theoretical comparison of optical traps created by standing wave and single beam · 2007. 3....
TRANSCRIPT
Theoretical comparison of optical traps created bystanding wave and single beam
Pavel Zem�aaneka,*, Alexandr Jon�aa�ssa, Petr J�aakla, Jan Je�zzeka,Mojm�ıır �SSer�yyb, Miroslav Li�sskab
a Institute of Scientific Instruments, Academy of Sciences of the Czech Republic, Kr�aalovopolsk�aa 147, Brno 612 64, Czech Republicb Faculty of Mechanical Engineering, Brno University of Technology, Technick�aa 2, Brno 616 69, Czech Republic
Received 4 June 2002; received in revised form 28 February 2003; accepted 3 April 2003
Abstract
We used generalised Lorenz–Mie scattering theory (GLMT) to compare submicron-sized particle optical trapping in
a single focused beam and a standing wave. We focus especially on the study of maximal axial trapping force, minimal
laser power necessary for confinement, axial trap position, and axial trap stiffness in dependency on trapped sphere
radius, refractive index, and Gaussian beam waist size. In the single beam trap (SBT), the range of refractive indices
which enable stable trapping depends strongly on the beam waist size (it grows with decreasing waist). On the contrary
to the SBT, there are certain sphere sizes (non-trapping radii) that disable sphere confinement in standing wave trap
(SWT) for arbitrary value of refractive index. For other sphere radii we show that the SWT enables confinement of high
refractive index particle in wider laser beams and provides axial trap stiffness and maximal axial trapping force at least
by two orders and one order bigger than in SBT, respectively.
� 2003 Elsevier Science B.V. All rights reserved.
PACS: 42.25.F; 42.50.Vk
Keywords: Single beam trap; Optical trapping; Optical tweezers; Standing wave; Mie scattering; Gaussian laser beam
1. Introduction
Since 1986 optical trapping has proved to be an
invaluable method for non-contact manipulation
of nanoobjects and microobjects, measurement of
extremely weak forces, study of single molecule
properties and surface properties [1–4]. The clas-sical set-up of optical tweezers uses a single laser
beam that is tightly focused by an immersion mi-
croscope objective of high numerical aperture [5].
This classical single beam trap (SBT) has been
gradually modified by using various transversal
beam profiles [6,7], interference of co-propagating
laser beams [8], self-aligned dual beam [9], or op-
tical fibres [10]. Multiple optical traps were createdusing several laser beams [11,12], diffractive optics
[13], and time sharing of a single beam [11,14,15].
Optics Communications 220 (2003) 401–412
www.elsevier.com/locate/optcom
* Corresponding author. Tel.: +42-05-4151-4202; fax: +42-05-
4151-4402.
E-mail address: [email protected] (P. Zem�aanek).
0030-4018/03/$ - see front matter � 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0030-4018(03)01409-3
The common characteristic of all the methods
mentioned above is that the axial force is smaller
than the radial one. This is given by physical limits
imposed upon intensity distribution in the focal
region of lens with limited numerical apertures.
Recently, it has been shown that optical trappingcan be also achieved in a standing wave created by
interference of two counter-propagating coherent
beams [16]. In this case the axial force is stronger
than the radial one because of inhomogeneous
optical intensity distribution in a periodical struc-
ture of standing wave nodes (minimums) and
antinodes (maximums). Therefore, particle con-
finement is achieved even in weakly focused oraberrated beams. An interesting problem, which
arises here, is to study the properties of this type of
optical trap (standing wave trap (SWT)) and
compare them with the classical SBT for different
parameters of a trapped object (size, refractive
index) and a trapping beam (waist size).
2. Theoretical background
Theoretical description of the interaction be-
tween a particle and an electromagnetic wavegenerally consists of description of an incident
beam, description of a field established due to the
interaction of the incident beam and the particle,
and description of forces acting on the particle. A
single beam optical trapping uses a focused laser
beam with a spot diameter comparable to the
trapping wavelength. To achieve this, high-quality
immersion objectives are usually used. In this casethe beam passes through a number of dielectric
interfaces (optical elements inside the objective,
immersion oil layer, coverslip, and water layer)
which makes a correct description of the field in
the focal region quite complex. Despite this fact,
the simple Gaussian beam is still the most fre-
quently considered form of the incident beam even
though that it is only a paraxial solution of thewave equation and it does not treat the polarisa-
tion components in the focused beam properly.
