CERN-ATS-2012-045
20/02/2012
EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH
CERN - ACCELERATORS AND TECHNOLOGY SECTOR
Advanced Accelerator Magnets for Upgrading the LHC
L. Bottura, G.de Rijk, L. Rossi, E. Todesco
CERN, Geneva, Switzerland
The Large Hadron Collider is working at about half its design value, limited by the defective splices of the magnet
interconnections. While the full energy will be attained after the splice consolidation in 2014, CERN is preparing a
plan for a Luminosity upgrade (High Luminosity LHC) around 2020 and has launched a pre-study for exploring an
Energy upgrade (High Energy LHC) around 2030. Both upgrades strongly rely on advanced accelerator magnet
technology, requiring dipoles and quadrupoles of accelerator quality and operating fields in the 11-13 T range for the
luminosity upgrade and 16-20 T range for the energy upgrade. The paper will review the last ten year of Nb
3
Sn
accelerator magnet R&D and compare it to the needs of the upgrades and will critically assess the results of the Nb
3
Sn
and HTS technology and the planned R&D programs also based on the inputs of first year of LHC operation.
Presented at the 22nd International Conference on Magnet Technology (MT-22)
12-16 September 2011, Marseille, France
Geneva, Switzerland
CERN-ATS-2012-045
February 2012
Abstract
This work was supported in part by the European Commission under FP7-EuCARD grant 227579.
2AO-5 EDMS 1165437
1
Abstract— The Large Hadron Collider is working at about
half its design value, limited by the defective splices of the magnet
interconnections. While the full energy will be attained after the
splice consolidation in 2014, CERN is preparing a plan for a
Luminosity upgrade (High Luminosity LHC) around 2020 and
has launched a pre-study for exploring an Energy upgrade (High
Energy LHC) around 2030. Both upgrades strongly rely on
advanced accelerator magnet technology, requiring dipoles and
quadrupoles of accelerator quality and operating fields in the 11-
13 T range for the luminosity upgrade and 16-20 T range for the
energy upgrade. The paper will review the last ten year of Nb
3
Sn
accelerator magnet R&D and compare it to the needs of the
upgrades and will critically assess the results of the Nb
3
Sn and
HTS technology and the planned R&D programs also based on
the inputs of first year of LHC operation.
Index Terms— Accelerator magnets, Large Hadron Collider,
large-scale systems, superconducting magnets.
I. INTRODUCTION
HE LHC is the largest scientific instrument ever built [1],
[2] and its performance critically relies upon its 1700
large superconducting magnets[3]. After the brilliant start-up
of 10 September 2008 and the severe setback due to the
incident of 19 September 2008[4], it has resumed operation on
22 November 2009. From 30 March 2010 LHC is regularly
working [5], producing particle collisions at energy of 3.5
TeV/beam, which is half its design value. Indeed the
consequences of the incident are such that the main dipoles are
operated at 4.15 T, which is half of the design field, exceeded
by all magnets during acceptance test. The physics run will
continue also in the next year before a long shutdown in 2013-
14, scheduled to fix all bad electrical splices in the magnet
interconnects.
Despite the setback of operating at reduced energy, LHC is
exploring new territory and first important results are
approaching. The machine is beating all records for hadron
accelerators in terms not only of energy (3.5 times the
Tevatron of Fermilab) but also in term of luminosity, an
important parameters proportional to the rate of particle
collision. Actually we are not far from the design luminosity,
L= 10
34
cm
-2
s
-1
, considering that luminosity scales linearly with
Manuscript received 12 September 2011. This work was supported in part
by the European Commission under FP7-EuCARD grant 227579.
L. Bottura, G. de Rijk, L. Rossi and E. Todesco are with CERN-
Technology Department, European Organisation for Nuclear Research,
Geneva 23, CH–1211 (corresponding author: Lucio Rossi, tel. +41-22-767-
1117 e-mail: lucio.rossi@cern.ch).
energy. The magnetic system is performing very well, with an
excellent reliability and with a field accuracy even better than
the design target [6], very much due to the strict Quality
Assurance and analysis during construction and test [7], [8].
The magnetic model of the machine [9], incorporating all
superconductivity effects, like persistent currents, decay, snap
backs, as well as iron yoke saturation and hysteretic effects, is
also performing very well, allowing LHC operators to forget –
almost – that the machine requires the adjustment of some 80
magnetic circuits, a good part of them needing to be precise in
term of field at better than 10
-4
.
