PHYSICAL REVIEW A 81, 063832 (2010)

Intense nonclassical light: Controllable two-photon Talbot effect

C. H. Raymond Ooi1,2,* and Boon Leong Lan2

1
Department of Physics, University of Malaya, 50603 Kuala Lumpur, Malaysia
2
School of Engineering, Monash University, 46150 Bandar Sunway, Selangor, Malaysia
(Received 9 February 2010; published 28 June 2010)
We study the spatial interference pattern of two-photon correlation function for a coherently phased linear
array of N emitters with a double-Raman scheme, each producing nonclassically correlated photon pairs. The
N 2 dependence in the two-photon correlation serves as a coherent amplification method for producing intense
nonclassical light. The spatial distribution of the correlation can be controlled by lasers, and depends on the
detection configuration. For two coincident detectors, the nonclassical correlation displays the spatial Talbot
pattern, but modulated by quantum interference effect. The image revival distance is found to be twice the usual
Talbot length. For symmetrically located detectors (X1 = −X2 ), the correlation displays a distorted Talbot pattern
with intricate features and lack of symmetry.

DOI: 10.1103/PhysRevA.81.063832 PACS number(s): 42.50.Ar, 42.50.St

I. INTRODUCTION [14], and in atomic systems [15]. The effect has been used
in laser phase locking and incorporated into high-field laser
Nonclassical light source is generated artificially and does
systems [16]. More exotic applications of the effect include
not exist in nature. Recently, it was found that correlated pho-
factorization of large numbers and evaluation of the Gauss sum
ton pairs can provide optical resolution beyond the diffraction
[17]. Here, we will study how the nonclassicality of correlated
limit through coincident photon detection [1] and have been
photons affects the Talbot pattern for arrayed emitters. We
proposed for quantum imaging [2]. The photon pairs can also
analyze the spatial structure of the two-photon correlation from
enhance the spectral resolution of atoms and molecules [3].
the array of emitters (see Fig. 1).
Particles with the double-Raman scheme can produce photon
In Sec. II, two-photon amplitude for multiple double-
pairs with nonclassical properties, such as photon antibunching
Raman emitters is obtained from the amplitude for a single
[4] in the correlation function G(2) . The nonclassicality of
emitter. In Sec. III, we consider a particular case of a one-
photon pairs is also found in the two-atoms case [5]. In the
dimensional (1D) array within the paraxial approximation.
many-atoms case or extended medium [6] the correlation
Two possible detection schemes are considered in Sec. IV,
function violates the Cauchy-Schwarz inequality. The results
each producing entirely different spatial structures. In the final
for extended medium are based on full quantum theory and
Sec. V, we discuss the features and the implications of the
agree with the experiment [7]. However, its applicability is
obtained results.
limited to photon pairs at moderate intensity. Nonclassical
light at high intensity has not been thoroughly studied both
experimentally and theoretically. A sufficiently bright source II. NONCLASSICAL CORRELATION FOR
of nonclassically correlated photons could become a new tool MULTIPLE EMITTERS
in optics. Several research groups claimed to have successfully We now consider an array of N quantum emitters (atoms or
produced nonclassical light sources with high brightness. quantum dots), each located in periodic sites, with the distance
Photonic crystal fiber [8] can generate up to 107 photon between two emitters larger than an optical wavelength such
pairs per second, 100 times higher compared to laser-cooled that they are noninteracting, as shown in Fig. 1. Otherwise,
atoms. The corresponding quantum theory was developed by dipole-dipole interaction due to proximity may contribute
Agrawal [9], valid for moderately high intensity. Recently, to the coherent effects [18]. Each emitter has three levels
Harris proposed a scheme to generate single-cycle photon driven coherently by two lasers forming the double-Raman
pairs [10] from periodic structure with nonlinear optics. scheme [Fig. 1(a)]. The pump laser (with Rabi frequency
The quest toward intense nonclassical light source moti-
p ) creates the Stokes photon with wave vector k via an
vates our present work to study photon correlation for an off-resonant Raman transition from level c to level b which is
array of quantum emitters. Coherently phased emissions can resonantly coupled by the control laser
c to level a, creating
produce intensity that scales quadratically with the number the anti-Stokes photon with wave vector q [19]. For atoms,
of emitters in linear as well as circular configuration [11]. Rb-87 can be used, with the levels a,b,c corresponding to
An interesting question follows: Would such scaling apply |52 P1/2 ,F = 1
,|52 S1/2 ,F = 2
,|52 S1/2 ,F = 1
, respectively
for correlated or joint coincident detection of nonclassically (D1 transition). For quantum dots, the upper level corresponds
correlated photons? It is interesting to investigate how the to the biexciton XX which is coupled to two (intermediate)
interference pattern repeats regularly with the distance from exciton levels X1 and X2 , split by the exchange interaction due
the emitters, the Talbot effect [12]. Recently, the effect was to anisotropy of the exciton wave function [see Fig 1(a)].
found in waveguide modes [13], in terms of Wigner function In order to study the interference effects of the two
sequential detection of the Stokes and anti-Stokes photons
[at (r1 ,t1 ) and at (r2 ,t2 ), respectively], we compute
*
[email protected] the two-photon correlation function G(2) = |Ê (−) (r1 ,t1 )

