Nuclear Instruments and Methods in Physics Research A 1059 (2024) 168938

Contents lists available at ScienceDirect

Nuclear Inst. and Methods in Physics Research, A

journal homepage: www.elsevier.com/locate/nima

Development of a 12 mm period cryogenic permanent magnet undulator for

time-resolved X-ray beamline on HEPS
X.Y. Li *, H.H. Lu **, Yuhui Li, S.C. Sun, W. Chen, L.L. Gong, Q. Guo, S.T. Zhao, X.Z. Zhang,
L. Zhang, Z.Q. Li, K.Y. Gong, Y.J. Sun
Key Laboratory of Particle Acceleration Physics and Technology, Institute of High Energy Physics, CAS, Beijing, 100049, China

A R T I C L E I N F O A B S T R A C T

Keywords: In order to obtain hard x-rays with the fundamental harmonic energy over 20 keV, a Pr2Fe14B based cryogenic
Insertion device permanent magnet undulator (CPMU) with the period length of 12 mm has been developed on High Energy
Undulator Photon Source (HEPS) for time-resolved beamline. Several technical difficulties arise with the further shortening
CPMU
of the CPMU period length such as magnets and poles clamping, constitutionally stable pole height shimming,
Synchrotron radiation photon source
Low emittance storage ring
multipoles shimming within the small horizontal area according to the dynamic apertures of ultra-low emittance
storage ring. Some specific treatments are adopted in response to these difficulties and proved to be effective.
Details of development process including design, assembling, methods of optimization, measurement and
shimming are presented.

1. Introduction six CPMUs to be constructed in the first phase of HEPS and is the most
challenging to develop due to its short period length. Although it is not
Synchrotron radiation has been used for over 50 years, and as necessary to strictly control the phase error for CPMU12 since only the
technology has advanced, users’ demands for brighter and more fundamental and third harmonic radiation are needed [12,13], a phase
coherent light sources have increased. This has led to the development error specification of less than 4◦ within the full range of operation gap
of fourth-generation synchrotron radiation facilities, such as Free Elec­ has been set to verify the manufacturing and adjusting techniques for all
tron Lasers (FELs) [1,2] and Diffraction-Limited Storage Rings (DLSRs) the CPMUs in the first phase of HEPS.
[3,4]. Compared to FELs, DLSRs provide transverse coherent photons CPMU12 is one of the insertion devices used for the B3 beamline,
but have higher average brightness and flux due to their high average which is specifically designed for structure dynamics experiments. The
current density. This makes DLSRs the preferred choice for upgrading photon energy range for the B3 beamline is 14.3–24 keV and 43–72 keV.
storage ring-based light sources, as they can support multiple users Two CPMUs are used in tandem for this beamline, one with a period
simultaneously. HEPS [5,6] is designed for this purpose. It has a natural length of 14.2 mm and the other being this CPMU12 with a period length
emittance of 34.5 p.m., two orders of magnitude lower than of 12 mm. The brilliance curve of CPMU12 applied in B3 beamline,
third-generation light sources. This ultra-low emittance allows for the calculated using the Gaussian-Schell approximation [14,15], is shown in
use of smaller undulators with smaller gaps, narrower good field re­ Fig. 1.
gions, and shorter period lengths. However, the reduction in undulator CPMU12 is designed to produce the fundamental harmonic photons
period length presents technical challenges, particularly in correcting with the energy range cover 20–24 keV and the brightness over 1022ph/
phase errors and multipole errors in the small good field region. (s.mm2.mrad2).0.1%B.W. Parameters of CPMU12 are present in
Based on the electron beam quality required for HEPS, users of time- Table 1.
resolved X-ray sciences [7] have requested a synchrotron radiation The development of this undulator began in September 2020. After
source that can produce photons with a fundamental energy higher than tuning and with long current coil compensation, the maximum phase
20 keV. In response to this requirement, CPMU12, a 2 m long CPMU error was successfully corrected below 3.2◦ [16] within the operating
[8–11] with a period length of 12 mm, has been designed. It is one of the gap range, meeting the specified requirements. However, a strong

