DESIGN OF A NOVEL FLOW-AND-SHOOT MICROBEAM
G.Garty1,*, M.Grad2, B. K.Jones2, Y .Xu1, J.Xu2,3, G.Randers-Pehrson1, D.Attinger2and D. J.Brenner1
1RARAF, Columbia University, 136 S. Broadway, Irvington, NY 10533, USA
2Department of Mechanical Engineering, Columbia University, 500 W 120th Street, New York, NY 10027, USA
3Present address: Department of Mechanical Engineering, Washington State University, Vancouver, W A 98686,
*Corresponding author: firstname.lastname@example.org
Presented here is a novel microbeam technology—the Flow-And-ShooT (FAST) microbeam—under development at RARAF.
In this system, cells undergo controlled fluidic transport along a microfluidic channel intersecting the microbeam path. They
are imaged and tracked in real-time, using a high-speed camera and dynamically targeted, using a magnetic Point and Shoot
system. With the proposed FAST system, RARAF expects to reach a throughput of 100 000 cells per hour, which will allow
increasing the throughput of experiments by at least one order of magnitude. The implementation of FAST will also allow the
irradiation of non-adherent cells (e.g. lymphocytes), which is of great interest to many of the RARAF users. This study pre-
sents the design of a FAST microbeam and results of first tests of imaging and tracking as well as a discussion of the achiev-
Current microbeam systems(1)
cells adhered to a thin membrane. The cells are
microbeam where they are individually targeted,
either by a precision mechanical stage or by deflect-
ing the beam slightly to hit individual cells (Point
and Shoot). There are two drawbacks of this pro-
cedure. First, it only allows irradiation of cells that
can be made to adhere to the membrane. Second,
throughputs to about 10 000 cells per hour(2, 3), lim-
iting the possibility to probe rare endpoints such as
mutagenesis and oncogenesis.
To expand irradiation throughput and capabilities,
the authors propose to build a novel microbeam
using Flow-And-ShooT (FAST) technology. In this
system, cells will undergo controlled flow along a
microfluidic channel intersecting the microbeam path.
They will be imaged and tracked in real-time, using a
high-speed camera and targeted for irradiation by
single protons or helium nuclei, using the existing
Point and Shoot system. With the proposed FAST
system, a throughput of 100000 cells per hour is
expected, allowing experiments with much higher
The implementation of FAST will also allow the
irradiation of non-adherent cells (e.g. cells of hema-
topoietic origin), which is of great interest to many
of the RARAF users. Current irradiation of lym-
phocytes is extremely difficult due to the low yield of
cells that can be attached to a surface(4).
The term ‘microfluidics’ pertains to the behaviour,
control and manipulation of fluids geometrically
constrained to small length scales, where ‘micro’
forces such as surface tension overtake ‘macro’
forces such as gravity(5). The field involves continu-
ous flow microfluidics, which deals with the flow in
channels with sub-millimetre critical dimensions,
and digital microfluidics, which deals with nanolitre-
to picolitre-sized droplets(6). At these small length
scales, viscous forces are much larger than inertia
and the Reynolds number of the flow is very low
(Re ,, 1). This leads to a laminar fluid flow that
does not exhibit the random oscillations associated
with the turbulent flow. Microfluidics is also natu-
rally linked to biotechnology: continuous flow in
microchannels can be used as a carrier for biological
cells with high throughputs, and droplets can be
used to encapsulate cells for isolation from external
factors(7). In this paper, we discuss both continuous
flow (FAST microbeam) and digital microfluidics
(cell encapsulation), and their impact on improve-
ments to the microbeam at RARAF.
DESIGN OF THE FAST MICROBEAM
The FAST end station is designed for mounting on
the Permanent Magnet Microbeam (PMM) at
RARAF(8). The PMM currently provides a focused
Heþþbeam (5.2 MeV/5 mm diameter/50–250 par-
ticles per second) with development under way for
providing a proton microbeam (4.5 MeV/5 mm).
# The Author 2010. Published by Oxford University Press.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/
licenses/by-nc/2.5), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is
Radiation Protection Dosimetry (2011), Vol. 143, No. 2–4, pp. 344–348
Advance Access publication 11 December 2010
The PMM is also equipped with a magnetic
‘Point and Shoot’ beam deflector, which can target
the beam anywhere in a 60?240 mm field-of-fire
with a targeting time of ,1 ms.
