Expand this Topic clickable element to expand a topic
Skip to content
Optica Publishing Group

Interference suppression of light backscattering through oblique deposition of a layered reflecting coating: bi-layer on a substrate

Open Access Open Access

Abstract

We theoretically demonstrate the possibility of entirely suppressing light backscattering towards the incident beam from a rough bilayer fabricated by deposition at an oblique angle, such that the substrate relief is replicated at a certain angle to the substrate surface. The necessary conditions of the scattering suppression were formulated. Interference suppression was observed experimentally (633 nm wavelength, 45° incidence angle, −35° scattering angle) with a set of SiO2-on-Ta2O5 bi-layers fabricated at different deposition angles. We detected a suppression of the scattering from the bi-layer fabricated at the deposition angle of 7.7° by a factor of about 30 compared with that from the sample deposited at normal incidence. The observed value of the scattering suppression was limited by the background noise of the measurements.

© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement

1. Introduction

Radiation scattering by surface and interface roughness is one of the crucial issues in advanced optical systems operating in the infrared to X-ray region. As an illustration, we mention only two examples from quite different branches of optics. The first example is the mirror reflection of coherent and extremely high-power beams of X-ray free electron lasers (XFELs). The long-scale (low-frequency) roughness of the mirror surface creates undesirable speckles in the focal plane [1] resulting in the appearance of artifacts and can even make it impossible to correctly reconstruct the object structure from analysis of the diffraction pattern. The short-scale (high-frequency) roughness increases scattering into the mirror depth leading to a significant enhancement (by a factor of several times) of the absorbed power [2], and thus a proportional increase in the thermal deformation of the mirror, thereby spoiling the optical performance of the outgoing beam.

The second example is the high-quality interferometer used for gravitational wave detection (GWD) [3]. The long-scale roughness distorts the ideal eigenmode of the interferometer, while the small-scale roughness scatters light out of the interferometer, thus limiting the power buildup in the resonant cavity. Both factors result in imperfect interference, distortion in the cavity field, and thus a decrease in the sensitivity of gravitational wave detection.

Over the past few decades, considerable advances have been made to improve polishing and deposition technologies allowing to minimize scattering from mirrors. Today, the root-mean-squared (rms) roughness of substrates of different materials can be as small as 0.07–0.1 nm in the [0.5, 100 μm−1] range of spatial frequencies [4]. The RMS roughness of the mirror installed in a European XFEL beamline and fabricated with elastic emission machining technology [5] is only 0.84 nm over the whole measurable range of spatial frequencies spanning eight orders of magnitude from 1.1·10−6 to 100 μm−1, where the lowest frequency corresponds to the mirror length of 900 mm [2].

According to the results presented in [3], the peak-to-valley value (PVV) of the surface relief variations of one of the mirrors for the advanced GWD interferometer does not exceed 1.5 nm over the central mirror area with a diameter of about 140 mm. The author of [3] indicated that the PVV is one order of magnitude lower than that of the mirrors for the previous generation of GWD interferometers.

To date, essential progress has also been achieved in the deposition of multilayer coatings. As an example, we can mention the prototype of the 500-mm-long mirrors used for synchrotron beamlines. The silicon substrate was coated with an Ru/Si multilayer structure with a period of approximately 2.8 nm, and the total number of bi-layers was equal to 200 [6]. The rms roughness of the substrate before deposition was about 0.1 nm, as determined by atomic force microscopy (AFM). After deposition of the multilayer structure, the rms roughness was unchanged over the whole mirror surface area. This result confirmed the high quality of the multilayer deposition process and gives evidence of the conformal growth of interfacial roughness, at least in the measurable range of the spatial frequencies.

Nevertheless, the optical performance of even atomically smooth mirrors may not be high enough for the needs of modern optics. The best known example in visible optics is probably the problem of backscattering toward the incident beam in laser gyroscopes. Backscattering results in a strong coupling of counter-propagating gyro modes, and thus results in errors in the rotation rate measurements and even the total disappearance of the gyroscopic effect [7–9]. The sensitivity and accuracy of gyroscopes essentially depend on the backscattering intensity; thus, it should be as small as possible and tend towards zero in the ideal case.