* Corresponding author.
** Corresponding author.
E-mail addresses: [email protected] (X.Y. Li), [email protected] (H.H. Lu).

https://doi.org/10.1016/j.nima.2023.168938
Received 29 August 2023; Received in revised form 15 November 2023; Accepted 15 November 2023
Available online 25 November 2023
0168-9002/© 2023 Elsevier B.V. All rights reserved.
X.Y. Li et al. Nuclear Inst. and Methods in Physics Research, A 1059 (2024) 168938

dependence of the phase error on the gap was observed. The reduction in Table 1
the undulator period length also leads to a proportional reduction in the Parameters of CPMU12.
width of the magnetic poles. This presents an additional challenge in Period length (mm) 12
adjusting the integral multipole field, which was not fully anticipated
deflection value K 0.92
beforehand. The integrated field variation in the good field region ex­ Gap range (mm) 5–7
ceeds the specified limits. Periods 164
After conducting beam dynamic studies and engaging in several Temperature (K) 80
discussions, the specification requirements for the multipole field of the Material PrFeB

insertion device have been finalized. Simultaneously, a second round of

assembly and adjustment for CPMU12 was conducted to address the
remaining issues related to multipole field performance and gap
dependence errors from the first round of adjustment.
This report mainly introduces the development process of CPMU12
including design, assembling, measurement and adjusting. User re­
quirements and the final performance are also presented.

2. Design and structure

To enhance the remanent magnetization at liquid nitrogen temper­

ature and prevent the occurrence of SRT [17], CPMU12 is designed as a
hybrid permanent magnet undulator. It utilizes PrFeB [18–21] as the
magnet material and FeCoV for the magnetic poles. Both the magnets
and poles are coated with titanium nitride to reduce the vacuum vent
rate at 80K. The magnetic performance of the magnet material at this
temperature is illustrated in Fig. 2.
The PrFeB material used in CPMU12 has a remanence of 1.71T at
80K, while maintaining a coercivity (Hcj) of over 5287 kA/m. This re­
sults in a 25% increase in the peak field of the undulator at low tem­
peratures compared to room temperature, as illustrated in Fig. 3. Fig. 2. Demagnetization curve of PrFeB used in CPMU12 magnets.
The magnetic structure of this undulator was designed using RADIA
[22], as depicted in Fig. 4. The configuration of 1:1/2:3/4 [23] was attached to the in-vacuum girder. The upper and lower in-vacuum
implemented in the end field. Due to the small period length of the girders are connected to the out-vacuum girders, which are fixed on a
undulator, determining the horizontal size of the magnets is a crucial ‘C’ type frame with 7 pairs of connecting rods (Fig. 5). Each rod’s length
consideration. Increasing the magnet width raises the risk of demagne­ can be independently adjusted to correct girder deformations during
tization. Additionally, the requirement for a good field region, where the cryogenic temperatures, crucial for addressing phase errors during
peak field strength decreases by less than 1 ‰, necessitates a horizontal operation. Locking nuts outside the vacuum tube stabilize the girder
magnet size larger than 10 mm. To meet the good field region require­ deformation after tuning. This mechanical structure design has been
ment, we opted for the minimum horizontal size for the magnets. successfully tested in the early stages [24,25]. Cryogenic cooling is
However, this decision introduced additional challenges in adjusting the primarily achieved through a double row liquid nitrogen cooling pipe­
integral multipole errors due to the small horizontal size of the magnets line that runs through the entire girder (Fig. 6). This setup maintains a
and poles. The dimensions of the magnets and poles are provided in temperature variation below 1 K from end to end [26]. To maintain a
Table 2. static vacuum degree below 10-10 torr/L/s at the operating tempera­
The magnets and poles are secured using aluminum modules and ture, a total of 9 pumps are utilized. Among these, 4 are ion pumps and 5

Fig. 1. Brilliance curve of CPMU12.

2
X.Y. Li et al. Nuclear Inst. and Methods in Physics Research, A 1059 (2024) 168938

longitudinal integral field distribution and minimize multipole errors.