Therefore, there have been several attempts to
‘‘improve’’ the Gaussian beam so that it better
fitted the wave equation for off-axis positions in
the focused beam [17–20]. For this purpose, field
expansion in the beam size parameter s ¼ 1=ðkw0Þwas adopted but its solutions to wave equation are
known explicitly only up to the fifth order in the
parameter s. Moreover, even though the average
error of solution to Maxwell�s equations drops for
bigger s [18], it is still not sufficient for the focusedbeam that is usually used in optical tweezers (s ’0:4, which corresponds to the diffraction limited
focus, gives the average error ’3.34% in the fifth
order correction). A theoretical treatment de-
scribing the incident beam as the Gaussian beam
corrected to the fifth order (CGB) should therefore
deal with wider beam waists w0 (s6 0:2 gives
w0 P 0:8k) to reduce the average and the maximalerror (for s6 0:2 the two errors are lower than
0.062% and 1.19%, respectively [18]).
Exact Gaussian-beam-like solutions of the wave
equation are known in the oblate spheroidal co-
ordinates or using the complex-point source
method [21–23]. However, up to our knowledge,
there is no experimental method that could deter-
mine the components of the optical field distribu-tion below the objective with sufficient precision to
determine which of the above mentioned field de-
scriptions better fit the experimental conditions.
Moreover, it seems that the above mentioned im-
provements of the beam profile are still negligible
compared to the effects caused by diffraction of the
beam on the objective back aperture and by
spherical aberration due to the refractive indexmismatch at the dielectric interfaces below the
objective [24].
Description of the field produced as the result
of the interaction of the incident beam and the
particle is a general problem studied by scattering
theory. An original description of such a field,
presented by Mie and Lorenz at the beginning of
the 20th century, assumed a spherical particleilluminated by a plane wave (Lorenz–Mie the-
ory). This theory has been gradually modified so
that it can be applied for spheres, spheroids,
multilayered spheres, infinite cylinders with cir-
cular and elliptical cross-sections, sphere with one
internal eccentrically located spherical inclusion
and aggregates particle placed into an arbitrary
field distribution [18,25–28]. It is commonly re-ferred to as the generalised Lorenz–Mie theory
(GLMT).
402 P. Zem�aanek et al. / Optics Communications 220 (2003) 401–412
Description of a force interaction between the
beam and a particle is based on the momentum
conservation principle (forces acting on any finite
volume in a material body can be expressed
through the forces applied to the surface of that
volume) and on assumptions that a fluid sur-rounding the particle is isotropic, non-magnetic,
linear in its response to the applied field, and in
hydrodynamic equilibrium. In this case the time-
averaged total electromagnetic force acting on the
particle is equal to integral of the dot product of
the outward-directed unit normal vector n and the
Maxwell�s stress tensor in Minkowski form TM
over the surface S enclosing the particle [29,30]:
hFii ¼IS
Xj
T Mij nj da
* +; ð1Þ
TMij ¼
�eEiEj þ l0HiHj �
1
2ðeE2 þ l0H
2Þdij
�; ð2Þ
where e is the permittivity of a surrounding medium,l0 is the permeability of the vacuum, Ei and Hi are
components of the vectors of electric and magnetic
intensities, respectively. The components of the
optical force in cartesian co-ordinate system can be
obtained after insertion of the total outer field
components into Eq. (1) (i.e., the sum of the incident
and scattered field obtained by the GLMT).
Theoretical methods for calculation of the opti-cal forces acting on a sphere placed into a single
focused Gaussiam beam corrected to the fifth order
and to the higher orders have been presented by
Barton et al. [29] and by Ren et al. [31], respectivelly.
For our study we have chosen the Barton�s ap-
proach applied on the CGB [25,29,32] with the
Gaussian beam waist wider than 0:75k. Under this
condition, the GLMT provides reasonable basis fora theoretical comparison of the SBT and SWT. We
assume that the standing wave is created as the in-
terference of the incident and retroreflected
Gaussian beams. If a sphere is inserted into such a
field a complex scattering event occurs (see Eq.