1050-2947/2010/81(6)/063832(7) 063832-1 ©2010 The American Physical Society

C. H. RAYMOND OOI AND BOON LEONG LAN PHYSICAL REVIEW A 81, 063832 (2010)

FIG. 1. (Color online) An array of quantum emitters, each producing correlated photons from the double-Raman scheme (inside orange
frame). (a) Double-Raman scheme, applied to a typical scheme in quantum dots, (b) array with central emitters (N = 31) and the g (2)
distribution, and (c) array with side emitters (N = 30) and the g (2) . Note that the peaks in the images occur at D = 2zT = 4d 2 /λ (nonclassical
2
Talbot length) which is twice the classical length zT = 2dλ . We use X2 = X1 ,
c = 2000, D2 = 1.001D1 , and d = 3N 2 λ with  = 108 s−1
and λ = 0.3 µm. Thus, the peaks emerge around D/λ = 36N 4 = 2.9 × 107 . All subsequent figures below are for case (a).


Ê (−) (r2 ,t2 )Ê (+) (r2 ,t2 )Ê (+) (r1 ,t1 )|
. The Schrödinger equa- is the sum over all the emitters, |
= √1N m |
m =
tion approach for the single m-th atom with the 
mkq Cmkq (∞)|cm ,1k ,1q
. Thus, the two-photon ampli-
√1
double-Raman scheme located at rm gives the analytical N
tude for the many-atoms state becomes
expression for the two-photon amplitude in the far field [20],

ψm(2) (1,2) = 0|Ê (+) (r2 ,t2 )Ê (+) (r1 ,t1 )|
m ψ (2) = 0|Ê (+) (r2 ,t2 )Ê (+) (r1 ,t1 )|
= ψm(2) (1,2)
m
Zm i(kc +kp )·rm 
= e (1) = e i(kc +kp )·rm
Zm K1m Q2m , (3)
rm1 rm2
Zm = iC12 e−iντm1 e−iωτm2 e−(/2)τm1 m

× e−(γ /4)(τm2 −τm1 ) sin
(τ

˜ m2 − τm1 ), (2) where
where (kc + kp ) · rm is the coherent phase factor (CPF),  
K1m = k f (x1m ) − g(x1m ) 12 sin2 α1m , (4)

