C. Johnstone
Fermilab/Particle Accelerator Corporation
2nd Workshop on Hadron Therapy
Ettore Majorana Foundation and
Center for Scientific Culture
Erice, Sicily
5/25/2011
Fermilab
The International FFAG Collaboration: International
Accelerator Laboratories and Universities
U.S.
Fermilab
Brookhaven National Lab
Lawrence Berkeley National Laboratory
University of California: L.A., Riverside
Michigan State University
Illinois Institute of Technology
Canada
TRIUMF
University of British Columbia
Switzerland
CERN
France
LPS
U.K.
Daresbury Laboratory.
Manchester, Liverpool, Leeds, and
Lancaster and Oxford University
Imperial College
Rutherford Appleton Laboratory
John Adams Institute, Oxford
Birmingham University
Clatterbridge Centre for Oncology
Beatson Oncology Centre
Gray Cancer Center
Japan
KEK
Kyoto University (KURRI)
Osaka University
Particle Accelerator
Corporation
Particle Accelerator
Corporation
Particle Accelerator Corp
Accelerator and component design – FFAGs and
synchrotrons, magnets, diagnostics
Employees and collaborators
Fred Mills (retired, Fermilab, built the first MURA FFAGs)
C. Johnstone (Fermilab
Richard Ford (Fermilab)
V. Kashikhin (Fermilab)
Shane Koscielniak (TRIUMF)
Martin Berz (Michigan State University)
Kyoko Makino (Michigan State University)
Pavel Snopok (University of California, Riverside)
Quick Field Guide to Fixed-Fielding Alternating Gradient FFAGs
Simplest Dynamical Definition:
FFAG is ~ a cyclotron with a gradient; beam confinement is via:
Strong alternating-gradient (AG) focusing, both planes: radial sector FFAG
normal/reversed gradients alternate (like a synchrotron)
Gradient focusing in horizontal, edge focusing in vertical: spiral sector FFAG
vertical envelope control is through edge focusing (like a cyclotron)
the normal gradient increases edge focusing with radius /momentum (unlike a cyclotron)
A cyclotron can be considered a class of FFAGs
Types of FFAGs:
Scaling:
B field follows a scaling law as a function of radius - ∝rk (k a constant;) presentday scaling FFAGs: Y. Mori, Kyoto University Research Reactor Institute
Nonscaling:
Linear (quadrupole) gradient; beam parameters generally vary significantly with
energy (EMMA FFAG, Daresbury Laboratory, first nonscaling FFAG)
Nonlinear-gradient; beam parameters such as machine tune can be fixed (as in a
synchrotron, PAMELA and a new generation of constant-tune, CW FFAGs)
Why a FFAG – the reason is not mysterious?
• The FFAG combines all forms of transverse beam
(envelope) confinement in an arbitrary CF magnet:
– For the horizontal, the three terms (in the thin lens approximation)
are
synchrotron
1/ f F = kF l +
cyclotron
ϑ
η
+
ρF ρF
with ϑ is the sector bend angle, η the edge angle (tangent is
approximated), length, l , is the F half - magnet length and
k F is the " local" gradient for an arbitrary order field.
– For the vertical only the quadrupole gradient, kDl, and the edge
term are available
– In a FFAG the focusing terms can be varied independently optimizing
machine parameters; i.e. footprint, aperture, and tune in a FFAG
To summarize beam envelope control
(in the thin Lens Limit):
1.
Centripetal (Cyclotrons + FFAGs) :
bend plane only, horizontally defocusing or focusing
Strength ∝ θ/ρ (bend angle/bend radius of dipole field component on reference orbit)
2.
Edge focusing (Cyclotrons + FFAGs) :
Horizontally focusing / vertically defocusing, vice versa, or no focusing
depending on field at entrance and entrance angle
Strength ∝ tan η/ρ , (or ~ η/ρ for reasonably small edge-crossing angles)
3.
Gradient focusing (Synchrotrons + FFAGs) :
Body gradient, fields components > dipole:
B= a + bx +cx2 + dx3 + …
B’= b + 2cx + 3dx2 + …
Linear field expansion, constant gradient
Synchrotrons + linear-field nonscaling FFAGs (EMMA!, muon accelerators)
Nonlinear field expansion up to order k, magnitude of gradient increases with r or energy:
Scaling FFAGs
Arbitrary nonlinear field expansion, magnitude of gradient increases with r or energy:
Nonlinear Non-scaling FFAGs
Edge crossing angles are kept deliberately small in large multi-cell synchrotron rings. This term becomes increasingly important for and
causes problems in small synchrotron rings.
