Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

TUTORIAL ON LOW BETA CAVITY DESIGN

A. Facco, INFN – LNL

Abstract
Low beta superconducting cavities are gaining more The first superconducting low-β resonators have been
and more interest in the particle accelerators community. used from the 70’s in linac boosters for tandem
Their use, once limited to low current, low energy heavy accelerators. The low current, cw heavy ion beams
ion beams, was recently extended to high current proton produced by tandems, with wide range of mass to charge
and deuteron linacs for energy production, spallation ratio and 0.05≤β≤0.2, were an ideal application for
neutron sources, radioactive beam production, nuclear superconducting structures with 2÷3 gap, that could be
waste transmutation and other applications. powered with a few tens of Watts only.
A large number of different resonator geometries is In the last decade, however, an important boost to low-
now available, with peculiar characteristics suitable for β cavity development was given by R&D programs on
different applications. In spite of their complex structures high intensity proton and ion drivers for nuclear waste
and construction difficulties, their performance, in terms transmutation, material irradiation, accelerator driven
of surface resistance and maximum achievable electric reactors, radioactive beam production and post-
and magnetic fields, is comparable to the one of high acceleration, and other applications (see table 1). This
gradient elliptical cavities. extended the resonator families and also their
An introduction on low-beta resonators types, performance.
characteristics and design techniques, including
accessories which are typical of this class of cavities, will
be given. Type βmax A/q current mA

1. INTRODUCTION Heavy ion linacs for

~ 0.2 <10 < 10-4
Low-β resonators are just cavities that accelerate nuclear physics
efficiently particles with velocity β<1. This set is usually
Post-accelerators for
further subdivided in low-, medium- and high-β ones, ~ 0.2 (0.5) 7÷ 66 < 10-6
RIB facilities
centered around β=0.1 in linac boosters for Tandem
accelerators, and around β=0.3÷0.6 in all other SC linacs HI drivers for RIB
communities. ~ 0.3÷0.9 ~1÷10 ~0.1÷10
facilities
Differently from β=1 cavities, that have all elliptical-
like profiles (see, e.g. ref. [1]), low-β ones have many
different shapes, sizes and EM modes to allow for a rather p,d linacs ~ 0.3 1÷2 ~1÷10
large variety of beams that differ in velocity, intensity and
A/q. Important differences are also the higher peak High Power Proton
~ 0.9 1 ~10÷100
surface fields Ep and Bp required to get the same Accelerators
acceleration, and the lower rf frequency.
High Power Deuteron
A typical Superconducting low-β linac consists of
Accelerators for material ~ 0.3 2 ~100
many short cavities, independently powered and phased
irradiation
(ISCL), with a small number of gaps (typically 2 or 3) and
relatively large aperture compared to normal conducting Table 1. Low-β resonator applications.
ones. This structure allows a large velocity acceptance,
different beam velocity profiles along the linac and thus 2. BASIC DEFINITIONS
the capability of accelerating efficiently particles with For tutorial purposes, in this introductory paper we will
different A/q. Moreover, this allows to some extent linac use beam dynamics and rf formulas in their simplest form.
lattices with cavity fault tolerance. For a deeper understanding and a more general approach
see ref. [2] and [3].
β=0.07 β=0.12
A resonator working in its accelerating mode has a time
dependent electromagnetic field of the form
G G
E = E ( x, y, z ) cos(ωt + ϕ )
G G
Figure 1. Example of low-β SC linac scheme: the B = B( x, y, z ) sin(ωt + ϕ )
SPIRAL-2 driver (GANIL), with cryostats housing 1- and To characterize superconducting resonators, parameters
2- double gap cavities each. are used that relate their measurable rf and

SUA04 21
 Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

electromagnetic characteristics. Some of the most ∆Wp lower than qVg, due to the change in the field during
important ones are shown in Table 1. All parameters in the finite time of transit.
the table are constants that depend on the cavity L/2
geometry, surface resistance and accelerating field [4][5]. ∆W p = q ∫ E z ( z p , t )dz p (3)
−L / 2
Using the constant velocity approximation, and
Stored energy U/ Ea2 J/(MV/m)2 choosing t=0 when the particle crosses z=0, then t=zp/βc
2
Shunt impedance Rsh=Ea L/P MΩ/m and the instantaneous accelerating field becomes
Quality Factor Q=ωU/P Ez(zp)cos(ωzp/βc+φ). Introducing the transit time factor
Geometrical factor Γ = Q Rs Ω L/2 ⎛ ωz ⎞
Peak electric field Ep/Ea
∫
−L / 2
E z ( z ) cos⎜⎜ ⎟⎟dz
⎝ βc ⎠ ≤ 1 (4)
Peak magnetic field Bp/Ea mT/( MV/m)
T (β ) =
Optimum β β0 L/2