= γ (
p / ) is the Raman pumping rate,
=
 2 ˜  
Q2m = q f (y2m ) − g(y2m ) 21 sin2 α2m , (5)

c − (γ /4) is the effective Rabi frequency, |
m =
2 2

kq Cmkq (∞)|cm ,1k ,1q
is the two-photon state with Cmkq are exact expressions valid for the near field, with f (x1m ) =
being the coefficient of the ground state c with one Stokes 1
+ i x12 − x13 , g(x1m ) = x11m + i x32 − x33 , x1m = kr1m , and
and one anti-Stokes photon, τmj = tj − rmj /c are emis- x1m 1m 1m 1m 1m

sion times, rmj = |rm − rj | with τm2 − τm1 > 0 and C12 = sin αj m = (xm − Xj )/rj m (j = 1,2). Similar expressions for


f (y2m ) and g(y2m ), with y2m = qr2m .
−
˜p c ℘ba ℘ca ( 4πε
ων 2
2 ) with ℘αβ the dipole transition matrix
oc
  The two-photon correlation can then be obtained from
elements, and Ê (+) (rj ,tj ) = k (ˆj · ε̂k ) 2εh̄νokV âk ei(k·rj −νk tj ) is G(2) = |ψ (2) |2 and Eq. (3), which is valid for arbitrary distances
the collective electric field operator at detector j (=1,2) due rj m . In the far field, Kj m → rj1m .
to the contributions of N arrayed emitters. For rectangular geometry the CPF can be expressed as
Since the emitters are identical, independent, and non- exp[iθm,x + iθm,y ] where θm,x = (kcx + kpx )xm and θm,y =
interacting, there is no correlation between the fields from (kcy + kpy )ym . For a two-dimensional (2D) arrayed system
different emitters. The collective state of all the particles we can replace xm → ma and ym → nb.

063832-2
INTENSE NONCLASSICAL LIGHT: CONTROLLABLE TWO- . . . PHYSICAL REVIEW A 81, 063832 (2010)

The correlation for the case of two atoms can be written A. Paraxial approximation
analytically as In the paraxial regime, Dj |md − Xj | leads to the
C12 constraint,
G(2) = 2e−τ1 e−γ τ21 sin2
τ
˜ 21
(r1 r2 )2 D1,2 |Xmax + N d|, constraint Ia. (8)
× {1 + cos[ν τ1 + ω τ2 − (kc + kp ) · r]}, (6) √ x2 x4
Substituting D 2 + x 2 by D + 2D − 8D 3 in Eq. (7) and

with τj = τj − τj , τj = tj − rj /c, τj = tj − rj /c, neglecting the quartic and higher order terms, we have
r = r − r is the vector joining the two emitters, and  eimKd
 
˜ τ+1 x2 x2
rj and rj are the distances between the two emitters to ψ (2) ∝ sin
D + 1m − 2m
detector j (=1,2). Analysis of Eq. (6) without the CPF can be m
r1m r2m c 2D1 2D2
found in Refs. [19,21] in the context of quantum microscopy.    
ν x2 ω x2
× exp i D1 + 1m + D2 + 2m ,
c 2D1 c 2D2
III. 1D ARRAY OF EMITTERS (9)
We focus on a 1D phased array of quantum emitters [see where D = D1 − D2 .
x4
Figs. 1(b) and 1(c)]. This structure can be realized by using At the center Xj = 0, the negligible term 8D 3
in paraxial
a holographic optical lattice for atoms [22] or a microtrap (ν+ω) (md)4 π (md)4
approximation implies max{ c 8D3  λ 2D3 } 2π .
array for ions [23] as regular trapping sites, as well as an
Taking m → N ( 1) we have the condition,
array of quantum dots [24]. Correlated photon pairs have
been produced by using quantum dot driven by two-photon (N d)4 (N d)4
absorption in cascade configuration [25]. For the optical k = 1, constraint Ib. (10)
8π D 3 4λD 3
trapping system it might be a challenge to create a large 2
number of sites, with each site containing only one atom, When D = dλ (in the order of the Talbot length) is substituted
although this has been demonstrated with seven atoms [26]. into Eq. (10), we have the constraint,
Certain sites may contain two or more atoms. In the case of λ 2
quantum dots, the decoherence rate γ (∼41012 s−1 ) is typically d> N constraint II. (11)
2
1000 times higher than in atoms, significantly reducing the
correlation time. Thus, cooling down to a few Kelvins is
B. Central amplitude
required. Quantum dots are usually grown at random locations.
Recently, a large array of regularly spaced quantum dots has At the center Xj = 0 the amplitude Eq. (7) gives
been produced by ion-beam irradiation [27].   0 
0
r1m − r2m
Using rm = md x̂, the amplitude goes as (since ti  ti ) ψ (0) ∝
(2)
sin
τ +
˜
c
 m
ψ (2)  iC12 e−iνt1 e−iωt2 e−(/2−γ /4)τm1 e−(γ /4)τm2 eimKd   
m × 0
exp i νr1m + ωr2m
0
c , (12)
r1m r2m
eimKd

