† Corresponding author. E-mail:
Project supported by the World Academy of Science (TWAS) research grant 2018 (Ref: 18-121 RG/PHYS/AS I-FR3240303643) and North South University (NSU), Bangladesh, internal research grant 2018-19 & 2019-20 (approved by the members of BOT, NSU, Bangladesh).
Considering the inhomogeneous or heterogeneous background, we have demonstrated that if the background and the half-immersed object are both non-absorbing, the transferred photon momentum to the pulled object can be considered as the one of Minkowski exactly at the interface. In contrast, the presence of loss inside matter, either in the half-immersed object or in the background, causes optical pushing of the object. Our analysis suggests that for half-immersed plasmonic or lossy dielectric, the transferred momentum of photon can mathematically be modeled as the type of Minkowski and also of Abraham. However, according to a final critical analysis, the idea of Abraham momentum transfer has been rejected. Hence, an obvious question arises: whence the Abraham momentum? It is demonstrated that though the transferred momentum to a half-immersed Mie object (lossy or lossless) can better be considered as the Minkowski momentum, Lorentz force analysis suggests that the momentum of a photon traveling through the continuous background, however, can be modeled as the type of Abraham. Finally, as an interesting sidewalk, a machine learning based system has been developed to predict the time-averaged force within a very short time avoiding time-consuming full wave simulation.
In 1908, physicist Hermann Minkowski proposed that the total momentum of an electromagnetic field inside any matter is equal to ∫
One of the recently popularized ideas in the area of photon momentum transfer or optical force is known as tractor beam.[9–11] A tractor beam is a customized light beam that exerts a counter intuitive negative force on a scatterer,[9–16] pulling it opposite to the propagation direction of light, in contrast to the conventional pushing force. Tractor beam experiments, which involve the material background,[12–14] can also be investigated in details to understand the persistently debated roles of different photon momenta such as those of the Abraham–Minkowski controversy.[3–5,17,18] In this article, we have made an attempt to investigate the Abraham–Minkowski controversy based on the tractor beam like effects.
We have investigated the light momentum transfer and related optical force on a scatterer (both absorbing and non-absorbing) floating on an interface of two or three different media (both absorbing and non-absorbing). Prior to this current work, for the interfacial tractor beam experiment, detailed calculations by ray tracing method and stress tensor equations showed that the pulling force is natural in an air–water scheme due to the linear increase of photon momentum in the infinitely long water medium.[12,14] Interestingly, non-Minkowski formulations[14] show a pushing force contradicting the experimental observation in Ref. [12]. However, the suggested interpretation in favor of Minkowski photon momentum for the interfacial tractor beam experiment has been questioned in Ref. [19]. Two different recent experiments have supported the observation of Abraham photon momentum for the air–water interface.[6,7] An old experiment has suggested the possible observation of Abraham force density,[20] which is also a partial support to the observation of the associated electromagnetic momentum density [
Considering the inhomogeneous or heterogeneous background, we have theoretically demonstrated that if the background and the half-immersed object are both non-absorbing, the transferred photon momentum to the object can be considered as the one of Minkowski exactly at the interface in time averaged scenario. Remarkably, we have shown that even if the background’s width is only a few nano-meter (extremely small in comparison with the object), the half-immersed object experiences an optical pulling force. Several intuitive thought experiments have also been put forwarded to establish this fact. An extremely small gap between the non-absorbing object and the non-absorbing background can fully nullify the increase of the transferred photon momentum at the interface of the object and the gap, which again supports in favor of our previous proposal (the change of momentum of photon exactly at the object and the touching background). Notably, these proposals have been verified based on both full wave simulations and analytical calculations.
In contrast, the presence of loss inside matter, either in the half-immersed object (i.e., plasmonic or lossy dielectric) or in the background, causes optical pushing of the object. Our analytical analysis suggests that for half-immersed plasmonic or lossy dielectric, the transferred momentum of photon can mathematically be modeled as the type of Minkowski and also of Abraham. However, according to a final critical analysis, we have found that the idea of Abraham momentum transfer of photon may lead to some notable problems and hence, it should better be rejected for the case of half-immersed absorbing object.
As a result, considering our analysis in this article (and also several previous works), it appears that the transferred photon momentum to a half-immersed object (absorbing or non-absorbing) or a fully immersed/embedded object[5] (or even to the free carriers[4] inside an absorbing object) can better be modeled as the one of Minkowski (where Minkowski momentum may arise exactly at the interface of two distinct media). Then a question arises: whence the Abraham momentum?
