Nanotechnology 22 (2011) 285503 (7pp)
Surface imaging using holographic optical
D B Phillips1, J A Grieve1, S N Olof1, S J Kocher1, R Bowman2,
M J Padgett2, M J Miles1and D M Carberry1
1H H Wills Physics Laboratory, University of Bristol, Tyndall Avenue, Clifton,
Bristol BS8 1TL, UK
2SUPA, Department of Physics and Astronomy, University of Glasgow,
Glasgow G12 8QQ, UK
Received 4 March 2011, in final form 19 April 2011
Published 7 June 2011
Online at stacks.iop.org/Nano/22/285503
We present an imaging technique using an optically trapped cigar-shaped probe controlled using
holographic optical tweezers. The probe is raster scanned over a surface, allowing an image to
be taken in a manner analogous to scanning probe microscopy (SPM), with automatic closed
loop feedback control provided by analysis of the probe position recorded using a high speed
CMOS camera. The probe is held using two optical traps centred at least 10 µm from the ends,
minimizing laser illumination of the tip, so reducing the chance of optical damage to delicate
samples. The technique imparts less force on samples than contact SPM techniques, and allows
highly curved and strongly scattering samples to be imaged, which present difficulties for
imaging using photonic force microscopy. To calibrate our technique, we first image a known
sample—the interface between two 8 µm polystyrene beads. We then demonstrate the
advantages of this technique by imaging the surface of the soft alga Pseudopediastrum. The
scattering force of our laser applied directly onto this sample is enough to remove it from the
surface, but we can use our technique to image the algal surface with minimal disruption while
it is alive, not adhered and in physiological conditions. The resolution is currently equivalent to
confocal microscopy, but as our technique is not diffraction limited, there is scope for significant
improvement by reducing the tip diameter and limiting the thermal motion of the probe.
(Some figures in this article are in colour only in the electronic version)
Numerous fields ranging from cellular biology through to
device fabrication use scanning probe microscopy (SPM) to
determine surface topography [1–3]. The broad applicability
of the technique stems from its ability to map out the
location of nanoscale features, such as surface defects in
semiconductors  or receptor sites on cell membranes
under physiological conditions [5–7]. Modification of probe
surface chemistry has enabled the interactions between various
functionalized surfaces to be investigated  and has spawned
techniques such as force–volume interaction mapping. SPM
instrumentation research is mainly aimed at increasing the
rates at which surface images are obtained [9, 10], and in
lowering the interaction forces . Optical tweezers  and
photonic force microscopy (PFM)  have also been used to
identify surfaces within a sample. Using PFM, probe particles
have been held in the optical trap and their inherent Brownian
motion used to map the cavities in which they are located .
Atomic force microscopy (AFM) ‘contact-mode’ functions
have been replicated within optical tweezers by translating a
trapped probe along a surface [15, 16], and an optically trapped
scanning near-field optical microscope (OT-SNOM) has also
been demonstrated .
Each of these techniques (SPM, PFM) has its own
strengths, but is also subject to its own set of constraints.
SPMs can record pixels at MHz rates, are able to use tips
which can taper to a single atom, and they are able to tune
the residual Brownian motion to amplitudes of less than 1 nm.
However, they require relatively flat samples and operate in
an orientation that is perpendicular to the surface. Contact
and tapping mode AFMs also exert relatively large forces
© 2011 IOP Publishing LtdPrinted in the UK & the USA
Nanotechnology 22 (2011) 285503D B Phillips et al
Figure 1. (a) An SEM image showing typical diatoms used in the experiments. (b) An SEM image of the tip of a single diatom. As can be
seen the tip has a radius of curvature of approximately 350 nm. This is within the natural tip variance of the diatoms: 200–500 nm.
on samples , of the order of hundreds of piconewtons,
which can damage sensitive biological specimens.
tweezers and PFMs are sensitive to very low forces, typically
in the range of tens of femtonewtons. This force sensitivity
comes at the price of lower positional resolution, which is
limited by Brownian motion and the size of the probe particle.