× exp[i(νr1m + ωr2m )/c] where rj0m = Dj2 + (md)2 . The paraxial approximation re-
r1m r2m
  duces Eq. (12) to
r1m − r2m  
× sin
τ +
˜ , (7) D
c ψ (0) ∝ e
(2) i(νD1 +ωD2 )/c
sin
τ +
˜
c
where −(N − 1)/2 < m < (N − 1)/2, K = kcx + kpx is the  eimKd
 wave vector, τ = t2 − t1 is the effective delay, and
coherent × exp[iSm2 d 2 ], (13)
r r
rj m = Dj2 + xj2m (with xj m = Xj − md and j = 1,2 for the m 1m 2m

detectors). +ω/D2 2 2
where S = ν/D1 2c . The quadratic dependence in eiSm d
In the following figures we have plotted normalized is the consequence of the Fresnel (near-field) diffraction.
correlation g (2) = |ψ (2) /C12 |2r1m, r2m =1 by neglecting the decay The series summation gives the usual Talbot effect. For
factors, the overall phase factors, and the overall constant. We convenience, we give a short review and tutorial on the Talbot
also set r1m ,r2m → 1 since it does not yield important physics. effect in the Appendix.
Figures 1(b) and 1(c) show two possible configurations of Letting ν/D1  ω/D2 (S  Dλ 2π
) the central amplitude
the array of particles with the corresponding plots of the  
becomes ψ (0) ∝ m exp[i Dλ m d ] = m exp[iπ m2 zDT ].
(2) 2π 2 2
g (2) for X1,2 = X, computed from Eq. (7). The correlation 2

patterns show that the Talbot images are revived at D = When D = zT = 2dλ , the series is a sum of equal numbers
2zT = 2d 2 /(λ/2) which is twice the usual Talbot length zT , in of terms with even and odd multiples of π for the case with
(N−1)/2
contrast to that reported in Ref. [28]. Our result implies that the central emitter [Fig. 1(b)], that is, m=−(N−1)/2 exp[iπ m2 ] =
effective wavelength is λ/2, as expected, due to the coincident 1 + 2(eiπ + ei4π + ei9π + ei16π ...) = 1 (N odd), giving
detection. In the subsequent analysis and results we focus on ψ (2) (0)  1 ( N ) with almost no signal due to destructive
the configuration with the central emitter only, as in Fig. 1(b), interference of photons from all emitters except the
where N is odd. central one. However, for the case with side emitters

063832-3
C. H. RAYMOND OOI AND BOON LEONG LAN PHYSICAL REVIEW A 81, 063832 (2010)

(N−1)/2
m=−(N−1)/2 exp[iπ m ] = 2(e + ei9π/4 +
2 iπ/4
[Fig. 1(c)],
e i25π/4
+e i49π/4
+e i81π/4
+ · · ·) (N even) would give a finite
value.

IV. TWO-PHOTON DETECTION SCHEMES

The use of two detectors for obtaining the two-photon corre-
lation provides a new method for acquiring information or sig-
nals, utilizing the novel properties of coherently phased non-
classical photon pairs with interference and diffraction effects.