A magnet sorting experiment is conducted to verify the superposition
of the integral field of magnetic blocks after assembly and the effec­
tiveness of the clamping mechanical design [31]. The experiment re­
veals that when an odd number of magnets and an even number of poles
are assembled on a magnetic module, the integral field distribution of a
single module follow a linear superposition relationship. Based on these
results, the magnet clamping structure for CPMU12 is determined. It
consists of 5 magnets and 4 or 6 magnetic poles, forming a magnetic
module as shown in Fig. 9. During the first step of sorting, multipole
errors are taken into consideration.
In-vacuum undulators such as CPMU need to consider the perfor­
mance of the multipole field in the operating gap range, as well as at the
maximum gap when used in a storage ring. Existing multipole correction
methods such as magic-fingers and end poles cutting are effective for
small gaps but have limited impact on large gaps. Sorting magnets
during the assembly stage is the only way to effectively correct these
Fig. 3. Peak field measurement result of CPMU12 after pole height shimming.
integral multipole errors. Therefore, we optimize the sorting process by
considering both the horizontal distribution of integral fields at the
are NEG pumps [27]. maximum gap of 40 mm and the on-axis first integral of single modules
simultaneously. The phase error and horizontal field integral
3. Sorting and assembling
Table 2
The initial step in controlling magnetic field errors in the undulator sizes of magnets and poles.
involves sorting the magnetic blocks. This sorting process primarily fo­
horizontal vertical longitudinal
cuses on optimizing the longitudinal distribution of the first and second
integrals [28,29]. However, when applied to a storage ring, it is magnet 39 mm 26 mm 4 mm
pole 27 mm 22 mm 2 mm
important to consider optimizing the multipole field errors within the
good field region. This entails simultaneously optimizing the horizontal
distribution of the first integral at the exit of the undulator.
Instead of using a 3-axis Helmholtz coil, the integrated field mea­
surements of magnetic blocks are employed for sorting. To optimize the
multipole field error, the integral field is measured at different hori­
zontal positions. Before assembling and adjusting CPMU12, the integral
field of each magnetic block is measured at horizontal positions of 0 and
±3 mm using a short stretching wire measurement bench [30] (Fig. 7).
The stretching wire is positioned 2.6 mm above the upper surface of the
magnet. The accuracy of stretching wire measurement is 4G.cm, or 0.6G.
cm when considering the average value of 20 repeated measurements.
The installation of magnetic blocks on the in-vacuum girder of CPMU
is done in two steps. First, magnets and poles are assembled on a
clamping fixture to create a magnetic module that clamps the small-size
magnetic structure. Second, these magnetic modules are installed on the
in-vacuum girder. The clamping structure of the magnetic modules is
shown in Fig. 8. The sorting process is also divided into two steps,
following the assembly process. First, the magnets are sorted during the
module assembly based on the integral field measurement data of each
individual magnet. Once the module assembly is completed, the integral
field measurement is conducted for each module, and the modules are
sorted based on the measurement results. This two-step sorting process
helps partially reduce the nonlinear effect of magnetic poles on the
Fig. 5. photo of CPMU12 without vacuum tube.

Fig. 4. Magnetic configuration designed by RADIA.

3
X.Y. Li et al. Nuclear Inst. and Methods in Physics Research, A 1059 (2024) 168938

Fig. 6. photos of cooling pipe at both ends of the girders.

Fig. 7. stretching wire measurement for single magnet.

distribution of CPMU12 after assembly and calibration are measured

and shown in Figs. 10 and 11, respectively.

4. Measurement and adjustment

4.1. Measurement system

To evaluate the magnetic field performance of CPMU12 at different

temperatures and obtain accurate field measurements, three magnetic
measurement benches are developed for in-air and in-vacuum
measurements.
The first measurement bench consists of a 3-m long titanium-
aluminium stretch wire fixed on two rods. These rods can pass
through the bellow of the vacuum chamber and move in two dimensions
Fig. 8. Mechanical clamping and pole height tuning structure.
[32,33]. To compensate for the vacuum suction during in-vacuum
measurements, each rod is supported by a strainer that can withstand

Fig. 9. Two types of magnetic modules with 5 magnets and 4 poles (left) or 6 poles (right).

