Abstract
For the measurement of Mueller matrix in an optical system with birefringence and small polarization-dependent loss or gain (PDL/G), we theoretically derive the statistical relationship between the Mueller matrix measurement error and three input states of polarization (SOP). Based on this theoretical relation and simulation results, it can be concluded that the three input SOPs, that are coplanar with an angle of 120° between any two of them in Stokes space, can be considered as a substitute for the best input SOPs which can statistically lead to the minimum measurement error. This conclusion is valid when the PDL/G of the optical system under test is less than 0.35dB.
©2009 Optical Society of America
1. Introduction
Birefringence and polarization-dependent loss or gain (PDL/G) are two fundamental polarization effects. In some optical systems, they may exist simultaneously. An example of such a system in the infrared band is an optical fiber communication link composed of single-mode optical fibers (SMF), passive and active optical components [1]. In the terahertz (THz) band, both polarization effects have also been found in certain materials. These polarization effects could be used as potential fingerprints in THz spectroscopy to characterize these materials [2]. In principle, an optical system, having both birefringence and PDL/G, can be completely depicted by a real Mueller-Jones matrix M, or equivalently, a complex Jones matrix J. Polarization properties of an optical system, such as PDL/G vector, polarization mode dispersion (PMD) vector and so on, can be extracted from M or J using appropriate algorithms [3–6]. Hence, the measurement of Mueller matrix or Jones matrix is of great importance in both basic and applied research. In this paper, we only consider the measurement of the Mueller matrix.
To date, several Mueller matrix measurement methods, such as dual rotating retarders polarimetry [7], null ellipsometry [8] and the Stokes methods [9], have been proposed. All these methods realize the measurement by setting some input states of polarization (SOP) and measuring the corresponding outputs. A potential technique for the fast infrared polarization modulation was also reported based on the high birefringence and low linear diattenuation of ferroelectric liquid crystals in some spectral bands [10]. In some applications based on Mueller matrix measurement, the smallest number of input SOPs should be used to take the measurements because a series of Mueller matrices are required to be measured in finite time. Two examples of such applications are PMD vector measurement [4] and polarization measurement in terahertz time-domain spectroscopy (THz-TDS) [11]. However, most of the above-mentioned methods are not good candidates in such applications.
Actually, it has been demonstrated that at least three input SOPs must be used to realize the Mueller matrix measurement when the system under test has both birefringence and PDL/G [12,13]. In 1947, R. C. Jones proposed a Jones matrix measurement approach using the three SOPs , and (“T” denotes the matrix transpose) [12]. We had also proposed a Mueller matrix measurement approach using the three input SOPs , and [13]. Theoretically, if the input and output SOPs are set and measured without any errors, the error-free Mueller matrix can be measured using three arbitrary input SOPs [13]. Unfortunately, errors arising from environmental perturbations, imperfect components and alignments, do exist in practice. Hence, the measured input and output SOPs definitely have errors. As a result, these errors will be transferred to the calculated Mueller matrix. The optimizations of polarimeters and noise influences on the SOP measurement have been reported in many papers [14–17]. Analysis of the Mueller matrix measurement error, induced by imperfect components and alignments, has also been presented [18,19]. On the other hand, for a pure birefringent system, we have demonstrated that the error of the Mueller matrix measurement is a function of two input SOPs and the statistically minimum error can be achieved when two input SOPs are orthogonal in Stokes space [20]. However, to the best of our knowledge, for a system with both birefringence and PDL/G, the relationship between the error of the calculated Mueller matrix and three input SOPs has not been presented until now.
A general analysis of this relationship, for an optical system with arbitrary PDL/G, is very complicated. In this paper, we present the theoretical and simulation results of this relationship in a simple case, where the system under test has a small PDL/G and the measurement errors of the output SOPs are far larger than those of the input SOPs. Some optical systems satisfy the first precondition. For example, an optical fiber system, composed of SMF, optical isolators and optical couplers, usually has a small PDL. Some measurement setups may approximately satisfy the second precondition. For example, when the input SOPs are generated by precisely rotating a high-quality polarizer and the output SOPs are measured by a fiber-type polarimeter, the measurement errors of the input SOPs must be less than those of the output SOPs.
