The HL-LHC Low-β Quadrupole Magnet MQXF: From Short Models to Long Prototypes

TLDR: The inner triplet (or low-β) quadrupole magnet (MQXF) as discussed by the authors was the first to operate at a gradient of 132.6 T/m, with a conductor peak field of 11.4 T.

Abstract: Among the components to be upgraded in LHC interaction regions for the HiLumi-LHC projects are the inner triplet (or low-β) quadrupole magnets, denoted as Q1, Q2a, Q2b, and Q3. The new quadrupole magnets, called MQXF, are based on Nb3Sn superconducting magnet technology and operate at a gradient of 132.6 T/m, with a conductor peak field of 11.4 T. Q1 and Q3 are composed of magnets (called MQXFA) fabricated by the U.S. Accelerator Upgrade Project (AUP), with a magnetic length of 4.2 m. Q2a and Q2b consist of magnets (called MQXFB) fabricated by CERN, with a magnetic length of 7.15 m. After a series of short models, constructed in close collaboration by the US and CERN, the development program is now entering in the prototyping phase, with CERN on one side and BNL, FNAL, and LBNL on the other side assembling and testing their first long magnets We provide in this paper a description of the status of the MQXF program, with a summary of the short model test results, including quench performance, and mechanics, and an update on the fabrication, assembly, and test of the long prototypes.

Figures

Fig. 1. A view of the MQXFS support structure with Al dummy coils.

Fig. 14. Quench current of MQXFS5. Ramps are at 20 A/s unless indicated. Data are compared to nominal, ultimate and short-sample current (estimated from witness samples).

Fig. 12. Quench current of MQXFS3a-b. Ramps are at 20 A/s unless indicated. Data are compared to nominal, ultimate and short-sample current (estimated from witness samples).

Fig. 13. Quench current of MQXFS3c. Ramps are at 20 A/s unless indicated. Data are compared to nominal, ultimate and short-sample current (estimated from witness samples).

Fig. 15. Quench current of MQXFS4. Ramps are at 20 A/s unless indicated. Data are compared to nominal, ultimate and short-sample current (estimated from witness samples).