A. Coincident detection
For detectors that coincide, X1,2 = X, Eq. (9) becomes
(neglect the damping terms)
 eimKd
ψ (2) ∝ ei(νD1 +ωD2 )/c
2
eiS(X−md)
m
r 1m r 2m
 2
˜ τ + D 1 − (X − md)
× sin
, (14)
c 2D1 D2
where S = ν/D1 +ω/D
2c
2
, D = D1 − D2 .
The Talbot effect is applicable for the region of X satisfying
2
the constraint Eq. (8). Since (X−md)
D1 D2
1, the final (quadratic)
term in the sine function is much smaller than the second term
and may be neglected, Eq. (14) becomes
 
D  eimKd iS(X−md)2
ψ ∝ sin
τ +
(2) ˜ e , (15)
c m
r1m r2m

where we have discarded the overall phase factor ei(νD1 +ωD2 )/c
which gives no important effect. Thus, the nonclassical sine
term does not depend on X within the Talbot region. For large
X (i.e., beyond the paraxial region), however, the nonclassical FIG. 2. (Color online) Two-photon correlation g (2) (normalized)
effect would have X dependence. Assuming ν = ω and plotted using Eq. (16) (for paraxial region) against X = X1 = X2
taking D2 = αD1 with D1 = D we have D = (1 − α)D, and D = D1 at around D = zT . Note that the images at D = zT
−1
S = π(1+αDλ
)
, and [thick (red) arrow] are shifted half-period compared to the images at
  D = 2zT in Figs. 1(b) and 1(c). (a) Without the sine term showing
ψ (2) ∝ sin
˜ τ + (1 − α)D (2)
peak intensity gmax = N 2  900 (N  30) and (b) with the sine term,
c corresponds to nonclassical Talbot carpet. The heights of the peaks
 eimKd π (1 + α −1 ) at Talbot length zT = 2d 2 /λ are below N 2 due to modulation by the
× exp i (X − md)2 . (16) sine term, due to quantum interference. The zero intensity channel
m
r r
1m 2m Dλ [thin (blue) arrow] in (b) is independent of X. It occurs at the location
D0 = π c/|1 − α|
˜ (α = 1.001) (i.e., inversely proportional to the
The term with quadratic m is important for small D while ˜ We use N = 31,
c = 2000, D2 =
effective Rabi frequency
).
the sine term is due to quantum interference. The CPF eimKd 1.001D1 , and d = 3N 2 λ.
gives the usual diffraction grating effect. It also has an effect
of shifting the entire carpet along X.
Based on the constraint, Eq. (11), the analytical formula where Xm (±)
= X ± md. Simpler expression is obtained by
Eq. (16) in the paraxial regime gives results (shown in Fig. 2) letting D1 = D2 and ν = ω,
that agree with the exact formula Eq. (7).  eimKd 2πm2 d 2  
2Xmd
ψ ∝
(2)
e i λD
sin
τ +
˜ . (18)
m
r1m r2m Dc
B. Symmetric detection
For detectors situated on symmetrically opposite locations 2 N
At D = zT /2 = dλ the amplitude takes the form
X1 = −X and X2 = X, Eq. (9) reduces to ˜ + m λX ), which vanishes at X = 0 when τ = 0.
m=−N
sin
(τ
 eimKd   cd
i ν (+)2 ω (−) 2 The dependency on X is governed by the nonclassical sine
ψ (2) ∝ exp Xm + Xm (interference) function. Both the argument in the sine function
m
r1m r2m 2c D1 D2
 (+)2 and the coherent phase factor depend linearly on m. We see
(−) 2 
˜ τ + D + 1 Xm − Xm that the X dependence is important when 2πNd ∼
˜ X or N d
× sin
, (17) λ c
c 2c D1 D2 is not too large. The relative magnitudes of the terms are,