(A.1) in Appendix A). To simplify the calculation
we omit multiply scattered fields between the re-
flective surface and the sphere (see Eq. (A.2) inAppendix A). In this case the incident beam is equal
to the standing wave created by the interference of
two independent counter-propagating focused laser
beams. Therefore, we can easily adapt the above
mentioned GLMT formalism and instead of a single
CGB we sum field components of two counter-
propagating CGB to get the initial field components
of the standing wave. To speed up the calculation,we assume that the spherical object is located on the
beam axis and we can thus employ the radial sym-
metry of the problem [32]. We also assume that in
the case of the SWT, its beam waist is placed on the
reflective surface and, therefore, the beam is re-
flected as a plane wave with equal reflectivity all
over the beam width. Although the adopted sim-
plifications (CGB, absence of spherical aberrationsand diffraction) could seem drastic, this model at
least provides correct qualitative description of the
behaviour of dielectric spheres in the SBT and SWT
[33]. We wrote the modified code for the force
evaluation ourselves but we do not present a de-
tailed mathematical description, because the meth-
od is well described in the literature [18,25,29,32].
3. Numerical study of optical trap properties
We are primarily interested in the study of the
optical trap properties for dielectric particles
smaller than the trapping wavelength which are
placed into the single focused Gaussian beam and
into the Gaussian standing wave (GSW). Figs. 1and 2 illustrate the differences between intensities,
axial forces, and trapping potentials of both types
of traps. It is assumed that the direction of the z-
axis follows the propagation of the reflected beam
and it is antiparallel to the direction of the incident
beam. Therefore, positive axial force acts against
the incident beam propagation. These plots also
define the maximal axial trapping force Fmax, axialtrap stiffness j, and potential depth of the trap
DW . In the following sections, we will study how
the parameters of both traps depend on the par-
ticle size, refractive index, and beam waist size.
3.1. Dependence of the axial force on particle size
and position in the GSW
An example of the GLMT calculation giving
the dependence of the GSW axial force on the
P. Zem�aanek et al. / Optics Communications 220 (2003) 401–412 403
sphere radius and its position in the GSW is pre-
sented in Fig. 3 in the form of a contour plot.
Force changes with the axial position and radius of
the sphere are clearly seen. Zero value of the forceand negative slope of the force with respect to the
z-co-ordinate determine a stable equilibrium po-
sition of the sphere, which is called the trap posi-
tion. Several trap positions can be found along the
optical axis for certain particle sizes. On the other
hand, there are particular radii of the sphere
(‘‘non-trapping radii’’) for which the axial force is
close to zero regardless of the position of thesphere in the GSW and, obviously, these particles
cannot be optically confined. This is caused by a
competition of the gradient forces coming from
the neighbour GSW antinodes, which pull the
sphere in opposite directions and thus cancel each
other [34]. Consequently, the weaker gradient
force caused by the focused beam envelope domi-
Fig. 2. Illustration of the SBT intensity distribution, axial in-
tensity profile, axial force profile, and potential energy profile.
The parameters used for the comparison are defined: F SBTmax –
maximal trapping force; jSBT – trap stiffness, DW SBT – potential
depth of the trap. Parameters were: beam waist w0 ¼ 0:75k;
relative refractive index m ¼ 1:19; incoming beam power P ¼ 1
W; wavelength of the beam in vacuum kvac ¼ 1064 nm.
Fig. 3. Dependence of the axial trapping force on the particle
radius a and the position z of the sphere in the GSW. Param-
eters were: w0 ¼ 1k, m ¼ 1:19, R ¼ 1, w ¼ 3p=2, P ¼ 1 W,
kvac ¼ 1064 nm, beam waist is placed on the surface.
Fig. 1. Illustration of the SWT intensity distribution, axial in-
tensity profile, axial force profile, and potential energy profile.
The parameters used for the comparison are defined: F SWTmax –
maximal trapping force; jSWT – trap stiffness, DW SWT – energy
depth of the trap. Parameters were: beam waist w0 ¼ 0:75k;
relative refractive index m ¼ 1:19; surface reflectivity R ¼ 1;
phase shift of the reflected wave w ¼ 3p=2; incoming beam
power P ¼ 1 W; wavelength of the beam in vacuum kvac ¼ 1064
nm; beam waist is placed on the surface.
404 P. Zem�aanek et al. / Optics Communications 220 (2003) 401–412
nates and accelerates the sphere towards the beam
waist placed on the reflective surface–mirror. It is
also seen that very small particles are confined
near the intensity maximums, but particles bigger
than about 0:32k are confined near the intensity
minimums. This means that the optical traps areshifted from the neighbourhood of the GSW ant-
inodes to the nodes or vice versa [34].