Cavity length L m ∫ Ez ( z )dz

−L / 2
Table 2. Main parameters that characterize a we obtain the usual expression for the energy gain:
superconducting cavity.
∆W p = qE a LT (β ) cos ϕ (5)
Here:
ω =2πf rf frequency
Ea =Average accelerating field (MV/m) The transit time factor for 1 gap of length g is
P =rf power losses in the cavity (W), proportional to Rs particularly simple in the approximation of constant Ez in
Rs= RBCS +RRES surface resistance of cavity walls Ω. the gap and zero outside:
⎛ πg ⎞ ⎛ πgf ⎞
sin ⎜⎜ ⎟⎟ sin ⎜⎜ ⎟
The residual resistance RRES depends on the quality and
⎝ βλ ⎠ ⎝ βc ⎟⎠
purity of the rf surface. T (β ) = ≡ (6)
RBCS is the theoretical value for pure Nb: ⎛ πg ⎞ ⎛ πgf ⎞
⎜⎜ ⎟⎟ ⎜⎜ ⎟⎟
f 2 ( MHz) ⎛ 17.67 ⎞ ⎝ βλ ⎠ ⎝ βc ⎠
RBCS (nΩ) ≅ 0.09 ⋅ ⋅ exp⎜ − ⎟ (1)
T (K ) ⎝ T ⎠
The accelerating field on the beam axis is usually The real field in the gap, however, is not constant and
axially symmetric and can be expressed by the beam port aperture allows the field to extend beyond
Ez(r,z,t)=Ez(r,z)cos(ωt+φ) g. This results in an “effective gap length” larger than g,
(For simplicity, we assume to be on axis, so that r=0, approximately g eff ≈ g 2 + (2b )2 where b is the bore
and Ez(0,z) ≡ Ez(z) ).
radius.
An example of a 1 gap transit time factor, using two
different aperture values, is shown in figure 3.

1
g 15 mm
T (β) 0.75
1 gap
30 mm
0.5 g=30 mm
b b=15 and 30 mm
0.25 f=350 MHz

0
0 2g/λ 0.2 0.4
β 0.6 0.8 1

Figure 3. Single-gap transit time factor curves, for the

same gap length and different apertures.
Figure 2. 1-gap resonator schematic and definitions.
Acceleration takes place efficiently only above some
We can define the average accelerating field critical β0~2geff/λ and it is maximum at β=1. Thus, to be
1 L/2 efficient at low-β, it is necessary to work at low rf
L ∫− L / 2
Ea = Ez ( z )dz (2) frequency, short gap length and small beam port aperture.
Rules of thumb for a good design are: 2gf<βc and 2b<g.
In this definition, EaL is the maximum gap voltage Vg. For 2 equal gaps at a distance d, working in the usual π-
A particle traversing the cavity with velocity βc, charge mode (i.e. where there is a phase delay of π between the
q and coordinate zp, however, has always an energy gain EM field in the two gaps) we have:

22 SUA04
 Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

⎛ πg ⎞ The requirements of low-β cavities carry some negative

sin ⎜⎜ ⎟ consequences. Short gaps provide little energy gain; low
⎝ βλ ⎟⎠ ⎛ πd ⎞
T (β ) = sin ⎜⎜ ⎟ (7)
rf frequency results in large resonators or complicated
⎛ πg ⎞ ⎝ βλ ⎟⎠ shapes; small aperture means low transverse acceptance
⎜⎜ ⎟⎟ for the beam. All these items result in cavities with rather
⎝ βλ ⎠ high peak-to-accelerating field ratios.
In general, for a resonator with 2 or more equal gaps, Superconducting resonators are particularly suitable for
T(β) consists of the single gap term plus a term counteracting these drawbacks, thanks to their low power
accounting for the sinchronization of the field with the dissipation and the high surface fields that they can
particle phase in different gaps. In this case, T has a achieve. Due to the low frequency, the surface resistance
maximum T(β0)<1 at some optimum velocity β0<1. in equation (1) is rather low already at 4.2K; differently
1 from most β=1 ones, low-β cavities usually do not need to
be cooled below this temperature.

0.5 Remark: different definitions of gradient

The resonator performance is usually represented in a Q
vs. Ea plot. With the many different shapes and
0
β0 geometries available for low beta cavities, it is sometimes
difficult to find a unique definition of the effective length
0.5
L which appears in the average gradient expression. The
0.2 0.4 0.6 0.8 most used definitions are (see figure 6) lint , Lmax or even
Figure 4. Example of transit time factor for a 2-gap nβλ/2, where n is the gap number.
resonator (π-mode). Red curve: 1-gap term; blue curve:
total T(β).

It is usually better to normalize the transit time factor

and the accelerating field expressions:

T (β )
T * (β ) = (8) lint
T (β 0 )

T (β 0 ) L / 2
L ∫− L / 2
E *a = E z ( z )dz (9) Lmax
nβλ/2
Thus T*(β0)=1, and E*a is not anymore the gap voltage n gap N.
but the real field seen by the particle at optimum β and ϕ.
(From now on we will use normalized quantities and we Figure 6. Different definitions of accelerating length in
will omit the asterisk). The energy gain expression (5) a 2-gap quarter wave resonator.
does not change.
1 All definitions are consistent and the energy gain of a
resonator has of course always the same value with all of
T 2 gap
0.75
them. However, the shorter L is defined, the larger Ea
appears in the graph; the discrepancy can be rather large
0.5 3 gap especially in 1 and 2 gap resonators. Thus, when
comparing gradient in different resonators, it is important
4 gap
to know how the cavity lengths are defined.
0.25