063832-4
INTENSE NONCLASSICAL LIGHT: CONTROLLABLE TWO- . . . PHYSICAL REVIEW A 81, 063832 (2010)

g ( 2) twin peaks

(a) (b) i (b) ii

FIG. 3. (Color online) Two-photon correlation g (2) (top view) beyond paraxial approximation for opposite detectors (X2 = −X1 = X)
[plotted using the exact expression and not the paraxial Eq. (18)] for: (a) without the sine term and (b) with the sine term, where (ii) is the
magnified view of the circled area in (i). Note that the peaks and regions of destructive interference now depend on X. The scale along X
is substantially larger than the X2 = X1 case (i.e., Xmax = 5500d. Here, N = 61 with other parameters
c = 2000, D2 = 1.001D1 , and
d = 3N 2 λ, the same as in Fig. 2.

mKd : 2πm
2 2
d ˜
:
md 2X
= 1 : Nd
˜X
:
ν D . Taking into account width of the peaks is about 50λ, thus the intensity is I =
P /σ ∼ 107 Wm−2 , which is close to the intensity of a typical
λD Dc D
˜ X
d ∼ λN and D ∼ d /λ ∼ λN we have 1 : N1 : νN
2 2 4

Nd
.
diode laser with 100-mW power, diameter of 5 mm. This
In Fig. 3 we plot Eq. (7), the exact expression for G(2) result shows that it is possible to engineer a coherent source of
(beyond paraxial region). Here, the scale along X has to be nonclassical light with high intensity by constructing arrayed
larger than the scale in Fig. 2 by a factor of N ν/
˜ in order
emitters that are coherently phased, providing the prospect for
to see the nonclassical pattern. Figure 3(a) shows how the two generation of intense nonclassical light, providing a coherent
detectors at the opposite sides of the y axis can produce such amplification method for two-photon lithography compared
a distorted Talbot pattern. The main feature here is the arc to existing parametric amplification schemes using an optical
structure, which is mainly due to the departure from the parametric amplifier (OPA) [29], an optical parametric oscilla-
paraxial region. The peaks of the structure vary with both tor (OPO) [30], and an image-forming projection system using
D and X, even without the sine term. The presence of the the optical transfer function [31].
nonclassical sine term makes the pattern less regular with
more complicated features [see Fig. 3(bi)]. A closer look
reveals a prominent twin-peaks feature, as shown in Fig. 3(bii). B. Laser-controlled Talbot effect
Further studies and analysis of this structure are required to
The nonclassical sine term modulates the entire Talbot
gain insights for new applications.
pattern in Fig. 2(b), lowering the peak intensity below N 2 level.
It gives the no-signal (NS) channel as the result of destructive
V. DISCUSSIONS interference at all values of X. This happens (for τ = 0) at
D1 satisfying
˜ |1−α|D1 = mπ , m is integer. The spatial period
We elaborate on several important aspects of the results, c
particularly for generating nonclassical light sources with high between two NS lines is
intensity and the significance of the nonclassical interference
“sine” term on the spatial structure of the nonclassical Talbot D0 = cπ/(
|1
˜ − α|), (19)
pattern.
which correctly gives the location of the destructive interfer-
ence in Fig. 2(b).
A. Toward intense nonclassical light This feature gives a useful laser-controlled Talbot effect.
Figure 2(a) shows the coherence effect of the two-photon The Talbot image at a specific location of Dimage can be
correlation (i.e., the peak intensity of G(2) scales with N 2 , erased by tuning the control field (through
) ˜ such that
regardless of whether the coherent phase factor exp[imKd] Dimage = mD0 . We also find the coherent phase factor (CPF)
is present). The result shows that the coherent method can has an effect of shifting the entire pattern horizontally, across
produce high-intensity photon pairs without using the nonlin- X. This provides another coherent control mechanism in
ear amplification process. If we let d = 10λ, large enough for another dimension.
the neglect of dipole-dipole interaction, with λ = 300 nm, an Since the resulting (modulated) two-photon Talbot pattern
array size of 3 cm would contain 104 emitters which provide contains essentially the same old features as the classical case,
N = 108 photons in one collective emission corresponding to it should be applicable for mode locking and factorization of
an energy of U = N h̄ν ∼ 7 × 10−11 J. The cyclic pumping numbers using two-photon light sources. Moreover, the laser-
duration corresponds to the spontaneous emission lifetime controlled Talbot effect can be developed into an innovative
1/  ∼ 10−8 s. Thus, the power is P = U  = 7 mW. The system of dynamically controlled phase locking.