If the current deposition technologies cannot provide a low enough backscattering intensity, it is necessary to analyze alternative approaches to scattering suppression. The fact is that the scattering intensity is determined by both statistical parameters of interfacial roughness and the stationary electric field distribution inside a multilayer structure. Therefore, by choosing the proper multilayer structure design leading to a shift of the field intensity maxima away from the rough interfaces, it is possible to reduce the total scattering, at least in the case of fully uncorrelated interfacial roughness [10].

At the same time, there are manifold justifications that the roughness of different interfaces is well-correlated. The comprehensive study of the roughness replication was performed in [11–13] as applied to single films and multilayer mirrors operating in the visible spectral range. The authors demonstrated that the substrate roughness was always replicated at each interface in the whole range of measurable spatial frequencies. Then a new possibility of scattering reduction arises; designing multilayer structures to provide destructive interference with coherent waves scattered from different correlated interfaces.

The possibility was firstly analyzed in [14], where the light scattering suppression in the specular direction as well as the scattering reducing in the whole back hemisphere was demonstrated theoretically and experimentally through a single quarter-wavelength or half-wavelength layer deposition on a substrate. While the paper was devoted to single layers, it initiated extensive studies of destructive interference in scattering from multilayer structures.

Similar idea was also proposed in [15] as applied to the scattering suppression in the specular direction in X-ray domain, when X-ray beam falls onto a substrate at extremely small grazing angle within the total external reflection region.

More recently, the idea suggested in [14] was applied to rough multilayer structures. Adding half-wavelength layer with properly chosen dielectric constant on a top of a multilayer structure was demonstrated in [11] to reduce scattering as the scattering amplitudes from the interfaces of the added layer were in phase opposition with those of the initial coating.

Generalization of this approach was realized in [16], where the destructive interference of the scattered waves was provided by adding Fabry–Perot cavity structures on top of the multilayer with conformal interfacial roughness. A 30% reduction in the total scattering from a high-reflectivity Ta2O5/SiO2 multilayer mirror was experimentally observed at normal incidence with 808-nm wavelength radiation.

The final goal of our long-term work is the development of highly reflective multilayer mirrors with suppressed backscattering in towards the incident beam rather than decreasing the total scattering. The problem is of extreme importance for laser gyroscope physics, while any attempts to solve it are unknown to us. As it is demonstrated in the next section, in the case of fully correlated (conformal) interfacial roughness, the only possible way to decrease the backscattering intensity toward the incident beam is to reduce the mirror reflectivity, which is unacceptable for laser gyros.

Therefore, we will analyze another approach to solve this problem, namely, the use of oblique deposition of materials resulting in the skewed replication of roughness within the multilayer [13,17–20]. In this case, the conditions of the destructive interference of the scattered waves are changed compared to the case of conformal roughness, so that the scattering pattern is asymmetric even at the normal incidence of the outgoing beam. The effect was firstly observed in the visible range [13] and, more recently, in the extreme ultraviolet region [20]. The effect of the skewed roughness was successfully used in [21] to decrease the total scattering of extreme ultraviolet radiation (13.5 nm wavelength, 10° incidence angle) from a Mo/Si multilayer mirror. The authors demonstrated a 70% reduction in the total scattering, and based on their data presented in [21] we can conclude that they also observed backscattering suppression towards the incident beam by a factor of about 5-7.

The present paper is only the first step in our study, and we limit our investigations to the analysis of scattering suppression from a single bi-layer with skewed roughness. The bi-layer can be considered as the simplest example of a layered structure demonstrating specific features of scattering due to oblique deposition. In Section 3, we demonstrate theoretically the possibility of entirely suppressing (down to zero) the backscattering in the direction towards the incident beam, if the development of uncorrelated roughness during multilayer structure growth is neglected. The necessary conditions for the complete suppression of backscattering towards the incident beam are also analyzed. The results of our first experimental observations of backscattering suppression are presented in Section 4, where, in addition, we discuss a number of factors that can adversely affect the scattering suppression. Finally, the main results obtained are summarized in Section 5.

