week ending 19 JULY 2013

PHYSICAL REVIEW LETTERS

Interferometric Spectroscopy of Scattered Light Can Quantify the Statistics of Subdiffractional Refractive-Index Fluctuations L. Cherkezyan, I. Capoglu, H. Subramanian, J. D. Rogers, D. Damania, A. Taflove, and V. Backman* Northwestern University, Evanston, Illinois 60208, USA (Received 12 February 2013; published 19 July 2013) Despite major importance in physics, biology, and other sciences, the optical sensing of nanoscale structures in the far zone remains an open problem due to the fundamental diffraction limit of resolution. e 2 ) of a far-field, diffraction-limited microscope We establish that the expected value of spectral variance ( image can quantify the refractive-index fluctuations of a label-free, weakly scattering sample at e for an arbitrary refractive-index subdiffraction length scales. We report the general expression of e distribution. For an exponential refractive-index spatial correlation, we obtain a closed-form solution of that is in excellent agreement with three-dimensional finite-difference time-domain solutions of Maxwell’s equations. Sensing complex inhomogeneous media at the nanoscale can benefit fields from material science to medical diagnostics. DOI: 10.1103/PhysRevLett.111.033903

PACS numbers: 42.25.Dd, 42.25.Fx, 42.25.Hz, 87.64.t

Do Maxwell’s equations permit determining the nature of three-dimensional (3-D) subdiffractional refractiveindex (RI) fluctuations of a linear, label-free dielectric medium in the far zone? Recently, by capturing high spatial-frequency evanescent waves, metamaterial-based lenses and grating-assisted tomography have achieved a resolving power no longer limited by the diffraction of light [1,2]. However, this super-resolution is confined to the transverse plane, which limits its ability to characterize 3-D inhomogeneous media. Whereas various nonlinear techniques have been proposed to image subdiffractional structures in 3-D [3–5], these techniques require exogenous labeling or intrinsic fluorescence and, thus, only image the spatial distribution of particular molecular species. Currently, elastic, label-free spectroscopic microscopy techniques are emerging that characterize the endogenous properties of a medium by utilizing the spectral content of a diffraction-limited microscopic image. Examples include multiple high-precision quantitative phase microscopy techniques [6–8], which measure the longitudinal integral of RI and, hence, are insensitive to longitudinal RI fluctuations. Alternatively, partial-wave spectroscopic microscopy [9], confocal light scattering and absorption spectroscopy [10], and spectral encoding of spatial frequency [11] analyze the light-scattering response of inhomogeneous materials to obtain information of their subdiffractional structure in both lateral and longitudinal dimensions. However, the reported theory behind these techniques involves strong assumptions such as one-dimensional light transport, approximation of the medium as solid spheres, or having a single length scale. Here, we establish that the spectral signature of scattered light in a far-zone microscope image contains sufficient information to quantify the 3-D RI fluctuations of weakly 0031-9007=13=111(3)=033903(5)

scattering media at deeply subdiffractional scales. We report three-dimensional light transport theory for linear, label-free weakly scattering media with an arbitrary form of RI distribution: continuous or discrete, random or deterministic, statistically isotropic or not. We consider the e 2 ) of a far-field, expected value of spectral variance ( diffraction-limited image registered by a microscope with a small numerical aperture (NA) of illumination and spece quantifies trally resolved image acquisition. We show that RI fluctuations at nanometer length scales limited only by the signal-to-noise ratio of the system. Under the single scattering approximation, we obtain an explicit expression e to the statistics of RI fluctuations inside the relating sample. Moreover, for the special case of an exponential form of the RI spatial correlation, we present a closed-form e and validate it via numerical simulations of solution for an experiment based on rigorous 3-D finite-difference timedomain (FDTD) solutions of Maxwell’s equations [12]. Consider a spatially varying RI object sandwiched between two semi-infinite homogeneous media (Fig. 1). The RIs of the three media are, from top to bottom: n0 , n1 ½1 þ n ðrÞ (as a function of location r), and n2 . To air sample glass

n0 n1(1+n∆ n2

0 -L

U(r) U(s)

I= U(r)+U(s)

2

FIG. 1 (color online). Sample: RI of the middle layer is random, and RIs of the top and bottom layers are constant; RI as a function of depth is shown in gray. The coherent sum of UðrÞ and UðsÞ is detected. Reflection from the bottom of the substrate (glass slide) is negligible, as its thickness (1 mm) is much larger than the microscope’s depth of field (for most setups, 0:5–15 m).