More on the role of the gradient in edge-focusing
Understanding the powerful interplay between gradient
and edge focusing is critical to understanding the potential
of FFAGs
In cyclotrons,
horizontal envelope control is through the centripetal term
Centripetal term increases with radius/pathlength/momentum in the cyclotron magnets
Vertical envelope control is through edge focusing / field shaping at the magnet
edges
proportional to the constant dipole field; more difficult to increase, much weaker than horizontal
focusing in cyclotrons
In FFAGs,
the gradient increases both the horizontal (centripetal) and vertical (edge
focusing) with radius/momentum
This last point is very important for FFAGs because it allows the field, orbit
location, and important machine parameters such as tune to be more
independent and strongly controlled than in cyclotrons
FFAGs and their
Variations
Scaling FFAGs (spiral or radialsector) are characterized by
geometrically similar orbits of
increasing radius, imposing a
constant tune (field and derivative
gradient scale identically with r).
Magnetic field follows the law B ∝
rk, with r as the radius, and k as
the constant field index.
Spiral Sector: example: more
compact; positive bend field
only. Vertical focusing controlled
by edge crossing angle.
Field expansion: k determines multipole order;
Comments: the lower the k value, the more slowly field increases
with r and the larger the horizontal aperture, but the more linear
the field composition and dynamics.
F
Radial Sector: example: This is a
triplet DFD cell; there are also
FDF, FODO and doublets. In a
radial sector the D is the negative
of the F field profile, but shorter.
Linear nonscaling FFAGs for
rapid acceleration
Linear-field, nonscaling FFAGs.
Ultra-compact magnet aperture,
proposed and developed for High
Energy Physics (Neutrino
Factories and Muon Colliders),
relaxes optical parameters and
aims only for stable acceleration.
In general they are not suitable
for an accelerator with a modest
acceleration system and
accelerate only over a factor of 23 range in momentum.
EMMA – world’s first nonscaling
FFAG, @Daresbury Laboratory,
NOW ACCELERATING BEAM !!!!!
Extraction
reference orbit
D
Injection
reference
orbit
Cartoon of orbit compaction: nonsimilar orbits,
nonconstant tune, resonance crossing
Characteristics– tune sweep/unit cell, parabolic pathlength on
momentum (small radial apertures); serpentine (rapid)
acceleration – beam “phase-slips”, crossing the peak 3 times,
accelerating between rf buckets
Tune-stable nonscaling FFAGs
1)
for slow acceleration
Tune-stable, nonscaling FFAGs
•Tune is strongest indicator of
stable particle motion – allowing
particles execute periodic motion
eventually returning to the same
transverse position relative to a
reference orbit. Constraining the
tune can be sufficient to design a
stable machine.
•Release of other linear optical
parameter allows flexibility and
optimization both in cost and
complexity of the accelerator
design; i.e. simpler magnets, strong
vertical focusing, for example
•Tune Stable Nonscaling FFAGs have
either linear or nonlinear field
profiles and/or edge contours
Two lattice approaches
Machida version - which uses a scaling law truncated at
decapole, rectangular magnets , (not discussed here,
see PAMELA project) and
2) Johnstone version – The most general form of a radial
sector : allowing independent, unconstrained field and
edge profiles between two combined-function magnets.
1
40c
m
~17
cm
Extraction
orbit
reference
2
Injection
orbit
reference
1st consideration
for a “new” type
of IBT
Bringing the Community a Step Forward
By combining attributes of the synchrotron and cyclotron the
FFAG has the following combined advantages:
The simplicity of fixed magnetic fields rather than pulsed operation
Increased duty cycle over the synchrotron (50 Hz to 1 kHz RF swept-frequency
cycling)
Potential for variable energy extraction
Elimination of degraders or substantial reduction in energy degradation
Kicker or resonant extraction (Yokoi, PAMELA project)
Cyclotron-like extraction - field shaping on extraction orbit
Strong focusing allows synchrotron-like straights
Lower losses associated with synchrotrons especially at extraction
Goal: lower operational overhead and system simplicity of the cyclotron
Are these Advantages Sufficient to Warrant a NEW type of Ion
Beam Therapy Facility?