4
X.Y. Li et al. Nuclear Inst. and Methods in Physics Research, A 1059 (2024) 168938

4.2. Adjustment process

The field performance adjustment process of CPMU12 involves two

cooling cycles to 80K. This divides the adjustment process into two
steps. The first step focuses on phase error adjustment through pole
height tuning before the initial cooling to 80K. Additionally, the defor­
mation of the girders at cryogenic temperatures is corrected using 7
pairs of link rods based on measurements from the in-vacuum hall probe.
Once the phase error correction is complete and the temperature is
restored to room temperature, integral multipole field adjustment is
carried out. To minimize the multipole field at 80K, the expected

Fig. 10. phase error before the pole height tuning.

an additional tensile force of 21 N. Two motion monitors are placed

away from the measurement area, near the ‘C’ type frame. This stretch
wire bench is used to measure the first and second field integrals at both
room temperature and cryogenic temperature.
Two hall probe measurement beaches are developed for the field
measurement outside the vacuum and inside the vacuum respectively. A
three dimensions hall probe is fixed on a rod and installed on the sliding
table which can move in three orthometric directions on a 6 m long
Fig. 12. in air hall probe measurement bench.
marble platform for the out-vacuum measurement as shown in Fig. 12. It
is mainly used in the pole height tuning process. Rotation angles along
the horizontal axis and longitudinal axis of the probing rod can be tuned
independently by two rotating motors. It is convenient to correct these
two rotation angles by motors according to the measurement result of
magnetic fields in horizontal and longitudinal directions. Rotation angle
along vertical direction is corrected by tuning two foundation bolts of
the frame during the horizontal magnetic alignment.
The in-vacuum hall probe measurement bench is installed inside the
vacuum tube, following the development at ESRF. The hall probe is fixed
on a 50 mm rod and connected to a shifter that moves along a rail in the
longitudinal direction, as shown in Fig. 13. To minimize transverse
displacement during movement, the rail has a straightness tolerance of
10 μm RMS. A laser DSP system is used to measure and compensate for
any transverse positional deviation of the shifter. After mechanical
alignment, the rotating angle of the hall probe is fixed to be less than 50
μrad. This ensures that the hall probe maintains a vertical offset of less
Fig. 13. in-vacuum hall probe measurement bench.
than 2.5 μm when moving along the rail [34,35].

Fig. 11. integral field distribution along the horizontal axis.

