levels. Compared with the nominal LHC, depending on the beam-screen operating temperature range, the heat loads on the beam-screen circuits increase by a factor 4 to 9, and those on the cold-mass circuits by a factor 1.6 to 3.6.

Concerning the heat loads on the cold-mass circuits, the present LHC cooling loop is locally limited to 0.9 W/m, i.e. it is not compatible with the HE-LHC specific heat load corresponding to the 85-100 K operating temperature range of the beam screens.

20

5.7 Continuous-cryostat cooling capacity

Assuming a continuous cryostat length of 2800 m and an overcapacity margin of 1.5, the required cooling capacity per continuous cryostat is given in Table 10 and is compared with the existing installed capacity of LHC sector cryogenic plants. Values in brackets correspond to the equivalent entropic capacity in kW at 4.5 K. Figure 13 shows the equivalent entropic capacity for the different temperature levels. Depending on the operating temperature range of the beam screen, the total equivalent entropic capacity of HE-LHC refrigerators varies from 31 to 19 kW at 4.5 K. Operating the beam screens between 4.6 and 20 K requires continuous-cryostat refrigerators about 1.7 times larger than the LHC sector refrigerators. Operating the beam screens between 40 and 60 K allows reducing the size of the continuous-cryostat refrigerators which becomes similar to the LHC sector refrigerators. Operating the beam screens between 85 and 100 K overloads the cold-mass temperature level. With a cold-mass operating temperature of 2 K, the optimum beam-screen temperature range is 40-60 K.

The electrical input power of the different scenarios, assuming a coefficient-of-performance of 250 W per W, is given in Table 11.

Table 10: Continuous cryostat cooling capacity per sector (values in brackets: equivalent entropic capacity in kW at 4.5 K)

Temperature level HE-LHC continuous cryostat LHC high load sector

BS @ 4.6-20 K

BS @ 40-60 K

BS @ 85-100 K

BS @ 4.6-20 K

TS (50-75 K) [kW] 32 (2.2) 33 (2.2)

HI (4.6-20 K) [kW] 33 (18.4)

1.0 (0.5) 1.0 (0.5) 7.7 (4.3)

BS [kW] 40 (3.5) 69 (3.3)

CM (2 K) [kW] 2.4 (7.8) 3.1 (10.0) 5.4 (17.4) 2.7* (9.3)

CL [g/s] 56 (2.5) 41 (1.8)

Total equivalent entropic capacity (30.8) (18.6) (25.8) (17.6)

*: 2.4 kW at 1.8 K plus 0.3 kW at 4.5 K

21

Figure 13: Equivalent entropic capacity

Table 11: Electrical input power for continuous-cryostat refrigerators

HE-LHC CC refrigerator LHC refrigerator

BS @ 4.6-20 K

BS @ 40-60 K

BS @ 85-100 K

Electrical input power per refrigerator [MW] 7.7 4.7 6.5 4.4

Number of refrigerator [-] 8 8 8 8

Total electrical input power [MW] 62 37 52 35

5.8 Closing remarks on cryogenics

In these present cryogenic studies, no contingency has been introduced in the numbers. A lot of assumptions have to be confirmed like the splice resistance and number, the main magnet current, the current-leads distribution and number, and the cryostat performance.

The optimization of the refrigeration cycle has still to be done. Transient heat loads (ramp/de-ramp, fast de-ramp, quench), have still to be considered in order to define the correct level of buffering. In addition, the LSS loads have still to be considered with probably new cryoplants for insertions accommodating experiments.

Depending on the cooling scenario, up to 9 temperature levels have to be distributed along the continuous cryostats to supply or recover the different cooling loops. A rationalization study has to be done for reducing the number of distribution headers like operating the beam screen and the thermal shield with the same temperature range and/or cooling the resistive part of HTS current lead with a helium flow at a higher temperature and pressure (e.g. 40 K, 20 bar).

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

HE-LHC CCBS @ 4.5-20 K

HE-LHC CCBS @ 40-60 K

HE-LHC CCBS @ 85-100 K

LHC sector

Equi

vale

nt e

ntro

pic

capa

city

[KW

@ 4

.5 K

]

Current leads

Cold-mass

Beam screen

Heat intercept

Thermal shield

22

At the end of the LHC (2030), the LHC cryogenics will be 30 to 40 years old. Taking into account the 20-year operation initially specified, major and wide overhauling has to be considered for the equipments reused for the HE-LHC project (cryoplants, QRL, distribution boxes…).