Given the technical advances on conventional accelerator facilities
I would say the answer is “NO”
One of the strongest arguments for the prevailing dominance of cyclotrons is continuous
(CW) beam. One paper which makes this specific case is:
“Fast Scanning Techniques for Cancer Therapy with Hadrons – A Domain of Cyclotrons”
J.M. Schippers, D. Meer, E. Pedroni, Proceedings of CYCLOTRONS 2010, Lanzhou, China
The Conclusion:
“In order to exploit the advantages of 3D pencil beam scanning, the scanning process
must be performed as fast as possible. This allows different strategies to prevent dose
errors due to tumour/organ motion. A reliable application of fast 3D scanning
necessitates firm specifications on the accelerator: a CW beam, with an intensity that
must be stable, quickly and accurately adjustable over a large dynamic range as well as a
fast and accurate energy modulation. Currently the combination of these specifications is
not possible to achieve with pulsed machine operation at repetition rates below 0.5-1
kHz.”
One can employ broad band, (~MHz) RF, to attain these repetition rates (PAMELA)
BUT --- this is a power-consumptive, often lossy solution with low acceleration
capability and not fully realized at this time.
Recent Innovation: Isochronous (CW) NS FFAGs -
Cyclotron limit
~ 1 GeV protons
P (MeV/c) or Bfield
An isochronous orbits must be proportional to velocity
Orbital path length at a given momentum follows the B field; therefore the
B field must also scale with velocity
At relativistic energies, momentum is an increasingly nonlinear function of
velocity; therefore B field transitions from a linear slope to nonlinear, nonrelativistic to relativistic.
THIS HAS BEEN ACHIEVED IN RECENT NONLINEAR NS FFAG DESIGNS
Arbitrary nonlinear field expansion+edge angle can also constrain the tune
The nonlinear gradient provides increased focusing at high energy in both
planes, relative to the cyclotron 3500
FFAG limit
3000
≥2 GeV protons
2500
2000
1500
1000
500
0
0,5
1
β or normalized
pathlength
0
1,5
Innovations in Nonlinear, NS FFAG designs
Isochronous (CW) FFAG designs have been successfully
designed and will be presented later in this talk.
Competitive BUT NOT COMPELLING
What more can an FFAG offer?
Dual Accelerator Patient Model (PSI data*)
Below is a model-based treatment using 70 MeV as the lower limit since many nozzles ( or energy
degraders) only work in the 70 to 250 MeV range. (Lower energies can be obtained by plastic range
shifters placed close to the skin and aperture, for example with breast, pediatric patients, and parotid
tumors in the jaw. )
Site Percentage Energy Range (MeV)
Lung
9%
70 - 170
Breast
3%
70 - 140
CNS
15%
70 - 150 ( central nervous system, i.e., base of skull & tumors around spinal
vertebrae)
Rectum 2%
70- 170 ( also cervical cancers may be a few % in this energy range)
Pediatric 8%
70 - 150
Head &
Neck
15%
70 - 150
Prostate 45%
200-250
Other
2%
This fits nicely with a dual energy accelerator system where E≤150 MeV can be used for roughly 50%
of the patients. At PSI all patients were treated with E< 180 MeV excluding prostates. *
*based on discussions with G. Coutrakon, 2009.
Two-Stage Proton Therapy Facility
Dual Accelerator Model:
Simultaneous treatment/setup can be conveniently split into 70-150
MeV, 150-250 MeV
Accelerator is a small part of the cost of a facility
Delivery beamlines are also a drop in the bucket
Could consider separate gantries: gantries with smaller energy
range requirements are conceivably cheaper – easier magnets;
lower energy gantry much smaller.