3.2. Comparative study of the maximal trapping
force
Let us define the maximal trapping force in the
GSW as the maximal optical force directed along
the z-axis (against the incident beam direction) for
the separation between the sphere surface and the
mirror smaller than k:
F SWTmax ¼ max F SWT
GLMTða�
6 z6 aþ kÞ�: ð3Þ
The sphere surface just touches the mirror at the
initial position. Let us define the maximal trappingforce in the SBT as:
F SBTmax ¼ max F SBT
GLMTð�
�16 z61Þ�: ð4Þ
The contour plots of the maximal trapping force
values (see Fig. 4) have been chosen as the most
comprehensive way to show the principal force
dependencies on the particle size a, relative re-
fractive index m, and beam waist size w0. The
contour labels represent levels of constant forces in
pN, the shaded regions correspond to the negative
maximal trapping force (non-trapping regions).For these parametric configurations the particle is
accelerated in the incident beam direction without
a chance of its confinement. In the GSW the width
of the non-trapping regions increases with particle
size. It is also seen that the shaded regions are
getting narrower with increasing beam waist since
the gradient of the intensity of the Gaussian en-
velope becomes smaller. The dot-and-dashed (da-shed) curves link the values of parameters m and a,
for which the maximal trapping force along the z-
axis reaches its maximum (minimum) with respect
to the sphere radius for considered value of w0.
It is interesting to compare maximal axial
trapping forces in the SWT (see Fig. 4) and in the
SBT (see Fig. 5) for the same particle sizes, relative
refractive indices, and beam waists. It is seen im-
mediately that the wider is the beam waist, the
narrower is the region of relative refractive indices
where the particle can be confined using SBT and
this region reduces to the unity value of the relative
refractive index for bigger particles. This means
Fig. 4. Contour plot of the maximal SWT axial trapping force
(in pN) with respect to the z-axis as a function of the relative
refractive index m, particle radius a, and beam waist radius w0
calculated using the GLMT. Dot-and-dashed and dashed lines
represent the curves of maximum and minimum values of
maximal trapping forces with respect to the particle size and
relative refractive index, respectively. Dotted lines denote the
borders of the trapping region, the shaded areas represent the
non-trapping regions. The following parameters were used:
P ¼ 1 W, R ¼ 1, w ¼ 3p=2, kvac ¼ 1064 nm, beam waist is
placed on the surface.
P. Zem�aanek et al. / Optics Communications 220 (2003) 401–412 405
that using wider beams, only those objects can be
trapped by SBT which have the refractive index
close to the value of surrounding immersion me-
dium. If the values of maximal trapping forces for
bigger particles are compared, one finds that the
SWT provides at least one order of magnitude
stronger trapping forces. For smaller particles
and wider beam waists (outside the non-trapping
regions) this disproportion becomes even more
pronounced.
3.3. Comparative study of minimal trapping power
Due to the competition of the scattering and
gradient forces, the total axial force is the mostlimiting factor of the optical confinement. The
stability of the trap along the optical axis can be
inferred from the axial profile of the potential en-
ergy. As on-axis sphere position is assumed here,
the change of the potential energy DWGLMT due to
the motion of the sphere in the field from the
starting position z1 to the end position z2 is given by:
DWGLMT ¼ �Z z2
z1
FGLMTðzÞdz: ð5Þ
Here, FGLMTðzÞ is the axial optical force calculated
by the GLMT and acting on the sphere placed at z.The integration of the GLMT force along the z-axis gives potential energy profile whose minimum
represents the equilibrium position of the sphere in
the trap.The knowledge of the trap depth can be em-
ployed to obtain minimal trapping power that is
necessary for axial confinement of a particle. It is
generally accepted that if the trap is deeper than
10kBT , the particle can be considered confined. We
defined the trap depth as the difference between the
potential energy of the lowest trap edge and the
trap bottom (see Figs. 1 and 2). The power thatprovides the trap depth equal to 10kBT will be
called the minimal trapping power. The minimal
trapping power was calculated as a function of
sphere radius a, relative refractive index m, and
beam waist size w0 (see Figs. 6 and 7). The contour
line labels give the minimal trapping power in mW.
The shaded areas correspond again to the ‘‘non-
trapping’’ configuration of parameters. The plotsreveal that the minimal trapping power in the
SWT can be smaller than 1 mW even for
w0 ¼ 1:25k if optimal sphere size and slightly
higher relative refractive index are chosen. In
contrast, much higher trapping power must be
used to confine smaller spheres or spheres close to
Fig. 5. Contour plot of maximal SBT axial trapping force (in
pN) with respect to the z-axis as a function of the relative re-
fractive index m, particle radius a, and beam waist radius w0
calculated using the GLMT. Dot-and-dashed lines represent the
curves of maximal value of the maximal trapping forces with
respect to the particle size and relative refractive index. The
dotted lines denote the borders of the trapping regions and the
shaded areas represent the non-trapping regions. The inserted
subplots in the last two plots magnify the regions of interest
with the same horizontal scale. The following parameters were
used: P ¼ 1 W, kvac ¼ 1064 nm.