063832-5
C. H. RAYMOND OOI AND BOON LEONG LAN PHYSICAL REVIEW A 81, 063832 (2010)

C. Other features result of the two-detector scheme, which is not present in

In addition to the main features discussed above, there are the single-photon detection case. The structure varies over
a few side points to note. On closer inspection of the Talbot a larger scale, beyond the paraxial region and displays a
patterns in Figs. 2 and 3, we find seemingly chaotic and irreg- distorted Talbot pattern. The most interesting result is the
ular background oscillations which look like sea waves around effect of the interference “sine term,” which makes the pattern
regularly revived peaks (images of the source). This peculiar more intricate and complex. Our results on the nonclassical
feature may provide insight in connection with the formation version of the two-photon Talbot effect could be useful for
of rogue waves [32]. Also, in the near field (kr1m ,qr2m ∼ 1), the development of nonclassical optics and diffractive optical
the N 2 dependence does not apply and we find no Talbot effect. elements based on nonclassical light sources.
Finally, the coherent phase factor (due to lasers) introduces an
overall shift that has little effect on the general features. ACKNOWLEDGMENTS
This work is supported by University of Malaya Research
VI. CONCLUSIONS Grant No. RG111/10AFR. C. H. R. Ooi and B. L. Lan thank
The present work shows that a coherently phased array Monash University for support through the internal research
of quantum emitters can produce large number of correlated fund.
photon pairs that scales with N 2 at localized spots, with
promising intensity close to a diode laser. This is a significant APPENDIX: TALBOT EFFECT
step forward to realizing new kinds of light sources with high
intensity and nonclassical properties which could lead to new The self-images of the original periodic structure would
2
possibilities, particularly in quantum nonlinear optics with appear at multiples of a distance zT = 2dλ , called the Talbot
intense nonclassical light. length. The same pattern would also recur at D = (2n +
We have also studied the spatial structures of the two-photon 1)zT /2, but shifted in space by half the spatial period. Higher
correlation function G(2) for the 1D array of nonclassical order fractional revivals occur at rational multiples of zT .
emitters with two detectors. When the detectors coincide, At D = (p/2)zT /q (particularly zT /4, 3zT /4, . . . , zT /6,
X1 = X2 , the G(2) distribution across the screen position X 3zT /6, 5zT /6, . . . zT /8, 3zT /8, 5zT /8, 7zT /8, . . .) with q =
and distance D displays patterns reminiscent of the classical 1,2,3, . . . ; p = 1,3,5 . . . , the images would recur but they are
Talbot carpet, but with effective Talbot length twice the inverted or shifted in space by half-period, each consists of q
classical Talbot length. The nonclassical interference “sine copies of the original grating, separated by d/q.
term” provides a possibility to tune the intensity of the At D = rzT /s (particularly zT /3, 2zT /3, . . . , zT /5, 2zT /5,
Talbot image by using the control laser. However, when 3zT /5, 4zT /5, . . .) with s = 3, 5, 7 . . . ; r = 1, 2, 3 . . . , the
the detectors are on the opposite sides of the y axis, such images are not shifted, but each consists of s copies of the
that X1 = −X2 , we find unique structure in the G(2) as the original grating, separated by d/s.