TABLE II

Full text

4LOr3A-01
Template version 8.0d, 22 August 2017. IEEE will put copyright information in this area
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
1
The HL-LHC Low-β Quadrupole Magnet MQXF:
from Short Models to Long Prototypes
P. Ferracin, G. Ambrosio, M. Anerella, H. Bajas, M. Bajko, B. Bordini, R. Bossert, N. Bourcey, D. W. Cheng, G.
Chlachidze, L. Cooley, S. Ferradas Troitino, L. Fiscarelli, J. Fleiter, M. Guinchard, S. Izquierdo Bermudez, S. Krave,
F. Lackner, F. Mangiarotti, M. Marchevsky, V. Marinozzi, J. Muratore, F. Nobrega, H. Pan, J.C. Perez, I. Pong, S.
Prestemon, H. Prin, E. Ravaioli, G. Sabbi, J. Schmalzle, S. Sequeira Tavares, S. Stoynev, E. Todesco, G. Vallone, P.
Wanderer, X. Wang, M. Yu
Abstract—Among the components to be upgraded in LHC inter-
action regions for the HiLumi-LHC projects are the inner triplet (or
low-β) quadrupole magnets, denoted as Q1, Q2a, Q2b, and Q3. The
new quadrupole magnets, called MQXF, are based on Nb
3
Sn super-
conducting magnet technology and operate at a gradient of 132.6
T/m with a conductor peak field of 11.4 T. The Q1 and Q3 are com-
posed by magnets (called MQXFA) fabricated by the US Accelera-
tor Upgrade Project (AUP) with a magnetic length of 4.2 m. The Q2a
and Q2b consists of magnets (called MQXFB) fabricated by CERN
with a magnetic length of 7.15 m. After a series of short models, con-
structed in close collaboration by the US and CERN, the develop-
ment program is now entering in the prototyping phase, with CERN
on one side and BNL, FNAL, and LBNL on the other side assem-
bling and testing their first long magnets. We provide in this paper
a description of the status of the MQXF program, with a summary
of the short model test results, including quench performance, and
mechanics, and an update on the fabrication, assembly and test of
the long prototypes.
Index Terms— High Luminosity LHC, Interaction Regions,
Low-β Quadrupoles, Nb3Sn magnets
I. INTRODUCTION
N order to reduce the beam size by a factor two in the inter-
action points, and to increase the rate of collisions by a factor
of five, the HL-LHC Project [1] is planning to install in the LHC
Interaction Regions (IR) new inner triplet (or low-β) quadru-
pole magnets, called MQXF [2]-[6]. With respect to the current
triplet quadrupole magnets, MQXF will feature a larger aper-
ture, from 70 to 150 mm, a higher peak field, from 8.6 to 11.4 T,
and a new superconducting material, Nb
3
Sn instead of Nb-Ti.
Out of the 30 triplets magnets (including spares) that will be
installed in the HL-LHC, 20 magnets, called MQXFA and
4.2 m long, will be fabricated by the US Accelerator Research
Program (AUP), a continuation of the LARP Program [7].
This work was supported by the U.S. Department of Energy, Office of Science,
Office of High Energy Physics, through the US LHC Accelerator Research Pro-
gram (LARP) and the US LHC Accelerator Upgrade Project (AUP), and by the
High Luminosity LHC project at CERN.
P. Ferracin, H. Bajas, M. Bajko, B. Bordini, N. Bourcey, S. Ferradas Troitino,
L. Fiscarelli, J. Fleiter, M. Guinchard, S. Izquierdo Bermudez, F. Lackner, F. Man-
giarotti, J.C. Perez, H. Prin, E. Ravaioli, S. Sequeira Tavares, E. Todesco, G. Val-
lone, are with CERN, CH-1211 Geneva 23, Switzerland (e-mail: paolo.ferra-
cin@cern.ch).
However, articles are selected for publication by the editors of
the special issue, after consideration of suitability and peer re-
view.
The remaining 10 magnets, called MQXFB and 7.15 m long,
will be fabricated by CERN. Both MQXFA and MQXFB,
which have identical cross-sections and 3D design, will have to
produce at a nominal gradient of 132.6 T. The fabrication of the
so-called “series magnets”, i.e. the ones to be installed in the
machine, will start in 2019, and it was preceded by the devel-
opment of short model magnets, 1.5 m long, and full-length pro-
totypes, constructed and tested to characterized magnet perfor-
mance. At the time of the submission of this paper, 4 short mod-
els (MQXFS1-3-5-4) and 2 MQXFA prototypes (MQXFAP1-
2) have been tested, while the first MQXFB prototype
(MQXFBP1) is being assembled. In addition, two single-coil
tests, called MQXFSM1 and MQXFAM1 for the short and long
coils, have been carried out. We provide here a description of
the conductor and coils used in the different magnets, the pre-
loading conditions, and a summary of the quench performance.
G. Ambrosio, R. Bossert, G. Chlachidze, S. Krave, V. Marinozzi, F. Nobrega,
S. Stoynev, M. Yu are with Fermi National Accelerator Laboratory, Batavia, IL
80510 USA.
M. Anerella, J. Muratore, J. Schmalzle, P. Wanderer, are with BNL, Upton, NY
11973-5000, USA.
D. W. Cheng, M. Marchevsky, I. Pong, S. Prestemon, G. Sabbi, X. Wang, are
with Lawrence Berkeley National Lab, Berkeley, CA 94720, USA.
L. D. Cooley is with the Applied Superconductivity Center, National High
Magnetic Laboratory, Tallahassee, FL 32310, USA
I
Fig. 1. A view of the MQXFS support structure with Al dummy coils.
This manuscript has been authored by Fermi Research Alliance, LLC under Contract No. DE-AC02-07CH11359 with the U.S.
Department of Energy, Office of Science, Office of High Energy Physics.