406 P. Zem�aanek et al. / Optics Communications 220 (2003) 401–412
the non-trapping radii. Quantitative comparison
of the SBT and SWT shows that for the samerelative refractive indices, the SWT needs lower
trapping powers for spheres smaller than about
one half of the trapping wavelength. Bigger
spheres are influenced by neighbouring antinodes
which tend to pull the sphere to opposite sides and
decrease the total axial trapping force. Thus, the
trap gets shallower. It is also seen that the differ-
ences between the minimal trapping power needed
in the SWT and in the SBT are not as big as the
differences in maximal trapping forces. This is
caused by the fact that – unlike the trapping force
Fig. 7. Contour lines of the minimal trapping power (in mW) in
the SBT as a function of the relative refractive index m, particle
radius a, and beam waist size w0 for kvac ¼ 1064 nm. Dot-and-
dashed lines represent the curves of maximal value of the
maximal trapping forces with respect to the particle size and
relative refractive index. Dotted lines denote the borders of the
trapping regions and the shaded areas represent the non-trap-
ping regions. The inserted subplots in the last two plots magnify
the regions of interest with the same horizontal scale.
Fig. 6. Contour lines of the minimal trapping power (in mW) in
the SWT as a function of the relative refractive index m, particle
radius a, and beam waist size w0 for R ¼ 1, w ¼ 3p=2,
kvac ¼ 1064 nm, and beam waist placed on the surface. Dot-
and-dashed lines represent the curves of maximal value of the
maximal trapping forces with respect to the particle size and
relative refractive index. Dotted lines denote the borders of the
trapping regions, the shaded areas represent the non-trapping
regions.
P. Zem�aanek et al. / Optics Communications 220 (2003) 401–412 407
– the trap depth does not depend on the intensity
gradient but rather on the difference between the
maximal and minimal intensity, which is compa-rable in both cases.
The knowledge of the minimal trapping power
for given sphere size, refractive index, and beam
waist enables estimation of the maximal trapping
force for this power. This is the weakest maximal
trapping force that can confine this particular
sphere. In the case of the SWT this force lies be-
tween 0.31 and 0.44 pN for all studied sphere radii,refractive indices, and beam waists except the
narrow regions close to the non-trapping radii
boundaries. Even though this value is very small, it
is still significantly higher (more than 10 times for
particles smaller than the trapping wavelength)
than the sum of gravity and buoyancy forces if
glass or polystyrene spheres are considered (see
Fig. 8). However, this conclusion is not generallyvalid if the SBT is analysed. In this case, the values
of the smallest maximal axial force lie in the range
from 0.01 to 0.07 pN and for glass spheres with
radius bigger than about half of the trapping
wavelength this force becomes comparable to the
sum of gravity and buoyancy forces.
3.4. Position of the trap in the GSW
We showed in the previous section that the
maximal SWT trapping force is at least 10 times
higher than the total force done by gravity and
buoyancy if the minimal trapping power is used.
We can therefore conclude that these forces, with
respect to the high value of the axial trap stiffness
in the SWT (see Section 3.5), do not influence
significantly the axial trap position. Since the
minimal trapping power is lower than the power
that is usually used for trapping, position of the
trap can be considered as a point where FGLMTðzÞ is
equal to zero and has a negative slope (minimum
in the potential energy profile). Using this as-sumption, we can analyse the trap position gen-
erally without a precise specification of the trapped
object material. On the contrary, this conclusion is
not generally valid if the SBT is used with low
trapping power; in this case the equilibrium posi-
tion is shifted in the direction of the gravity.
The results of the SWT equilibrium position
analysis are shown in Fig. 9 for very small and bigrefractive indices as a function of the sphere size.
For the analysis we assumed that the distance
between the sphere surface and the reflective sur-
face is smaller than one trapping wavelength.
Therefore, only two trap positions nearest to the
reflective surface could be shown here because the
trap separation is close to k=2 (the minor differ-
Fig. 8. The sum of gravity and buoyancy forces (Fgrav, Fbuoy) (in
pN) as a function of the particle radius a (in kvac ¼ 1064 nm) for
polystyrene (full line) and glass (dashed line) microspheres in
water.