0
5 gap 3. LOW-β CAVITIES DESIGN
0.5 1 1.5 2
β/β0 A good SC low-β resonator must fulfil the following
Figure 5. Normalized transit time factor curves vs. principal requirements:
• large energy gain
normalized velocity β/β0, for cavities with different
• large shunt impedance (thus low power dissipation)
number of equal gaps.
• easy and reliable operation
• easy installation and maintenance
Of course, the larger the number of gap of a resonator,
• low cost-to-performance ratio
the larger is the energy gain that we can expect, but the
narrower the velocity acceptance (figure 5). The main specifications determining the initial cavity
design characteristics derive from the beam.

SUA04 23
 Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

beam energy → β0, gap length Resonators performance limitations

velocity acceptance → n. of gaps
Performance limitations in low-β superconducting
beam size, transv. → bore radius
beam long. size & freq. → rf frequency resonators are similar to the ones of β=1 cavities, with
beam power → rf coupling type some peculiarities generated by low frequency and
cw or pulsed → mech. Design different geometry.
Critical surface properties. The present achievements
Other technical choices in the cavities (thus in the linac) in critical surface properties of low-β superconducting
can be prompted by requirements on budget, space, cavities, in laboratory test and in operation, are shown in
flexibility, availability, fault tolerance, and by other table 3. These results seem not to depend too much on
aspects that sometimes make a less performing resonator resonator frequency and shape.
a better choice for a successful project. Symbol achieved reliable spec.
Max. surface Ep(MV/m) ~60 30÷35
Resonators technology electric field
The available resonator technologies are substantially Max. surface Bp (mT) ~120 60÷70
three. magnetic field
1. Bulk Nb (by far the most used) Residual surface Rres(n Ω) ~1 5÷10
Resonators are made of full niobium with high RRR resistance
(150÷300); sometimes they include, as part of the Table 3. Critical surface properties reported for
structure, a He vessel made of normal grade niobium or Niobium low-β resonators and values reliably achieved in
stainless steel. The resonator parts are obtained by operation.
machining Nb sheets, rods, plates,etc., and are joined
together by electron beam welding in high vacuum. Even Q-slope. The so-called Q-slope is observed in almost
rather complicated shapes can be obtained with this all high-Q low-β resonators, where it usually shows a
technology, which is well mastered by industry and roughly exponential behavior (a straight line in the
allows the highest performance. The surface treatment logarithmic plot for Q, see figure 9). Differently from the
techniques developed for electron cavities (chemical Q decay originated by field emission, this effect is not
polishing, electropolishing, high pressure water rinsing associated with production of x-rays and cannot be
etc.) have been fully transferred to low-β ones with reduced by rf conditioning. The Q-slope is not yet
excellent results. completely understood [21][22][23]. It gives an important
2. Sputtered Nb on Cu (mostly at INFN-Legnaro[6]) dependence of the surface resistance Rs on accelerating
The resonator is made of thick OFHC Copper, possibly field, that must be taken into account in resonator design
with no brazing, with rounded shape optimized for specifications at high field.
sputtering and no holes in the high current regions. A thin
layer of Nb (~1÷2 µm) is deposited on its rf surface by
sputtering. This allows to merge the excellent thermal
properties of OFHC Copper and the superconducting ones
of the Nb. Compared to bulk Nb ones, sputtered cavities
not always have comparable performance but can allow
cost savings, especially for large production. This
technology is well suited for regular geometries, with
large openings for cathode insertion and large volumes to
maintain sufficient distance between cathode and cavity
walls. Until now, sputtering was fully exploited in low-β
cavities at INFN Legnaro on 160 Mhz quarter-wave
resonators, but R&D is going on to extend the technique Figure 9. Q-slope in high-Q quarter-wave resonators.
to other kinds of resonators like RFQs. Cavity 4 shows a pure Q-slope behaviour. Cavities 1,2
3. Pb platet on Cu (in some tandem boosters) and 3 show also typical field emission behavior at high
OFHC Copper resonators are plated with Pb by means field, that can be usually eliminated by rf conditioning.
of electrochemical processes. Compared to the previous
case, this older technique has lower cost, can be Q-disease. Q disease [4] affects also low-β cavities. In
performed in-house with minimum equipment and can be low frequency ones it has been observed only recently. It
applied to a larger variety of geometries. The causes a strong reduction of Q in Nb resonators with large
superconducting properties of Pb, however, are not bulk hydrogen concentration, when they are kept for a
comparable to the ones of Nb; this results in significantly long time in the 150>T>60 temperature range during
lower gradients (less than 50%). cooldown. The Q can be fully recovered by heating the
The first low-β superconducting linacs have been made cavity above 150 K and then applying a fast cooldown
this way, and some of them are still in operation. The Pb (tens of minutes, the shorter the better). Large bulk
technology is being replaced with the Nb one. hydrogen concentration can be avoided with a careful