2
II. SUPERCONDUCTING STRAND AND CABLE
The MQXF coils are wound with a cable composed by 40
strands with a diameters of 0.85 mm. For all the MQXFA series
magnets, RRP 108/127 strands from Bruker will be used. The
same strand will be utilized for 8 of the 10 MQXFB series mag-
nets, while in the remaining two Bruker’s PIT 192 with bundle
barrier will be used [8]. In addition to these two types of strands,
the short models and the prototypes employed also different
strand architectures, namely RRP 132/169, RRP 144/169, and
PIT 192 (without bundle barrier). In Fig. 2, the strands’ cross-
sections are shown, while the strand specification are given in
Table I. Both strands must have a critical current >331 A at 15
T, while a 7% lower critical current at 12 T was set for the PIT
strand (>590 A instead of >632 A).
Fig. 2. Superconducting strands used for the MQXF program (left to right):
RRP 108/127, RRP 132/169, PIT 192, PIT 192 with bundle barrier.
TABLE I
STRAND SPECIFICATIONS
Parameter
Unit
RRP
PIT
Strand diameter
mm
0.85
Sub-element diameter
µm
≤55
Filament twist pitch
mm
19±3
Cu/SC
1.2±0.1
RRR
>150
I
c
(12 T, 4.2 K), no self-field corr.
A
>632
>590
I
c
(15 T, 4.2 K), no self-field corr.
A
>331
>331
Non-Cu J
c
(12 T, 4.2 K), no self-field corr.
A/mm
2
>2450
>2290
Non-Cu J
c
(15 T, 4.2 K), no self-field corr.
A/mm
2
>1280
>1280
TABLE II
CABLE SPECIFICATIONS
Parameter
Unit
Number of strands in cable
40
Cable bare width (before/after HT)
mm
18.150/18.363
Cable bare mid-thick. (before/after HT)
mm
1.525/1.594
Cable bare inner-thick. (before/after HT)
mm
1.462/1.530
Cable bare outer-thick. (before/after HT)
mm
1.588/1.658
Cable width expansion during HT
%
1.2
Cable mid-thick. expansion during HT
%
4.5
Keystone angle
Deg.
0.40
Pitch length
mm
109
Cable core width
mm
12
Cable core thickness
µm
25
Cabling I
c
degradation
%
<5
RRR after cabling
>100
Insulation thickness per side at 5 MPa
µm
145±5
The cable, whose parameters are given in Table II, is 18.150
mm wide, and it contains a 316L stainless steel core 25 µm thick
to reduce dynamics effects during magnet ramping. Both the
design of the coil fabrication tooling (in particular the curing/re-
action/impregnation cavity size) and the coil cross-section have
been defined assuming a cable expansion during heat treatment
of 4.5% in thickness and 1.2% in width [9]. The critical current
degradation due to cabling is set as <5%.
In Fig. 3 the strand critical current specifications, including
self-field correction and 5% cabling degradation, are fitted with
a parameterization curve [10] and compared with the magnet
load-line (conductor peak field vs. magnet current). The short
sample currents I
ss
, which represents the magnet’s current limits
and are obtained from the intersection of the magnet load-line
with the 1.9 K critical curves, are 21.26 kA for the RRP and
20.89 kA for the PIT. This means that, at the nominal current
I
nom
of 16.47 kA, the magnet will operate at a percentage of
short sample of 77% (RRP) and 79% (PIT)
.
Also, the according
to the HL-LHC project requirements, the MQXF magnet must
be able to reach an ultimate current I
ult
8% higher than I
nom
, that
is 20.89 kA. At this current, the percentages of I
ss
are 84%
(RRP) and 86% (PIT).
Fig. 3. Strand critical currents vs. total magnetic field (including self-field
correction and 5% cabling degradation): specifications for RRP 108/127 and
PIT 192 with bundle, and magnet load-line.
III. COIL FABRICATION
Since the beginning of the development of the MQXF mag-
net, the fabrication of the coils has proceed in parallel in the US
laboratories and at CERN [11]-[16]. Short model coils from US
and CERN were identical, and therefore usable in the same
magnet. Prototype and series coils are instead different in
lengths, consistently to the magnetic lengths of MQXFA
(4.2 m) and MQXFB (7.15 m). The main parameters of the coils
implemented in MQXF magnets are given in Table III. Both
short coils and MQXFA coils were fabricated using two differ-
ent cable design: a 1
st
generation cable with keystone angle of
0.55, and a 2
nd
generation cable, where the angle was de-
creased to 0.40 to reduce the critical current degradation due
to cabling for both RRP and PIT strands. A second modifica-
tion, which took place during the coil fabrication, was the in-
crease of the magnetic length for MQXFA/B from 4.0/6.8 m to
4.2/7.15 m; the first MQXFA coils were still 4.0 m long.