Fig. 9. Quantitative comparison of the trap positions above the
mirror in the GSW for two beam waist sizes, two values of
refractive index, and for the separation between the sphere
surface and the mirror smaller than k. The beam waist is placed
on the mirror, R ¼ 1, w ¼ 3p=2, kvac ¼ 1064 nm. The dashed
lines denote the position of the trapped sphere centre where it
touches the mirror.
408 P. Zem�aanek et al. / Optics Communications 220 (2003) 401–412
ences are caused by the Guoy phase shift). The size
of the trapped sphere limits the distance between
the trap and the reflective surface (dashed line in
Fig. 9, because the trap cannot be closer to the
reflective surface than the sphere radius. This is the
reason of abrupt breaks in the trap position curvesnear to the reflective surface, which are caused by
contact of the sphere with the reflective surface.
The absolute position of the trap array is strongly
affected by phase shift of the reflected wave. If the
sphere size crosses the non-trapping region, the
trap position is changed approximately by k=4. It
is immediately seen from Fig. 9 that the value of
the relative refractive index does not influencenoticeably the position of the existing optical trap
but it only influences the width of the region of the
sphere sizes where the trap exists. Trap positions
for spheres with parameters close to the non-
trapping region (see Fig. 4) could be shifted in the
direction of gravity for smaller powers because the
trapping force could be smaller than the sum of
gravity and buoyancy forces.
3.5. Trap stiffness
The knowledge of the trap position enables es-timation of the trap stiffness. It is assumed that for
small deviations of the trapped particle from its
equilibrium position the restoring force is linearly
dependent on this deviation and the proportion-
ality constant defines the trap stiffness (see Figs. 1
and 2). The stiffness was obtained as the derivative
of FGLMTðzÞ with respect to the z co-ordinate at the
trap position. The results for the SWT are sum-marised in Fig. 10. The contour labels represent
the stiffness values in nN/lm. We consider only
those trap positions that form continuous curve
which is not broken by the contact of the sphere
with the surface and by the end of the sphere po-
sition range where the calculation was performed
(see Fig. 9). Non-trapping regions are shaded,
except the upper left corner where the trap posi-tions were not analysed.
Using the same procedure, the trap stiffness in
the SBT was calculated. These results, presented in
Fig. 11, are valid only for such sphere radii and
refractive indices where the maximal trapping
force (see Fig. 5) is significantly higher that the
total force produced by gravity and buoyancy (see
Fig. 8). Having this in mind, the comparison of
Figs. 10 and 11 reveals that the stiffness produced
by the SWT is at least by two orders of magnitude
bigger (the units used for the contour labels of the
SBT stiffness in Fig. 11 are in pN/lm but those for
Fig. 10. Contour lines of the SWT stiffness (in nN/lm) for
beam waist placed on the surface, P ¼ 1 W, R ¼ 1, w ¼ 3p=2,
kvac ¼ 1064 nm. Dot-and-dashed lines represent the curves of
maximal value of the maximal trapping forces with respect to
the particle size and relative refractive index. Dotted lines de-
note the borders of the trapping regions and the shaded areas
represent the non-trapping regions.
P. Zem�aanek et al. / Optics Communications 220 (2003) 401–412 409
the SWT in nN/lm). Therefore, the particles in the
SWT are more precisely confined axially.
If similar calculation is repeated for the minimal
trapping power, instead of 1 W, trap stiffness jSWT
is in the range 4.3 pN/lm < jSWT < 10:5 pN=lm
for w0 ¼ 0:75k and 5 pN=lm< jSWT < 5:2 pN=lm
for w0 ¼ 2:5k, where the lower values are valid for
smaller spheres and the bigger ones for bigger
spheres. Only a weak dependence of the stiffness
on the relative refractive index was found. The
value of the stiffness is almost constant over the
trapping regions for small sphere sizes. In the case
of the SBT similar procedure would give inaccu-rate results because the trap position is shifted in
the direction of the beam propagation due to the
gravity force. Therefore, the equilibrium position
is influenced by one more parameter – density of
the microsphere and the results cannot be gener-
alised in the way used above.