The following three design criteria are important
for designinga microfluidic
microbeam system. First, a laminar flow pattern is
sought; so the position of cells can easily be pre-
dicted. Second, the flow rate of the cells past the
irradiation chamber must match the available beam
flux. For example, given the 60-mm wide ‘field of
fire’ of the Point and Shoot system and a beam flux
of 100 particles per second, this translates into a
required cell velocity of 6 mm s21to deliver a dose
of 1 particle per cell. A proportionally slower vel-
ocity is required if multiple cells are present in the
field-of-fire or if a higher dose is desired. Finally, the
height of the channel and surrounding material must
be limited; so it does not induce excessive beam scat-
tering or prevent the beam from reaching the cells
and the detector above the channels. This is not a
problem for 4.5-MeV protons (295-mm range in
water) but is a significant limitation for 5.2-MeV
helium nuclei (40-mm range in water)
MANUFACTURING OF CHANNELS
The microfluidic channels used in this work were
manufactured using soft lithography(9). Since the
channel width is larger than 100 mm, moulds were
made of polymethylmethacrylate (PMMA) slabs
using a computer numerical controlled micromilling
lathe (Minimill 3, Minitech Machinery). The accu-
racy of the lathe is between 2 and 5 mm, with the
surface roughness being on the order of 100 nm.
Micro-endmills are commercially available with a
diameter down to 25 mm. The length scales of
milled microgeometries (50 mm to several milli-
metres) bridge the gap between the conventional
macroscale machining (millimetre to metre) and
lithography (nm to 100 mm). The advantages of
micromilling over UV lithography are that (1) a
clean room is not required; (2) the cumbersome
process of drawing and ordering photolithographic
masks is replaced by the software generation of
numerical commands to the milling machine out of
a computer drawing and (3) the possibility to mill
non-planar surfaces. A picture of a mould created
with the micromilling machine is shown in Figure 1.
Once the mould is machined, Polydimethylsiloxane
(PDMS)—a transparent silicone-based rubber—is
poured over and allowed to harden to form three
sides of the microfluidic channels. Both the PDMS
and the planar surface used to seal the channel
(made, for example, of glass, silicon, Si3N4, mica)
were treated with oxygen plasma for 30 s, and then
For preliminary testing of cell flow and targeting,
the authors have manufactured a PDMS microfluidic
chip (Figure 2), featuring a flow-through channel, as
described previously. For testing of the tracking
system, the channels were sealed by plasma-bonding
a standard glass cover slip. For cell irradiations,
channels will be sealed to a 1-mm thick Si3N4
window, which will also serve as the beam-line
vacuum window. The cross section of the channel
has respective width and height of 200 and 20 mm,
so that the cells, when targeted by the microbeam,
flow within 20 mm of the exit window. This should
be contrasted with the distance of 50–100 mm cur-
rently maintained in the PMM between the exit
window and the (moving) cell dish.
The flow rate is controlled by a syringe pump
(KDS210, KD Scientific). Following irradiation, the
cells will be either delivered into a capillary tube for
storage or interfaced with other microfluidic chips
(flow-through or cell dispensers) using modular
through the microfluidic channel, first, it is necessary
to image and track the cells, and then shoot par-
ticle(s) at them. Due to geometry limitations of the
Figure 1. A mould for microchannel machined into
PMMA using a micromilling machine.
Figure 2. Photograph of flow-through chip used for testing
microbeam end station, it is not possible to illumi-
nate the sample from below. Consequently the
RARAF microbeams have focused on epi-fluor-
escent imaging, using low concentrations of vital
stains (e.g. Hoechst 33342). Epi-fluorescent imaging
provides high-contrast images, enabling easy auto-
mation of the cell locating algorithm(3). The beam
(a lymphocyte nucleus is about 7 mm in diameter)
specify that a resolution of about 1 mm per pixel is
sufficient for tracking.
In their preliminary studies, the authors used a 25
frame-per-second (fps) image-intensified camera(3)
to take pictures of a fluorescent bead, then analysed
the images and located its position at different times.
In these experiments, the effective pixel size was 1.3
mm. Figure 3 shows an image of 5 mm diameter flu-
orescent beads flowing through the channel. Because
of the laminar flow, the bead motion is fairly
smooth and linear(11), so that the cell position at
subsequent time points can be accurately predicted.