24 SUA04
 Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

cavity construction and chemical polishing, but not in

electropolished cavities (a definitive cure, if necessary, is
elimination of the hydrogen by baking the resonator at
700÷900 °C). To avoid problems, fast resonator cooling
in the dangerous range of temperatures should be foreseen
at the design stage.
MP region
Multipacting (MP) is the resonant field emission of
electrons from the cavity walls [4]. It sets up under the
following conditions:
• stable electron trajectories ending on cavity walls Figure 8. Qualitative graph of the secondary electron
• electron time of flight multiple of ½ rf cycle (the emission coefficient in Nb.
multiplicity order depends on the particular MP
trajectory)
• secondary electron emission coefficient δ >1 Resonator design
• presence of an initial electron impinging the right As in the high-β case, the resonator design requires a
surface at the right field and phase intervals, to start sequence of steps and includes electromagnetic,
the process. mechanical and thermal aspects, as well as beam
dynamics ones that must be integrated together. This
involves the use of different simulation codes and
different design skills.

EM design. This requires the design of the cavity

geometry with the specified frequency, modes and
mechanical constraints. Its optimization includes:

1. minimization of the peak surface fields Ep/Ea and

Bp/Ea;
2. maximization the shunt impedance Ea2/(P/L);
3. optimization the fields for beam dynamics;
4. optimization of the geometry for elimination of
multipacting;
5. positioning of apertures suitable for rf coupling and
for resonator cleaning.
6. Positioning of a suitable tuning section in the
resonator walls.

Rf coupling. Inductive couplers are used for low rf

power applications (<1 kW) and low f (<300 MHz); there
is sufficient experience in these devices. For high-power,
medium- and high-β elliptical cavities, successful
capacitive couplers are also available [7]. There are only a
few examples, however, of couplers designed for high
power low-β cavities. These couplers can be larger than
resonators and require a well integrated design [8]. The
know-how developed for electron cavities can be
fruitfully used for low-β ones.

Figure 7. MP in a half–wave resonator. Top: 1-wall

“horseshoe” MP in a high magnetic field region. Bottom:
2-walls MP in a high E field (courtesy of ACCEL).

Multipacting is almost impossible to avoid in low-β

cavities, due to the presence of involute geometries. Most
resonators, however, show only multipacting levels at low Figure 10. Low-power inductive and high-power
gradient, that can be conditioned with low rf power in a capacitive couplers (not in the same scale).
reasonable time (up to about one day).

SUA04 25
 Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

Beam steering. Non symmetric cavities can have non-

symmetric EM fields in the beam region. This can
produce undesirable beam steering .
The magnetic field gives usually the dominant
contribution especially in quarter-wave resonators
(QWRs) with large aspect ratio (i.e. approximately for
β0≅L/λ>0.1); this can give serious beam dynamics
problems, especially for high intensity beams. QWRs
working above ~100 MHz often need some correction.
Using the approximation of constant field in the gap,
the vertical kick in a QWR can be represented as follows:
⎛ ⎞
⎜ ⎟
qE LT (β ) ⎜ K EY ⎟ (10)
∆y ' = a sin ϕ ⎜ − cK BX ⎟
βγmc 2 ⎜ β ⋅ tg ⎛⎜ πd ⎞⎟ ⎟
⎜ ⎜ βλ ⎟ ⎟
⎝ ⎝ ⎠ ⎠
where KEY=|Ey/Ez| and KBX=|Bx/Ez|, calculated in one
gap.
During acceleration, for typical values of ϕ ~20÷30
deg, Ey is symmetric and tends to cancel in the 2 gap (see
figure 11), while Bx is antisymmetric and adds up,
becoming dominant. Steering is proportional to
qEaLT(β)sinϕ : this expression is rather similar to the rf Figure 12. Steering correction by drift tube shaping in
defocusing effect in misaligned cavities, proportional to QWRs.
qEaLT(β)sinϕ ⋅r, where r is the displacement of the beam
axis from the beam port axis. Multipacting must be also considered in cavity design.
Computer simulation codes can be used for shaping the
cavity in order to avoid stable MP trajectories with
3.00E+06
Ez
5.00E-03 electron impact energy in of the δ>1 region. Lack of
Bx, T

Ey (x10)
symmetry of most low-β geometries often requires three-
2.50E-03 dimensional codes and very long computing time. Lack of
symmetry, however, also prevents MP to extend to too
E, V/m

0.00E+00 0.00E+00 large areas: old low-β cavities, even if designed with no
Bx possibility of MP simulations, have almost always been
made to work with reasonable conditioning time.
-2.50E-03
Rf power losses calculations are required to maximize
cavity shunt impedance. The maximum rf power density
-3.00E+06 -5.00E-03 on the cavity wall should be kept below ~1 W/cm2 at
-120 -60 0 60 120 4.2K, to allow efficient heat removal by liquid Helium
z, mm
[4]. A large safety margin is required in operation
Figure 11. Profiles of the accelerating field and of the conditions, since local defects can increase power losses
steering Ey and Bx components in a QWR. significantly and cause cavity quench.