II. THE CERN MAGNET UPGRADE PROGRAM
Meanwhile the LHC will continue improving and producing
new physics, CERN has defined a few projects requiring the
use of SC accelerator magnets beyond 10 T:
Upgrade of the background field of the 30 kA current test
station, FRESCA; the station is based on a 10 T@1.9 K - 80
mm aperture dipole about 1 m long. The upgrade aims at a
dipole capable to produce 13 T in a 100 mm useful aperture
dipole [10]. The magnet, called FRESCA2, will have a coil
aperture of 120 mm, therefore the jump in energy and
forces beyond the present magnet is considerable.
A new 11 T dipole for improving the beam collimation
system, capable to generate a bending strength equal to
LHC main dipoles: 8.4T14.2m120 Tm, with a 3 m
shorter length, i.e., 11T11m [11]. Despite that its field is
30% higher, this dipoles must respect many constraints
imposed by their use as LHC main dipole: i) minimum 56
mm aperture, 570 mm yoke outer diameter; ii) transfer
function in Tm/A equal to the main dipole; iii) field
harmonic content very near (within few 10
-4
) to the LHC
main dipoles despite the very different iron saturation
behavior. The number of such magnets is between 10 and
20 units, on the horizon 2017-2021, according to various
scenarios for collimation upgrade.
New magnets for upgrading the Interaction Regions (IRs)
around the two high luminosity insertions (ATLAS and
CMS experiments). The most important change will
concerns sixteen low-β quadrupoles that govern luminosity
[12]. They will have all main parameters strongly enhanced
over the present ones: peak field of 13 T (+60%), aperture
120-150 mm (+100%), 8-10 m of length (+30%): the jump
in forces and stored energy is striking. Other sixteen new
magnets, with higher field and/or larger apertures, are
requested by the IRs upgrade: two types of dipoles and two
types of quadrupoles, some of them requiring probably A15
Advanced Accelerator Magnets
for Upgrading the LHC
Luca Bottura, Gijs de Rijk, member IEEE, Lucio Rossi, senior member IEEE, Ezio Todesco
T
2AO-5 EDMS 1165437
2
conductors. All will have to cope with an increased
radiation environment and must be ready by 2020 at latest.
A new twin aperture 20 T dipole for a future possible
upgrade in energy of the LHC. A preliminary study
indicated that 20 T is close to the maximum compatible
with the boundary imposed by the LHC tunnel [13]. The
challenge of such a magnet are multiple: superconductors
(not yet available), multiple grading by use of Nb-Ti, Nb
3
Sn
and HTS sections independently powered, very large forces
and inductances, huge stored energy with severe protection
issues. The mass production, eventually 20 km of twin
dipoles, demands also an affordable cost, especially for the
Nb
3
Sn and HTS superconductors. A design and possibly a
prototype must be ready on the horizon 2016-17.
All these studies and projects has been regrouped under the
project called High Luminosity LHC (HL-LHC), recently
formed at CERN with the scope to study and to implement the
necessary changes in the LHC to increase its luminosity by a
factor five around 2022. The program, which counts on the
participation of many EU partners, includes a basic R&D on
Nb
3
Sn superconductor initiated in 2004 [14] and on high field
magnet technology, initiated in 2007 and then delayed by two
years because of the LHC incident [4].
The magnet program for the LHC upgrade is more
advanced in the USA, thanks to the long term program LARP
(Lhc Accelerator Research Program) [15], [16] and the basic
programs of the various DOE laboratories. In Fig. 1 the
historic of superconducting magnets for hadron accelerators is
traced showing the objectives for the High Luminosity and the
High Energy upgrades of the LHC, while in Table I a
summary of the new magnets, of their main parameters and
installation time is reported.
Fig. 1. Field progress for main dipoles used for large colliders and the region
of interest for the next CERN projects. Main Ring and Tevatron are at
Fermilab (USA), HERA at Desy (D) RHIC at Brookhaven (USA), SPS and
LHC at CERN, Geneva (CH). For LHC the date of September 2008 is
considered, since all magnets passed nominal field, however the accelerator
will operate at maximum field after 2014.