5
X.Y. Li et al. Nuclear Inst. and Methods in Physics Research, A 1059 (2024) 168938

amount of compensation is predicted based on the difference between After three iterations, the phase error is eventually reduced to less than
the integral field measurements taken by the in-vacuum stretching wire 1◦ in all operational gaps, as shown in Fig. 15. The second field integral
at cryogenic and room temperatures. Magic-fingers and end pole cutting demonstrates a significant reduction in gap dependence. This result is
techniques are then applied to correct the multipole field within the obtained after the collimation and flatness adjustment of the girders.
operational gap range using this expected compensation. The entire During the first stage of adjustment out of the vacuum, only the
adjustment process concludes after the second cooling to 80K. phase error was corrected. At cryogenic temperature, the magnetic field
After the girder assembly, a flatness measurement and correction adjustment aims to correct the phase error by tuning the deformation of
process are carried out for both the upper and lower girders before the the upper and lower girders using gap motors and connecting rods.
phase error tuning. A capacitive position sensor, installed on the Hall Before the phase error correction, the longitudinal distribution of the
motion platform as shown in Fig. 14, is used to measure the parallelism magnetic center is fitted based on the peak field measurements taken at
of the inner surface of each girder. This allows for independent adjust­ different vertical positions and then corrected. The deviation in gap
ment of the inclination angle and flatness of the upper and lower girders. distribution along the longitudinal direction is also derived and cor­
Four motion motors and 7 pairs of connecting rods are installed on each rected based on the on-axis field measurement results.
girder for this purpose. The adjustment process began with a gap of 5.2 mm. The correction
After adjustment, the longitudinal deformation of each girder is of the magnetic center distribution along the beam direction reducing
reduced from 400 μm to 50 μm. Horizontal and vertical magnetic the RMS deviation to less than 20um. After compensating for the taper
alignment is performed for all operating gaps by measuring the magnetic angle of the girder using the gap motor, the phase error was reduced to
field distribution at off-axis positions using a hall probe. With more 6.4◦ . It took five iterations between the taper angle tuning and the girder
precise calibration and girder deformation adjustment, the gap depen­ deformation tuning to achieve a magnitude of gap variation below 5um.
dence error is significantly reduced in the subsequent adjustment. The phase error was reduced to 2.97◦ , as shown in Fig. 16. The adjust­
Phase error has been adjusted by tuning the height of each pole out of ment accuracy of the connecting rods, determined by the magnetic field
the vacuum. The phase error is defined by the phase slip factor between measurement, reached up to 1um.
each poles as [36]. The second round of performance adjustment focuses on correcting
∫ z2 the multipole field error, specifically addressing the excessive slope of
1 e2
S(z2 )− S(z1 ) = 2 1− 2 2 I(z)2 dz the horizontal integral field distribution in the good field region when
2γ z1 mc
operating at a 40 mm gap. Based on the experience gained from the
previous adjustment, it was observed that adjusting the phase error at
φi = 2π(Pi − i / 2),Pi = S(zi ) / λ0 (1)
both room temperature and cryogenic temperature significantly affects
the multipole field errors. In contrast, the multipole field adjustment
Where φi is the phase error at the ith pole, zi is the longitudinal position
mainly takes place at the end of the undulator, which does not impact
of the ith pole, S is the slippage between the radiation and the electron in
the phase error. As a result, the multipole field error adjustment is
different longitudinal positions, and I is the longitudinal distribution of
scheduled after the phase error adjustment.
the first field integral. The phase error of the undulator is determined by
Following the completion of phase error correction at a temperature
the RMS value of the phase error at each pole. Optimizing the phase
of 80K, the horizontal and vertical integral fields in the horizontal good
error is equivalent to optimizing the distribution of the phase integral
field region were measured using a stretch wire. The measurements were
component related to the second term in equation (1) at each pole. A
taken at an operational gap ranging from 5.2 mm to 7 mm, with a
linear iteration method was developed to optimize the product of Ii .ΔI
measurement step of 0.5 mm. Fig. 17 illustrates the comparison between
and ΔI while maintaining a constant effective deflection value of K.
the integral field distributions at a temperature of 80K and at room
Here, Ii represents the first field integral at each pole, and ΔI represents
temperature when operating at the minimum gap.
its deviation in half a period during the pole height changes by 100 μm.
The variation range of the horizontal and vertical integration fields is
more than 500 G cm and nearly 150 G cm, respectively. To correct the
multipole field errors over such a large range, a combination of magic
finger [37] and end pole cutting [38,39] techniques was employed. The
correction process involved three steps. Firstly, the magic finger was
used to correct the slope of the integral field at room temperature,
aiming to achieve a zero slope in the integral field at cryogenic tem­
perature. Secondly, the end pole continuous cutting was used to mini­
mize the expected fluctuation in the integral field at cryogenic
temperature. Finally, the height and angle of the standard magnetic
poles adjacent to the end poles were adjusted to correct the overall offset
of the expected integral field at cryogenic temperature.
To implement the correction, two groups of magic fingers were used.
These magic fingers were installed at the exit of the upper and lower
girders. Each magic finger frock consisted of two rows of column holes,
with each hole capable of holding up to three magnetic columns, as
shown in Fig. 18.
The magic fingers used in the adjustment are made from the same
material as the undulator magnets. An optimization process was carried
out using a random algorithm to determine the optimal height, position,
and direction of the magnetic columns for all 186 holes. Each magnetic
column has a diameter of 2 mm, and the distance between adjacent
installation holes is 1.5 mm. Due to the requirement of the horizontal
good field region being limited to ±5 mm in the horizontal direction, the
magic finger method is only capable of correcting the offset and slope of
the horizontal integral field distribution. To address the integral field
Fig. 14. Capacitive sensor (top) and its clamping structure (bottom). fluctuation in smaller transverse areas, the end pole cutting method is