In low-β QWRs with small aspect ratio, a slight The mechanical design can be performed with
misalignment, if properly done, is usually sufficient to standard techniques using the powerful codes which are
cancel most of the steering thanks to rf defocusing. In commercially available. The procedure includes:
cavities where the required misalignment would become • Statical analysis (stability against He pressure,
too large compared to the beam port aperture, radiation pressure, mechanical tuning, …)
compensation can be obtained by gap shaping: the • Dynamical analysis (mechanical modes…)
magnetic deflection can be cancelled by an artificial • Thermal analysis (cooling, T distributions,…)
enhancement of the electric deflection [11] (see fig. 12). • Construction procedure.
Higher order modes. In high power pulsed linacs, The thermal conductivity at 4.2 K is k=RRR/4
higher order modes should be also considered; at present, (W/m)/K. Thus high RRR Nb material is required, which
only multi-cell elliptical cavities with β>0.6 are used in has poorer mechanical properties compared to normal
such conditions and the techniques developed for β=1 grade Nb (RRR~40), and higher cost. A typical good
cavities, like higher order mode dampers, can be applied choice for rf cavities is RRR~200÷300. Accumulation of
[12]. He gas bubbles on the cavity walls can limit heat

26 SUA04
 Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

exchange on the Nb-He interface. Good ways for liquid displacements with opposite effects to the frequency, to
He transport and for gas He removal must be provided: obtain a balance. Double wall structures allow to null the
this can be rather complicated in low-β geometries, net force on cavity walls and allow to expose large
sometimes requiring forced He flow, syphons and other surfaces to He pressure without making them collapse.
accessories. A clever “self-compensating” resonator design, where
Low-power density surfaces (e.g. tuning plates) can be every displacement in high E region is associated in an
cooled by He gas and by thermal conduction through rf equivalent one in a high B region, can keep df/dP close to
joints, where experience made in many laboratories zero [13].
suggests not to exceed a few mT magnetic field. The
effect of a possible super- to normal-conducting transition
in such regions should be checked: sometimes it is not
critical for cavity operation and low quality plates can be
used without problems.
Tuning and mechanical instabilities. Mechanical
tuning requires cavity deformation. Wall displacement Figure 14. (Left) SPOKE cavity end plate with “self-
toward high electric (magnetic) field areas cause rf compensating” design obtained by gusset shaping. (Right)
frequency decrease (increase). Many designs are Cross-section of double wall structure in a reentrant
available for slow tuning that use motors (inside and/or cavity.
outside cryostats). The same mechanisms cause cavity
detuning in the presence of mechanical noise and Lorentz force (or radiation pressure), produced by the
undesired forces acting on the cavity walls. EM fields on cavity walls, gives a quadratic detuning:
δf∝-δ(Ea2). Typical values are a few Hz/(MV/m)2. This
frequency shift can be easily compensated by mechanical
tuning once the operation field is reached. However, a
large coefficient δf/δ(Ea2) can generate the so-called
ponderomotive oscillations, where small Ea errors, that
can be initially induced by any source, cause detuning
through Lorentz force and start a self-sustained
mechanical vibration [14]. This makes cavity operation
Figure 13. QWR tuning plates with electromechanical difficult and unreliable. Mainly for this reason df/d(Ea2)
(left, TRIUMF design) and gas pressure actuation (right, should be kept as low as possible, by reinforcing the
ANL design). cavity structure and by means of a careful EM design.

Cavity detuning around the center frequency is

proportional to He pressure changes: δf∝δP. The effect on
cavity ability to maintain the operating conditions depend
on the resonator bandwidth and rf system capabilities.
Cu-Pb and Cu-Nb QWRs with thick walls are rather
insensitive to pressure variation. For Nb low-β cavities,
typical df/dP values are from a few to a few tens of
Hz/mbar. A good mechanical design can reduce
drastically these values. Mechanical tuners can be used
for slow frequency compensation in feedback with the
frequency error, and the cryogenic system should be
carefully designed with sufficient pressure stability.
Source: Solution:
Helium pressure mechanical tuning in feedback,
variations mechanical strengthening Figure 15. Example of Lorentz force transfer function
Lorentz Force slow tuning and rf feedback in a 109 MHz QWR (ANL).
detuning
microphonics fast tuners, mechanical design, Resonant modes are the most dangerous mechanical
noise shielding, etc. instabilities. A small vibration can cause large
deformation, and consequently large detuning that can
resonant vibrations mechanical damping exceed the resonator rf bandwidth. The deformation is
Table 4. Mechanical instabilities sources and cures. usually too fast to be recovered by standard mechanical
tuners. After making the detuning range narrower by
Since it is impossible to eliminate completely careful mechanical and rf design, possible solutions to the
deformations caused by He pressure fluctuations, the problem are:
resonator can be designed in order to produce • Widening the rf bandwidth by overcoupling;