2. Backscattering from conformal roughness

In this section, we demonstrate the fundamental problem preventing backscattering suppression in the direction toward the incident beam if the interfacial roughnesses are conformal (identical), and they are replicated perpendicular to the substrate surface during the deposition of a multilayer structure. We will show that in this case, the only possible way to decrease the backscattering intensity is to decrease the mirror reflectivity.

We will analyze the mostly general case of a random layered media with conformal irregularities. Suppose that ε0(z) and ε(r) is the dielectric permittivity of an ideal homogeneous layered medium and the same medium with random irregularities, respectively. Z-axis is directed into the medium depth, so thatε(r)1, ifz, andε(r)εsub, ifz+. An s-polarized plane wave (the electric field vector is perpendicular to the incidence plane) is assumed to fall onto a surface of inhomogeneous medium. Below we will analyze the case of s-to-s scattering only, what is of the prime interest for laser gyroscope physics and corresponds to the condition of our experiment described below. Notice that s-to-p scattering intensity is typically very small compared with the s-to-s one.

The angle-resolved scattering (ARS) is defined as the ratio of the radiation power dQs scattered into a small solid angle to the incident power Qi; this is expressed via the amplitude of the scattered wave A as

ARS(θs,φs,θi)=1QidQsdΩ=|A(θs,φs,θi)|2Scosθi.
where the scattered field in vacuum far from the sample is expressed as Escat(r)=Aexp(ikr)/r (k = 2π/λ is the wave number), S is the illuminated area on the sample surface, triangular brackets mean ensemble averaging, the incidence and scattering angles θi[0,π/2] and θs[0,π/2] are measured from the Z-axis, and φs[0,2π] is the azimuth scattering angle measured from the incident plane XZ.

The scattering amplitude in the approximation of one-fold scattering (the first order of the Distorted Wave Born Approximation) has the following form (see, e.g., [22])

A(θs,φs,θi)=k24πei(q0q)ρΔε(r)E0(z,θi)E0(z,θs)d3r,Δε(r)=ε(r)ε0(z),q0=k(sinθi,0),q=k(sinθscosφs,sinθssinφs),ρ=(x,y),k=2π/λ.

Here, Δε(r) is a perturbation of the dielectric permittivity, q0 and q are the projections of the incident and diffracted wave vectors in the XY-plane, λ is the wavelength in vacuum, and E0(z,θ) is the amplitude of an unperturbed wave obeying the one-dimensional wave equation and conventional boundary conditions

E0(z,θ)+k2(ε0(z)sin2θ)E0(z,θ)=0,E0(z,θ)=eikzcosθ+r(θ)eikzcosθ,E0(z+,θ)=t(θ)eikzεsubsin2θ.
where r and t are the reflected and transmitted amplitudes, respectively.

We consider an inhomogeneous random medium with the specific dielectric permittivityε(r)=ε0(zζ(ρ)), where the random function ζ(ρ) describing the irregularities of the medium is independent of the z-coordinate. These irregularities are called conformal, because a surface of constant value of ε is determined by the same equation zζ(ρ)=const over the whole medium depth. Assuming the scattering from the medium irregularities to be low and expanding the perturbation of the dielectric permittivity as Δε(r)=ε(r)ε0(zζ(ρ))ζ(ρ)ε0(z), we obtain from Eqs. (1)-(2) that in frame of the first order perturbation theory, the ARS distribution is determined by the following equation

ARS(θs,φs,θi)=k4(4π)2cosθ0|E0(z,θi)E0(z,θs)ε0(z)dz|2PSD(ν),PSD(ν)=ζ(ρ)ζ(0)exp(2iπρν)d2ρ,2πν=qq0.
where PSD is the power spectral density of the medium irregularities.