We have demonstrated that though the transferred momentum to a half-immersed Mie object (either lossy or lossless) can better be considered as the Minkowski momentum, Lorentz force analysis along with our full wave simulation results suggests that the momentum of a photon traveling through a host dielectric (i.e., in a continuous background), however, can be considered as the type of Abraham momentum. Interestingly, this conclusion also supports the previous conclusion drawn in a notable work[4] for the lossy semiconductor.
Finally, a machine learning algorithm has been applied to predict the time averaged force on half-immersed objects spending very short simulation time (instead of time-consuming full wave simulations). Apart from our motivation of getting a quick prediction of the force type without a time-consuming full wave simulation, this artificial intelligence-based investigation may open up a novel research avenue by providing us with yet another useful tool/approach to investigate other problems related to optical force and photon momentum inside a matter.
Throughout this paper, we refer to exterior (interior) or outside (inside) forces as those evaluated outside (inside) the volume of the macroscopic objects. We have done all the full wave electromagnetic simulations using COMSOL MULTI PHYSICS software[33] (and few of them have also been verified using Lumerical FDTD software[34]).
Two of the possible proposed set-ups are illustrated in Figs.
In the next few subsections, we will demonstrate three specific interesting facts regarding photon momentum inside matter based on both full-wave simulation and analytical analysis: (a) the change of linear momentum of photon exactly at the interface of two distinct media, (b) the change of photon momentum in presence of loss, and (c) the traveling momentum of photon inside a continuous medium.
Though the interfacial tractor beam experiment supports the linear increase of photon momentum (i.e., Minkowski momentum), the following question is still a matter of investigation: inside the matter, does the photon momentum always increases, and even if it increases, does it increase throughout the whole medium or just at the interface? In Ref. [43] it has been argued that the increase of linear photon momentum usually occurs at the interface of two different media due to the reduced impedance mismatch. But no conclusive proof is available regarding this very interesting argument given in Ref. [43]. Based on a few proposed thought experiments, we shall first check this proposal for the interfacial tractor beam set-up. However, it should be noted that although we shall show some strong evidence in favor of such proposal (increase of photon momentum specifically at the interface of two different medium[43]), it does not mean that the traveling momentum of photon decreases inside the continuous material medium. Our main focus in the next section is to judge whether the photon momentum really increases at the interface of two distinct media or not (instead of the long continuous material medium). Based on full wave simulations (both 2D and 3D simulations) in COMSOL Multiphysics (FEM method)[33] and Lumerical FDTD (FDTD method),[34] we have made the following three interesting observations.
Even if the refractive index of the lower long background is increased up to the value of 1.87 (one of the maximum values of index possible for liquid materials[44]), the force is always found to be a pushing force on the object due to the small gap based on the full wave simulations (both 2D and 3D) shown in Fig. s1 of the
But the notable fact is that even this pushing force still supports the initial proposal, i.e., the change of linear momentum of photon exactly at the interface of two distinct non-absorbing media and hence the time averaged pushing or pulling force based on the initial and final momenta of photon for the non-absorbing object. We are explaining this matter analytically in the next sub-section.
One well accepted theory to explain the time averaged total force on any immersed object is the theory of Minkowski on optical force and photon momentum. It is well known that Minkowski stress tensor[27–31,35–37] has a divergence free nature, which suggests that, if the Minkowski stress tensor is applied inside a non-absorbing object employing the internal field of the embedded object, it would lead to zero time-averaged total force. The connection between Eqs. (
From the non-diagonal (ND) components of the Minkowski stress tensors given in Eq. (
So, the difference between the internal Minkowski stress tensor and the external Minkowski stress tensor, exactly at the object and background interface, leads to the Helmholtz’s surface force. The time averaged pulling force observed in the interfacial tractor beam experiment can be explained/calculated solely based on this Helmholtz’s surface force given in Eq. (
One may point a fact that the Helmholtz surface force density also arises exactly at the interface of two distinct media for another well-known force density: the Abraham force density[17,20,25] (but not for other well-known force densities like Einstein–Laub force density[25] and Chu force density[25]). But the main focus of this article is to figure out the appropriate momentum of photon associated with the distinct theories of Abraham and Minkowski (or any other theory), instead of figuring out the correctness of Abraham and Minkowski (or any other) stress tensors or force densities. It is important to note that the electromagnetic momentum density of Abraham (and hence the Abraham momentum of photon) is also associated with two other optical force density formulations:[25] the Einstein–Laub force density and the Chu force density. In the next section and later sections, we shall demonstrate that at least for time averaged scenario, the appropriate version of the transferred momentum of photon appears as the one of Minkowski (exactly at the interface or surface regions) instead of the one proposed by Abraham. The above calculations in favor of Helmholtz surface force (which does not appear in Einstein–Laub or Chu force formulations[25]) are nothing but an additional support in favor of the force calculated (at the surface regions) by the corpuscular momentum of photon[45] in the next section (which will support the Minkowski momentum of photon instead of the one of Abraham).