033903-1

Ó 2013 American Physical Society

PRL 111, 033903 (2013)

begin with, we assume n1 ¼ n2 , approximating the case of fixed biological media on a glass slide [13,14]. A unit amplitude plane wave with a wave vector ki is incident normally onto a weakly scattering sample. Under the Born approximation, the field inside the sample is uniform and has an amplitude T01 ¼ 2n0 =ðn0 þ n1 Þ (transmission Fresnel coefficient). In the far zone, the scattering amplitude of the scalar field UðsÞ , scattered from the RI fluctuations n ðrÞ in theR direction specified by the wave 0 vector ko , is fs ðks Þ¼T01 ðk2 =2Þn ðr0 Þeiks r d3 r0 , where ks ¼ ko ki is the scattering wave vector (inside the sample) [15]. The scalar-wave approximation is used here as it sufficiently describes the intensity image formed by a microscope with a moderate NA [15]. Its further justification by full-vector 3-D FDTD results is discussed below. When the sample is imaged by an epi-illumination bright-field microscope, the back-propagating field reflected from the sample’s top surface, UðrÞ , returns to the image plane. Meanwhile, only the part of UðsÞ that propagates at solid angles within the NA of the objective is collected. For a microscope with magnification M, moderate NA (kz k), ignoring the angular dependence of the Fresnel coefficient T10 ¼ 2n1 =ðn0 þ n1 Þ, UðsÞ focused at a point (x0 , y0 ) in the image plane is [16] ðsÞ 0 0 Uim ðx ; y ; kÞ ¼

kT10 ZZ k ky 0 0 TkNA fs eiðkx x þky y Þ d x d ; i2jMj k k (1)

where TkNA is the microscope’s pupil function—a cone in the spatial-frequency space with a radius kNA [Fig. 2(a)]. Thus, the objective performs low-pass transverse-plane spatial frequency filtering, with the cutoff corresponding to the spatial coherence length. With substitution of fs into Eq. (1) and the introduction of a windowing function Tks ðsÞ that equals 1 at k ¼ ks and 0 at k ks [Fig. 2(a)], Uim is T T Z1 ðsÞ 0 0 ðx ; y ; kÞ ¼ 10 01 kn ðrÞei2kz dz; Uim ijMj 1 1D 0

week ending 19 JULY 2013

PHYSICAL REVIEW LETTERS

(F ) of TkNA Tks in the transverse plane (xy, ? ), n1D ðrÞ ¼ F ? fTkNA Tks g ? n ðrÞ. Equation (2) presents a new treatment of the Born approximation, which is here extended to include the optical imaging of a scattering object in the far zone. Mathematically, Eq. (2) signifies that to describe a microscope-generated spectrum (a 1-D signal), the 3-D problem of light propagation can be reduced to a 1-D problem where the RI is convolved with the Airy disk in the transverse plane. The microscope image intensity (normalized by the image of the source) is an interferogram Iðx0 ; y0 ; kÞ ¼ 201 2 Im

Z þ1 1

kn1D ðrÞei2kz dz ;

(3)

where 01 ¼ ðn0 n1 Þ=ðn0 þ n1 Þ is the Fresnel reflectance coefficient, ¼ 01 T01 T10 , Im denotes ‘‘the imaginary part of,’’ and n1D is zero at z 2 = ðL; 0Þ. Here, Oðn2 Þ terms are neglected. We quantify the spatial distribution of n via 2 , the spectral variance of the image intensity within the illumination bandwidth k. Since the expectation of the spectrally averaged image R intensity equals 201 , 2 ðx0 ; y0 Þ is 2 0 0 defined as ðx ; y Þ ¼ k ðIðx0 ; y0 ; kÞ 201 Þ2 dk=k. For convenience, we introduce a windowing function Tks that is a unity at k ¼ ks for all ki with magnitudes within the k of the system and is zero elsewhere [jki j between k1 and k2 in Fig. 2(a)]. On denoting kc as the value of the central wave number of illumination bandwidth inside the sample, approximation of k kc , and application of the convolution and the Parseval’s theorems [for mathematical details see the Supplemental Material [17]], 2 ðx0 ; y0 Þ equals 2 ðx0 ; y0 Þ ¼