6
X.Y. Li et al. Nuclear Inst. and Methods in Physics Research, A 1059 (2024) 168938

Fig. 15. phase errors (left) and 2nd integral field errors (right) at room temperature after pole height tuning.

assumes that the cutting depth of the pole is small compared to the
magnetic gap, resulting in an approximately linear effect on the mag­
netic field at the central plane. To maintain linearity, the maximum
cutting depth is limited to below 0.6 mm. Nonlinear effects are mini­
mized through multiple iterations using field simulation with RADIA.
The shape of the end poles is optimized based on the integral field dis­
tribution at a temperature of 80K. Fig. 20 illustrates the comparison of
the integral field distribution in the horizontal direction before and after
the end pole cutting at room temperature. The comparison at the
operating temperature is shown in Fig. 21.

5. Summary

CPMU12 is the first of 14 insertion devices developed for the HEPS

project. During its development, several key methods were tested to
address the challenges posed by the short period length of the undulator
Fig. 16. Phase errors at the temperature of 80 K after rods tuning. and the need to reduce phase error and integral field errors, including
integral multipole errors. A two-step cryogenic testing process was
implemented to tackle these objectives simultaneously.
utilized. This method, depicted in Fig. 19, results in a narrower hori­
The short period length of the undulator presented the first difficulty,
zontal distribution of the integral field deviation. However, it is
as it resulted in magnets and poles that were too small to be clamped
important to note that the end pole cutting method has a reduced ability
independently. To address this, a unique clamping structure was
to correct field integral errors compared to magic fingers. Additionally,
developed, allowing for the independent adjustment of the height of
this ability diminishes rapidly as the magnetic gap increases.
each pole. This structure demonstrated good stability when tested at
The surface cutting of the end poles is carried out using a line cutting
varying temperatures, ranging from room temperature to cryogenic
machine with a cutting step length of 0.5 mm. The optimization of the
temperatures.
pole surface is achieved through a linear optimization algorithm. This

Fig. 17. First integral field distributions at room temperature and at the temperature of 80 K.

7
X.Y. Li et al. Nuclear Inst. and Methods in Physics Research, A 1059 (2024) 168938

Fig. 18. photos of magic fingers (left) and end poles cutting (right).

Fig. 19. RADIA model and photo of end poles cutting.

Fig. 20. Field integral distribution at room temperature.