4. Conclusion
In this paper the trapping properties of a re-
flection-generated GSW and a single beam werestudied theoretically using the GLMT. The
GLMT was employed to calculate the axial optical
forces acting on dielectrics spheres smaller than the
trapping wavelength and made of different dielec-
trics with the relative refractive index in the im-
mersion (water) smaller than 1.21. In the SWT, the
beam waist was placed on 100% reflective surface
and the sphere-surface distance was smaller thanone trapping wavelength. It was proved that par-
ticles can be trapped either in the GSW antinodes
(intensity maximums) or nodes (intensity mini-
mums) depending on their size and refractive in-
dex. Spheres smaller than about 0:3k are trapped
in the antinodes, spheres bigger than about 0:3kbut smaller than about 0:55k in the nodes, etc. This
can be explained by a general tendency of thesphere with relative refractive index higher than
unity to compass the high intensity region as much
as possible. When the size of such object covers
two standing wave intensity maximums, this can
only be fulfilled if the sphere centre is placed be-
tween these intensity maximums, i.e., in the GSW
node. For some sphere radii, the optical trapping
in the GSW is disabled. The precise value of theboundary non-trapping sphere radius decreases
with increasing refractive index of the trapped
sphere.
The maximal trapping force, minimal trapping
power, trap position, and trap stiffness of the
Gaussian SWT have been studied as a function of
Fig. 11. Contour lines of the SBT stiffness (in pN/lm) for P ¼ 1
W, kvac ¼ 1064 nm. Dot-and-dashed lines represent the curves
of maximal value of the maximal trapping forces with respect to
the particle size and relative refractive index. Dotted lines de-
note the borders of the trapping regions and the shaded areas
represent the non-trapping regions. The inserted subplots in the
last two plots magnify the regions of interest with the same
horizontal scale.
410 P. Zem�aanek et al. / Optics Communications 220 (2003) 401–412
the beam waist size, object size and refractive in-
dex and have been compared to the SBT param-
eters. It was found that the maximal SWT
trapping force is at least by one order of magni-
tude stronger for spheres comparable with the
trapping wavelength. For smaller spheres (outsidethe non-trapping regions) and wider beam waists,
this disproportion still increases. On the contrary
to the SBT, the SWT enables confinement of high
refractive index particles even for wider beam
waist. Comparison of minimal trapping power
revealed that a stable SWT could be obtained with
much lower trapping power than a SBT for
spheres smaller than about one half of the trappingwavelength. For bigger particles influenced by
neighbour antinodes, which tend to pull the par-
ticle to the opposite sides, the GSW effect becomes
less visible. The differences between minimal
trapping power needed in the SWT and SBT are
not as big as differences in maximal trapping forces
because of much larger axial extent of the SBT.
However, the maximal SWT optical force inducedby the minimal trapping power is still 10 times
bigger than the total force produced by the gravity
and buoyancy force. Therefore, the SWT axial
position can be considered as unaffected by these
forces, which is not the case if the SBT with min-
imal trapping power is used. Axial SWT stiffness is
at least by two orders of magnitude bigger than in
the SBT and provides much more precise axialparticle confinement.
To summarise, we proved theoretically that the
SWT enables dielectric sphere confinement in a
wider range of sphere refractive indices and beam
waists compared to the SBT. Apart from small
regions of the sphere sizes and refractive indices,
where the GSW effect is cancelled, the force, stiff-
ness, and depth of the SWT for comparable pa-rameters of the sphere and the laser beam are
generally bigger than corresponding properties of
the SBT.
Acknowledgements
The authors thank the Grant Agency of the
Czech Republic for the financial support (Grant
No: 101/00/0974).
Appendix A. Multiple scattering between the sphere
and reflective surface
Let us assume, that the GSW is created by in-
terference of an incident wave and a wave reflectedat a dielectric surface [35]. An object placed into
such a GSW scatters the field incident upon it. Due
to the presence of the dielectric surface, this scat-
tered field is retro-reflected towards the object
where it is scattered again and the process is then
repeated. These multiple reflections and scattering
make the description of the total field much more
complex. Let us divide the process into two paths.Along the first one incident field Ei is scattered by
the object giving rise to scattered field SEi, where S
is a general scattering operator. Field SEi is re-
flected by the surface (RSEi, R is a general reflec-
tion operator) and then it is scattered again by the
object (SRSEi). This process goes on in a similar
way. Along the second path the incident field is not
scattered directly by the object but it is first re-flected by the surface (REi) and then scattered
(SREi) and reflected (RSREi), etc. The resulting
field is done by the superposition of all these par-
tial fields:
E ¼ Ei þ REi þ SEi þ SREi þ RSEi þ : ðA:1ÞThe partial fields due to the multiple scattering are
divergent and can be omitted if the particle is not
in the close vicinity of the surface. The total field
outside the particle then takes the form:
E ’ Ei þ REi þ SðEi þ REiÞ: ðA:2ÞThis result can be understood as the superposition
of two fields. The first one is the standing wave
created by incident and reflected waves and the
second one arises due to the scattering of thisstanding wave by the particle.