SUA04 27
 Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

• Using fast tuners with suitable range (electronic, Low-β resonators performance. The peak fields of
piezoelectric, magnetostrictive, electromechanic) >60 MV/m and >120 mT, achieved in on-line resonators,
[15][16][17]; and residual surface resistance values <1 nΩ achieved in a
• Using mechanical dampers [18] test cryostat are not very far from the ones achieved by
• Using rf damping techniques [19] β=1 cavities in similar conditions at 4.2K. This in spite of
Usually the solution requires the combination of many geometries that are not always favorable for surface
different techniques. preparation (numerous welds, small apertures, etc). On
the other hand, the low power dissipation (related to low
Cavity integration in cryostats. Design objectives are rf frequency) and the small areas where peak fields are
easy installation and maintenance, stable and reliable concentrated (related to cavity geometry), facilitate field
operation. In addition to the usual design requirements of emission conditioning and the achievement of high
integration of all systems in high vacuum, widely treated gradients. The recent application of the most advanced
during this conference in a dedicated tutorial [20], it preparation techniques has raised low-field Q’s to very
should be mentioned that in some low-β cryostats the high values.
vacuum volumes inside the resonators (for beam The frequent appearance of strong Q-slopes that limit
transport) and outside (for thermal isolation) are not the possible operation at high gradient, however,
separated, but communicating through the beam ports, represents one of the main open problems in low-β
and pumped out by the same units. In spite of that, high Q cavities development.
can be maintained for years in on-line resonators. The
clean resonator apertures are kept closed during assembly, 4. LOW-β CAVITY TYPES AND
to protect the inner surfaces; during the first pumping out
no contamination happen due to the outgoing gas flow.
CHARACTERISTICS
Moreover, dust cannot be transported in vacuum and no In this paragraph a synthetic review of different low-β
contamination happens. However, Q degradation resonator families will be given. Families are
(although usually recoverable with high pressure water characterized by the particular EM mode that is excited in
rinsing) can appear after the cryostat undergoes venting: the cavities. Only a few references will be given here; for
dust, deposited on the internal surfaces of the cryostat, is more references and for a recent review of low-β
transported inside cavities through beam ports and other resonators models and performances see, e.g., ref [24].
apertures. This can be avoided by venting the resonators
first, with pure and filtered gas, and maintaining an Quarter-wave structures
outgoing flow. This was not always possible in old-design The mode excited is a TEM one (Transverse Electric
low-β cryostats, but it is being implemented in new ones. Magnetic) in a λ/4 long transmission line section. This
Although common vacuum has some advantages in allows to obtain cavities with small g/λ, and relatively
simplicity and cost, most specialists are still in favour of small size [25].
the well established scheme with separate vacua.
L ~λ/4
I
I0 V V CL
1

V( z )
0.5
I( z )

0
0 0.2 0.4 0.6 0.8

Figure 17. Top: scheme of a quarter-wave structure

with loading capacitance. Bottom: Voltage and current
amplitude along the line.

Split-ring resonators
. The split ring arms (fig. 19) and the outer conductor
wall form a pair of coupled QW transmission lines, with 2
principal modes: push-push (drift tubes at equal voltage)
and push-pull (drift tubes at opposite voltage), the latter at
Figure 16. Inner part of a common-vacuum cryostat lower frequency and used for acceleration [26].
(TRIUMF ISAC-2) in a clean-room assembly. Application range: 90≤f≤150 MHz, 0.05≤ β0≤0.15
“pro” :
• relatively large energy gain

28 SUA04
 Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

• good efficiency “contra”:

“contra”: • Dipole steering above β0~0.1 (unless corrected)
• mechanical stability • Mechanical stability below ~100 MHz
• beam steering
• high peak fields Comment: In operation. Very successful, developed in
• more expensive and difficult to build than many models by many laboratories.
QWRs
Comment: In operation since many years but not Double quarter-wave resonators
developed any longer. Application range: f~225 MHz, β0 ~0.16
“pro” :
• Most of the pros of QWRs
• Larger voltage gain than 2-gap QWRs
“contra”:
• Dipole steering
• Mechanical stability below ~100 MHz

Comment: Originally named “half-wave” for its

fundamental mode, with two QWRs operating in push-
pull. Prototyped in 1985 for Pb plating and never put in
Figure 19. Schematic of a split-ring resonator operation [27]. Alternative to Split-ring.

Quarter-wave resonators.
Application range: 48≤f≤160 MHz, 0.001≤ β0≤0.2
“pro” :
• Compact
• Modular
• High performance
• Low cost Figure 20. Double QWR schematic.
• Possibility of large openings, easy access
• Down to very low beta Half-wave structures
A half-wave structure is equivalent to 2 quarter-wave
ones facing each other. The same accelerating voltage of
Superconducting QWRs is obtained with almost 2 times larger power
LNL 2-gap QWRs familyQ
dissipation. The symmetry of the structure, however,
cancels steering and allows to use HWRs at higher β than
QWRs.
~λ/2
I V0 CL I

V( z )
ANL 4-gap QWR family
0
I( z )

1
0 0.5 1 1.5 2
Figure 21. As in Fig. 19, for a half-wave structure.

Half-Wave coaxial resonators

Application range: 160≤f≤352 MHz, 0.09 ≤ β0≤ 0.3
“pro” :
• No dipole steering
• High performance
• Lower Ep than QWRs
• Very compact
Figure18. Families of 2- and 4-gap QWRs. “contra”:
• Not easy access

SUA04 29
 Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

• Difficult to tune Comment: Still under development. Multigap cavity

• Less efficient than QWRs with easy access, alternative to CH structures.