Dual accelerators are facilitated using FFAGs by
Multiple long straights
Ease of injection/extraction
CW operation in both stages
CW 18-150 MeV H- (p + 2e) NS FFAG
Stripping wire Htreatment beam
proton for a low energy
nv
3.0
n h, n v
0.5
2.5
0.4
2.0
0.3
1.5
0.2
1.0
0.1
cA
4.5
cB
A
200
4.0
B
350
400
450
500
550
P, @MeV êcD
200
250
300
350
400
450
500
550
P, @MeV êcD
cC
Tune per cell and ring tune. Extended fringe
field will split tunes
cD
E
2.5
2.0
300
C
3.5
3.0
250
D
F
G
cF
H
cG cH
0.5 1.0 1.5 2.0 2.5 3.0
cE
DRad
Ravg
4.5
250
300
350
400
450
500
550
P, @MeV êcD
4.0
-0.01
3.5
3.0
-0.02
2.5
-0.03
2.0
200
Comments and further work
250
300
350
400
450
500
550
P, @MeV êcD
-0.04
Average radius vs momentum and deviation in
% from isochronous average radius
Peak field 0.8T @extraction in F - can take this design to 180 MeV
Increase compactness: 3-sector or no reverse bends, low losses – reduced shielding
Extraction at 150 MeV through field shaping of high-energy orbit
Particle Accelerator
Corporation
CW 150-250 MeV SC Proton Ring
0,02
(Riso-Ravg) /Riso
nv
3.0
0,015
2.5
2.0
cA
2.4
2.2
2.0
1.8
1.6
cB
A
cC
B
F C
cD
G
D
cE cF
H
cG
cH
0.2 0.4 0.6 0.8 1.0 1.2 1.4
E
0,005
1.5
1.0
0,01
Deviation from
isochronicity
0
Ring tune
600
650
1,6
1,55
1,5
1,45
700
550
P, @MeV êcD
600
650
700
750
Momentum MeV/c
Ravg (blue) and
Risochronous (red)
1,4
1,35
1,3
1,25
1,2
Comments and further work
550
600
650
Peak field 3.5T @extraction in F;
small 25 cm aperture
Low losses
Kicker or resonant extraction
CW extraction through field shaping (as in a cyclotron) + degrader
700
750
Particle Accelerator
Corporation
Summary of Operational Advantages in a Dual Stage FFAG
Proton Therapy Facility
CW operation with ….
Variable energy
Limited use of degraders if needed; 250 – 150; or 150 – 70 MeV.
Low loss operation, comparable to synchrotrons
Multi- accelerators/multi-room operation is perhaps the most
important next step in patient throughput.
Less susceptible to hysteresis, smaller tuning window
Less susceptible to energy variations in beam phase space
Amenable to most advanced hardware developments in cyclotrons and
synchrotrons
Much of the ground-breaking work in PAMELA can be adapted (the higher
energy ring, for example uses up to octupole field expansion)
Research Facility : Funding for Proton Therapy
In the U.S. non-private funding for medical facilities is essentially NonExistent .
Funding tends to be confined to R&D not “publicly offered” for routine service
Medical is particularly difficult
One caveat is university-based facilities,
Generally do not have the resources to fund large scale, “experimental” facility
No strictly “Local Competence” centers based on government funding
Requires a broader R&D justification –~1 GeV CW proton beam is
required by
Accelerator Driven Subcritical Reacter (ADS):
Homeland Security (cargo inspection)
A 1.2 GeV proton machine can be used for 400 MeV/u C6+
Basis for a Research and Possible Competence Facility in the U.S.
Advanced design and simulation of an
Isochronous 250-1000 MeV Nonscaling FFAG
2m
nv
3.0
DRad
0.07
2.5
0.06
cA
A
5.0
4.5
2.5
1.5
C
3.5
3.0
0.04
cC
B
4.0
E
0.05
2.0
cB
0.03
cD
D
F
G
cF H
cG
cH
0.5 1.0 1.5 2.0 2.5 3.0 3.5
0.02
1.0
cE
0.01
800
1000
1200
1400
1600
P, @MeV êcD
1000
1200
1400
1600
P, @MeV êcD
Ravg
5.0
General Parameters of an initial 0. 250 – 1 GeV non-scaling, nearisochronous FFAG lattice design
Parameter
Avg. Radius (m)
250 MeV
3.419
585 MeV
4.307
1000 MeV
5.030
Cell ν x / νy (2π rad)
0.380/0.237
1.520/0.948
1.62/-0.14
1.17/0.38
0.400/0.149
1.600/0.596
0.383/0.242
1.532/0.968
2.06/-0.31
2.35/-0.42
1.59/0.79
1.94/1.14
Ring.
