O

light’s wavelength. The reference beam’s electric field can

E Quantum-noise-limited be defined by
sensitivity of an
interferometer using a
冋
E r ⫽A r exp j ␻ o t⫺2k o L r ⫺2k o 冉冊␭
4 册
sin共 ␻ c t 兲 , 共1兲

phase generated where A r is the reference field amplitude, ␻ o is the light’s

angular frequency, k o is the light propagation factor in free
L carrier demodulation space, and L r is the distance from the beam splitter to the
reference mirror. The signal beam’s electric field can be
scheme defined by
E E s ⫽A s exp关 j ␻ o t⫺2k o L s ⫺2k o ␦ 共 t 兲兴 , 共2兲
Tristan J. Tayag, MEMBER SPIE
T Texas Christian University where A s is the signal field amplitude, L s is the distance
from the beam splitter to the target structure, and ␦ (t) is the
Department of Engineering
TCU Box 298640 surface displacement to be measured. If we assume identi-
T Fort Worth, Texas 76129
E-mail: [email protected]
cal polarization states of the two fields, then coherent inter-
ference at the photodetector produces a resultant electric
field whose magnitude is given by
Abstract. The quantum-noise-limited sensitivity is derived
E for a two-beam interferometer, which uses a phase gener- 兩 E Resultant兩 ⫽A r 兵 1⫹K 2
ated carrier demodulation scheme. For surface-disturbance-
detecting sensors, this sensitivity limit is expressed in terms ⫹2K cos关 2k o ␦ 共 t 兲兴 cos关 2k o ⌬⫺ ␲ sin ␻ c t 兴
R of the smallest displacement that may be detected if the
⫺2K sin关 2k o ␦ 共 t 兲兴 sin关 2k o ⌬⫺ ␲ sin ␻ c t 兴 其 1/2,
signal-to-noise ratio is unity. We determine the quantum-
noise-limited displacement of the classical Michelson inter- 共3兲
S ferometer to be 5 pm. © 2002 Society of Photo-Optical Instrumenta-
tion Engineers.
where K is the field contrast factor (A s /A r ) and ⌬ is the
[DOI: 10.1117/1.1431971]
quiescent path-length difference between the signal and ref-
Subject terms: interferometry; metrology; sensors; signal-to-noise erence arms, (L r ⫺L s ). To express the limit of the interfer-
ratio; noise.
ometer’s sensitivity, only small displacements are consid-
Paper OEL 010040 received July 20, 2001; accepted for publication ered 共i.e., k o ␦ (t)Ⰶ1兲. Using small angle approximations
Nov. 7, 2001; appeared online Nov. 7, 2001. yields

兩 E Resultant兩 ⫽A r 关 1⫹K 2 ⫹2K cos共 2k o ⌬⫺ ␲ sin ␻ c t 兲

1 Introduction
⫺4Kk o ␦ 共 t 兲 sin共 2k o ⌬⫺ ␲ sin ␻ c t 兲兴 1/2. 共4兲
A wide variety of demodulation techniques have been de-
veloped for interferometric sensors. Homodyne demodula- The last term within the square brackets of Eq. 共4兲 is the
tion with a phase generated carrier 共PGC兲 achieves passive only term functionally related to the surface displacement,
operation, large dynamic range, and high sensitivity. Pas- ␦ (t). This is the only component of the resultant field,
sive operation obviates the need for the large piezoelectric which contributes to the signal portion of the photodetector
phase modulators and fast reset circuitry required in ac- current, i s . The entire resultant field contributes to the
tively stabilized homodyne systems.1 In addition, a large quantum noise portion of the photodetector current, i q .
dynamic range on the order of 109 has been realized using Now that we have defined the origins of signal and noise
digital signal processing to perform the demodulation.2 In in the system, we may express the performance of the sys-
this paper, we derive the quantum-noise-limited sensitivity tem in terms of the SNR. In the quantum noise limit, the
of interferometers using a PGC demodulation scheme. photodetector current’s SNR is given by
Quantum noise 共also called shot noise兲 in the photodetec-
tion process sets a fundamental limit in the performance of
the interferometric sensor.3
The classical Michelson interferometer is the optical
SNR⫽ 冋 册
具 i s2 典
具 i 2q 典
1/2
, 共5兲

configuration chosen for this analysis. However, the follow-

ing sensitivity analysis is applicable to all two-beam inter- where 具"典 indicates a time averaging operation. With respect
ferometers. Figure 1 shows the system layout. We form the to the optical detection process, the signal current is defined

冉 冊
PGC by dithering the position of the reference mirror with as
␩q
a piezoelectric transducer. This interferometric configura- i s⫽ P s , 共6兲
tion detects changes in the position of the target structure. h␯
2 Ultimate System Resolution where P s is that part of the instantaneous optical power
The PGC is imposed on the interferometer signal by sinu- incident on the photodetector, which contributes to the sig-
soidally dithering the reference mirror at an angular fre- nal, ␩ is the detector’s quantum efficiency, q is the carrier’s
quency of ␻ c and at an amplitude of ␭/4, where ␭ is the charge, h is Planck’s constant, and ␯ is the light’s optical

276 Opt. Eng. 41(2) 276–277 (February 2002) 0091-3286/2002/$15.00 © 2002 Society of Photo-Optical Instrumentation Engineers
Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 05/18/2015 Terms of Use: http://spiedl.org/terms
 Tayag: Quantum-noise-limited sensitivity . . . O
E

L
Fig. 1 Michelson interferometer.