In this paper, under the two above-mentioned preconditions, an approximate upper limit of the mean of the Mueller matrix measurement error is derived. This upper limit is determined by three angles among three input SOPs in Stokes space. It shows that the three best input SOPs should be coplanar in Stokes space and the angles between any two of them should be 120°. In fact, this upper limit is strictly valid only when there is no PDL/G. Generally, the Mueller matrix measurement error should also be related to the PDL/G vector. Thus, the best input SOPs should also depend on the PDL/G in the system. However, as long as the PDL/G is small enough, the three best input SOPs are very close to the relationship mentioned above. Therefore, if the system under test has a small PDL/G, we can use the inputs, and as a substitute for the best inputs.
This paper is organized as follows: firstly, some useful equations are deduced based on the properties of the Mueller matrix in Section 2; secondly, the equations governing Mueller matrix measurement using three arbitrary input SOPs and error propagations are derived in Section 3; the statistical properties of Mueller matrix measurement error are investigated in Section 4 and Section 5; finally, some simulation results are used to verify the theoretical finding.
2. Some properties of Mueller-Jones matrix
It has been demonstrated that the Mueller matrix of a system having both birefringence and PDL/G satisfies the Lorentz transformation [21,22]. Thus, we can express such a Mueller matrix in the form of a complex matrix
where and are the elements of the real Mueller matrix M. Based on this definition and the property of Lorentz transformation [22], we havewhere denotes the determinant of Mueller matrix and I stands for the identity matrix.It has been demonstrated that such a Mueller matrix can be decomposed as [23]
where is the polarization-independent loss or gain (PIDL/G); ; is the PDL/G vector and . PDL/G in dB can be calculated using [23].Sub-matrix is a orthogonal matrix, which can be expressed as [23]
where ϕ denotes the rotation angle in Stokes space; is a unit vector, standing for the rotation axis; is a dyadic; is a cross-product operator.Sub-matrix is a symmetric matrix, which can be written as [23]
where .From Eqs. (3), (4) and (5), the determinant and the Frobenius matrix norm of , which is defined as , can be calculated as
where “Tr” denotes the trace of a square matrix and “H” denotes the conjugate transpose. To be compatible with the complex Mueller matrix , a 4-dimensional complex Stokes vector can be defined as and the dot product of two vectors is defined as . Here, α is the angle between and . The input and output 4-dimensional complex Stokes vectors are linked by .From Eq. (3), we can easily obtain
where and is the angle between the PDL/G vector and input SOP in Stokes space.Another useful equation between input and output SOPs is [13]. Based on this equation, we have
3. Measurement approach and measurement error
In an experimental setup [13], three input SOPs , , and three output SOPs , , can be measured. Then, the complex Mueller matrix can be calculated using the following two equations [13]
andwhere and .Elements in and are defined as , , and .Firstly, is calculated using Eq. (9). Secondly, can be calculated by substituting in Eq. (10).When the measured output SOPs have errors , and , the exact form of Eq. (9) becomes
The measurement error can be calculated asPlease note that we ignore all high-order error terms in this paper.Similarly, when errors exist, the exact form of Eq. (10) should be
and the Frobenius matrix norm of the Mueller matrix error iswhere denotes the inverse matrix of F. Equations (12) and (14) are the starting point of the theoretical analysis.For three input SOPs, α, β and γ are the angles between and , and , and in Stokes space, respectively. Please note that we have ignored the subscript “in” for the three angles. These angles are bounded by and . In fact, these angles are not independent. They should satisfy . Then we have the following relations
Then, can only take values on the surface and the inner space of a tetrahedron enclosed by four planes, ,, as shown in Fig. 1 .4. Statistical properties of
From Eq. (12), the smallest may be achieved corresponding to different input SOPs in different tests because the measurement errors of the output SOPs vary. Therefore, the relationship between the statistical parameter of and input SOPs should be considered. Since a good polarimeter should have a completely random measurement error with zero mean, then . stands for the mean of a random variable. Obviously, . Thus, we need to use the variance to evaluate the uncertainty. A smaller variance means the larger possibility of of having a smaller measurement error. In this paper, we assume all SOP measurement errors follow the Gaussian distributions, that is, . Then, can be calculated as
Obviously, this variance depends not only on the relative relationship among three input SOPs, namely α, β and γ, but also on the relative relationship between three input SOPs and the PDL/G vector, namely , and . It also depends on the magnitudes of the PDL/G and the PIDL/G. When the PDL/G is small, that is, and are tenable. Thenwhere .Equation (17) means that is completely determined by α, β and γ if is not taken into consideration. From Eq. (17), it is easy to know that is the smallest when takes its minimum value. Actually, all , with the same , form a curved surface as shown in Fig. 2 . From Fig. 2, has the minimum value when , which means three input SOPs are coplanar and have angles of 120° between any two of them in Stokes space.