2 k2c Z 1 jF fTks TkNA g n ðrÞj2 dz: k 1

(4)

(2)

0

where r is (x =M, y =M, z) inside the sample, and n1D is the n ðrÞ convolved ( ) with the unitary Fourier transform

FIG. 2 (color). Spatial-frequency space with kz axis antiparallel to ki . (a) Cross section of Tks , TkNA , and their interception T3D ; (b) PSD of the RI fluctuation (blue) and T3D (gray) when lc L and (c) lc & L.

Physically, Tks accounts for the limited bandwidth of illumination and serves as a bandpass longitudinal spatial-frequency filter of RI distribution with its width related to the temporal coherence length l ¼ 2=k. The interception of the two frequency filters associated with the spatial and temporal coherence, TkNA and Tks , signifies the frequency-space coherence volume centered at kz ¼ 2kc : T3D ¼ Tks TkNA [Fig. 2(a)]. Given an infinite bandwidth, one could reconstruct the full 3-D RI from Iðx0 ; y0 ; kÞ. However, since k and kc are finite, detects the variance of an ‘‘effective RI distribution,’’ i.e., of n ðrÞ F fT3D g [Eq. (4)]. Note that 2 ðx0 ; y0 Þ is random since n ðrÞ is random. Hence, to characterize the sample statistics, we compute e 2 . Using the Wienerits expected value, denoted as e 2 from Eq. (4) as Khinchine relation, we obtain

033903-2

e2¼

week ending 19 JULY 2013

PHYSICAL REVIEW LETTERS

PRL 111, 033903 (2013)

2 k2c L Z n ðkÞd3 k; k T3D

(5)

where n ¼ jF fn ðrÞgj2 is the power spectral density (PSD) of n . Equation (5) establishes the general quadrature-form e 2 for an arbitrary n ðrÞ. Note that while expression for the 3-D structure of complex inhomogeneous materials cannot be described by a single measure of size or RI, the PSD fully quantifies the magnitude, spatial frequency, and orientation of all RI fluctuations present within the e 2 measures the integral of sample. As seen from Eq. (5),

e2 the tail of the PSD within T3D . Hence, as shown later, presents a monotonic measure of the width of the PSD. e 2 has a different preWhen n1 n2 , the expression for factor and a deterministic offset, specified in the Supplemental Material [17]. e 2 for a We further obtain a closed-form expression for special case when n ðrÞ has an exponential form of spatial correlation with a variance n and correlation distance lc . Since lc can only be defined for a random medium with a physical size much larger than the correlation distance, we define lc as the correlation distance of an unbounded medium n1 ðrÞ and the sample as a horizontal slice of ðrÞ with thickness L: n ðrÞ ¼ TL n1 n1 ðrÞ where TL is a windowing function along the z axis with width L. The PSD of such sample is an anisotropic function of lc 2 and L: n ðkÞ ¼ jF fTL g F fn1 gj [Figs. 2(b) and 2(c)]. e 2 is found by independently computing the Alternatively, contributions from (i) scattering from within the sample e 2 ) and (ii) reflectance at z ¼ L ( e 2 ), ( R

L

e2 þ e2 : e2¼ R L

(6)

e is fully described by the RI contrast at the Here, L e 2 ¼ 2 2 ðn Þ=4, where 2 ðn Þ is the bottom surface L ? 1D ? 1D −1

10

10

variance of the effective n1D in the transverse plane [details e , in turn, is shown in the Supplemental Material [17]]. R defined by n1 , which is independent of L when L * l ; e 2 is obtained by integrating the PSD of an exponentially R

e2 correlated n1 ðrÞ according to Eq. (5). Substituting R and e 2 into Eq. (6) and introducing a unitless size parameter L x ¼ kc lc , we obtain the following closed-form solution for e 2 for an exponential form of the spatial RI correlation: 2 2 kc Lx3 NA2 e 2 ¼ 2 n 2 ½1 þ x ð4 þ NA2 Þð1 þ 4x2 Þ qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ þ 2 2n ½1 1= 1 þ ðxNAÞ2 =4:

(7)

Two assumptions were made to derive Eq. (7) from Eq. (5): (1) we approximated the top and bottom surfaces of T3D as planes perpendicular to the kz axis, and (2) we e from 1 , not considering the extreme case calculated n

R

of L l . Both assumptions are not crucial from the theoretical perspective and are there only to obtain a relatively simple closed-form solution of Eq. (5). e preTo confirm these approximations, we evaluate dicted by the general quadrature-form expression [Eq. (5)] using MATLAB computing software (MathWorks Inc.). We e calculated from obtain an excellent agreement between Eq. (5) and the closed-form expression [Eq. (7)] derived from it (Fig. 3). This validates the closed-form solution for e for an exponential RI correlation. We support the present theory by simulating a physical experiment using the rigorous 3-D FDTD solution of Maxwell’s equations [18–20]. Our technique accurately synthesizes microscope images of arbitrary inhomogeneous samples under various imaging parameters, incorporating RI fluctuations as fine as 10 nm. We synthesized bright-field, plane-wave epi-illumination microscope images of samples with a RI distribution resembling that of biological cells: n1 ¼ 1:53 [13,14], n1 n ¼ 0:05 [21].

−1

−1

FDTD:

10

−3

L =.5µ m

10

10

(a)

2

lc (nm)

10

−2

L=1.5µ m 10

(b)

0.6 NA 0.3 NA

−2

2

lc (nm)

Quadrature−form, Eq. 5:

01

~ 2 Σ/Γ

−2

10

01

~ Σ/Γ2

~ 2 Σ/Γ

01

0.6 NA 0.3 NA

10

Closed−form, Eq. 7:

L =2.0µ m 2

10

(c)

lc (nm)

0.6 NA 0.3 NA

e dependence on l predicted by the quadrature-form [Eq. (5)] and the closed-form [Eq. (7)] analytical FIG. 3 (color). Illustration of c e expressions for (circles and solid lines, respectively) and by FDTD (solid lines with error bars representing standard deviation between 20 realizations of each statistical condition), calculated for (a) L ¼ 0:5 m, k ¼ 4:9 m1 , kc ¼ 16:8 m1 , (b) L ¼ 1:5 m, k ¼ 4:9 m1 , kc ¼ 16:8 m1 , and (c) L ¼ 2:0 m. k ¼ 11:9 m1 , kc ¼ 18:1 m1 (wave number values inside the sample). Data are shown normalized by 201 , the image intensity in the absence of RI fluctuations inside the sample.

033903-3

c

e to Second, as opposed to s ðlc Þ, the sensitivity of changes at smaller length scales is not obscured by changes e Þ for at larger lc . We note that the functional form of ðl c lc < 1=kc can be roughly approximated as linear [r2 values e Þ presented in Fig. 3 range of linear regressions for ðl c

e is independent of l for l from 0.86 to 0.91]. Finally, c c e 1=kc , and therefore ðlc Þ exhibits predominant sensitivity to subdiffraction length scales that is only limited by the signal-to-noise ratio (SNR). The larger structures are naturally resolved in the microscope image. In addition, whereas the above mentioned scattering parameters are e is / (confirmed by FDTD with r2 ¼ 0:99, / 2 , n

data not shown), which substantially improves the SNR. Results of an FDTD-simulated experiment are shown in Fig. 4. As expected, the bright-field microscope images of samples with lc ¼ 20 and 50 nm [Figs. 4(a) and 4(b)] are essentially indistinguishable. However, a drastic difference between the two samples is revealed in the respective ðx0 ; y0 Þ images [Figs. 4(c) and 4(d), where color bar limits match the ordinate range in Fig. 3(c)]. Figures 4(e) and 4(f) illustrate that a smaller amplitude of

2

2

I (x’,y’) / Γ01

I (x’,y’) / Γ01 0.75

1µ m

0.75

1µ m

( x’o , y’ o)