Comment: Prototyped and under production. Best in the

frequency range of 160÷250 MHz, too high for QWRs
(steering) and too low for SPOKE resonators (large size).

Figure 24. Ladder resonator model.

TM mode cavities
TM010 (Transverse Magnetic) mode: B is always
Figure 22. Half-wave coaxial resonators. perpendicular to the EM wave propagation axis (and to
the beam axis) [2].
Single-SPOKE resonators
Application range: 345≤f≤ 805 MHz, 0.15 ≤ β0≤ 0.62
“pro” :
• No dipole steering
• High performance
• Higher Rsh than HWRs
• Larger aperture than HWRs
“contra”:
• Not easy access (but better than HWRs)
• Difficult to tune
Figure 25. Examples of TM structure: pillbox cavity
• Larger size than HWRs
Elliptical resonators
Comment: Prototyped in many models and β0, but not
Application range: 352≤f≤805 MHz, 0.47≤ β0≤ 1
yet in operation. Best results around 350÷400 MHz,
“pro” :
where it is the most popular choice. Rather large size at
• Highly symmetric field
lower frequency. Multi-cell possibility.
• High performance
• Low Ep and Bp
• Multi cell possibility
• Large aperture
“contra”:
• Not suitable for β<0.4
• Mechanical modes

Comment: β0>0.6 in operation, β0<0.6 prototyped. Very

successful and well developed in many laboratories. Best
Figure 23. Spoke resonators. around 700÷800 MHz.
Ladder SC cavities
Application range: 350MHz, 0.1≤ β0≤ 0.3 [32]
“pro” :
• Efficient
• large energy gain
• They can be made for rather low β
• Easy access
“contra”:
• small aperture
• β acceptance
• not easy to build Figure 26. β0=0.5, 5-cell elliptical resonator.

30 SUA04
 Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

• High efficiency
Reentrant cavities • Large energy gain
Application range: 352≤f≤ 402 MHz, 0.1≤ β0 • Lower frequency than elliptical
“pro” : • Mechanically more stable than elliptical
• Highly symmetric field “contra”:
• Very Compact • Large size
• Low Ep and Bp • Not easy access
• Widest velocity acceptance • Difficult to tune
• Possibility of large aperture • smaller aperture than elliptical
“contra”, • More expensive than elliptical
• short accelerating length, little E gain
• mechanical stability Comment: Prototyped. Very promising, especially
• inductive couplers only around β~0.5, alternative to elliptical cavities. It allows
linac lattices with large longitudinal acceptance [31].
Comment: Prototyped. One of the oldest low-β
superconducting cavities [29], recently renewed. For
special applications.

Figure28. Triple-spoke resonator sketch.

Figure 27. β>0.1 reentrant cavity
CH multi-gap SC cavities
IH and CH multi-gap structures Application range: 174≤f≤ 800 MHz, 0.1≤ β0≤ 0.3
These interdigital and crossed electrode structures have “pro” :
3 or more gap. The electrodes are either 2 (IH) or 4 (CH) • Very efficient
creating dipolar or quadrupolar field distributions, • large energy gain
respectively. The magnetic field is parallel to the beam • suitable for rather low β
axis. These structures can host either accelerating drift • It allows a very compact linac
tubes or RFQ electrodes for simultaneous acceleration and “contra”:
focusing at very low-β [30]. • velocity acceptance
IH-RFQ 4-vane RFQ • difficult to have large aperture
• not easy to build
B- B-
• High peak fields
• cost

Comment: Recently prototyped. New technology, very

E- E-
promising for beams with fixed velocity profile.

IH-Structure CH-Structure

Figure 28. Schematic of IH and CH structures and their

application in Drift Tube Linacs and RFQs [30].
Multi-SPOKE resonators
Application range: 345≤f≤ 805 MHz, 0.15≤ β0≤ 0.62
“pro” : Figure 29. CH 19-gap superconducting resonator.
• High performance