6. Vacuum system

Similarly to the LHC, the bending sections (arcs) operating at a cryogenic temperature will rely on the cryo-pumping (gas condensation) and the long straight sections, operated at ambient temperature, will rely on NEG coatings. The later has shown its efficiency to provide a continuous pumping while reducing significantly the desorption yields for photo-electrons, photons and secondary electrons as well as providing a low secondary electron yield (1.1 after NEG activation), a key parameter for the electron-cloud build up.

However, the parameter list of the HE-LHC has several changes compared to the LHC, which can significantly affect the vacuum performances and its stability:

- The increase of the beam energy, of the synchrotron radiation power and of the critical photon energy will impact the effects linked both to the total intensity and to the beam bunch structure. These effects are detailed in the next subsection.

- The decrease of the number of bunches will reduce the effects linked to the total beam intensity and partly compensate the increase of the bunch population for the effects linked to the bunched structure of beams.

- Finally, the increase of the beam potential and the reduction of the magnet aperture will impact on the vacuum stability and electron cloud build up.

6.1 Operating temperature and beam screens

To ensure a proper pumping of hydrogen, the dominant residual gas in the beam vacuum, an operating temperature for the cryomagnets below 2-3 K is recommended. At higher temperatures, the hydrogen released will be condensed up to an equivalent of a monolayer and then, the equilibrium pressure (hydrogen partial pressure) will start to increase very rapidly with the temperature i.e. 10-9 Pa at 2 K and up to 10-4 Pa at 4.2 K [Ben76]. Similarly to what was done in the LHC, a beam screen will be required to shield the condensed gases from the beam induced effects (electrons, ions and photon-stimulated desorption). Above 2-3 K, the use of cryo-absorbers will be required to ensure the required hydrogen pumping speed and capacity. The option of a higher operating temperature of the beams screen exists. Two alternatives can be considered for the beam-screen temperature: 40-60 K and 85-100 K. The option of 40-60 K provides better cryogenic performance, since at lower temperatures the heat exchange between the beam screen and the cold bore is reduced and the higher helium densities will increase the cooling efficiency.

6.2 Beam induced desorption effects

The static pressure in the beam vacuum is dominated by the thermal outgassing which is not beam related and results from the design, choice of materials and procedures (firing and bake-out).

The beam induced desorption effects dominate at high energies and high intensities. These effects belong to three categories.

6.2.1 Effects linked to the total beam intensity

- Primary beam losses induced desorption The cryogenic sections are the most critical due to the potentially large quantities of condensed gases which can be released resulting from a local heat load. However, the cryosections are “protected” by the quench limit of the cryomagnets since the magnets

23

will quench well before the beam-loss induced desorption rates would be high enough to induce a vacuum limitation.

- Primary ionisation with circulating beams The primary ionisation of the residual gas induced by the beams is linearly dependent on the ionisation cross section (about constant) and on the total intensity (x0.6). A similar effect as in the LHC is expected.

- Ion induced instability The ion induced instability is linearly dependent on the desorption yield (about constant), the ionisation cross-section (also approximately constant), and the total intensity (x0.6) and is inversely proportional to the effective pumping speed. As the aperture of the beam pipe is smaller (factor 1/1.4 on the diameter, and so a factor 1/2.8 for the conductance), the pumping speed through the pumping slots (4.4% transparency for the LHC) shall be maintained, in order to ensure vacuum stability. Considering the new beam screen aperture, the resulting transparency must be increased to 6.2%, which could imply impedance and HOM issues.

- Synchrotron radiation power The synchrotron radiation power is proportional to the 4th power of the energy and to the beam intensity. An increase by a factor 17.3 is expected as compared to the LHC. In the LHC, this heat load is intercepted by the beam screen. To keep such a design, an evaluation has to be made to ensure that the existing size of the cooling capillaries will be large enough to provide the cooling required. Any increase in the capillary diameter would lead to a further beam-aperture reduction. An alternative could be to install photon absorbers in the cryomagnet interconnecting bellows (Plug In Modules), which would intercept the heat load outside the cryomagnets, in order to minimise the heat deposition onto the beam screens. The residual fraction of heat still incident on the beam screen would be determined by the length, aperture and bending angle of the dipole magnets.

- Linear photon flux The photon flux per unit length depends linearly on the energy and the intensity. It increases by +30% with respect to the nominal LHC. Similarly to the LHC, a sawtooth structure shall be used in the beam screen to reduce the photon reflection and the photo-electron yield.