References
[1] A. Ashkin, IEEE J. Sel. Top. Quantum Electron. 6 (2000)
841.
[2] K.-O. Greulich, Micromanipulation by Light in Biology
and Medicine, Birkhauser, Basel/Boston/Berlin, 1999.
[3] M.P. Sheetz, L. Wilson, P. Matsudaira (Eds.), Laser
Tweezers in Cell Biology, vol. 55 of Methods in Cell
Biology, Academic Press, San Diego, 1998.
[4] A.D. Mehta et al., Science 283 (1999) 1689.
P. Zem�aanek et al. / Optics Communications 220 (2003) 401–412 411
[5] A. Ashkin, J.M. Dziedzic, J.E. Bj€oorkholm, S. Chu, Opt.
Lett. 11 (1986) 288.
[6] S. Sato, M. Ishigure, Electron. Lett. 27 (1991) 1831.
[7] K.T. Gahagan, J.G.A. Swartzlander, J. Opt. Soc. Am. B 16
(1999) 533.
[8] A.E. Chiou et al., Opt. Commun. 133 (1997) 7.
[9] W. Wang, A.E. Chiou, G.J. Sonek, M.W. Berns, J. Opt.
Soc. Am. B 14 (1997) 697.
[10] A. Constable, J. Kim, Opt. Lett. 18 (1993) 1867.
[11] K. Visscher, S.P. Gross, S.M. Block, IEEEJ. Sel. Top.
Quantum Electron. 2 (1996) 1066.
[12] E. F€aallman, O. Axner, Appl. Opt. 36 (1997) 2107.
[13] E.R. Dufresne, D.G. Grier, Rev. Sci. Instrum. 69 (1998)
1974.
[14] K. Sasaki et al., Opt. Lett. 16 (1991) 1463.
[15] C. Mio, T. Gong, A. Terray, D.W.M. Marr, Rev. Sci.
Instrum. 71 (2000) 2196.
[16] P. Zem�aanek, A. Jon�aa�ss, L. �SSr�aamek, M. Li�sska, Opt. Lett. 24
(1999) 1448.
[17] L. Davis, Phys. Rev. A 19 (1979) 1177.
[18] J.P. Barton, D.R. Alexander, J. Appl. Phys. 66 (1989) 2800.
[19] W.L. Erikson, S. Singh, Phys. Rev. E 49 (1994) 5778.
[20] G. Gouesbet, J. Opt. 27 (1996) 35.
[21] M. Couture, P.-A. Belanger, Phys. Rev. A 24 (1981) 355.
[22] G.A. Deschamps, Electron. Lett. 7 (1971) 684.
[23] B. Landesman, H.H. Barrett, J. Opt. Soc. Am. A 5 (1988)
1610.
[24] A. Rohrbach, E.H.K. Stelzer, J. Opt. Soc. Am. A 18 (2001)
839.
[25] J.P. Barton, D.R. Alexander, S.A. Schaub, J. Appl. Phys.
64 (1988) 1632.
[26] G. Gouesbet, B. Maheu, G. Gr�eehan, J. Opt. Soc. Am. A 5
(1988) 1427.
[27] B. Maheu, G. Gouesbet, G. Gr�eehan, J. Opt. 19 (1988) 59.
[28] G. Gouesbet, G. Grehan, Atom Sprays 10 (2000) 277.
[29] J.P. Barton, D.R. Alexander, S.A. Schaub, J. Appl. Phys.
66 (1989) 4594.
[30] F.N.H. Robinson, Phys. Rep. 16 (1975) 313.
[31] K.F. Ren, G. Grehan, G. Gouesbet, Appl. Opt. 35 (1996)
2702.
[32] S.A. Schaub, J.P. Barton, D.R. Alexander, Appl. Phys.
Lett. 55 (1989) 2709.
[33] A. Jon�aa�ss, P. Zem�aanek, E.L. Florin, Opt. Lett. 26 (2001)
1466.
[34] P. Zem�aanek, A. Jon�aa�ss, M. Li�sska, J. Opt. Soc. Am. A 19
(2002) 1025.
[35] P. Zem�aanek, A. Jon�aa�ss, L. �SSr�aamek, M. Li�sska, Opt.
Commun. 151 (1998) 273.
412 P. Zem�aanek et al. / Optics Communications 220 (2003) 401–412