3
A. Winding, Curing, Reaction, Impregnation
The MQXF coils are composed by 50 turns, wound in 2 layers
around a Ti-alloy pole with a single unit length of cable (no in-
ternal splices). Each layer is divided in 2 blocks per quadrant.
The coil winding is performed by keeping a tension on the cable
of 25 kg. After the first layer is wound, polymer-derived ce-
ramic binder CTD-1202 is applied to the S2 glass insulation of
the cables; the layer is then cured in two steps, first at 80 °C for
2 h, then at 170 °C for 3 h. The same operation is applied to the
second layer, after its winding on top of the first layer. Once the
winding and curing is completed, the coil is placed in a reaction
mold and heat treated in an oven under argon flow. The heat
treatments are based on the following schedules: 48 h at 210
(ramp 25 °C/h), 48 h at 400 °C (ramp 50 °C/h), 50 h at 665 °C
(ramp 50 °C/h) for the RRP, and 40 h at 415 °C, 120 h at 620
°C, 200 h at 645 °C (all with ramp 30 °C/h) for the PIT with
bundle. Before transferring the reacted coil in the impregnation
mold, printed circuits (traces) with quench heaters and voltage
taps are placed on top of the outer layer and connected. At the
same time, the splicing operation, consisting in soldering Nb-Ti
cables to the Nb
3
Sn coil leads, is executed. The impregnation
process consists in inserting the coil, placed inside a dedicated
mold, inside a vacuum tank, and injecting CTD-101K epoxy
resin system at atmospheric pressure (MQXFA) or at 2 bar
(MQXFB). The epoxy is injected with resin and mold at a tem-
perature of 60 °C. The epoxy curing is done in two plateaus, the
first at 110 °C for 6 h and the second at 125 °C for 16 h.
B. Coil Dimensional Measurements
Before the magnet assembly, coils dimensions are measured
using a Coordinate Measurement Machine (CMM). Data are ac-
quired on the outer radius and the mid-planes of the coils in
different locations along the longitudinal direction [17]. By
aligning the data on the nominal outer radius of the impregnated
coil (113.376 mm), it is possible to estimate the deviations of
the azimuthal dimensions with respect to the nominal values.
Fig. 4. Azimuthal coil size deviation (left + right mid-plane) with respect to
nominal dimension for short model 1
st
(I) and 2
nd
(II) generation coils: from left
to right, coils for MQXFS1-3-5-4.
Fig. 5. Azimuthal coil size deviation (left + right mid-plane) with respect to
nominal dimensions for prototype coils: from left to right, coils for MQXFAP1-
2 and MQXFBP1.
For the short model coils, 7 cross-sections at a 150 mm dis-
tance along the coil straight section are analyzed, while for the
MQXFA (MQXFB) the measurements are taken on respec-
tively 11 (32) locations 420 (200) mm apart. In Fig. 4 and 5, the
TABLE III
PARAMETERS OF COIL USED IN SHORT MODELS AND PROTOTYPES
Coil
Laboratory
a
Strand
Cross-section
L
b
Magnet
2
LARP/AUP
RRP 108/127
1
st
gen.
1.19
MQXFSM1
103
CERN
RRP 132/169
1
st
gen.
1.19
MQXFS1a-d
104
CERN
RRP 132/169
1
st
gen.
1.19
MQXFS1a-d
3
FNAL/BNL
RRP 108/127
1
st
gen.
1.19
MQXFS1a-d
5
FNAL/BNL
RRP 108/127
1
st
gen.
1.19
MQXFS1a-d
105
CERN
RRP 132/169
2
nd
gen.
1.20
MQXFS3a-c
106
CERN
RRP 132/169
2
nd
gen.
1.20
MQXFS3a-c
107
CERN
RRP 132/169
2
nd
gen.
1.20
MQXFS3a-c
7
FNAL
RRP 108/127
2
nd
gen.
1.20
MQXFS3a-b
8
FNAL/BNL
RRP 144/169
2
nd
gen.
1.20
MQXFS3c
203
CERN
PIT 192
2
nd
gen.
1.20
MQXFS5
204
CERN
PIT 192
2
nd
gen.
1.20
MQXFS5
205
CERN
PIT 192
2
nd
gen.
1.20
MQXFS5
206
CERN
PIT 192
2
nd
gen.
1.20
MQXFS5
108
CERN
RRP 108/127
2
nd
gen.
1.20
MQXFS4
109
CERN
RRP 108/127
2
nd
gen.
1.20
MQXFS4
110
CERN
RRP 108/127
2
nd
gen.
1.20
MQXFS4
111
CERN
RRP 108/127
2
nd
gen.
1.20
MQXFS4
QXFP01
FNAL/BNL
RRP 108/127
1
st
gen.
4.00
MQXFAM1
QXFP02
FNAL/BNL
RRP 132/169
1
st
gen.
4.00
MQXFAP1
QXFP03
FNAL
RRP 144/169
2
nd
gen.
4.00
MQXFAP1
QXFP04
FNAL/BNL
RRP 132/169
2
nd
gen.
4.00
MQXFAP1
QXFP05
FNAL
RRP 108/127
2
nd
gen.
4.00
MQXFAP1
QXFA102
FNAL
RRP 108/127
2
nd
gen.
4.20
MQXFAP2
QXFA104
FNAL/BNL
RRP 108/127
2
nd
gen.
4.20
MQXFAP2
QXFA105
FNAL
RRP 108/127
2
nd
gen.
4.20
MQXFAP2
QXFA106
FNAL/BNL
RRP 108/127
2
nd
gen.
4.20
MQXFAP2
104
CERN
RRP 108/127
2
nd
gen.
7.15
MQXFBP1
105
CERN
RRP 108/127
2
nd
gen.
7.15
MQXFBP1
107
CERN
RRP 108/127
2
nd
gen.
7.15
MQXFBP1
108
CERN
RRP 108/127
2
nd
gen.
7.15
MQXFBP1
a
Laboratory where the coil was produced. The case “FNAL/BNL” refers to
coils wound/cured at FNAL, and reacted/impregnated in BNL.
b
Magnetic length (m).