- Photon stimulated pressure rise The photon stimulated pressure rise is increased by a factor 7.4 since it grows with the third power of the beam energy and linearly with the intensity. This large increase is the major limitation of the HE-LHC parameters for the vacuum system. Indeed, to ensure pressure stability, the pumping should be increased by at least an order of magnitude which would bring the equivalent transparency of the beam screen to 41 %!

The photon desorption yield 𝜂 scales as 𝜂 ∝ 𝜀𝑐

2/3 and the critical photon energy𝜀𝑐 as 𝜀𝑐 ∝ 𝐸3, so that the dynamic pressure increase rises with the third power of beam energy and linearly with the beam current: Δ𝑃 ∝ 𝐸3𝐼.

24

6.2.2 Effects linked to the bunched structure of the beams

The electron and ion cloud build-up are two avalanche phenomena which can take place in the beam pipe. As compared to the LHC design, the bunch population has been increased (+12 %) i.e. 1.29x1011 p/bunch, well above the electron cloud threshold measured in the SPS i.e. 3x1010 p/bunch in a dipole field. The beam potential has also been increased (+30 %). Based on these new parameters, an electron cloud build up can be expected. However, the reduction of the number of bunches by a factor of 2 and the resulting bunch spacing of 50 ns has shown its efficiency to reduce the electron cloud build up e.g. reduction by a factor 10 measured in the SPS.

Two other parameters playing a major role in the electron build up are varying: the beam screen height is decreased from 36.8 to 26 mm and the magnetic field is increased by a factor 2.4. Changing the beam screen aperture could bring the system out of resonance conditions. Increasing the beam potential will increase the energy of the primaries and finally, the small Larmor radius (few micrometers for a 100 eV electron) can also change the SEY yield.

Following the measurements made in the SPS, the LHC beam screen was equipped with shielding baffles placed between the beam screen and the cold bore and attached to the cooling capillaries. These baffles aim to intercept the electrons from the cloud, escaping from the beam screen through the pumping slots, to prevent the heat deposition onto the cold bore. Right from the design stage, the same configuration can be modified to convert the shielding baffles into clearing electrodes by insulating them from the cooling capillaries and polarising them to about 1 kV. New coatings (carbon a-C) to be applied to the inner surface of the beam screen are being investigated as potential electron-cloud mitigation solution.

As the beam will ionise the residual gas and due to the slow motion of the ions and enhanced by the secondary ionisation effect by the trapped electrons from the cloud (if any), an ion-induced positive space charge can take place. This phenomenon opens the risk for feedback effects. However, the reduction of the beam pipe aperture will probably cancel this effect.

6.2.3 Feedback effects

In presence of an electron cloud, part of the cloud electrons can be trapped by an ion space charge. These electrons will spiral along the magnetic field and contribute to an additional ionisation of the residual gas. This secondary ionisation effect can lead to ion instability.

6.3 Closing remarks on the vacuum system

The design parameters have to consider the long running-in period to increase the energy and intensities. Indeed, this results in a progressive reduction of the dynamic effects due to the decrease of the yields (photons, ions, electrons, SEY, protons). The initial yields can be reduced by orders of magnitude as observed in LEP and in other accelerators world-wide. Other effects will not improve with time like reflectivity, pumping speed and conductance and shall be addressed right at the design phase.

As in the LHC, the use of a beam screen will be required and the operating temperature of both the cold bore and beam screen will be determinant for the vacuum stability. An operating temperature of the cryomagnets of, or below, 2-3 K is highly recommended to provide the required pumping speed and capacity for hydrogen. Similarly and to minimise the risk of helium leaks which will stop the operation of the accelerator, transitions (welds and brazing) between liquid helium circuits and the beam vacuum shall not be permitted.

The synchrotron radiation and the critical energy of the photons are an issue. The significant increase of the induced heat load onto the beam screens has to be intercepted by the beam screen

25

without significant temperature variations. The introduction of photon absorbers in the magnet interconnections (PIMs) and of clearing electrodes between the beam screen and the cold bore could be options to reduce the effect of the photons and limit the risk of an electron cloud build-up.

7. Collimation issues

The collimation of the high intensity LHC beams poses many challenges already for the present LHC. An energy upgrade of the LHC makes it necessary to revisit collimation performance from various important aspects. Some are discussed in the following sections. A more detailed analysis must follow.