We are now going to apply the idea of initial and final momenta of photon[45] for a simplified case to support our previous three observations by considering a slab object for which the first half is immersed in air but the second/lower half is immersed in water or a material medium.
Let us consider a lossless magneto-dielectric slab of sides at z = 0 and z = d, embedded (fully immersed) in a magneto-dielectric medium, illuminated at normal incidence by a linearly polarized plane wave propagating along the z direction with time harmonic dependence ei(kz − ωt). There is an interesting alternative way to calculate the time averaged force known as Lorentz force[46–48]
An important issue is to verify the agreement of the total Lorentz force with the external Minkowski ST. To simplify, let us consider
This last equation is the fundamental equation of the interfacial tractor beam (ITB)[12,14] concept. If the background of the input (light) interface of the slab is air and that of the output interface is water, then only within the Minkowski’s approach one will have
Based on this same approach, it can also be shown that if the rectangular slab (of length d) is fully placed in air just above the water medium (an extremely small/short air layer between the slab interface and the water interface), the time averaged force will be found positive instead of negative. Such an analysis will also explain the results given in Figs.
All these analytical (and full wave simulation based) observations are supporting the proposal that the interface of the two distinct media (and the entering and leaving momenta of photon at those interfaces) plays a vital role on the transfer of photon momentum. Notably, the linear increase of photon momentum [i.e.,
Although for the non-absorbing object and non-absorbing background, the transferred momentum of photon usually increases, this scenario changes when applied in the presence of absorption, either in the sub-merged object or in the embedding lower background. At first, we can check the case of half immersed absorbing object. If we replace a half immersed non-absorbing dielectric by a lossy dielectric or a plasmonic object, it is clearly observed that the pulling force fully vanishes as shown in Figs.
Now the first question is what actually happens in Fig.
It is well known that Minkowski stress tensor[37–41,46–48] has a divergence free nature, which suggests that, if the Minkowski stress tensor is applied inside a non-absorbing object employing the internal field of the embedded object, it would lead to zero time-averaged total force. When absorption is introduced inside the half-immersed object, the force inside the absorbing object has a non-zero value equal to Eq. (
The connection between previously discussed Eqs. (
To illustrate these issues by using a simple mathematical analysis, let us consider an absorbing dielectric rectangular slab half embedded in a non-absorbing liquid dielectric medium. The pushing force has been verified by full wave simulation for a half-immersed slab in Figs.
(i) The first way is the straightforward one: by employing the Abraham momentum of photon in the transmitted medium (the lower touching background), which directly suggests that the pushing force occurs because of Nipi + Nr pr > Nt pt (always),
(ii) The second way is not the straightforward one: by employing the Minkowski momentum of photon in the transmitted medium (the lower touching background), which does not directly suggest how the pushing force occurs following this equation:
At the very first look, it appears that: process (i), which suggests in favor of Abraham momentum of photon, may have no trouble. But this conclusion leads to two notable problems: (1) the mathematical form along with the behavior of an equation of physics [i.e., Eq. (
Hence, an obvious question arises: when the Abraham momentum of photon may appear. Is it a wrong momentum of photon? This issue is addressed in the next section.
According to Ref. [37], inside the absorbing dielectric slab, the conducting force of Eq. (
However, for our proposed set-ups, the doubt arises for the lower continuous background medium. The reason is obvious: though the transferred momentum from this lower (touching) background to the half-immersed object is always the one of Minkowski (according to our previous detail discussions), does the Lorentz force analysis lead to the Abraham momentum of photon (traveling one) for the lower continuous background? This is analyzed next based on a simple thought experiment.
