Abstract
It is well known that in general the spectrum of a beam that is generated by a partially coherent source will change on propagation. Here we derive necessary and sufficient conditions under which the often-used Gaussian Schell-model sources can produce beams whose normalized spectrum is invariant everywhere, or is invariant just along the beam axis. These sources are not necessarily quasi-homogeneous or obeying the scaling law.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Due to the pioneering work of Wolf it is now generally appreciated that, in general, the spectrum of the field that is generated by a partially coherent source changes on propagation [1–4]. An exception are quasi-homogeneous sources [5, Sec. 5.3.2] that obey the scaling law [6–8]. These are planar, secondary sources that, at points ${\boldsymbol \rho }_1$ and ${\boldsymbol \rho }_2$ and frequency $\omega$, are characterized by a spectral degree of coherence $\mu ^{(0)} ({\boldsymbol \rho }_1,{\boldsymbol \rho }_2;\omega )$ that depends on ${\boldsymbol \rho }_1$ and ${\boldsymbol \rho }_2$ only through their difference, i.e.,
Apart from the researches we just mentioned, most studies do not explore the possibility of the source width or the spectral degree of coherence being frequency-dependent. Notable exceptions are [10–12] in which frequency-dependent source parameters are shown to cause significant spectral changes. Taking the opposite approach, we investigate if such a frequency dependence can lead to spectrally invariant beams. A related investigation was reported in [13]. There conditions were derived under which the spatial distribution of the spectral density remains invariant on propagation, apart from a transverse scale modification.
We report two new classes of Gaussian Schell-model (GSM) sources. One generates beams whose normalized spectrum is invariant everywhere, the other produces beams whose spectrum is invariant just along the propagation axis. These sources need not be of the quasi-homogeneous type, and they do not necessarily satisfy the scaling law. GSM sources are the workhorse of coherence theory [5, Sec. 5.4] and numerous studies have been dedicated to the fields that they generate. Because of their ubiquity, and their use as being representative of a general source, a better understanding of the spectral changes that they may or may not produce is useful. Furthermore, GSM beams are candidates for telecommunication applications because they are known to be more resistant to atmospheric turbulence than their fully coherent counterparts [14]. The GSM sources that we describe in this study would also display this robustness, but with the additional advantage of having a spectrum that remains invariant.
2. Gausian Schell-model sources
A scalar, planar, secondary GSM source (see Fig. 1) is described by a cross-spectral density function
The spectral density at an arbitrary point of observation $\textbf {r}= ({\boldsymbol \rho },z)$ in the half-space $z \ge 0$ follows from the cross-spectral density function through the relation
It is important to distinguish the spectral density and the normalized spectral density. The latter is defined asThe constant $C$ cannot be chosen arbitrarily. The left-hand side of Eq. (18), being positive, implies that
for all wavenumbers $k$ that are present in the source spectrum. Together with the beam condition (6) we thus find two constraints for $C^2$, namely It is possible to construct a GSM source that generates a beam with an invariant normalized spectrum not just on the axis, but in the entire half-space $z \ge 0$. This only happens when Eq. (18) is satisfied, and in addition the source width $\sigma ^2(k)$ does not depend on the wavenumber, i.e., when with $B$ a constant length. Only in that case, namely, the exponential in Eq. (13), and therefore the entire right-hand side of that equation, is independent of frequency. On substituting from Eq. (21) into Eq. (18) we obtain On using this result in Eq. (3) we thus find that a secondary, planar GSM source will generate a beam whose normalized spectrum everywhere in the half-space $z \ge 0$ is invariant if and only if its cross-spectral density function is of the formAs mentioned above, a source described by (23) does not necessarily satisfy the scaling law (2), and furthermore it need not be quasi-homogeneous. The latter can be seen as follows. The inequalities (19) must hold for all wavenumbers present in the source spectrum $S_0(k)$. If we denote the lowest of them by $k_\textrm {min}$, then
Let us now choose a value for $C^2$ such that with $k_0 < k_\textrm {min}$. We can then use Eq. (22) to plot $\delta (k)$ as a function of the wavenumber $k$. An example is shown in Fig. 2 for a band-limited source with $k_\textrm {min} < k < k_\textrm {max}$. It is seen that the coherence radius $\delta (k)$ and the source width $\sigma =B$ are comparable in magnitude across the entire spectral range. Therefore the source is clearly not quasi-homogeneous, but it nevertheless generates a beam that is spectrally invariant in the entire half space into which the source radiates.3. Spectral invariance only along the beam axis
We saw above that a necessary and sufficient condition for spectral invariance along the beam axis is given by Eq. (17). One possible alternative solution (not the most general) can be found by assuming that the effective source width and the transverse coherence length have a $k$ dependence that is of the form
respectively, with $a$ and $b$ two positive constants that are independent of frequency, and with the powers $\alpha$ and $\beta$ to be determined. On using these two Ansätze in Eq. (18) and collecting identical powers in $k$ it follows that $\Delta (z;k)$ will be $k$-independent if Thus we readily find that and hence with $a$ and $b$ two constant lengths related by Eq. (31). When the source width $\sigma$ and the coherence radius $\delta$ satisfy Eqs. (32) and (33) the expansion coefficient $\Delta (z;k)$ is again independent of frequency, ensuring that the normalized spectrum along the beam axis is invariant. On making use of Eq. (3) we thus conclude that a sufficient condition for a secondary, planar GSM source to generate a beam whose normalized spectrum everywhere on the propagation axis is the same is for its cross-spectral density function to be of the formIt is interesting to note that for the case of GSM sources the beam expansion factor $\Delta ^2$ is equivalent to the transverse scale factor $M$ that is used in [13] to discuss the so-called shape invariance of polychromatic fields. It can be shown that for such sources the concept of shape invariance is equivalent with spectral invariance on the beam axis.