The list of Table I deserve some comments since it is rather
inhomogeneous, comprising both R&D prototypes and
magnets that have to operate in the accelerators:
All magnets for HL-LHC must have the quality to operate
in the accelerator. The tolerance to deviation from
specification is almost zero; their reliability must be as high
as the LHC magnets to avoid downgrading performance.
The current density is almost the same for all type of
magnets, around 400 A/mm
2
at their operational field and
1.9 K. This feature is intrinsic in the optimization of the
accelerator magnets when pushed toward their limit and
when practical conditions and cost are taken into account.
For the HE-LHC for the next years we will focus on
prototypes: the issue for the cost however is critical since,
eventually, some 1200 15m-long dipoles and about 500 4m-
long quadrupoles will be needed for the project. Cost issues
are much more important for the Energy upgrade than for
the Luminosity upgrade.
In addition to the list of Table I, a number of corrector
magnets, which might also be in Nb
3
Sn, will be needed to be
designed and integrated in the main magnet cold mass.
The ambitious program of Table I is complemented by two
more programs in similar domain:
1. The construction of HTS round cables capable of 100-200
kA@5kV d.c.; this project is mainly driven by the HL-LHC
and aims to remove the power converters feeding the
magnets in the IRs or other high radiation zone from the
100 m deep tunnel up to the surface [17]; each cables is
300-600 m long and will be cooled by He gas at 4-20 K.
About 3 km of cable will be needed starting from 2014 until
2021.
2. The construction of a small prototype of a Fast Cycling
Magnet (FCM). This small prototype [18] employs a hollow
Nb-Ti cable and is used in super-ferric configuration to
yield about 2 T with a continuous field ramp of 2 T/s.
This dipole might be the prototype for a renovation of the
PS accelerator in view of its upgrade for the HE-LHC,
while a magnet that could serve for the SPS accelerator
upgrade has been manufactured by the INFN-GSI
collaboration [19] for the FAIR project.
III. SUPERCONDUCTOR DEVELOPMENT
The timely availability of a superconductor with high
current density in the targeted field range (10-15 T and above),
precise and stable geometry (2 m tolerance), tolerance to
mechanical stress and strain (150 MPa pressure), controlled
magnetization in DC and AC conditions (smaller than 100
kA/m at 1 T), and, last but not least, acceptable cost, is a
necessary condition for the success of the magnet R&D with
the ambitious targets described above. Therefore a large effort,
has been allocated to the development of Nb
3
Sn for high field
magnets in the range of 15 T, while in future we intend to
dedicate a similar effort also to HTS development for magnets
targeting the 20 T. Apart for the inherent difference among the
two technologies, the level of maturity of Nb
3
Sn is higher than
for HTS materials. For this reason the conductor program
unfolds in two directions: i) in the case of Nb
3
Sn the aim is to
demonstrate that the technology is sufficiently mature for its
first application as a main optics element in a running
accelerator, including issues of beam control, reliability and
long term operation; ii) for HTS materials the aim of the
conductor program is to explore the technology options and
verify the feasibility for accelerator application.
2AO-5 EDMS 1165437
3
TABLE I MAGNETS FOR LHC UPGRADES
Magnet type
Name
Scope
Quantity
Peak
Field (T)
Coil bore
(mm)
Length
(m)
Energy
(MJ)
F
x
(MN/m)
Deadline
(year)
Race track
SMC
R&D
10
12.5
=
0.4
0.35
2
=
Dipole
FRESCA
2
Ic Test station
1-2
13
120
1.5
3.6
15
2013
Twin dipole
LHC
=
=
8.3
56
14.3
7
3.4
=
Twin Dipole
11T
HL-LHC DS
10-20
11
60
11
(25.5)
11
7.3
2017-
2020
Quad
Low-β
Q1-Q3
HL-LHC IR
16
12
120-150
8-10
12
=
2018-
2020
Dipole
D1
HL-LHC IR
4
6-8 ?
120-150
5 ?
6
7
2019
Two-in-One
Dipole
D2
HL-LHC IR
4
3-5?
100 ?
5-10
?
?
2019
Twin Quad
Q4
HL-LHC IR
4
8
85
4.5
1.2
=
2019
Twin Quad
Q5
HL-LHC IR
4
8 ?