8
X.Y. Li et al. Nuclear Inst. and Methods in Physics Research, A 1059 (2024) 168938

Fig. 21. Field integral distribution at the temperature of 80K

The second difficulty involved shimming the multipole field integrals [11] J. Chavanne, M. Hahn, R. Kersevan, C. Kitegi, C. Penel, F. Revol, E.G. France,
Construction of a cryogenic permanent magnet undulator at ESRF, in: Proceedings
within a small horizontal good field range. The small size of the magnets
of the 11th European Particle Accelerator Conference, 2008, p. 2243. Italy.
and poles resulted in a wide variation in field integrals. Additionally, [12] R.P. Walker, Interference effects in undulator and wiggler radiation sources, Nucl.
multipole components fitted within a 10 mm horizontal range were Instrum. Methods A 335 (1993) p328–p337.
found to be more sensitive to fluctuations in field integral errors [13] R.P. Walker, Phase errors and their effect on undulator radiation properties, Phys.
Rev. ST Accel. Beams 16 (2013), 010704.
compared to those fitted in a larger range. To compensate for these [14] M. Idir, M. Cywiak, A. Morales, M.H. Modi, X-ray optics simulation using Gaussian
fluctuations, end poles cutting proved to be more effective than the superposition technique, Opt Express 19 (2011), 19050.
magic finger technique within this small horizontal good field region. A [15] R. Coisson, S. Marchesini, Gauss-schell sources as models for synchrotron
radiation, J. Synchrotron Radiat. 4 (1997) 263.
polyline pole shape cutting method was applied to shim the multipole [16] X.Y. Li, Y. Jiao, H.H. Lu, S.K. Tian, Status of HEPS insertion devices design, in:
integral field, yielding expected results at cryogenic temperatures. Proceedings of IPAC2021, 2021, pp. 339–341. Brazil.
[17] D. Givord, H. Li, R.P. de la Bâthie, Magnetic properties of Y2Fe14B and Nd2Fe14B
single crystals, Solid State Commun. 51 (1984) p857.
Declaration of competing interest [18] C. Kitegi, P. Cappadoro, O. Chubar, T. Corwin, D. Harder, P. He, H. Fernendez,
G. Rakowsky, C. Rhein, J. Rank, et al., Development of a PrFeB cryogenic
undulator at NSLS-II, in: Proceedings of IPAC2012 vol. 12, 2012, p. 762. New
The authors declare the following financial interests/personal re­ Orleans, Louisiana, USA.
lationships which may be considered as potential competing interests. [19] C. Benabderrahmane, et al., Nd2Fe14B and Pr2Fe14B magnets characterisation
Huihua lu reports administrative support was provided by National and modelling for cryogenic permanent magnet undulator applications, Nucl.
Instrum. Methods A 669 (2012) 1.
Development and Reform Commission.
[20] J. Bahrdt, H. Bäcker, M. Dirsatt, W. Frentrup, A.Gaupp, D. Just, D. Pflückhahn, M.
Scheer, B. Schulz et al., “Cryogenic Design of a PrFeB-Based Undulator”, in
Data availability Proceedings of IPAC2010, Japan, p. 3111.
[21] C. Benabderrahmane, et al., Development and operation of a Pr2Fe14B based
cryogenic permanent magnet undulator for a high spatial resolution x-ray beam
No data was used for the research described in the article. line, Phys. Rev. Accel. Beams 20 (2017), 033201.
[22] P. Elleaume, et al., Computing 3D magnetic field from insertion devices,
Proceedings PAC97 (1997) 3509–3511.
Acknowledgments
[23] K. Halbach, Some concepts to improve the performance of DC electromagnetic
wigglers, Nucl. Instrum. Methods A 250 (1986) p115–p119.
This work is supported by High Energy Photon Source (HEPS), a [24] H. H. Lu et al., “Delelopment of a PrFeB Cryogenic Permanent Magnet Undulator
major national science and technology infrastructure in China (Grant (CPMU) Prototype at IHEP”, in Proc.IPAC2017, Copenhagen, Denmark,
TUPAB064, p.1469-1471.
No. (2017)2173). [25] S. C. Sun et al., “Mechanical Design of a Cryogenic Permanent Magnet Undulator at
IHEP”, in Proc. IPAC’17, Copenhagen, Denmark, TUPAB066, p.1475-1477.
References [26] Y.C. Zhang, et al., in: German Dresden (Ed.), Preliminary Design of Cooling System