We have considered 2D set-ups given in Figs.
Finally, we take an interesting sidewalk from our primary focus and investigate whether machine learning based data analytics can provide us an accurate prediction on the time averaged optical force for the more generic cases based on the discussed complex set-ups in this article. The motivation behind this rather unusual investigation is as follows. A lot of parameters actually contribute towards determining the optical force type (i.e., pulling vs. pushing) and a full wave simulation[33,34] takes long time. In contrast, a quick accurate result in this regard could be extremely useful for experimental physicists at least to make quick primary decisions in many situations. So, we have made an attempt to develop a machine learning based system, using WEKA workbench,[56] which can accurately predict the outcome extremely faster. However, the accuracy of this prediction is naturally dependent on the dataset used for training the system. The dataset has been prepared based on some full wave simulations and is provided in chart 1s in the
We have used the WEKA workbench[56] to develop the machine learning based classifiers and to conduct the computational experiments. The data given in the supplement s4 (chart 1s of the supplement) has been used to train a system that is able to predict the optical force type. In what follows, this data will be referred to as the training dataset. The varying parameters (i.e., features in the context of a machine learning model) used in the dataset are as follows: particle radius/lambda, particle loss, refractive index of the particle, background and short background, loss at long background and short background, and the short background width. Thus, chart 1s in the
For our classification task, we have used a number of classifier algorithms, namely, Naïve Bayes,[57–59] simple logistics,[57,59,60] J48,[57,61] random forest,[57,62] random tree,[57,62] and so on. Following the machine learning literature, we have employed the 10-fold cross validation scheme: the training dataset is randomly partitioned into 10 equal sized subsets, of the 10 subsets, a single subset is retained as the validation data for testing the model, and the remaining 9 subsets are used as training data. The cross-validation process is then repeated 10 times, with each of the 10 subsets used exactly once as the validation data. The 10 results can then be averaged to produce a single estimation. Table
So, given the values of the required parameters (i.e., features), our machine learning based predictor tool can be used to accurately and very quickly predict the force type. Apart from our motivation of getting a quick prediction of the force type without a time-consuming full wave simulation, this artificial intelligence-based investigation may open up a novel research avenue[63] by providing us with yet another useful tool/approach[63] to investigate other problems related to optical force[64] and photon momentum inside a matter.[65]
In this work, we have considered a distinct case — a particular configuration of light–matter interaction, i.e., an object (both non-absorbing and absorbing) is half immersed in an inhomogeneous or heterogeneous background, which can be very useful to conduct some very simple experiments to determine the transferred momentum of photon along with the traveling momentum of photon in a material medium. Considering our detail analysis in this work (and also several previous works), it appears that the transferred photon momentum to a half-immersed object (absorbing or non-absorbing) or a fully immersed/embedded object (or even to the free carriers inside an absorbing object) can better be modeled as the one of Minkowski (where Minkowski momentum may arise exactly at the interface of two distinct media according to our analysis). But at the same time, we have demonstrated that though the transferred’ momentum to a half-immersed Mie object (either lossy or lossless) can better be considered as the Minkowski momentum, optical Lorentz force analysis along with our full wave simulation results suggests that the momentum of a photon traveling through a host dielectric (i.e., in a continuous long background), however, can be considered as the type of Abraham momentum. Interestingly, this conclusion also supports the previous conclusion drawn in a notable work[4] for lossy semiconductor. Finally, based on several parameters, a machine learning based technique has been applied to quickly predict the time averaged total force, which may open up a novel research avenue by providing a useful tool/approach (i.e., artificial intelligence) to investigate other problems related to optical force and photon momentum inside a matter.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] | |
[25] | |
[26] | |
[27] | |
[28] | |
[29] | |
[30] | |
[31] | |
[32] | |
[33] | |
[34] | |
[35] | |
[36] | |
[37] | |
[38] | |
[39] | |
[40] | |
[41] | |
[42] | |
[43] | |
[44] | |
[45] | |
[46] | |
[47] | |
[48] | |
[49] | |
[50] | |
[51] | |
[52] | |
[53] | |
[54] | |
[55] | |
[56] | |
[57] | |
[58] | |
[59] | |
[60] | |
[61] | |
[62] | |
[63] | |
[64] | |
[65] |