70
4.5
0.6
=
2019
Dipole
LHC2D
HE-LHC demo
1
20
40
1-2 m
5
20
2016
Twin Dipole
LHC2T
HE-LHC demo
1
20
40
1-2 m
10
20
2017
Although both LTS and HTS technologies have great
challenges, the program is naturally biased towards industrial
procurement of Nb
3
Sn. Overall, the conductor development
and procurement for the high-field magnet program is
expected to require approximately 25 tons of Nb
3
Sn and
funding at the level of 20 M€. For HTS conductors it is too
early to provide a forecast. For this reason, below we focus on
the work on Nb
3
Sn. The main activities of CERN on HTS
materials are summarized elsewhere [17].
At present, HL-LHC program is capitalizing on the
achievements of the development in US (DOE Conductor
Development Program (CDP) and on EU-FP6 program NED
[14] .
The US CDP, complemented by basic program of the various
DOE labs, has managed to raise the critical current density in
the non-copper cross section to values well in excess of 3000
A/mm
2
on usable piece lengths (1 km and longer) of wires
with a diameter in the range of 0.7 to 1 mm. To date, these
high J
C
wires have filament diameter of 50 m at 0.7 mm
strand diameter, or 75 m at 1 mm strand diameter [latest
OST]. The NED wire R&D culminated in the best
performance PIT 1.25 mm strand that achieved a critical
current density of 1500 A/mm
2
at 15 T and 4.2 K,
corresponding to 2700 A/mm
2
at 12 T and 4.2 K. This was
achieved at a moderate reaction temperature (625 °C) that
maximizes the final fraction of fine-grained Nb
3
Sn in the
initial Nb tube. This wire has a geometric filament diameter of
50 m, and an RRR of 200 [20].
The spectacular increase of J
C
achieved over the past 10 years
is a great success, but has also brought a number of riddles. In
some cases, magnet performance was found to be below
expectations, affected by instabilities that could be reproduced
in single strands and cables both experimentally and
theoretically [21], [22]. The basic explanation lies in the well-
known effect of flux jumps and self-field instability. Indeed,
very high J
C
is only accessible in strands of modest diameter
(typically 1 mm and smaller) if the filament diameter is small
(typically below 50 m) and the RRR is large (typically above
100). Achieving simultaneously high J
C
with small filaments
and high RRR is challenging for any of the leading wire
manufacturing routes, see Fig. 2.
Fig. 2 The performance parameter space for Nb
3
Sn.
In particular, the demand of high J
C
implies that the filament
cross section must be reacted almost completely, with the risk
of a Sn leak in the stabilizer matrix and a catastrophic drop of
RRR. In practice, a fixed thickness of Nb barrier is left
unreacted (a few m), which is essentially a lost percentage of
the filament cross section. A demand for high RRR hence
limits the maximum achievable J
C
. Similarly, reducing the
filament diameter while maintaining the thickness of
unreacted barrier, also reduces the real estate available for
reaction, and causes a reduction of the final J
C
. In summary,
critical current density J
C
, effective filament diameter and
RRR have a simple but very delicate interplay, that requires a
careful compromise in the cable design.
The above elements were instrumental in determining the
target specifications for the CERN HFM strands. Two strands
are presently on the palette, namely a large diameter strand (1
mm) for the production of the high current cable for
FRESCA2 [10], and a moderate diameter strand (0.7 mm) for
the 11 T Twin dipole (see next sections). The strand for
FRESCA2 is an evolution of the NED strand, with smaller
diameter (1 mm vs. 1.25 mm) and reduced critical current
density (1250 A/mm
2
vs. 1500 A/mm
2
at 15 T and 4.2 K), to
limit the risk of self-field instability. In the case of the strand
for the 11 T dipole, a smaller diameter is mandatory to satisfy
the constraints on available space and operating current. The
reduced filament diameter (30 m) in this case is beneficial as
it brings better field quality at injection. For both strands we
relaxed the NED specifications on RRR in view of the recent
2AO-5 EDMS 1165437
4
experimental and analytical results indicating that a lower
limit of 100 is appropriate [23].
Both leading manufacturing routes are considered for the
HFM strands, i.e. the RRP of Oxford OST, and the PIT of
Bruker-EAS. A cross section of two samples from wires
procured recently is shown in Fig. 3. The RRP wire used at
present is identical to the wire developed within the scope of
US-LARP, i.e. 108 superconducting sub-elements in a 127
stack arrangement (108/127). A new architecture is in
production, based on a 169 stack arrangement that will reduce
the sub-element dimension to approximately 50 m at a strand
diameter of 1 mm and 35 m at a strand diameter of 0.7 mm.