Additionally, PFMs exert high laser intensities near the sample
and are ineffective in environments where samples are likely to
cause beam interference or occlusion.
The aim of this work is to combine the strengths of SPM
and optical tweezers to image any surface.
uses a probe held by two optical traps (figure 2) which
can be dynamically reconfigured to reorient the probe using
holographic beam shaping [19, 20]. This enables the imaging
of cells with low elastic moduli, minimizing membrane
disruptionandallowingtheinvestigationof cellsnotadhered to
a substrate. Using this technique, we can vary the orientation
and position of our probe, scan along any path or axis, and
remove the probe tip from the laser focus—limiting the laser
intensity which is incident on the sample.
grants access to samples which are highly curved or have
challenging surface geometries in their native environments.
2. Experimental details
2.1. Holographic optical tweezers setup
Our holographic optical tweezers are slightly modified from
those reported in . The system consists of an inverted
brightfield microscope with a 1.4 NA, 100× objective (Plan-
Neofluar, Zeiss), a motorized xy stage (MS-2000, ASI) and
a closed loop piezoelectric objective lens positioning system
(Piezosystem Jena, Mipos 140 PL). The trapping beam is
provided by a Nd:YAG laser (Laser Quantum) emitting up to
3.2 W at 1064 nm. The beam is expanded to fill a spatial light
modulator (Hamamatsu, X10468-07), the hologram on which
the back aperture of the objective lens. A polarizing beam
splitter is used before the objective to transmit the laser beam
while reflecting half of the illumination light onto a Firewire
CMOS camera (Prosilica, EC1280).
feedback control were implemented in LabVIEW (National
Instruments)  running on a quad core PC, which also
contained the graphics processor (GPU) used for hologram
calculation (nVidia, Quadro FX 5600).
hologram generation and use of the GPU can be found in .
Image analysis and
Further details on
2.2. Preparation of probes
Probes can take a variety of forms—from the microtools
Here we have elected to use the diatom Nitzschia acicularis
(K¨ utzing) W Smith as the optically trapped probe as it can
be easily trapped and manipulated in 3D. SEM images of
typical diatoms are shown in figure 1. The diatoms are stored
in diatom medium, which promotes cell division and acts as
the trapping medium. The diatom is primarily composed of
a large, central cylindrical mass, which tapers to a rounded
point between 400 and 1000 nm in diameter. This structure
provides several advantages over a pure cylinder: the tapering
of the silica shell gives a slender probe with a small contact
area which is ideal for probe microscopy, and the larger size
of the central cylinder has a higher trapping constant than a
microrod of constant width equal to the tip diameter. Optical
images show two elliptical structures where light is focused,
which are used as the optical trapping locations.
Tracking the two high-intensity elliptical structures
using a centre-of-mass algorithm, and utilizing the methods
described in [28, 26], enables the translational and rotational
motion within the focal plane to be recorded at approximately
250 Hz. Equipartition is then used to calibrate the stiffness
of the probe in the trap resulting in κ = (κz,κx,κθ) =
(1.8×10−5N m−1, 5.7×10−5N m−1, 1.6×10−15N m rad−1),
where z is parallel to the probe’s long axis, x is normal to the
probe’s long axis and θ is the rotational deviation within the
focal plane about the optical stress centre. As position tracking
was only possible within the focal plane, we elected to employ
an image comparison algorithm to detect probe translations
Nanotechnology 22 (2011) 285503D B Phillips et al
Figure 2. A schematic showing the key points during the imaging of a sample. Dimensions and translations have been exaggerated for clarity.
At State 1 the probe is positioned such that there is no physical contact with the sample and the base image is recorded. The probe is advanced
in a stepwise manner until it makes contact in State 2, causing the probe to rotate or translate relative to the optical traps. This results in a
change in the image comparison routine, allowing the interaction to be detected and the coordinates of the pixel recorded. The probe is moved
back to its original configuration, to State 3, and then moved to the next pixel’s starting location. These steps are repeated in a raster scan to
obtain a whole image.