4
azimuthal deviations (left + right mid-plane) for each of the
tested coils with respect to nominal dimensions are given in the
form of a box plot: the horizontal lines indicate the minimum,
the 25% percentile, the median, the 75% percentile, and the
maximum deviations. The short coils have a size variation
along the length up to 0.250 mm, and a median value ranging
from -0.200 to +0.250 mm. No significant difference is found
between RRP of PIT coils, or between first and second genera-
tion coils. In the case of the prototype coils, the medians range
between -0.100 and +0.050 mm, but MQXFA coils show a sig-
nificant smaller variation along the axis with respect to
MQXFB coils.
IV. MAGNET ASSEMBLY AND LOADING
The measurements of the coil dimensions provide the inputs
to define a shimming plan, the first step of the coil-pack assem-
bly. In order to compensate for size deviations, coils are
shimmed radially and/or on the mid-plane, so that the final outer
radius of the four coils coincides. A 2D magnetic analysis is
performed to determine the coil locations within the four quad-
rants to minimize the un-allowed harmonics. Then, the assem-
bly of the MQXF structure, described in detailed in [18]-[21]
and showed in Fig. 6, and its pre-load with water pressurized
bladders are carried out.
Fig. 6. MQXF cross-section (top), and side view of MQXFB (bottom). The
red and blue markers (top) indicate the locations of the strain gauges on coils
and shell, and the vertical lines (bottom) their longitudinal positions.
In Fig. 7, a summary of the pre-load of the tested MQXF mag-
nets is depicted: the azimuthal stress measured on the coil is
plotted as a function of the azimuthal stress measured on the
shell [22], [23]. Data from all the magnets are compared with
numerical computations simulating the case with full contact
between collars and pole keys and the case without pole keys.
For a given tension in the shell, the coil compression can be
increased by applying a gap between the pole key and the col-
lars, thus reducing the compressive force intercepted by the col-
lars.
Fig. 7. Azimuthal stress measured in the winding pole vs. azimuthal stress
measured in the aluminum shell, both at 293 K. The gaps/interferences in the
legend are between collars and pole keys, per side. Data from tested magnets
are compared with computed values considering the case with full contact be-
tween collars and pole keys and the case without pole keys.
Fig. 8. Azimuthal stress measured in the winding pole vs. total longitudinal
force applied by the axial load system, after cool-down. Data from tested mag-
nets are compared to target values to (y axis) prevent azimuthal unloading of
the coil and to (x axis) equal the longitudinal electro-magnetic forces.
Several configurations have been explored in the short mod-
els and the MQXFA prototypes, ranging from an interference
of 0.100 mm in MQXFS3a-b to a gap of 0.200 mm in
MQXFS3c, resulting in a coil pre-load varying from -60 MPa
to -110 MPa. Measured data are consistent with the computa-
tions: the larger the pole key gap, the closer are the data to the
“no pole key” computed line.
After cool-down, an increase of shell stress, caused by its
high thermal contraction, produces an increase of coil azimuthal
compression. The data for the tested magnets are shown in Fig.
8 (y axis) and compared with target values: the two horizontal
lines indicate the levels of coil compression that, according to
computations and strain gauge measurements [24], prevent at
I
nom
and I
ult
an azimuthal unloading of the coil pole turns due to
the electro-magnetic (e.m.) forces. In addition, the plot provides