First, we discuss collimator robustness. The stored beam energy will be increased from 362 MJ to 478.5 MJ during the energy upgrade. At the same time the transverse cross-section of the beam reduces from 0.156 mm2 to 0.046 mm2 at a hypothetical location with βx=βy=100 m (even though hypothetical, this represents typical optics functions at a present collimator). The increase in energy density is due to the reduction in geometric emittance, which at 16.5 TeV can become as small as 1 Angstrom vertically.

The energy density for a single bunch at a typical collimator location then increases from 0.8 MJ/mm2 at 7 TeV to 7.4 MJ/mm2 at 16.5 TeV. This represents almost a factor 10 increase in energy density per bunch and is not compensated by a factor 2 fewer bunches. Collimator robustness must be revisited with the significantly increased energy densities per bunch. New collimators and advanced techniques might be required to provide sufficient robustness against impact of a single or few bunches. The highly robust primary and secondary collimators for 7 TeV rely on fiber-reinforced carbon (CFC). Additional theoretical studies and experimental tests should be performed to explore the robustness limit of the CFC material. If required, new composite materials must be considered for upgraded collimators.

Second, we consider cleaning efficiency. The four-stage collimation system of the LHC was designed for high performance at a beam energy of 7 TeV. Cleaning relies on a staged process of scattering protons, then converting them into showers (inelastic interactions) and then absorbing the energy carried originally in the protons. Various nuclear processes are carefully balanced to obtain best performance. These nuclear processes depend on the beam energy and the balance will be very different at 16.5 TeV. For example, while RMS scattering angles from Multiple-Coulomb scattering (MCS) are strongly reduced with higher energy, the cross section for single-diffractive scattering increases with beam energy. As a result, cleaning efficiency is decreased with energy.

The bad impact of higher beam energy on cleaning efficiency has been shown for collimation with up to 7 TeV and is shown in Fig. 14. A modified system design, additional collimators and/or lower allowed beam losses are possible consequences of this dependence. Detailed studies are required to quantify the loss of cleaning efficiency for beam energy of 16.5 TeV.

26

Figure 14: Simulated cleaning inefficiency (leakage from the collimation system) versus beam energy up to 7 TeV. The worse performance (higher inefficiency) with high beam energy is due to the increasing probability of single-diffractive scattering for halo particles passing through the collimators over a larger number of turns, while MCS scattering angles are reduced. The effect was simulated for two different settings of collimation (tight and intermediate).

As a third issue, we discuss operational settings and impedance. The horizontal beam aperture in the dipoles is reduced from 2.2 cm to 1.3 cm after the upgrade. Assuming a 200 m horizontal beta function in the dipoles, the normalized aperture reduces from 69 σx at the present 7 TeV machine to 63 σx at 16.5 TeV after the energy upgrade (expressed in betatronic beam size only). Regarding losses in the dispersion suppressors, and if no other countermeasures are taken, it may be a reasonable assumption that the normalized collimator settings must remain at similar values as at 7 TeV, for example with primary collimators set to 6 σ, secondary collimators to 7 σ, and so on. The required full collimation gaps can then become as small as 1 mm with most full gaps around 1.6 mm. These small collimation gaps will require higher precision in collimator control, setup and reproducibility, as well as tighter orbit control. Changes to the mechanical actuation system of the collimators might be required. With the smaller gaps the collimator-induced impedance will be increased by factors between 3 and 8. This could induce additional requirements for beam stabilization with fast transverse feedbacks. However, at the higher beam energies of the HE-LHC the tightest collimator settings, at a value of 6 σ or higher, might be determined by the IR optics and the IT apertures, and not by the much larger normalized aperture of the arcs.

As a fourth area of concern, the layout and design of the cleaning insertions in IR3 and IR7 are mentioned. The use of room-temperature magnets in the cleaning insertions was required to withstand the showers induced by the collimators. Heat loads to the warm magnets can reach several kW at 7 TeV (500 kW impacting at primary collimators). This is far beyond quench levels of super-conducting magnets. The energy upgrade of the cleaning insertions requires stronger magnets. As the present warm magnets are close to technological limits, new warm technologies or solutions with large bore super-conducting magnets must be studied. The new magnets must leave sufficient space and overall phase advance for collimation, while being highly resistant to the 16.5 TeV beam losses. A number of additional absorbers might be required to sufficiently protect the new warm or super-conducting magnets.