Field F/D (T)
Magnet Size F/D Inj
•
4.5
4.0
1000
1200
1400
1600
P, @MeV êcD
Clockwise: Matematica: Ring tune, deviation from
isochronous orbit (%), and radius vs. momentum
Comments and further work
–
–
Tracking results indicate ~50-100π mm-mr; relatively insensitive to errors
Low losses
Particle Accelerator
Corporation
Advanced Modeling: Simulations in COSY INFINITY
In general conventional accelerator*
codes provide too-little flexibility in
field description and are limited to
low order in the dynamics, new
tools were developed for the study
and analysis of FFAG dynamics
based on transfer map techniques
unique to the code COSY INFINITY.
HARD EDGE
Various methods of describing
complex fields and components are
now supported including
representation in radius-dependent
Fourier modes, complex magnet
edge contours, as well as the
capability to interject calculated or
measured field data from a magnet
design code or actual components.
FULL FRINGE FIELDS
Arbitrary shapes, field content, contours
Other FFAG codes: ZGOUBI, F. Meot, some implementation in CYCLOPS (Baartman, TRIUMF)
Particle Accelerator
Corporation
Example of dynamics studies of fixed-field accelerators
using new tools in COSY
Below is a sample FFAG having sixfold symmetry, with focusing stemming from an
azimuthal field variation expressed as a single Fourier mode as well as edge focusing. The
system is studied to various orders of out-of-plane expansion with the results for orders
three and five shown below (typical of a conventional out-of-plane expansion in codes
like Cyclops).
Tracking in a model non-scaling six-fold symmetric FFAG for horizontal (left pairs) and vertical (right, pairs) with 3rd (top), 5th(middle),
and 11th-order (bottom) out of plane expansion, with focusing from an azimuthal field variation expressed as a single Fourier mode as
well as edge focusing . With (left) and without (right) Expo symplectification is shown.
Field Map and Tracking: 250-1000 MeV Isochronous Proton FFAG:
COSY INFINITY: Order 11
Immediate large DA aperture:
+/- 0.25 m horizontal
Larger than vertical gap of magnet
50-100 mm-mr without correction
0.1-1% error tolerance –typical magnet
tolerances
Final isochronous optimization will be
performed using advanced optimizers in
COSY
Dynamic aperture at 250, 585, and1000 MeV – step
size is 1.5 cm in the horizontal (left) and 1 mm in the
vertical (right).
1-GeV Proton Driver Modeled in CYCLOPS: fine mesh and fringe
fields (Y.N Ray and M. Craddock, TRIUMF)
B field with Enge function fall off
tune per cell (radial, horizontal)
tune/cell (z, vertical)
frequency change, isochronous to
+/- 3% using simple hard-edge
model: progress will require the
advanced codes; agrees with COSY
results
Recent ZGOUBI Results S. Sheehy: DA even larger
Very Preliminary Hardware Concepts for a
SC Nonlinear NS FFAG (ADS application)
• An innovative new approach to a
combined-function 4T magnet is
under design based conventional
NbTi superconducting magnet
technology and construction
techniques.
Preliminary rf parameters have
been investigated With a small
vertical aperture of only a few cm,
the large horizontal magnet
apertures do not present a serious
technical issue in terms of an rf
structure.
In Summary – Case for Carbon
FFAGs can reach the highest energies in CW mode (lower energy injection
would likely be via C4+ or 5+ ).
Carbon requires a synchrocyclotron
Small (~1/2 the footprint of 60 m circumference synchrotron)
12 magnets in four periods, i.e. four 2m straights – can be expanded to 3-4 m.
Injection, extraction, RF, one spare straight
Multi-ion capability: charge/mass ratio of ½ is supported (He, H2+)
Can use the preceding ring designs as injectors – staged proton/carbon facility
with multi-port treatment capability
Variable energy is still possible or minimal use of degrader
Avoids the fragmentation problem of carbon with a cyclotron/full degrader
Delivers a low, constant dose: easy to monitor and integrate
CW sets the stage for PBS using carbon
Fulfills and presents the strongest argument for pursuing a new IBT
facility – satisfying Development, Operation, Research and potentially
Local Competence – particularly since carbon facilities are still in a
development stage.
Nesting Rings for Compact Footprint
Embedded rings support a compact multi-ion therapy facility.