To determine the error in tracking, the authors:
(1) Located all beads in each image (In), typically
two to three beads,
(2) Located, for each bead, the corresponding
(closest) bead in the previous image (In21).
(3) Predicted, by linear extrapolation, the position
of that bead in the next image (Inþ1).
(4) Compared the position of the prediction with
the actual position in the image Inþ1.
Figure 4 shows the distribution of prediction errors
based on tracking 10 beads over 270 frames. More
than 90 % of the errors are smaller than 1 mm and
98 % are smaller than the 2.5-mm beam radius avail-
able in the PMM.
throughput of about 3000 cells per hour. To further
increase the throughput, the authors are in the
process of incorporating a 30-times faster imaging
based on a scientific CMOS camera(12). The newly
introduced sCMOS camera line combines high sen-
sitivity with extremely fast data transfer rate (100 fps
at 5 Megapixel and 900 fps at 320?240). The
6.5 mm pixel size coupled with a 10? objective will
yield sufficient resolution for cell tracking, while cov-
ering the entire width of the channel even at a resol-
prototype of the pco.edge, loaned to us by the
Cooke Corporation, indicated its suitability for this
application. Further tests will be performed when
the commercial version
October 2010. The use of a fast camera along with
the implementation of a hardware-accelerated track-
ing algorithm will allow both increasing of cell-flow
speeds and flowing of a larger number of cells in
parallel, with the aim of reaching an irradiation
throughput of 100 000 cells per hour.
Of prime importance in a microbeam system is the
minimisation of beam scattering. This is achieved by
the close proximity of the cells to be irradiated to
the beam exit window. As noted above, the channel
to be used is 20 mm thick, with the accelerator exit
window serving as its bottom surface. This con-
strains the cells to be within 20 mm of the exit
window, minimising the path of scattered particles to
no more than 20 mm, during which they can only
move a fraction of a micron laterally.
If necessary, in order to further push the cells
down towards the irradiation window, the channel
height can be locally reduced by incorporating a per-
pendicular pneumatic channel on top of the flow
channel(13); a constant pressure of about 100 mbar
applied with a pressure regulator would force the
Figure 3. A fluorescent image of two beads flowing
through a microfluidic channel. The beads are denoted by
arrows, and the measured trajectories are overlaid. The
dashed lines denote the boundaries of the channel.
Figure 4. Error distribution of the predicted position.
G. GARTY ETAL.
cells to pass in close contact with the irradiation
SRIM simulations of this geometry indicate a
beam broadening of ,1 mm at the top of the
channel, due to scattering, for a Heþþbeam and
negligible scattering for a proton beam. This will, of
course, be verified using the standard knife edge
technique(14), using a TEM grid in the channel. By
scanning the beam across the entire grid area and
monitoring the residual beam energy as a function
of targeted position, the authors can measure both
the targeting accuracy and the spot size (the deflec-
tion required to transit between fully occluded and
In order to allow particle detection and counting,
the chip was designed to be thin enough to be pene-
trated by 5-MeV Heþþparticles, with a residual
energy of .1 MeV (?8 mm range in air). This will
allow the authors to couple the chip to a thin-
window gas proportional counter, similar to the one
the authors are currently using(3)(this will, of
course, not be an issue with the longer range
INTEGRATION WITH OTHER
The addition of microfluidics into the microbeam
end station offers additional flexibility for integrat-
ing the beam with other technologies, including cell
sorting(15)and cell encapsulation(16).
Currently, in microbeam systems, cell sorting is per-
formed by manual cell manipulation with pipettes
(at a speed of about 60 cells per hour). This process
is very laborious and impractical, when dealing with
larger numbers of cells. The integration of microflui-
dics into the microbeam end station allows for the
easy integration of microfluidic cell sorters(15).
Sorting can take place before the irradiation, e.g.
sorting cells by cell-cycle stage, to form a synchro-
nised cell population for irradiation. Alternatively,
cellscan besorted either
irradiation or cells can be irradiated and incubated
together with normal cells before sorting, as required
for bystander experiments(17).
The authors have recently demonstrated(15)manu-
ally controlled cell sorting at a rate of about 3000
cells per hour. By automating the cell recognition
and switching, throughputs of 100 000 cells per hour
are easily achievable.