The spectral density of the field at off-axis points is described by Eq. (13). Even when the expansion coefficient $\Delta ^2$ is independent of the wavenumber, the normalized spectrum at those points will not be invariant, due to the $k$-dependence of the product $\sigma ^2(k) \Delta ^2(z;k)$ in the exponential. The presence of this factor gives rise to a red-shifted spectrum. The magnitude of this shift can be remarkably small as we now show. As $z$ tends to infinity, the first term on the right-hand side of Eq. (10) may be be neglected, and the exponential in Eq. (13) becomes
4. Concluding remarks
Although in general the field that is generated by a Gausian Schell-model source will have a spectrum that changes on propagation, we have derived two conditions under which such a source generates a beam that is spectrally invariant. Both conditions prescribe a certain frequency dependence of the source width and its spatial correlation length. The first condition describes a source with a normalized spectral density that is the same in the entire half space into which the source radiates. The second condition describes a source that produces an invariant normalized spectral density along the beam axis. Such sources do not necessarily belong to the previously studied class of quasi-homogeneous sources that satisfy the scaling law.
Our results may be useful in applications where the advantages of partial coherence, such as reduced speckle and increased robustness with respect to atmospheric turbulence, are required but where the Wolf effect is detrimental.
Appendix 1
Consider a source that produces a field whose normalized spectral density, as defined by Eq. (15), is invariant along the beam axis. If we then equate $s(\textbf {r};k)$ at $\textbf {r} = (\textbf {0},0)$ and $\textbf {r} = (\textbf {0},z)$, we obtain
Funding
National Key Research and Development Program of China (2019YFA0705000); National Natural Science Foundation of China (11525418, 11904211, 11974218, 91750201); Innovation Group of Jinan (2018GXRC010); Air Force Office of Scientific Research (FA9550-16-1-0119).
Acknowledgments
We wish to thank an anonymous reviewer for bringing Ref. [13] to our attention. Funding was provided by the National Key Research and Development Program of China (2019YFA0705000); the National Natural Science Foundation of China (NSFC) (11904211, 11974218, 91750201, 11525418), and the Innovation Group of Jinan (2018GXRC010). T.D.V. acknowledges support by the Air Force Office of Scientific Research under Award No. FA9550-16-1-0119.
Disclosures
The authors declare no conflicts of interest.
References
1. E. Wolf, “Non-cosmological redshifts of spectral lines,” Nature 326(6111), 363–365 (1987). [CrossRef]
2. E. Wolf, “Redshifts and blueshifts of spectral lines caused by source correlations,” Opt. Commun. 62(1), 12–16 (1987). [CrossRef]
3. E. Wolf and D. F. V. James, “Correlation-induced spectral changes,” Rep. Prog. Phys. 59(6), 771–818 (1996). [CrossRef]
4. E. Wolf, “Spectral invariance and non-invariance of light generated by partially coherent sources,” Appl. Phys. B 60(2-3), 303–308 (1995). [CrossRef]
5. L. Mandel and E. Wolf, Optical Coherence and Quantum Optics (Cambridge University, 1995).
6. E. Wolf, “Invariance of the spectrum of light on propagation,” Phys. Rev. Lett. 56(13), 1370–1372 (1986). [CrossRef]
7. E. Wolf, “Two inverse problems in spectroscopy with partially coherent sources and the scaling law,” J. Mod. Opt. 39(1), 9–20 (1992). [CrossRef]
8. F. Gori, G. L. Marcopoli, and M. Santarsiero, “Spectrum invariance on paraxial propagation,” Opt. Commun. 81(1-2), 123–130 (1991). [CrossRef]
9. I. Moreno and C. C. Sun, “Modeling the radiation pattern of LEDs,” Opt. Express 16(3), 1808–1819 (2008). [CrossRef]
10. A. Gamliel and E. Wolf, “Spectral modulation by control of source correlations,” Opt. Commun. 65(2), 91–96 (1988). [CrossRef]
11. E. Shchepakina and O. Korotkova, “Spectral Gaussian Schell-model beams,” Opt. Lett. 38(13), 2233–2236 (2013). [CrossRef]
12. M. Luo and D. Zhao, “Propagation of electromagnetic spectral Gaussian Schell-model beams in atmosphere,” Opt. Commun. 336, 98–102 (2015). [CrossRef]
13. F. Gori and R. Grella, “Shape invariant propagation of polychromatic fields,” Opt. Commun. 49(3), 173–177 (1984). [CrossRef]
14. G. Gbur and E. Wolf, “Spreading of partially coherent beams in random media,” J. Opt. Soc. Am. A 19(8), 1592–1598 (2002). [CrossRef]
15. E. Wolf, Introduction to the Theory of Coherence and Polarization of Light (Cambridge University, 2007).