Equation (4) is also valid for multilayer structures with step-like variation of the dielectric constant at interfaces. The appearance of delta functions in the expression for ε0(z) results in no problems, as the field functions E0(z) are continuous.

Next, we demonstrate that the integral contained in Eq. (4) can be calculated exactly for arbitrary ε0(z) in the case of backward and forward scattering, when θs=θi, while φs = π or 0, and E0(z,θ) is independent of φs. Integrating by parts, we obtain

S+E02ε0dz=limL1,2{(E02ε0)z=L1z=L2L1L2ε0dE02dzdz}.

Multiplying Eq. (3) bydE0/dz, we find that

ε0dE02dz=sin2θidE02dz1k2ddz(dE0dz)2.

Then we can integrate Eq. (5) for arbitrary ε0 and obtain

S=limL1,2{(ε0sin2θi)E02+1k2(dE0dz)2}z=L1z=L2,=1k2limL1,2{E0d2E0dz2+(dE0dz)2}z=L1z=L2.

Substituting an asymptotic representation from Eq. (3) for the field function, we find that S=4cos2θir(θi) and the final expression for the backscattering intensity (θs = θi, φs = π) is written as

ARS(θi,π,θi)=k4π2cos3θi|r(θi)|2PSD(2ksinθi).

An expression for the forward scattering intensity (θs = θi, φs = 0) is formally the same, if the argument of the PSD-function is set to zero, although we do not discuss the physical sense of the PSD-function at a zero spatial frequency here.

Equation (8) demonstrates that, in the case of conformal roughness, the backscattering intensity could be only reduced through a decrease in the reflectivity. Therefore, it is impossible to design a highly reflective multilayer mirror (periodic or aperiodic, having abrupt or smooth interfaces, consisting of two or more materials) providing even a slight decrease in the backscattering intensity, if the interfacial roughnesses are conformal.

Notice that our consideration is not in a contradiction with the results obtained in [14,15], where the total scattering suppression in the specular direction was predicted. The fact is that Eq. (8) has been deduced assuming the both interfaces roughness to be fully conformal (identical) in contrast to [14,15], where a deposited film was supposed to smooth substrate roughness, while the roughness of both interfaces was still correlated. The roughness smoothening was necessary to guarantee an equality of the amplitudes of waves scattered from different interfaces.

An example is presented in Fig. 1, where the calculated ARS in the incident plane is shown for two high-reflectivity Ta2O5/SiO2 multilayer mirrors operating at an incident angle of θi = 45° and a wavelength of 633 nm characterized by the same model PSD function. The first mirror (HR45) is a standard quarter-wavelength multilayer mirror, while the second one (LSHR45) is an aperiodic mirror designed to decrease the scattering intensity caused by the conformal roughness. The approach used for designing this mirror is described in [16]. While the scattering was significantly reduced for the LSHR45 mirror, the backward and forward scattering intensities (at|θs|=θi=45) are the same for both mirrors according to Eq. (8), because their reflectivity is almost unity.

 figure: Fig. 1

Fig. 1 The calculated ARS (at λ = 633 nm) in the incident plane (φs = 0) from two high-reflectivity Ta2O5/SiO2 multilayer mirrors: HR45 and LSHR45 designed to suppress scattering into the back hemisphere. Here and below the ARS was calculated with the computer program described in [23], Elson et al.

Download Full Size | PDF

The calculations performed above clearly demonstrate the complexity of the problem of backscattering suppression. In the next section, we demonstrate that the problem may be solved through the oblique deposition of a layered reflecting coating. Moreover, we show that backscattering can be fully suppressed if we neglect the effect of uncorrelated intrinsic film roughness arising during the deposition of a multilayer structure

3. Backscattering from the bi-layer fabricated with oblique deposition

In the present paper we will analyze the simplest case of a single bi-layer deposited onto a rough substrate at oblique incidence. A sketch of the bi-layer and its parameters are shown in Fig. 2. We will suppose that the particles strike the sample at the angles α1 and α2 during deposition, and the incident flux is parallel to the ZX plane. The substrate relief is transferred during film growth in this plane in the direction of angles β1 and β2. We will consider the general case of different growth angles β1 and β2 during deposition of the bi-layer materials. These angles are measured from the Z-axis in the clockwise direction, so that β2 > 0 and β1 < 0 in Fig. 2. The boundaries between the materials are enumerated from the sample top to the bottom.

 figure: Fig. 2

Fig. 2 Sketch of the bi-layer fabricated by oblique deposition of materials.