5
the total longitudinal pre-load given to the coils by the axial
support system, and compare it with the axial e.m. forces (see
Table IV) at I
nom
and I
ult
(vertical lines). It can be noticed that
after cool-down, a conservative approach with low pre-load was
chosen for MQXFS1, and a progressive increase toward higher
pre-loads was pursued in the following magnets. For the
MQXFA prototypes, a low azimuthal and an intermediate axial
pre-load were chosen.
TABLE IV
COIL AND MAGNET PARAMETERS
Parameter
Unit
Coil clear aperture diameter
mm
150
Magnet (LHe vessel) outer diameter
mm
630
No. turns in layer 1/2 (octant)
22/28
Operational temperature T
op
K
1.9
Magnetic length (Q1-Q3)/(Q2)
M
4.20/7.15
Nominal gradient G
nom
T/m
132.6
Nominal current I
nom
kA
16.47
Nominal conductor peak field B
op
T
11.4
I
nom
/ I
ss
at 1.9 K for RRP/PIT (specs.)
%
77/79
Ultimate gradient G
ult
T/m
143.2
Ultimate current I
ult
kA
17.89
Ultimate conductor peak field B
ult
T
12.3
I
ult
/ I
ss
at 1.9 K for RRP/PIT (specs.)
%
84/86
Stored energy density at I
nom
(Q1-Q3)/(Q2)
MJ/m
1.17
Differential inductance at I
nom
mH/m
8.21
Stored energy at I
nom
(Q1-Q3)/(Q2)
MJ
4.91/8.37
F
x
/ F
y
(per octant) at I
nom
MN/m
+2.47/ -3.48
F

layer1/layer2 (per octant)
MN/m
-1.84/-2.14
F
z
(whole magnet) at I
nom
MN
1.17
V. QUENCH PERFORMANCE
Both short models and MQXFA prototypes test campaigns
started with so-called single-coil tests, where individual coils
were assembled inside an iron structure (so-called mirror con-
figuration) without pre-load and powered at 1.9 K. Although
not representative of the mechanical conditions of the full quad-
rupole magnets, the single coil configuration is characterized by
a load-line comparable to the MQXF quadrupole magnet, and
therefore it can provide an early feed-back on the coil design
and fabrication process. The test of the first MQXFS coil
(MQXFSM1) was carried out at FNAL in May 2015, while the
test of the first MQXFA coil (MQXFAM1) was performed at
BNL in December 2016. In both tests (Fig. 9) the coils passed
the ultimate current and reached about 87% of I
ss
at 1.9 K. After
the single coil tests, four short model magnets (MQXFS1-3-5-
4) and two MQXFA prototype magnets (MQXFAP1-2) have
been powered. Coil parameters and loading conditions for these
magnets can be found in Table III and Fig. 7 and Fig. 8. Their
quench performance are provided in the next sub-sections, and
compared with I
nom
, I
ult
and I
ss
. The latter was evaluated by
measurements of witness samples, that is strands extracted from
the cables used to wind the coils, and reacted with the coils.
A. MQXFS1
The first short model magnet tested is MQXFS1. The test
started at FNAL in March 2016 [25]-[27]. The magnet used first
generation coils, with RRP 108/127 and 132/169 conductor,
fabricated at FNAL and BNL (2 coils), and at CERN (2 coils).
In term of pre-load, a conservative approach was adopted, with
about -80 MPa on the coil after cool-down, and a longitudinal
pre-load of about half of the axial e.m. force (see Fig. 8). The
magnet had a first quench at 14.2 kA, reached I
nom
in 8 quenches
and I
ult
in 16 quenches. At 4.5 K, MQXFS1 maintained the same
quench current and after thermal cycle exhibited perfect
memory.
Fig. 10. Quench current of MQXFS1a-b. Ramps are at 20 A/s unless indi-
cated. Data are compared to nominal, ultimate and short-sample current (esti-
mated from witness samples).
Fig. 11. Quench current of MQXFS1c-d. Ramps are at 20 A/s unless indi-
cated. Data are compared to nominal, ultimate and short-sample current (esti-
mated from witness samples).
Since the pole gauges clearly indicated unloading during
powering, after the thermal cycle MQXFS1 was warmed-up, re-
loaded to a higher azimuthal pre-load and tested at 1.9 K as
MQXFS1b. Despite about 6 detraining quenches, the magnet