The 1 mm PIT strand procured has 192 tubes of 48 m
diameter. A 0.7 mm version of PIT is presently in R&D phase,
with qualification for production expected by early 2012.
Fig. 3 Nb
3
Sn layouts from leading ITD Nb
3
Sn manufacturers.
At present, three types of Nb
3
Sn Ruhterford cables are being
manufactured at CERN, using the cabling machine inherited
by the LHC project: the large size FRESCA2 rectangular cable
made of 40 strands of 1 mm diameter, its sub-scale prototype
for the SMC program, made of 18 strands of the same
diameter, and a keystoned cable of 40 strands of 0.7 mm
diameter for the MB-DS program.
The HEP-grade strands described above are delicate
material, as a general rule the cable compaction should be kept
in the range of 85 % to avoid excessive deformation and shear
of the sub-elements at the cable edges. This is much lower
than the 90 % compaction typically used for Nb-Ti Rutherford
cables and we count on a maximum cabling degradation of 10
% of the virgin strand I
c
. In practice, the cabling degradations
observed on the SMC and 11 T dipole cables are around 3 %
on average, which is a very good result. Larger degradation is
presently obtained in the FRESCA2 cable (around 18 % on
average) which is why we are still exploring the range of
cabling parameters to reduce this undesired effect.
IV. HFM R&D AT CERN AND FRESCA2 DESIGN
The aim of CERN High Field Magnet R&D program is to
develop the HFM technology for the magnets needed for the
LHC upgrade and future machines. In a first phase (2004-
2012) we focused on the development of Nb
3
Sn conductor
suitable for accelerator, base magnet construction technology,
and training of the personnel. Then in a second phase (2009-
2014) we aim at upgrading the cable test facility to 13 T
(FRESCA2), see Fig. 4, and we also aim working on design
concepts for magnets in the 15 T - 20 T domain. In this second
phase we also put the ground for design and construction of
models and later of prototypes for 11 T dipole and 13 T
quadrupole necessary for the upgrade (see next sections).
Fig. 4 Side section of the 13 T – 100 mm bore FRESCA2 dipole.
The European FP6-CARE-NED joint research activity
(2004-2008) [14], with a budget by Europe of less than 1 M€,
and more than 2 M€ provided as matching funds by the
collaborating Institutes, hosted the first phase, in which we
developed a 1.25 mm diameter Nb
3
Sn strand with European
industry, described in the previous section. In the frame of
NED design concepts for high field accelerator magnets and
insulation schemes were studied.
Beyond the official NED program, CEA-CERN-RAL-
LBNL, formed a collaboration to design small magnet with
racetrack coils: the so called “Short Model Coil”, SMC. The
main scope was to test the NED cable and provide a “fast
turnaround” test bed to qualify SC cables of new types and
new strands. The SMC program, which relies strongly on the
expertise in the US and in particular of LBNL, has already
produced 2 small magnets; the second one has been recently
tested with great success, confirming the good performance
for the NED cable. The SMC reached the design field on the
coil [24].
The second phase of the CERN high field magnet program
is carried out in the framework of the FP7-EuCARD project
[10]. The development of HFM technology is the subject of
EuCARD work-package 7 (WP7) shared by 12 partner
institutes. It runs from 1st April 2009 for 4 years with a total
budget of 6.4 M€ from which 2.0 M€ will be contributed by
the EC. Beside the technological development, the main tasks
of the WP7 is the design and construction of a 1.5 m long 13 T
dipole with an aperture of 100 mm and the development of an
High Temperature Superconductor insert with a flux density
contribution of B=6 T to be used inside the 13 T dipole. The
13 T dipole, intended to upgrade the CERN cable test facility
to higher fields (FRESCA2), features a coil block layout,
rather than a cos one. Block layout helps to limit the stress
build up, however it requires more conductor and requires the
development of flare ends, see Fig.4.
The radiation hardness of the Nb
3
Sn superconductor itself is
being assessed by CERN. Irradiation of samples with different
particle types is on-going or planned at the Atominstitut in
Vienna and the Kurchatov institute in Moscou. In parallel a
task in EuCARD is investigating the radiation resistance of the
coil insulation, aiming to produce a list of candidate radiation
resistant insulation schemes for the LHC upgrade magnets.