In summary, it is noted that no obvious collimation show-stopper for an energy increase to 16.5 TeV was identified in a preliminary analysis. However, a few difficult challenges of 16.5 TeV beams for collimation have been discussed: collimator robustness with increased energy densities, decreased

27

cleaning efficiency at higher energies, operation with much smaller gaps and higher impedance, new cleaning insertion layouts with stronger magnets. Other important issues have not been discussed but are noted: activation at 16.5 TeV, environmental impact from irradiated collimators and absorption of cleaning-induced showers. Future studies must address all these issues in detail and will define appropriate solutions.

8. Conclusions and outlook

This report has presented a first look at a Higher-Energy LHC, or HE-LHC. Given the short amount of time available for the studies described, it should be considered an early status report. Nevertheless a number of conclusions can be drawn.

The proposed HE-LHC beam parameters appear less demanding than those of the present LHC. The significant beam shrinkage due to synchrotron radiation can be modified by controlled emittance blow up, so as to stay at the beam-beam tune shift limit – a simple form of luminosity leveling. The HE-LHC with parameters as presented here should be able to deliver up to 1 fb-1 per day, at a stored beam energy comparable to the one of the nominal LHC.

The cryogenics cooling capacity required is near the limit of the capacity of the existing LHC cryo-plants, which in any case would need a major overhauling after 20 years of operation. Synchrotron radiation will lead to enhanced gas desorption and the transparency of the pumping slots must be increased accordingly. A cold-bore temperature of no more than 2-3 K is advantageous for the vacuum system. Collimation becomes more challenging, with regard to collimator robustness, cleaning efficiency, impedance, operational settings, and the warm quadrupoles of the cleaning insertions.

The key ingredient of the HE-LHC is the 20-T magnets and the underlying superconductor. A block-coil geometry is presently favoured for the magnets, combining sections of Nb-Ti, Nb3Sn and HTS. There are many design options for the magnets. The performance and availability of Nb3Sn is essential for the success. Use of HTS is necessary to go beyond 16-17 T. However, decisive progress in R&D and production of HTS material and cabling (including a strong reduction of the price) is critical.

The HE-LHC study has started in April 2010 and the considerations reported here are of a preliminary nature, and so are the parameters shown. Many other important items, including machine protection and magnet protection, have not yet been looked at.

Looking forward, a vigorous R&D programme should be pursued for superconductor development (HTS and Nb3Sn), in parallel to magnet design and to short-model construction. The maximum beam energy might need to be revised in light of these future studies. More detailed investigations will be needed on how to make the collimation system work, on how to protect the machine and the magnets, on the synchrotron radiation impact on vacuum and cryogenics, etc.

28

References

[ANE03] M. Anerella et al., “The RHIC magnet system“, Nucl. Instrum. Meth. A499 (2003) 280-315.

[BEN76] C. Benvenuti et al. J.Vac.Sci. 13(6), Nov/Dec 1976, 1172-1182

[BRU04] O. Brüning et al., “LHC Design Report, v.1 : the LHC Main Ring”, CERN-2004-003-V-1 2004.

[CON85] M. Conte and M. Martini, “Intrabeam Scattering in the CERN Anti-Proton Accumulator,” Part.Accel.17:1-10, 1985.

[HAN83] R. Hanft, et al., “Magnetic field properties of Fermilab Energy Saver Dipoles”, TM-1182, 1630, 03/1983.

[LEE] P. Lee, web pages of the Applied Superconductivity center, Florida State University,

[McI05] P. McIntyre and A. Sattarov, “On the feasibility of a tripler upgrade for the LHC”, Particle Accelerator Conference (2005) 634.

[PER98] R. Perin, in “Encyclopedia of Applied Superconductivity”, (IOP, London, 1998) 919-950.

[STR91] J. Strait, et al., “Mechanical design of the 2D Cross-section of the SSC collider dipole magnet”, Particle Accelerator Conference (1991) 2176-8.

[ROS02] L. Rossi “State of the art superconducting accelerator magnets”, IEEE Trans. Appl. Supercond. 12 (2002) 219.

[TAV03] L. Tavian, “Beam Screen Regenerative Heating: Cryogenic Impact and Feasibility,” LHC-PROJECT-NOTE-330, 2003.

[WOL95] S. Wolff, “The superconducting magnet system for HERA“, proceedings of MT19, ed. By C. Marinucci and P. Waymuth (Zurich, SIN, 1995).

[ZIM01] F. Zimmermann, “Luminosity Limitations at Hadron Colliders,” HEACC’2001, CERN-SL-2001-009-AP.

[ZIM06] F. Zimmermann, “Intrabeam Scattering with Non-Ultrarelativistic Corrections and Vertical Dispersion for MAD-X,” CERN-AB-2006-002.