Download Full Size | PDF

For simplicity of analysis, we will neglect the surface relaxation during deposition as well as the development of intrinsic film roughness uncorrelated with the substrate relief. Then, a feature on an underlying surface is simply shifted along the X-axis by Δx=djtanβj; therefore, the interface relief is written as

ζ2(ρ)=ζsub(ρnxd2tanβ2),ζ1(ρ)=ζsub(ρnxd2tanβ2nxd1tanβ1).
where nx is the unit vector along the X-axis and ζsub(ρ) describes the substrate relief. The Fourier transform of the surface relief has the form
ζ2F(ν)=ζsubF(ν)eiη2,ζ1F(ν)=ζsubF(ν)ei(η1+η2),ηj2πνxdjtanβj.
whereζsubF(ν) is the Fourier harmonic of the substrate surface and ν=(νx,νy) is the 2D spatial frequency. We introduce the notation ηj to shorten the formulas.

We will analyze the partial case, when the incident plane of a light beam coincides with that of the particle flux, and will be interested in the scattering in the incident plane to perform our theoretical analysis. Therefore, we will not consider the azimuth scattering angle and will measure the incidence angle θi and the scattering angle θs from the Z-axis in anticlockwise and clockwise directions, respectively, so that they are positive for the scattering geometry G1 shown in Fig. 2 and negative for geometry G2. Then, assuming abrupt interfaces between layers, we write the scattering amplitude A and the ARS as

A(θs,θi)=k24π[(ε1ε0)E0(z1,θ0)E0(z1,θ)ei(η1+η2)+(ε2ε1)E0(z2,θ0)E0(z2,θ)eiη2+(ε3ε2)E0(z3,θ0)E0(z3,θ)]ζsubF(ν),ηj2π/λ(sinθisinθs)djtanβj.
ARS(θs,θi)=k4(4π)2cosθ0|(ε1ε0)E0(z1,θ0)E0(z1,θ)ei(η1+η2)+(ε2ε1)E0(z2,θ0)E0(z2,θ)eiη2+(ε3ε2)E0(z3,θ0)E0(z3,θ)|2PSDsub(ν).
where the functions E0(z,θ0) and E0(z,θ) are the solutions of the wave equation describing the wave field inside an ideal bi-layer, so that all manifold specular reflections of the scattered waves from the bi-layer interfaces are taken into account, while the re-scattering effects are neglected.

Next, assuming a nonzero field amplitude on the substrate surface, we factor out the third summand from the parenthesis in Eq. (11) for the scattering amplitude and obtain the following expression well suited for analysis

A(θs,θi)~1+ε2ε1ε3ε2E0(z2,θi)E0(z2,θs)E0(z3,θi)E0(z3,θs)Cexp(iγ2)eiη2+ε1ε0ε3ε2E0(z1,θi)E0(z1,θs)E0(z3,θi)E0(z3,θs)Bexp(iγ12)ei(η1+η2).

Our goal is to find the condition of total scattering suppression A = 0 in the direction θs and φs = 0. Then, Eq. (13) represents an elementary trigonometric equation

1+Сeiс+Beib=0.
where C > 0 and B > 0 are the modulus, and c=γ2+η2 and b=γ12+η1+η2 are the phases of the summands in Eq. (13).

Equation (14) has two solutions, if the following necessary condition is fulfilled

|1B|C1+B.

Assuming phases c and b in Eq. (10) lie in the [π,+π] interval, it is easy to check that the solutions are written as

cη2+γ2=±arccosB2C212C,bη1+η2+γ12=arccosC2B212B.