Frequently asked questions

Q1. What contributions have the authors mentioned in the paper "The hl-lhc low-β quadrupole magnet mqxf: from short models to long prototypes" ?

Among the components to be upgraded in LHC interaction regions for the HiLumi-LHC projects are the inner triplet ( or low-β ) quadrupole magnets, denoted as Q1, Q2a, Q2b, and Q3. 4 T. The Q1 and Q3 are composed by magnets ( called MQXFA ) fabricated by the US Accelerator Upgrade Project ( AUP ) with a magnetic length of 4. 2 m. The authors provide in this paper a description of the status of the MQXF program, with a summary of the short model test results, including quench performance, and mechanics, and an update on the fabrication, assembly and test of the long prototypes. 

Q2. What was the first test of the MQXFA prototype?

Both short models and MQXFA prototypes test campaigns started with so-called single-coil tests, where individual coils were assembled inside an iron structure (so-called mirror configuration) without pre-load and powered at 1.9 K. 

Q3. What is the optimum preload for the MQXFA coil?

It can be noticed that after cool-down, a conservative approach with low pre-load was chosen for MQXFS1, and a progressive increase toward higher pre-loads was pursued in the following magnets. 

Q4. What is the composition of the MQXF coils?

The MQXF coils are composed by 50 turns, wound in 2 layers around a Ti-alloy pole with a single unit length of cable (no internal splices). 

Q5. What is the process of transferring the reacted coil?

Before transferring the reacted coil in the impregnation mold, printed circuits (traces) with quench heaters and voltage taps are placed on top of the outer layer and connected. 

Q6. What is the strand diameter of the MQXF coil?

Strand diameter mm 0.85 Sub-element diameter µm ≤55 Filament twist pitch mm 19±3 Cu/SC 1.2±0.1 RRR >150 Ic (12 T, 4.2 K), no self-field corr. 

Q7. What is the critical current for the PIT 192 strand?

Both strands must have a critical current >331 A at 15 T, while a 7% lower critical current at 12 T was set for the PIT strand (>590 A instead of >632 A). 

Q8. What are the strands used for the MQXF program?

2. Superconducting strands used for the MQXF program (left to right): RRP 108/127, RRP 132/169, PIT 192, PIT 192 with bundle barrier. 

Q9. How many quenches did the magnet achieve?

As shown in Fig. 12, the training started at 14.9 kA, and Inom was achieved in 7 quenches, but after quench #19 a progressive drop in current was recorded. 

Q10. What was the axial preload of the MQXFS3?

MQXFS3 was assembled with 2nd generation coils, fabricated at FNAL and BNL and at CERN with RRP 108/127, 132/196, and 144/169 conductor. 

Q11. What is the strand size of the MQXF coil?

The cable, whose parameters are given in Table II, is 18.150 mm wide, and it contains a 316L stainless steel core 25 µm thick to reduce dynamics effects during magnet ramping.