E
Fig. 2 Normalized SNR versus path-length difference between the
frequency. Applying the signal bearing portion of Eq. 共4兲 to signal and reference arms of the interferometer.
Eq. 共6兲 results in
T
i s ⫽CA r2
␩q
h␯ 冉 冊
关 ⫺4Kk o ␦ 共 t 兲 sin共 2k o ⌬⫺ ␲ sin ␻ c t 兲兴 , 共7兲
where P l is the laser’s output power. This is the quantum-
noise-limited SNR of an interferometer, which uses a PGC
demodulation scheme. T
where C is a constant related to the intrinsic impedance of
free space and the photodetector area. Trigonometric ma- 3 Results and Discussion
nipulation and use of the Fourier-Bessel expansion reveals Notice that the quantum-noise-limited SNR is a function of E
the mean-squared signal current: ⌬. Figure 2 illustrates the variation of the SNR with respect

冉 冊
to the quiescent path-length difference, ⌬, between the sig-
␩q 2
具 i s2 典 ⫽C 2 A r4
h␯
关 8K 2 k 2o ␦ 2 共 t 兲兴关 1⫺J 0 共 2 ␲ 兲 cos共 4k o ⌬ 兲兴 . nal and reference paths. Note that the SNR is nonzero for
practical values of ⌬. This result is in contrast to homodyne
R
共8兲 interferometers, which do not use a PGC. In these systems,
To determine the ultimate system resolution, we con- the SNR goes to zero at ⌬ equal to an integer multiple of
␭/2.3 Active stabilization is therefore required to stabilize
S
sider the quantum noise process as the limiting noise
source. The detector’s quantum noise is related to the time- the system at quadrature 共i.e., ⌬ equal to an odd multiple of
averaged photodetector current, 具 i d 典 , as follows4 ␭/4兲. The open-loop interferometric system, however, does
not actively stabilize the system at quadrature. Instead, this
具 i 2q 典 ⫽2q⌬ ␯ 具 i d 典 , 共9兲 system uses a PGC to extract the desired amplitude infor-
mation at any path-length difference, ⌬.
where ⌬␯ is the detection electronics bandwidth. This lim- The ultimate resolution of the system is calculated di-
iting mean-squared noise current may be expressed as rectly, using the results of the SNR analysis. We set the

冉 冊
requirement that the SNR equal unity. The surface displace-
␩q ment, ␦ (t), then assumes its value at the detection limit,
具 i 2q 典 ⫽2q⌬ ␯ •C•A r2 • 关 1⫹K 2 ⫹2KJ 0 共 ␲ 兲 cos共 2k o ⌬ 兲
h␯ which we denote by ␦ min . Let us assume the worst-case
scenario for the mean path-length difference. Figure 2 in-
⫺4Kk o ␦ 共 t 兲 J o 共 ␲ 兲 sin共 2k o ⌬ 兲兴 , 共10兲
dicates that minima in the SNR occur at ⌬ equal to an odd
multiple of ␭/4. If the laser power, P l , is 1 mW, the wave-
where the entire resultant field of Eq. 共4兲 contributes to the
quantum noise. length of light, ␭, is 0.6328 ␮m, the detector quantum ef-
The SNR is found by substituting Eqs. 共8兲 and 共10兲 into ficiency, ␩, is 45%, and the detection electronics band-
Eq. 共5兲, which yields the following expression: width, ⌬␯, is 1 MHz, then the minimum detectable
displacement is ␦ min⫽5 pm.
SNR⫽ 冉 冊
CA r2 ␩
h␯⌬␯
1/2
k o␦共 t 兲 Acknowledgments

⫻冋 4K 2 ⫺4K 2 J 0 共 2 ␲ 兲 cos共 4k 0 ⌬ 兲
1⫹K 2 ⫹2KJ 0 共 ␲ 兲 cos共 2k o ⌬ 兲 ⫺4Kk o ␦ 共 t 兲 J 0 共 ␲ 兲 sin共 2k o ⌬ 兲 册 1/2
.
This material is based upon work supported by the U.S.
Army Research Laboratory and the U.S. Army Research
Office under grant number DAA655-98-1-0477.
共11兲
References
If we use a 50/50 beam splitter and assume a lossless op- 1. A. Dandridge, A. B. Tveten, and T. G. Giallorenzi, ‘‘Homodyne de-
tical system, then the field contrast factor is unity and the modulation scheme for fiber optic sensors using phase generated car-
SNR becomes rier,’’ IEEE J. Quantum Electron. QE-18共10兲, 1647–1653 共Oct.

冉 冊
1982兲.
P l ␩ 1/2 2. P. G. Davis, I. J. Bush, and G. S. Maurer, ‘‘Fiber optic displacement
SNR⫽ k o␦共 t 兲 sensor,’’ Proc. SPIE 3489, 18 –22 共Sep. 1998兲.
h ␯ •⌬ ␯ 3. J. W. Wagner and J. B. Spicer, ‘‘Theoretical noise-limited sensitivity

冋 册
of classical interferometry,’’ J. Opt. Soc. Am. B 4共8兲, 1316 –1326
1⫺J 0 共 2 ␲ 兲 cos共 4k o ⌬ 兲 1/2
共Aug. 1987兲.
⫻ , 4. A. Yariv, ‘‘Noise in optical detection and generation,’’ Chap. 10 in
2⫹2J 0 共 ␲ 兲 cos共 2k o ⌬ 兲 ⫺4k o ␦ 共 t 兲 J 0 共 ␲ 兲 sin共 2k o ⌬ 兲 Optical Electronics, 3rd ed., pp. 315–317, Holt, Rinehart, and Win-
共12兲 ston, New York 共1985兲.

Optical Engineering, Vol. 41 No. 2, February 2002 277

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 05/18/2015 Terms of Use: http://spiedl.org/terms