Condition (15) means that the modulus of the arc cosine function arguments in Eqs. (16) do not exceed unity.

Finally, using Eq. (11) for the parameters ηj, we find the growth angles β1 and β2; thus, the problem of total scattering suppression in the desired direction is solved. Some examples are given below.

In the model calculations, we set the radiation wavelength to λ = 633 nm and used Ta2O5 (refraction index n = 2.1739) and SiO2 (n = 1.4777) as the bi-layer materials as well as BK7 (n = 1.5148) as a substrate, radiation absorption being neglected. These materials were used in our first experiment, as described below. For definiteness, we apply the following model of the two-dimensional PSD function of an isotropic substrate for the theoretical analysis:

PSDsub(ν)=σ2ξ2hπ(1+ξ2ν2)1+h.
where the root-mean-squared (rms) roughness σ = 3 nm, the correlation length ξ = 100 μm, and the Hörst parameter h = 0.5. Notice that the conditions of the interference suppression of backscattering are independent of the type and parameters of the PSD function.

We set the total optical thickness D = D1 + D2 of the bi-layer to λ/2, while introducing the thickness ratio 0 < Γ < 1 according to

D1=d1ε1sin2θi=λ2Γ,D2=d2ε2sin2θi=λ2(1Γ).
where θi is the incidence angle and d1,2 is the geometrical layer thickness varying with Γ, i.e., the optical thickness of each layer is not equal to λ/4. Then the problem of backscattering suppression is characterized by the only free parameter Γ. Recall that material of thickness d1 is placed on the top of the bi-layer (see Fig. 2).

As a rule, we set below the incidence angle θi = 45° and analyze the possibility of fully suppressing the backscattering at the scattering angle θs=θi=45, which is of prime interest for laser gyroscopes.

We consider first the bi-layer SiO2/Ta2O5/sub with the SiO2 layer on the top. An area limited by green lines in Figs. 3(a) and 3(b) show the range of the parameters C and B in Eqs. (13)-(14), where the conditions of Eq. (15) for backscattering suppression are fulfilled. Varying the thickness ratio Γ results in a change of the field amplitudes at the interfaces, and thus in a movement of the point (C, B) along the curve EF in Fig. 3(a) from the point E at Γ = 0 to the point F at Γ = 0.5 and then back to the point E as Γ tends to 1. The curve EF lies entirely inside the area delimited by green lines, i.e., total backscattering suppression can be achieved at any thickness ratio Γ. Notice that the values of the parameters C and B are on the order of unity in the curve EF, i.e., the scattering amplitudes from different interfaces are close to each other.

 figure: Fig. 3

Fig. 3 Green lines limit the area of the parameters B and C, where the necessary conditions of backscattering suppression in Eq. (15) are fulfilled. Red curves EF show the change of these parameters with the variation of the thickness ratio Γ from zero to unity according to Eq. (18). Calculations were performed for (a) SiO2-on-Ta2O5 and (b) Ta2O5-on-SiO2 bi-layers deposited onto BK7 substrates. The radiation wavelength, incidence angle, and scattering angle corresponding to total scattering suppression are set to 633 nm, 45°, and −45°, respectively.

Download Full Size | PDF

Using Eq. (16), we find two possible solutions of Eq. (14), which are shown in Fig. 4(a) by curves 1 and 2 representing two continuous sets of the growth angles β1 and β2, providing total backscattering suppression. The straight dashed line corresponds to the case β1 = β2, which is the most suitable condition from the fabrication point of view. As can be seen, total suppression can be achieved at both negative (−10.11°) and positive ( + 12.10°) values of the growth angle β1 = β2.

 figure: Fig. 4

Fig. 4 Two solutions of Eq. (12) found with varying thickness ratio Γ are indicated by colored solid curves. The straight dashed line corresponds to the optimal practical case β1 = β2. Calculations were performed for (a) SiO2-on-Ta2O5 and (b) Ta2O5-on-SiO2 bi-layers deposited onto BK7 substrates. Radiation wavelength, incidence angle, and scattering angle corresponding to total scattering suppression are set to 633 nm, 45 degrees, and −45 degrees, respectively.