To illustrate more clearly, two special cases are plotted in Fig. 3 : (a) and (b) . In Fig. 3(a), “orthogonal inputs” is equivalent to. In Fig. 3(b), “Jones inputs” is equivalent to .
5. Upper limit of
When , the matrix in Eq. (5) is almost equal to I. Then, the complex Mueller matrix is close to . Since [23], from Eq. (14), we have
whereIn Eq. (19), and . By substituting Eq. (19) into Eq. (18), we havewhere . Obviously, depends on the actual noise realization in a single test. Similarly, the statistical parameter should be used to characterize its uncertainty. Unfortunately, we cannot directly calculate the mean of because of the difficulties in mathematics. As an alternative, we can calculate its upper limit asUnder the condition that, we can calculate the four terms in Eq. (21) as In Eqs. (22), (23) and (24), are the elements of the matrix F, which arewhereIn Eq. (23), the terms of are given byIn Eq. (24), the terms of are given byFinally, we havewhereApparently, this upper limit is completely determined by α, β and γ. From the calculation, we can find that has its minimum value also when . All , with the same , form a curved surface as shown in Fig. 4 .To illustrate more clearly, two special cases are plotted in Fig. 5 : (a) and (b) . Strictly speaking, the upper limit shown in Eq. (29) is valid only when there is no PDL/G (). However, when , it still approximately governs the statistical relationship between three input SOPs and the Mueller matrix measurement error .
6. Simulation results
To verify the theoretical finding in Section 5, simulations are performed. The parameters of the system under simulation are 1) Birefringence:, ; 2) PDL/G: , varies on the whole Poincaré sphere and 3) PIDL/G: . In Section 5, theoretical result shows that the upper limit does not depend on the PIDL/G . This is because we assume and the variance is not a function of . Then, we take in the following simulations. Further, we take the first two input SOPs asand . In this paper, we only show the simulation results in the same two special cases as those in Section 4 and Section 5: (a) and (b) .
In (a),
and in (b),In the simulations, is calculated using 1000 independent noise realizations with . For three 4-dimensional Stokes vectors, this means that 12000 random values have been generated. In Fig. 6 , simulation results, with and different values of D, show that the theoretical upper limit (ULimit) is valid when (0.35dB). When D is less than 0.35 dB, can lead to a measurement error that is very close to the minimum. In fact, the real minimum measurement errors, corresponding to different values of D, have been shown in Fig. 6(a). However, to achieve these minimums, the PDL/G vectors must be known before the measurement. If the PDL/G vector is unknown, can be considered as the substitute for the best inputs when.
Next, we need to confirm the validity of the above conclusion for all when . In Fig. 7 , and varies on the whole Poincaré sphere. Results show that can be considered as the substitute for the best inputs whatever is. Based on these results, we can use, and as the standard input SOPs in the measurements. From the results shown in Fig. 6 and Fig. 7, these standard input SOPs can result in obviously better measurement accuracy than Jones inputs and orthogonal inputs.
7. Conclusion
We presented the statistical relationship between three input SOPs and the Mueller matrix measurement error in a system having small PDL/G. This statistical relationship is expressed as an upper limit of the measurement error, which is approximately valid when the PDL/G is small. Based on this upper limit, the minimum Mueller matrix measurement error will be statistically achieved when three input SOPs are coplanar with an angle of 120° between any two of them in Stokes space. From the simulations, these input SOPs are confirmed to lead to a measurement error that is very close to the minimum when the PDL/G of the system under test is less than 0.35 dB. The standard input SOPs , and can be suggested since they have obviously larger probability to result in better Mueller matrix measurement accuracy than Jones inputs and orthogonal inputs.
Further, if the PDL/G is not very small, three optimum input SOPs will depends on the PDL/G vector. They will be not coplanar any more, but still equally separated. A detailed analysis will be presented in another paper.
Acknowledgements
This work is supported by Singapore A-star, Singapore Bioimaging Consortium, SBIC Grant Ref: SBIC RP C-014/2007.