Figure 3. (a) An example of an interaction curve showing coarse and fine approaches to detect contact with the sample. The coarse approach
(blue crosses) allows approximate detection of the sample. A coarse withdrawal to the start of the fine approach (red circles) ensues. The fine
approach then detects the surface more accurately, resulting in contact at point A. (b) A flow diagram describing the feedback control loop
used to detect the height of each pixel in the image.
and rotations in and out of the focal plane. While this does
not provide quantitative forces acting on the sample, it does
provide an accurate surface profile.
2.3. Imaging methodology
The height z of each pixel in the surface image is obtained in a
manner analogous to AFM force–volume mapping. The probe
is positioned at the desired (x, y) pixel coordinate such that it
is not in contact with the sample, State 1 in figure 2. A region
of interest (ROI) is defined (red box in figure 2), and a base
image recorded. The positions of the ROI and optical traps are
updated to move the probe a step in the z direction towards the
sample. Once the probe has moved to its new position another
image of the probe is recorded, the stepped image. To reduce
the effect of the probe’s thermal motion, both the base and
stepped images are an average of five consecutively recorded
images—a compromise between noise reduction and imaging
speed. The average square of the difference D in intensity
between the base image and the stepped image is calculated
i,j(xi,j− ˆ xi,j)2
where xi,j is the greyscale value of the pixel in row i and
column j ofthebaseimage, ˆ xi,jisthesamepixelinthestepped
image, and N is the total number of pixels in each image. As
the ROI is translated the same distance as the traps, provided
the probe has not yet contacted the sample, the images will be
similar and D will be small. If the probe has made contact
with the sample it will translate and/or rotate relative to the
optical traps (State 2 in figure 2). The resulting difference in
the base image and the stepped image increases the relative
magnitude of D. When D exceeds a nominated threshold the
probe is defined as in contact with the sample, and retracted.
The probe control and image analysis is performed in real time
by an automated LabVIEW routine at approximately 30 Hz.
To maximizeimagingspeed andresolutiontwointeraction
curves were recorded for each pixel. A coarse approach was
performed with large steps to quickly detect the sample (blue
crosses in figure 3(a)). The maximum normal force can be
estimated by assuming a full coarse step occurs after contact,
resulting in a maximum normal force of 14 pN. When D
Nanotechnology 22 (2011) 285503 D B Phillips et al
exceeded thepre-determinedthresholdvaluea twotothreestep
coarse withdrawal occurs. The new probe location is closer
to the sample than the original probe’s starting position, for
example the first red point in figure 3(a). A fine approach
is then performed with smaller steps, thereby increasing the
z resolution.When D exceeds the threshold value in the
fine scan the probe is coarsely withdrawn and moved to the
coordinate defining the start of the next pixel. The flowchart in
figure 3(b) illustrates this process.
Values of D are recorded for every stepduringtheimaging
process, therefore each pixel has an associated coarse and
fine interaction curve—akin to AFM force–volume mapping.
Sample images are reconstructed from these curves in post-
processing, by evaluating the point in the interaction curve
at which the image starts to differ significantly from the
background noise (point A in figure 3(a)).
The pixel measurement order follows a raster scan pattern.
After each pixel is detected and the probe fully retracted, the
traps are stepped in the x direction to the adjacent pixel. When
a row is completed, the objective lens positioning system is
used to step the focal plane to the next row of pixels to be
scanned. This simultaneously changes the height of the probe
and the focal plane relative to the sample. All scan parameters
can be modified by the operator. This imaging method offers
the potential to reorient the probe during the imaging from
pixel to pixel. For example, the scan path could follow a raster
scanning around a sample, or other arbitrary trajectories.
3. Results and discussion
To characterize the technique, a selection of 8 µm latex
spheres were rigidly adhered to a coverslip. The sample was
searched until two microspheres were found in contact, and
our technique used to image the contact region. Figure 4(a)
shows a 3D surface plot of the contact area between the two
microspheres. It also indicates the x (lateral) and y (vertical)
probe stepsize used to discretely sample the surface, and the
fine approach stepsize z towards the surface for each pixel.