( x’o , y’ o)

1.00

1.00

l =20nm

(a)

c

Σ ( x’,y’)/ Γ01

l =50nm

(b)

1.25

2

c

1.25

Σ ( x’,y’)/ Γ012

0.01

0.01

0.05

l =20nm c

l =50nm c

c

c

0.10 l =20nm c

1.2

o

o

c

l =50nm

l =20nm

1.6

l =50nm

(d)

0.10

1.0

o

o 1

0.05

I(x’ , y’ , k)

(c)

n (x , y , z)

The spatial RI correlation was set to be exponential, and the RIs of the top and bottom media were n0 ¼ 1 and n2 ¼ 1:53. e predicted by the present theory Referring to Fig. 3, the [either by the quadrature-form Eq. (5) or the closed-form Eq. (7)] exhibits an excellent agreement with the FDTDsimulated experimental results over a wide range of lc , L, spectral bandwidth, and NA. The agreement is such that e values by both Eqs. (5) and the theoretically predicted (7) lie within the standard deviation bars of the FDTD results at all points tested. Whereas the present derivation assumes k kc , in fact, the closed-form analytical solution is robust for k that includes the full range of visible wavelengths [Fig. 3(c)]. This match also justifies the employed scalar-wave approximation as well as that the single scattering approximation applies to RI fluctuations typical for fixed biological cells. e and compare its We next describe the lc dependence of key aspects to those of the commonly used scattering parameters: the backscattering (b ) and the total scattering (s ) cross sections. The value of b manifests a nonmonotonic dependence on lc , which makes the inverse problem ambiguous [22], whereas s increases steeply / l3c and thus is relatively insensitive to structural changes at small e Þ is distinguished by three length scales [23]. In turn, ðl c important properties illustrated in Fig. 3. First, unlike b , e Þ can be monotonic. This property is apparent for thin ðl c samples [L < 2 m, Figs. 3(a) and 3(b)]. For thicker samples, a smaller collection NA can be chosen so that e Þ remains monotonic [e.g., NA ¼ 0:3 in Fig. 3(c)]. ðl

n

week ending 19 JULY 2013

PHYSICAL REVIEW LETTERS

PRL 111, 033903 (2013)

1.5

z (µ m)

(e) 0

1

2

0.8 (f) 7.9

−1

k (µ m ) 11.8

15.7

FIG. 4 (color). 40 magnification, 0.6 NA microscope images of samples with L ¼ 2 m were synthesized by FDTD. Bright-field images of samples with (a) lc ¼ 20 nm and (b) lc ¼ 50 nm; ðx0 ; y0 Þ=201 obtained from the wavelength-resolved image of (c) the sample with lc ¼ 20 nm and (d) lc ¼ 50 nm; (e) RI of the two samples as a function of z along central voxels (x0 , y0 ), and (f) image spectra of the corresponding pixels (x00 , y00 ).

spectral oscillations in the wavelength-resolved microscope image indicates a higher spatial frequency of the sample’s RI fluctuations. Recognizing that the experimental n ðrÞ may not be exponentially correlated, one may attempt to (a) use the validated approximations to obtain a closed form solution for a different functional form of the PSD from Eq. (5), (b) represent the correlation function of n as a superposition of exponentials, or (c) evaluate Eq. (5) numerically (no explicit functional form of the PSD is required for the latter two). e does not probe spatial We emphasize that whereas frequencies above 2k, the subdiffraction-scale structural alterations change the width of PSD and, therefore, the e Thus, e provides a monotonic measure for the value of . width of the 3-D PSD of RI fluctuations with a high sensitivity to subdiffractional length scales, without actually imaging the 3-D RI. We have established that despite the diffraction limit of resolution, the interferometric spectroscopy of scattered

033903-4

PRL 111, 033903 (2013)

PHYSICAL REVIEW LETTERS

light can quantify the statistics of RI fluctuations at deeply e subdiffractional length scales. We have shown that obtained from an elastic, label-free, spectrally resolved far-field microscope image quantifies RI fluctuations inside weakly scattering media at length scales limited by the SNR of the detector. We have derived a closed-form anae that yields results that agree with lytical solution for numerical solutions of Maxwell’s equations over a wide tested range of sample and instrument parameters. Potential applications include semiconductors, material science, biology, and medical diagnostics. This work was supported by National Institutes of Health (NIH) Grants No. R01CA128641, No. R01EB003682, and No. R01CA155284 and National Science Foundation (NSF) Grant No. CBET-0937987. The FDTD simulations were made possible by a computational allocation from the Quest high-performance computing facility at Northwestern University.