[1] M. D’Angelo, M. V. Chekhova, and Y. Shih, Phys. Rev. Lett. 87, [15] A. E. Kaplan, P. Stifter, K. A. H. van Leeuwen, W. E. Lamb Jr.,
013602 (2001). and W. P. Schleich, Phys. Scr. T 76, 93 (1998); A. E. Kaplan,
[2] Y. Shih, J. Mod. Opt. 49, 2275 (2002). I. Marzoli, W. E. Lamb Jr., and W. P. Schleich, Phys. Rev. A 61,
[3] M. O. Scully, U. W. Rathe, C. Su, and G. S. Agarwal, Opt. 032101 (2000).
Commun. 136, 39 (1997). [16] J. R. Leger, Appl. Phys. Lett. 55, 334 (1989); Y. Kono et al.,
[4] M. O. Scully and K. Druhl, Phys. Rev. A 25, 2208 IEEE J. Quantum Electron. 36, 607 (2000).
(1982). [17] D. Bigourd, B. Chatel, W. Schleich, and B. Girard, Phys. Rev.
[5] C. H. Raymond Ooi, Phys. Rev. A 75, 043817 (2007). Lett. 100, 030202 (2005).
[6] C. H. Raymond Ooi, Q. Sun, M. S. Zubairy, and M. O. Scully, [18] C. H. Raymond Ooi, Phys. Rev. A 75, 043817 (2007); C. H.
Phys. Rev. A 75, 013820 (2007). Raymond Ooi, B.-G. Kim, and H.-W. Lee, ibid. 75, 063801
[7] V. Balic, D. A. Braje, P. Kolchin, G. Y. Yin, and S. E. Harris, (2007).
Phys. Rev. Lett. 94, 183601 (2005). [19] M. O. Scully and C. H. Raymond Ooi, J. Opt. B: Quantum
[8] X. Li, J. Chen, P. l Voss, J. Sharping, and P. Kumar, Opt. Express Semiclass. Opt. 6, S816 (2004).
12, 3737 (2004); J. Fulconis, O. Alibart, W. J. Wadsworth, [20] C. H. Raymond Ooi, A. K. Patnaik, and M. O. Scully, Noise
P. St. J. Russell, and J. G. Rarity, ibid. 13, 7572 (2005). and Information in Nanoelectronics, Sensors, and Standards III,
[9] Q. Lin, F. Yaman, and G. P. Agrawal, Phys. Rev. A 75, 023803 in Proceedings of the SPIE 5846, 1, Bellingham, Washington
(2007); Opt. Lett. 31, 1286 (2006). (The International Society for Optical Engineering, Bellingham,
[10] S. E. Harris, Phys. Rev. Lett. 98, 063602 (2007). 2005).
[11] C. H. Raymond Ooi, J. Appl. Phys. 107, 043110 (2010). [21] M. O. Scully, Concepts of Physics 2, 261 (2005).
[12] W. H. F. Talbot, Facts Related to Optical Science, No. IV, Philos. [22] S. Bergamini, B. Darquié, M. Jones, L. Jacubowiez,
Mag. 9, 401 (1836). A. Browaeys, and P. Grangier, J. Opt. Soc. Am. B 21, 1889
[13] R. Iwanow et al., Phys. Rev. Lett. 95, 053902 (2005). (2004).
[14] O. M. Friesch, I. Marzoli, and W. P. Schleich, New J. Phys. 2, 4 [23] J. Chiaverini and W. E. Lybarger Jr., Phys. Rev. A 77, 022324
(2000). (2008).

063832-6
INTENSE NONCLASSICAL LIGHT: CONTROLLABLE TWO- . . . PHYSICAL REVIEW A 81, 063832 (2010)

[24] Q. Sun et al., Nature Photonics 1, 717 (2007). [29] G. S. Agarwal et al., J. Opt. Soc. Am. B 24, 270 (2007).
[25] C. Santori, D. Fattal, J. Vuckovic, G. S. Solomon, and [30] H. Cable, R. Vyas, S. Singh, and J. P. Dowling, New J. Phys. 11,
Y. Yamamoto, Fortschr. Phys. 52, 1180 (2004). 113055 (2009).
[26] D. G. Grier, Nature 424, 810 (2003). [31] N. Fukutake, J. Mod. Opt. 53, 719 (2006).
[27] M. Buljan et al., Phys. Rev. B 81, 085321 (2010). [32] N. Akhmediev, J. M. Soto-Crespo, and A. Ankiewicz, Phys.
[28] K. H. Luo et al., Phys. Rev. A 80, 043820 (2009). Rev. A 80, 043818 (2009).

063832-7