SUA04 31
 Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

Superconducting RFQ’s [4] [Link], J. Knobloch and T. Hays, “Rf

Application range: f~80 MHz, 0.001≤ β0≤0.035 superconductivity for accelerators” (Wiley).
“pro” : [5] K. Saito, “Basic Principles of SRF”, these
• Compact proceedings.
• CW operation [6] A. Porcellato, “Niobium sputtered QW resonators”,
• High efficiency these preoceedings.
• Down to very low beta [7] W. Moeller, “Input couplers for superconducting
• large acceptance cavities - design and test”, these proceedings.
“contra”: [8] F.L. Krawczic, “Cavity and Power Coupler
• Mechanical stability Integration”, Workshop on High-Power Couplers for
• Not easy to build Superconducting Accelerators, Jefferson Lab,
• MP and FE Newport News, Virginia, 2002.
• Rather high cost [9] R.L. Geng, “Multipacting Simulations for
Superconducting Cavities and RF Coupler
Comment: In operation. Technologically challenging Waveguides”, Proc. of the 2003 Particle Accelerator
[33]. Future applications in high power ion linacs could Conference.
be possible, due to its large aperture and high fields. [10] A. Facco, V. Zviagintsev, “Study on Beam Steering in
Intermediate-β Superconducting Quarter-wave
Resonators”, Proceedings of the 2001 Particle
Accelerator Conference, Chicago
[11] P.N. Ostroumov, K.W. Shepard, Physical Review
Special Topics - Accelerators and Beams, vol. 4,
110101, 2001.
[12] Brian Rusnak, “RF Power and HOM Coupler
Tutorial”, proc. of SRF03.
[13] K. W. Shepard, M. Kedzie, and M. P. Kelly, T.
Schultheiss, “Superconducting Intermediate-Velocity
Drift-tube Cavities for the RIA Driver Linac”,
Proceedings of the 2001 Particle Accelerator
Conference, Chicago.
[14] J. Delayen, “Ponderomotive instabilities and
microphonics”, these proceedings.
Figure 27. Superconducting RFQ. [15] [Link], [Link] and [Link], "Upgraded
Phase Control System For Superconducting Low-
Velocity Accelerating Structures" Proc. of 16th
FINAL REMARKS
LINAC Conf., August 24-28, 1992, Ottawa, Ontario.
The activity in superconducting low-β resonators [16] P. Sekalski, S. Simrock, A. Napieralski,
development is presently higher than ever before, mainly “Piezoelectric stack based system for Lorentz force
for the new proposed applications in high power linacs compensation”, these proceedings.
and in radioactive beams post-accelerators. Very good [17] [Link], K. Fong, S. Koscielniak, R.E. Laxdal, G.
perfomances have been achieved. Only recently, the very Stanford, “A Mechanical Tuner for the ISAC-II
successful technology of Nb treatment developed for β=1 Quarter Wave Superconducting Cavities”,
cavities was fully exploited in low-β ones, and Proceedings of the 2003 Particle Accelerator
performances are expected to improve further. Numerous Conference.
projects, some of them funded, foresee a large variety of [18] A. Facco, “Mechanical mode damping in
low-β resonators. There are different models of cavities superconducting low-β resonators”, Particle
developed for the same application; we will be able soon Accelerators, Vol. 61, 1998, pp. 265-278.
to evaulate and compare the results to find optimal [19] J. R. Delayen, “Electronic Damping of Microphonics
choices. There are still problems not completely solved, in Superconducting Cavities”, Proc. of the 2001
like Q-slope and cures for mechanical instability. New Particle Accelerator Conference, Chicago.
ideas and new inventions will be needed to fulfill all [20] C. Pagani, “Cryomodule design, assembly,
requirements. alignment”, these proceedings.
[21] E. Palmieri, “Advancement in comprehension of the
REFERENCES Q-slope for superconducting cavities”, these
[1] S. Belomestnykh and V. Shemelin, “High-beta cavity proceedings.
design”, these proceedings. [22] J. Halbritter, “The Nb-oxide system”, these
[2] T. Wangler, “Rf linear accelerators” (Wiley). proceedings.
[3] [Link], “Low and Intermediate Beta Cavity [23] G. Eremeev, “New results on "high-field Q-slope"”,
Design – a Tutorial”, proc. of SRF03. these proceedings.

32 SUA04
 Proceedings of the 12th International Workshop on RF Superconductivity, Cornell University, Ithaca, New York, USA

[24] A. Facco, “Low and Medium Beta Superconducting [30] R. Eichhorn and U. Ratzinger, “Superconducting H-
Cavities”, Proc. of EPAC ’04, Lucerne, Switzerland, mode Structures for Medium Energy Beams”, [Link]
2004. the XX International Linac Conference, Monterey,
[25] I. Ben-Zvi and J.M. Brennan,Nucl. Instr. And Meth. California,2000.
212 (1983) 73. [31] K.W. Shepard, P. N. Ostroumov, J. R. Delayen,
[26] [Link], [Link], and [Link], IEEE “High-energy ion linacs based on superconducting
Trans. Nucl. Sci. NS-22, 1179 (1975). spoke cavities”, PHYSICAL REVIEW SPECIAL
[27] J. R. Delayen, “Superconducting Accelerating TOPICS - ACCELERATORS AND BEAMS,
Structures for High-Current Ion Beams,” Proc. VOLUME 6, 080101 (2003).
LINAC’88, Newport News, October 1988, p. 199. [32] V. Andreev et al., “Study of a Novel
[28] M.P. Kelly, K.W. Shepard, M. Kedzie, G. Zinkann, in Superconducting Structure for the Very Low Beta
Proc. 2001 Particle Accelerator Conf., June 18-22, Part of High Current Linacs”, Proc. of EPAC 2002,
2001, Chicago, IL (2001), p. 1047. Paris, France, 2002.
[29] P.H. Ceperly, J.S. Sokolowski, I. Ben-Zvi, H.F. [33] G. Bisoffi et al., “Superconducting RFQ”, Proc. of
Glavish and S.S. Hanna, “Beam test of a SRF2001, Tsukuba, Japan, 2003.
superconducting niobium cavity for a heavy ion
accelerator”, Nuclear Instruments and Methods136
(1976) 421-423.

SUA04 33