The colourbar indicates the depth z on the surface plot. The
axes show the orientation of the coordinate system used in all
subsequent plots. The dark blue base of the surface represents
where the probe has continued past the gap between the
microspheresundisturbed,andreached adesignatedzero point.
Figure 4(b) demonstrates the experimental configuration with
the diatom probe approaching the two latex spheres. The red
box around the centre of the probe indicates the border of the
region of interest that was maintained around the trap positions
as the probe was translated towards the sample.
In a manner analogous to SPM, the number of samples
per line and the sampling stepsize may be changed to image
different sized areas at varying resolution. This allows the
user to zoom in on areas of interest. Figure 5(a) shows a
zoomed in image of a similar 8 µm latex sphere interface,
where the sampling stepsize has been reduced from 320 to
160nm. Figure5(b) showsthelinescanfrom thisimagearound
the equator of the microsphere. It demonstrates that the radius
of curvature of the diatom tip is approximately 250 nm, within
Figure 4. (a) A 3D surface scan of the interface between two 8 µm
latex microspheres. The x (lateral), y (vertical) and z (approach)
probe stepsize used to discretely sample the surface are also
indicated, along with the coordinate system used. The colourbar
indicates the depth z on the surface. Background spheres have been
added to the image to indicate the position of the microspheres.
(b) An optical image of the experimental configuration, with the
diatom probe approaching the microspheres. The red box indicates
the region of interest defining the portion of the image used to
determine when the probe is in contact with the sample.
the natural variance, and confirms that the microspheres are
8 µm in diameter.
After characterizing the imaging quality using micro-
spheres, a biological sample was imaged.
structure and soft cellular walls of the green alga Pseudope-
diastrum boryanum (Turpin) E. Hegewald prevents traditional
SPM from being used, and the strongly scattering structure
prevents PFM and optical tweezers from imaging the surface.
It represents a challenging sample on which to demonstrate our
imaging technique. The Pseudopediastrum was cultured and
stored in 3N-BBM + V (Bolds’ basal medium with three-fold
nitrogen and vitamins). Samples were prepared by adding an
aliquot of diatom and Pseudopediastrum medium to a glass
chamber. Figure 6 shows an optical image and confocal slice
of a typical Pseudopediastrum colony.
The probe was first positioned alongside a single
Pseudopediastrumcolonywhich had settledonto thecoverslip,
as shown in figure 7(b). As the Pseudopediastrum was not
adhered to the substrate, it could be easily moved if sufficient
laser power was directly applied. Our technique was utilized
to obtain images of a single spike in the algal structure
(figure 7(a)) and an image of the entire wall of a single
Pseudopediastrum unit cell (figure 7(c)).
figure 7(d) shows an image of a similar sample obtained by
brightfield confocal microscopy.
The 3D crown
Nanotechnology 22 (2011) 285503D B Phillips et al
Figure 5. (a) A higher resolution 3D surface representing the interface of two 8 µm microspheres. (b) A single line scan from (a) through the
equator of the two 8 µm microspheres. The measured gap between the spheres is approximately 500 nm, demonstrating that the probe’s
radius of curvature is 250 nm, within the natural variance of the diatoms.
Figure 6. (a) A bright field image of a Pseudopediastrum colony using an optical microscope. The crown structure and unit cells are well
defined. (b) A slice of a bright field confocal image of the same Pseudopediastrum colony. The structure of the colony is best defined on the
underside of the sample, before interference from other layers distorts the image.