3.5 m
12C 6+
p
Development of New Scaling FFAGs and NS FFAGs: EMMA
Revival of the FFAG is completely due to Y. Mori, presently at
Kyoto University – scaling FFAGs
Does the NS FFAG work?
Very long list of detractors
Short list of advanced designers – most of whom have moved into
this field – many are in the auditorium
Most important supporter: A. Sessler
~2002 proposed building an electron model of a NS FFAG
The U.K. collaboration took the lead (Daresbury Laboratory, AsTeC,
Oxford, U of Manchester, and many others) submitted a NEST
proposal and EMMA and PAMELA projects were “born” circa 2004
S. Berg, BNL did the final doublet lattice
S. Koscielniak, TRIUMF, first predicted the existence of the
serpentine acceleration channel – acceleration between fixed rf
buckets
Operating Scaling FFAGs or Scaling FFAGs
under construction (not updated)
- clearly not an experimental machine Ion
E
(MeV)
Cells
Spiral
Angle
Radius
(m)
First
Beam
Technical
Details
KEK-PoP
p
1
8
0º
0.8-1.1
2000
KEK
p
150
12
0º
4.5-5.2
2003
100 Hz, 90% ext.
KURRI
-ADSR
(Figure 1)
p
2.5
8
40º
0.6-1.0
2006
Initial spec: 120 Hz, 1µA
Now: 1kHz, 100µA, @200 MeV
p
20
8
0º
1.4-1.7
2006
p
150
12
0º
4.5-5.1
(2007)
NEDO-ERIT
p
11
8
0º
2.35
(2008)
70mA ionization cooling ring
PRISM study
α
0.8
6
0º
3.3
(2008)
Phase space rotator
Radiatron
e
5
12
0º
0.3-0.7
(2008)
24 kW, 10 kHz, betatron acc.
FFAGs in Industry
Mitsubishi – hand-held 1 MeV electron accelerator (spiral
sector FFAG)
NHV Corporation
Developing several commercial electron FFAGs to
replace Rhodotron, for example
Number of proton FFAG designs have been completed for
medical use (RACCAM, PAMELA projects)
Passport Systems : compact electron FFAGs for security
applications
Radiabeam – compact electron FFAG development
Last Word: EMMA – Does the NS FFAG work?
The definitive answer is YES !
PREDICTION
VS
MEASURED!
Measured orbital period in the EMMA ring as a function of momentum, measured cell tune based on BPM data, and
measured phase space trajectories of beams starting from 12.0 MeV/c equivalent momentum with 5 different initial phases
and accelerated to more than 18 MeV/c equivalent momentum. At each empty circle, the phase is measured in the 23rd cell
and measured BPM orbit position.
Nesting Rings for Compact Applications
FFAG designs are advancing rapidly internationally, particularly for
medical and high-energy applications with an isochronous nonscaling
FFAG now designed and verified
The first demonstration of Accelerator Driven Subcritical Reactor was
performed this year at KURRI using the FFAG.
Embedded rings supporting a compact multi-ion therapy facility are an
exciting new direction for FFAGs.
Highly advanced design and simulation tools have been developed and
tested for FFAGs and cyclotrons and are ready for public distribution
12C 6+
3.5 m
p
Periodic Beam Transport
combined emittance/ESS collimation
Vertical half-ring, 360 deg phase advance
Achromat
Dispersion wave peaks in mid-straights
Geometric emittance + energy selection
4.4
4.2
4.0
3.8
3.6
3.4
3.2
3.0
cA
EA
cE
cB
cC
n h, n v
0.5
0.4
0.3
FB
0.2
cD
C
G
0.0
690
HD
cF
0.1
700
710
720
730
P, @MeV êcD
90 deg cells: achromatic
across large energy range
cG cH
0.5
1.0
1.5
Particle Accelerator
Corporation
Example: 50 keV to 9 MeV Compact Electron Accelerator
nv
3.0
2.5
2.0
1.5
1.0
2
4
6
8
P, @MeV êcD
COMPARISON Tune Results:
hard edge vs. extendedfFringe field
Mathematica® Full 9-MeV ring with injection
and extraction orbits displayed.
TOSCA Magnet Design and Field profiles.
COSY INFINITY Tune Results
COSY Results: Electron Ring
COSY results and tracking at 3 energies: 50 keV, 4.4 MeV, and 9 MeV machine. Except for the tune change
required at horizontal injection, the DA is very large.