It has recently been shown that cells can be encapsu-
lated in aqueous droplets submerged in microfluidic
channels filled with oil, isolating the cells from the
‘beakers’(6). The encapsulation of cells in droplets
can easily be integrated before the cells flow into the
FAST microbeam, and the cells can be irradiated in
the droplets as they flow past the microbeam. This
technology has profound impacts on single-cell
microbeams, because factors secreted by the cell
during irradiation can be captured in a small volume
and studied. Also, two or more cells can be encapsu-
lated within the same droplet, one of which can be
irradiated and kept in close proximity to the other
cells. It is expected that, the bystander signal will be
enhanced, due to the close proximity of the cells and
the inability of secreted factors to diffuse away.
This study presents the preliminary design and
testing of a novel microbeam technology under
development at RARAF—the FAST microbeam. In
this system, cells undergo controlled fluidic transport
at rates of 1–10 mm s21along a 200-mm-wide/20-
microbeam path. Cells are imaged and tracked in
real-time, and dynamically targeted, using a mag-
netic Point and Shoot system. With the proposed
FAST system, this study expects to reach a through-
put of 100 000 cells per hour, which will allow
increasing the throughput of microbeam experiments
by at least one order of magnitude.
This work was supported by the national institute of
NIBIB, grant #5P41EB002033).
1. Bigelow, A. W ., Brenner, D. J., Garty, G. and Randers-
Pehrson, G. Single-particle/single-cell ion microbeams
as probes of biological mechanisms. IEEE Trans.
Plasma Sci. 36, 1424–1431 (2008).
2. Bigelow, A., Garty, G., Funayama, T., Randers-
Pehrson, G., Brenner, D. J. and Geard, C. Expanding
the question-answering potential of single-cell microbe-
ams at RARAF, USA. J. Radiat. Res. 50, A21–A28
3. Randers-Pehrson, G., Geard, C. R., Johnson, G.,
Elliston, C. D. and Brenner, D. J. The Columbia
University single-ion microbeam. Radiat. Res. 156,
4. Kadhim, M. A., Marsden, S. J., Malcolmson, A. M.,
Folkard, M., Goodhead, D. T., Prise, K. M. and
Michael, B. D. Long-term genomic instability in human
lymphocytes induced by
Radiat. Res. 155, 122–126 (2001).
5. Stone, H. A., Stroock, A. D. and Ajdari, A.
Engineering flows in small devices. Annu. Rev. Fluid
Mech. 36, 381–411 (2004).
6. Xu, J. and Attinger, D. Drop on demand in a microflui-
dic chip. J. Micromech. Microeng. 18, (2008).
7. Berthier, J. andSilberzan,
Biotechnology. Artech House, Inc (2006).
8. Garty, G., Ross, G. J., Bigelow, A. W ., Randers-
Pehrson, G. and Brenner, D. J. Testing the stand-alone
Dosimetry 122, 292–296 (2006).
9. Xia, Y . and Whitesides, G. M. Soft lithography. Annu.
Rev. Mater. Sci. 28, 153–184 (1998).
10. Yuen, P . K. SmartBuild-A truly plug-n-play modular
microfluidic system. Lab Chip 8, 1374–1378 (2008).
11. Bruus, H. Theoretical Microfluidics. Oxford University
12. Holst, G. Scientific CMOS imaging sensors: the new all-
14. Bigelow, A. W ., Ross, G. J., Randers-Pehrson, G. and
Brenner, D. J. The Columbia University microbeam II
endstation for cell imaging and irradiation. Nucl.
Instrum. Methods Phys. Res. B 231, 202–206 (2005).
15. Grad, M., Jones, B., Smilenov, L., Brenner, D. and
Attinger, D. A Microfluidic Cell Sorter based on
Hydrodynamic switching. In: Proceedings of International
16. Xu, J. and Attinger, D. Dispensing individual fluid par-
ticles on demand in a microfluidic chip. Proceedings of
MicroTAS, San Diego, pp. 1378–1380 (2008).
17. Hei, T. K., Zhou, H., Ivanov, V . N., Hong, M.,
Lieberman, H. B., Brenner, D. J., Amundson, S. A.
and Geard, C. R. Mechanism of radiation-induced
Pharmacol. 60, 943–950 (2008).
S. J.and Quake,
Congress and Exposition,
G. GARTY ETAL.