The resolution of these images is currently comparable to
the theoretical (δx,δy,δz) = (140,240,140) nm achieved
with the confocal system. However, our technique does
not subject the sample to the focused laser light required
in fluorescent confocal imaging, and does not suffer from a
reduction in resolution caused by light scatter through the
sample. Whileopticalandconfocalmicroscopyare diffraction-
limited techniques, the resolution of our technique in x and
y is limited by the probe’s 250 nm radius of curvature and
its residual Brownian motion (±2 standard deviations =
60 nm). The resolution in z is limited by the fine step size
and Brownian motion, not tip size. Further reductions in the
size of the probe could be achieved by identifying a biological
probe with a sharper tip or by using some of the micro- or
nanotools previously reported [24–26], therefore reducing tip
convolution artefacts. Improvements in the trapping technique,
such as position clamping the probe in 3D , could
reduce the amplitude of the thermal motion of the tip. The
technique itself is also easily parallelized. Using holographic
optical tweezers , or other methods for rapidly generating
numerous 3D-configurable optical traps, several probes could
be scanned simultaneously. This would have the advantage of
decreasing the imaging time and opens up the possibility of
imaging a sample while external stimulation is occurring, for
example, surface conformationscouldbeimagedwhilealipase
is used to disrupt the cell membrane.
In summary, we have developed a form of SPM which utilizes
holographicallycontrolled probes whose trapping locationsare
largely removed from the sample–probe interaction area. The
in 3D, bypassing the planar scanning constraints of traditional
SPM. This enables complex, highly curved samples to be
scanned with a resolution comparable to confocal imaging.
As the technique is not diffraction limited, there is scope for
improvements in scanning resolution.Also, the technique
Nanotechnology 22 (2011) 285503D B Phillips et al
Figure 7. (a) The 3D surface image of the spike encircled in (b). (b) An optical microscope image showing an optically trapped diatom in the
starting configuration (equivalent to State 1 in figure 2). The two optical traps are centred at each end of the elliptical handles. (c) The surface
image following the scan of the side wall of a Pseudopediastrum unit cell. (d) A bright field confocal image of a similar sample, for
comparison with (c).
imparts substantially less force than traditional contact SPM
techniques. We have demonstratedthisbyimagingthesideofa
Pseudopediastrumunitcellwhichwas notadhered toasurface.
This technique could be used when certain limitations in
existing technology are encountered and, in particular, for the
investigation of significantly curved or soft cellular surfaces.
This work is funded by a Basic Technology Translation Grant
through the Research Councils of the United Kingdom. MJM
acknowledges a Royal Society Wolfson Merit Award. This
work was carried out with the support of the Bristol Centre for
Nanoscience and Quantum Information. We would also like to
thank M Yallop, H Rosenkranz and D Fagan for culturing the
original biological samples.
 Garcia R and P´ erez R 2002 Dynamic atomic force microscopy
methods Surf. Sci. Rep. 47 197–301
 Santos N C and Castanho M A R B 2004 An overview of the
biophysical applications of atomic force microscopy
Biophys. Chem. 107 133–49
 H¨ orber J K H and Miles M J 2003 Scanning probe evolution in
biology Science 302 1002–5
 Morita S, Abe M, Yokoyama K and Sugawara Y 2000 Defects
and their charge imaging on semiconductor surfaces by
noncontact atomic force microscopy and spectroscopy
J. Cryst. Growth 210 408–15
 Haberle W, H¨ orber J K H and Binnig G 1991 Force microscopy
on living cells 5th Int. Conf. on Scanning Tunneling
Microscopy/Spectroscopy(Boston, MA) vol 9, AVS,
 Schaer-Zammaretti P and Ubbink J 2003 Imaging of lactic acid
bacteria with AFM—elasticity and adhesion maps and their
relationship to biological and structural data
Ultramicroscopy 97 199–208
 Cross S E, Jin Y-S, Rao J and Gimzewski J K 2007
Nanomechanical analysis of cells from cancer patients Nat.