References and Links
1. N. Gisin and B. Huttner, “Combined effects of polarization mode dispersion and polarization dependent losses in optical fibers,” Opt. Commun. 142(1-3), 119–125 (1997). [CrossRef]
2. J. Xu, J. Galan, G. Ramin, P. Savvidis, A. Scopatz, R. R. Birge, S. J. Allen, and K. Plaxco, “Terahertz circular dichroism spectroscopy of biomolecules,” Proc. SPIE 5268, 19–26 (2004). [CrossRef]
3. R. M. Craig, S. L. Gilbert, and P. D. Hale, “High-resolution, nonmechanical approach to polarization-dependent transmission measurements,” IEEE J. Lightwave Technol. 16(7), 1285–1294 (1998). [CrossRef]
4. H. Dong, P. Shum, M. Yan, J. Q. Zhou, G. X. Ning, Y. D. Gong, and C. Q. Wu, “Generalized Mueller matrix method for polarization mode dispersion measurement in a system with polarization-dependent loss or gain,” Opt. Express 14(12), 5067–5072 (2006), http://www.opticsinfobase.org/abstract.cfm?URI=oe-14-12-5067. [CrossRef] [PubMed]
5. B. L. Heffner, “Deterministic, analytically complete measurement of polarization-dependent transmission though optical devices,” IEEE Photon. Technol. Lett. 4(5), 451–454 (1992). [CrossRef]
6. B. L. Heffner, “Automated measurement of polarization mode dispersion using Jones matrix eigenanalysis,” IEEE Photon. Technol. Lett. 4(9), 1066–1069 (1992). [CrossRef]
7. D. H. Goldstein, “Mueller matrix dual-rotating retarder polarimeter,” Appl. Opt. 31(31), 6676–6683 (1992). [CrossRef] [PubMed]
8. S.-M. F. Nee, “Depolarization and principal mueller matrix measured by null ellipsometry,” Appl. Opt. 40(28), 4933–4939 (2001). [CrossRef]
9. B. J. Howell, “Measurement of the polarization effect of an instrument using partially polarized light,” Appl. Opt. 18(6), 809–812 (1979). [CrossRef] [PubMed]
10. D. B. Chenault, R. A. Chipman, K. M. Johnson, and D. Doroski, “Infrared linear diattenuation and birefringence spectra of ferroelectric liquid crystals,” Opt. Lett. 17(6), 447–449 (1992). [CrossRef] [PubMed]
11. H. Dong, Y. D. Gong, and M. H. Hong, “Polarization state and Mueller matrix measurements in terahertz time domain spectroscopy,” Opt. Commun. (Accepted).
12. R. C. Jones, “A new calculus for the treatment of optical systems VI. Experimental determination of the matrix,” J. Opt. Soc. Am. 37(2), 110–112 (1947). [CrossRef]
13. H. Dong, P. Shum, M. Yan, J. Q. Zhou, G. X. Ning, Y. D. Gong, and C. Q. Wu, “Measurement of Mueller matrix for an optical fiber system with birefringence and polarization-dependent loss or gain,” Opt. Commun. 274(1), 116–123 (2007). [CrossRef]
14. A. Ambirajan and D. C. Look, “Optimum angles for a polarimeter: part I,” Opt. Eng. 34(6), 1651–1655 (1995). [CrossRef]
15. A. Ambirajan and D. C. Look, “Optimum angles for a polarimeter: part II,” Opt. Eng. 34(6), 1656–1658 (1995). [CrossRef]
16. J. S. Tyo, “Design of optimal polarimeters: maximization of signal-to-noise ratio and minimization of systematic error,” Appl. Opt. 41(4), 619–630 (2002). [CrossRef] [PubMed]
17. Y. Takakura and J. E. Ahmad, “Noise distribution of Mueller matrices retrieved with active rotating polarimeters,” Appl. Opt. 46(30), 7354–7364 (2007). [CrossRef] [PubMed]
18. S.-M. F. Nee, “Error analysis for Mueller matrix measurement,” J. Opt. Soc. Am. A 20(8), 1651–1657 (2003). [CrossRef]
19. D. H. Goldstein and R. A. Chipman, “Error analysis of a Mueller matrix polarimeter,” J. Opt. Soc. Am. A 7(4), 693–700 (1990). [CrossRef]
20. H. Dong and P. Shum, “Effect of input polarization states on the error of polarization measurement,” Opt. Eng. 47(6), 065007 (2008). [CrossRef]
21. R. Barakat, “Theory of the coherency matrix for light of arbitrary spectral bandwidth,” J. Opt. Soc. Am. 53(3), 317–323 (1963). [CrossRef]
22. R. Barakat, “Bilinear constraints between elements of the 4x4 Mueller-Jones transfer matrix of polarization theory,” Opt. Commun. 38(3), 159–161 (1981). [CrossRef]
23. S.-Y. Lu and R. A. Chipman; “Interpretation of Mueller matrices based on polar decomposition,” J. Opt. Soc. Am. A 13(5), 1106–1113 (1996). [CrossRef]