*Corresponding author. [email protected] [1] D. Lu and Z. Liu, Nat. Commun. 3, 1205 (2012). [2] A. Sentenac, P. C. Chaumet, and K. Belkebir, Phys. Rev. Lett. 97, 243901 (2006). [3] S. W. Hell, Science 316, 1153 (2007). [4] B. Huang, W. Wang, M. Bates, and X. Zhuang, Science 319, 810 (2008). [5] D. W. Piston, Trends Cell Biol. 9, 66 (1999). [6] G. Popescu, Quantitative Phase Imaging of Cells and Tissues, McGraw-Hill Biophotonics (McGraw-Hill, New York, 2011). [7] Z. Wang, L. Millet, M. Mir, H. Ding, S. Unarunotai, J. Rogers, M. U. Gillette, and G. Popescu, Opt. Express 19, 1016 (2011). [8] B. Bhaduri, H. Pham, M. Mir, and G. Popescu, Opt. Lett. 37, 1094 (2012).

week ending 19 JULY 2013

[9] H. Subramanian, P. Pradhan, Y. Liu, I. R. Capoglu, J. D. Rogers, H. K. Roy, R. E. Brand, and V. Backman, Opt. Lett. 34, 518 (2009). [10] I. Itzkan, L. Qiu, H. Fang, M. M. Zaman, E. Vitkin, I. C. Ghiran, S. Salahuddin, M. Modell, C. Andersson, L. M. Kimerer, P. B. Cipolloni, K.-H. Lim, S. D. Freedman, I. Bigio, B. P. Sachs, E. B. Hanlon, and L. T. Perelman, Proc. Natl. Acad. Sci. U.S.A. 104, 17 255 (2007). [11] S. A. Alexandrov, S. Uttam, R. K. Bista, K. Staton, and Y. Liu, Appl. Phys. Lett. 101, 033702 (2012). [12] A. Taflove and S. C. Hagness, Computational Electrodynamics: The Finite-Difference Time-Domain Method (Artech House, Norwood, 2005), 3rd ed. [13] D. Cook, Cellular Pathology: An Introduction to Techniques and Applications (Scion, Bloxham, 2006). [14] G. C. Crossmon, Stain technology 24, 241 (1949). [15] M. Born and E. Wolf, Electromagnetic Theory of Propagation, Interference and Diffraction of Light, edited by M. Born and E. Wolf (Cambridge University Press, Cambridge, England, 1998). [16] J. Goodman, Introduction To Fourier Optics, McGraw-Hill Physical and Quantum Electronics Series (Roberts & Co., Englewood, 2005), pp. 126–154. [17] See the Supplemental Material at http://link.aps.org/ supplemental/10.1103/PhysRevLett.111.033903 for a detailed derivation. [18] I. R. Capoglu, ANGORA: A free software package for finite-difference time-domain (FDTD) electromagnetic simulation (2012), date accessed: April 2012, http:// www.angorafdtd.org. [19] I. R. Capoglu, J. D. Rogers, A. Taflove, and V. Backman, in Progress in Optics, Vol. 57, edited by E. Wolf (Elsevier, New York, 2012), pp. 1–91. [20] I. R. Capoglu, A. Taflove, and V. Backman, IEEE Trans. Antennas Propag. (to be published). [21] J. M. Schmitt and G. Kumar, Appl. Opt. 37, 2788 (1998). [22] A. Ishimaru, Wave Propagation and Scattering in Random Media, IEEE Press Series on Electromagnetic Wave Theory (Wiley, New York, 1999). [23] A. J. Radosevich, J. Yi, J. D. Rogers, and V. Backman, Opt. Lett. 37, 5220 (2012).

033903-5