Nanotechnol. 2 780–3
 Hinterdorfer P, Baumgartner W, Gruber H J, Schilcher K and
Schindler H 1996 Detection and localization of individual
antibody–antigen recognition events by atomic force
microscopy Proc. Natl Acad. Sci. USA 93 3477–81
 Picco L M, Bozec L, Ulcinas A, Engledew D J, Antognozzi M,
Horton M A and Miles M J 2007 Breaking the speed limit
with atomic force microscopy Nanotechnology 18 044030
Nanotechnology 22 (2011) 285503D B Phillips et al Download full-text
 Ando T, Uchihashi T and Fukuma T 2008 High-speed atomic
force microscopy for nano-visualization of dynamic
biomolecular processes Prog. Surf. Sci. 83 337–437
 Antognozzi M, Ulcinas A, Picco L, Simpson S H, Heard P J,
Szczelkun M D, Brenner B and Miles M J 2008 A new
detection system for extremely small vertically mounted
cantilevers Nanotechnology 19 384002
 Ashkin A, Dziedzic J M, Bjorkholm J E and Chu S 1986
Observation of a single-beam gradient force optical trap for
dielectric particles Opt. Lett. 11 288
 Florin E-L, Pralle A, H¨ orber J K H and Stelzer E H K 1997
Photonic force microscope based on optical tweezers and
two-photon excitation for biological applications J. Struct.
Biol. 119 202–11
 Rohrbach A, Tischer C, Neumayer D, Florin E-L and
Stelzer E H K 2004 Trapping and tracking a local probe with
a photonic force microscope Rev. Sci. Instrum. 75 2197–210
 Ghislain L P and Webb W W 1993 Scanning-force microscope
based on an optical trap Opt. Lett. 18 1678
 Friese M E J, Truscott A G, Rubinsztein-Dunlop H and
Heckenberg N R 1999 Three-dimensional imaging with
optical tweezers Appl. Opt. 38 6597–603
 Nakayama Y, Pauzauskie P J, Radenovic A, Onorato R M,
Saykally R J, Liphardt J and Yang P 2007 Tunable nanowire
nonlinear optical probe Nature 447 1098–101
 Gnecco E, Bennewitz R, Gyalog T, Loppacher Ch,
Bammerlin M, Meyer E and Gntherodt H-J 2000 Velocity
dependence of atomic friction Phys. Rev. Lett. 84 1172
 Reicherter M, Haist T, Wagemann E U and Tiziani H J 1999
Optical particle trapping with computer-generated
holograms written on a liquid-crystal display Opt. Lett.
 Dufresne E R, Spalding G C, Dearing M T, Sheets S A and
Grier D G 2001 Computer-generated holographic optical
tweezer arrays Rev. Sci. Instrum. 72 1810–6
 Gibson G, Carberry D M, Whyte G, Leach J, Courtial J,
Jackson J C, Robert D J, Miles M J and Padgett M 2008
Holographic assembly workstation for optical manipulation
J. Opt. A: Pure Appl. Opt. 10 044009
 Gibson G M, Leach J, Keen S, Wright A J and Padgett M J
2008 Measuring the accuracy of particle position and force
in optical tweezers using high-speed video microscopy Opt.
Express 16 14561–70
 Preece D, Bowman R, Linnenberger A, Gibson G, Serati S and
Padgett M 2009 Increasing trap stiffness with position
clamping in holographic optical tweezers Opt. Express
 Rodrigo P J, Gammelgaard L, Boggild P, Perch-Nielsen I and
Gl¨ uckstad J 2005 Actuation of microfabricated tools using
multiple GPC-based counterpropagating-beam traps Opt.
Express 13 6899–904
 Ikin L, Carberry D M, Gibson G M, Padgett M J and
Miles M J 2009 Assembly and force measurement with
spm-like probes in holographic optical tweezers New J.
Phys. 11 023012
 Carberry D M et al 2010 Calibration of optically trapped
nanotools Nanotechnology 21 175501
 Gl¨ uckstad J 2007 Manipulating microtools with nanofeatures
using light in 3d real-time iNANO Seminar (Arhus)
 Simpson S H and Hanna S 2009 Thermal motion of a
holographically trapped spm-like probe Nanotechnology
 Grier D G 2003 A revolution in optical manipulation Nature