TMP 91 F4
TMP 91 F4
TMP 91 F4
The spatial backscattering impulse-response at length of px; y. However, as we will demonstrate, the depen-
scales shorter than a transport mean free path ls is highly dencies are more complex:
sensitive to the shape of the scattering phase function
due to the fact that rays exiting within this regime have I EBS θx ; θy F fpx; y · pcx; y · sx; y · cx; y
undergone relatively few scattering events. Typically this
measurement is accomplished using a lens to focus the · mtfx; yg; 1
illumination beam to a near impulse on the sample sur- where F denotes the Fourier transform operation, pc
face. A CCD camera then detects the spatial distribution specifies the degree of phase correlation between the for-
of backscattered light at the sample surface. With this de- ward and reverse path, s represents the effective distri-
sign, the full azimuthal backscattering distribution can bution of rays that remain within the illumination spot, c
easily be measured for each polarization channel [1]. is the spatial coherence function, and mtf is the modula-
Still, a true impulse can never be achieved and informa- tion transfer function of the optical instrument. Within
tion at short length scales is lost to an optical mask (∼mm the F operation, p and pc represent instrinsic sample
diameter) needed to reject specular reflections. Thus in properties while s, c, and mtf represent extrinsic instru-
media such as biological tissue with long ls on the order ment properties. Each of these functions will be de-
of 1 mm, information about the shape of the phase func- scribed in further detail in the following paragraphs.
tion is extremely difficult to measure experimentally. As The experimentally measurable peff is found by com-
an alternative to overcome these difficulties, here we re- puting the inverse Fourier transform of the EBS peak:
port a simple bench-top instrument using polarized en-
hanced backscattering (EBS) to first measure the
backscattering distribution in the angular domain and peff F −1 fI EBS θx ; θy g. (2)
subsequently convert to the spatial domain through in-
verse Fourier transformation. For sample characterization, the functions p and pc are of
EBS is an angular intensity peak centered in the exact prime interest. In principle these intrinsic parameters can
backscattering direction that is formed by the summed be found by dividing peff by functions s, c, and mtf. How-
diffraction pattern from all sets of time-reversed path- ever, in practice this will amplify noise and a reliable re-
pairs within a scattering medium (a direct consequence sult can be difficult to obtain. Alternatively, a more stable
of the reciprocity theorem) [2]. Conceptually, it is useful approach that we follow in this Letter is to convert theory
to think of each time-reversed path pair as a single into the experimentally observable peff before compari-
Young’s double pinhole experiment. For a single time- son with experiment.
reversed path pair with rays exiting at a particular spatial Our experimental instrument is shown in Fig. 1. An il-
separation, the diffraction pattern is the Fourier trans- lumination source (xenon lamp with finite spatial coher-
form of two delta functions (cosine pattern). For a ence length Lsc or broadband laser with Lsc ∞) is
semi-infinite medium under plane wave illumination the collimated by lens L1 and polarized by linear polarizer
distribution of spatial separations for all time-reversed P before being directed onto the scattering sample via
path-pairs is given by the spatial backscattering impulse- a nonpolarizing 50∕50 beam splitter B. Backscattered
response for multiply scattered light in the exact back- light is then collected by B and passes through a linear
ward direction px; y. As a result, in an idealized case analyzer A before being focused onto a CCD camera with
the EBS peak I EBS θx ; θy is simply the Fourier transform a lens L2 . A liquid crystal tunable filter attached to the
0146-9592/11/244737-03$15.00/0 © 2011 Optical Society of America
4738 OPTICS LETTERS / Vol. 36, No. 24 / December 15, 2011
Fig. 4. Sensitivities to the spatial coherence of the illumina- Fig. 5. Sensitivity of EBS to the phase function. (a) Phase
tion. (a) Function c for different Lsc . (b) Comparison between function for two microsphere samples with the same g but dif-
experiment (symbols) and simulation (lines) for different Lsc ferent ka. (b) Corresponding peff for polarization shows a
using polarization. Sample: 0.65 μm diameter sphere with large difference in shape at short length-scales. ka ∼ 4: 0.65 μm
ls 205 μm at 633 nm. Spot size 6000 μm and Lsc ∞. diameter sphere at 633 nm. ka ∼ 16: 2.1 μm diameter sphere at
Simulation scaled by 0.65. 558 nm. Spot size 2000 μm and Lsc ∞. Simulation scaled by
0.65.