for a PrFeB-Based Cryogenic Permanent Magnet Undulator Prototype at IHEP”,
WEPRI105, Proc. IPAC’14, 2014, pp. 2743–2745. June.
[1] W.B. Colson, Theory of a free electron laser, Phys. Lett. v59 (3) (1976) p187–p190. [27] Lei Zhang, Huihua Lu, Shuchen Sun, “Design of the CPMU Vacuum System at the
[2] C. Pellegrini, A. Marinelli, S. Reiche, The physics of x-ray free-electron lasers, Rev. HEPS”, in Proc. IPAC’17, Copenhagen, Denmark, TUPAB068, p.1482-1484.
Mod. Phys. 88 (2016), 015006. [28] Yuhui Li, Bart Faatz, Joachim Pflueger, “Magnet sorting for the XFEL hybrid
[3] W. Eberhardt, Synchrotron radiation: a continuing revolution in X-ray undulator comparing study”, in Proceedings of FEL 2007, Novosibirsk, Russia,
science—Diffraction limited storage rings and beyond, J. Electron. Spectrosc. Relat. p326-p329.
Phenom. 200 (2015) 31–39. [29] G. Rakowsky, “A simple model-based magnet sorting algorithm for planar hybrid
[4] Mikael Eriksson, J. Friso van der Veen, Christoph Quitmann, “Diffraction-limited undulators”, in Proceedings of IPAC’10, Kyoto, Japan, p3201-p3203.
storage rings – a window to the science of tomorrow”, J. Synchrotron Radiat. v21 [30] C.S. Hwang, C.H. Hong, F.Y. Lin, et al., Stretch-wire system for integral magnetic
(2014) p837–842S. field measurements, Nucl. Instrum. Methods Phys. Res. 467 (2001) p194–p197
[5] Jiao Yi, Weimin Pan, High energy photon source, High Power Laser Part Beams 34 (part-P1).
(2022), 104002. [31] L.L. Gong, et al., Single-block measurement for the cryogenic permanent magnet
[6] Yi Jiao, et al., The HEPS project, J. Synchrotron Radiat. 25 (2018) p1611–p1618. undulator sorting, Rad. Detect Technol. Methods 5 (2021) p83–p89.
[7] T. Graber, et al., BioCARS: a synchrotron resource for time-resolved X-ray science, [32] J. Chavanne, G. Lebec, C. Penel, F. Revol, Recent progress in insertion devices at
J. Synchrotron Radiat. 18 (2011) p658–p670. the ESRF, Proc. IPAC’2011, San Sebastián THPC153 (2011) 3245–3247. Spain,
[8] T. Hara, T. Tanaka, H. Kitamura, T. Bizen, X. Maréchal, T. Seike, T. Kohda, Sep.
Y. Matsuura, Cryogenic permanent magnet undulators, Phys. Rev. ST Accel. Beams [33] Z. Li, W. Chen, H. Lu, et al., High-accuracy stretched-wire measurement system for
7 (2004), 050702. cryogenic permanent magnet undulator (CPMU) in High Energy Photon Source
[9] T. Tanabe, J. Ablett, L. Berman, D.A. Harder, S. Hulbert, M. Lehecka, G. Rakowsky, (HEPS), Rad. Detect Technol. Methods 4 (2020) p492–p496.
J. Skaritka, A. Deyhim, E. Johnson, et al., X-25 cryo-ready in-vacuum undulator at [34] L.L.Gong et al., “Hall element relative position and angle calibrations for the
the NSLS, Proceedings of AIP Conf 2007 (2007) p283, 879. cryogenic permanent magnet undulator”, in Proc. IPAC’16, Busan, Korea,
[10] T. Tanaka, R. Tsuru, T. Nakajima, H. Kitamura, Magnetic characterization for WEPMR048, pp. 2386-2388.
cryogenic permanent magnet undulators: a first result, J. Synchrotron Radiat. 14
(2007) p416.

9
X.Y. Li et al. Nuclear Inst. and Methods in Physics Research, A 1059 (2024) 168938

[35] Ling-Ling Gong, et al., Hall sensor angle error and relative position calibrations for [38] J. Chavanne, C. Benabderrahmane, G.L. Bec, C. Penel, “Recent Developments in
cryogenic permanent magnet undulator of high energy photon source test facility Insertion Devices at the ESRF: Working toward Diffraction-Limited Storage Rings”,
(HEPS-TF), Rad. Detect Technol. Methods 1 (2017). Article number: 17. vol. 28, Synchrotron Rad. News, 2015, pp. p15–p18.
[36] Takashi Tanaka, Universal representation of undulator phase errors, Phys. Rev. [39] M.F. Qian, Q.G. Zhou, An effective method for compensating the field integral
Accel. Beams 21 (2018), 110704. errors of hybrid in-vacuum undulators, J. Instrum. 14 (2019), T03002.
[37] E. Hoyer, S. Marks, P. Pipersky, R. Schlueter, Multiple trim magnets, or ‘magic
fingers’, for insertion device field integral correction, Rev. Sci. Instrum. 66 (1995)
1901.

10