Download Full Size | PDF

The ARS from the SiO2/Ta2O5/sub bi-layer calculated in the incident plane is shown in Fig. 5(a) for the growth angles β1,2 = −10.11° (curve 2) and β1,2 = 12.10° (curve 3). The thickness ratios for these angles were found to be Γ = 0.2844 and 0.8037, respectively. The ARS from a quarter-wavelength bi-layer (Γ = 0.5) deposited at normal incidence (β = 0) is also shown for comparison (curve 1). Curves 2 and 3 demonstrate the total suppression of the backscattering achieved at the scattering angle θs = −45°. The reflectivity of the bi-layers was proven to be 6.4% and 8.3%, respectively, i.e., nonzero.

 figure: Fig. 5

Fig. 5 ARS in the incident plane (φs = 0) from the bi-layers with the (a) SiO2 or (b) Ta2O5 layer placed on the top. The radiation wavelength, incidence angle, and scattering angle corresponding to total scattering suppression in the incident plane (φs = 0) are set to 633 nm, 45°, and −45°, respectively. The value of the optimal thickness ratio Γ and the growth angle β are indicated in the graphs. Curve 1 shows ARS from a quarter-wavelength bi-layer deposited at normal incidence.

Download Full Size | PDF

In contrast to curve 3, curve 2 lies significantly (2–3 orders of magnitude) below curve 1 in a wide angular interval. Therefore, we would expect that different adverse factors (development of intrinsic film roughness uncorrelated with the substrate roughness or inaccuracies in the thickness of the deposited layers as well as in the growth angle β, which is poorly controlled in experiments) can keep the backscattering intensity low for curve 2, while the scattering suppression effect can disappear fully for curve 3. Thus, we can conclude that the solution with the negative growth angle is more preferable for experimental study.

The calculated two-dimensional ARS is shown in Figs. 6(a) and 6(b) for two SiO2-on-Ta2O5 bi-layers: (a) a quarter-wavelength bi-layer deposited at normal incidence and (b) an obliquely (β = −10.11°) deposited bi-layer. The corresponding ARS curves in the incident plane are shown in Fig. 5(a) as curves 1 and 2, respectively. Figure 6(b) demonstrates that there is a rather large solid angle where the scattering intensity is 2–3 orders of magnitude lower compared to Fig. 6(a).

 figure: Fig. 6

Fig. 6 The calculated 2D ARS from SiO2/Ta2O5/sub bi-layers. The scattering intensity is presented on a logarithmic scale. Calculations were performed for a quarter-wavelength bi-layer grown perpendicular to the substrate surface (β = 0, Γ = 0.5) and a bi-layer deposited at an oblique incident angle (β = −10.11°, Γ = 0.2844).

Download Full Size | PDF

For completeness of analysis, we consider now Ta2O5-on-SiO2 bi-layers with the Ta2O5 layer on the top. As above, the bi-layer is deposited onto the BK7 substrate. Figure 3(b) demonstrates that the parameters B and C are large, i.e., the partial scattering amplitude from the substrate is much lower compared to the other two, because the variation of the dielectric constant at the SiO2-BK7 interface is small. As a result, there is a very narrow interval of Γ values where total backscattering suppression is possible. Moreover, Fig. 4(b) demonstrates that the growth angles providing suppression are large, which can be impractical. Note that only one solution is presented in Fig. 4(b), because the second solution requires significantly larger growth angles. Therefore, although the Ta2O5-on-SiO2 bi-layer can be designed to fully suppress backscattering with coincident growth angles of β1 = β2 = −34.46° (Fig. 5(b)), we conclude that the SiO2-on-Ta2O5 bi-layer characterized by closer values of partial scattering amplitudes is more preferable for experimental study.

In the first-order perturbation theory approximation, the total scattering amplitude is merely a sum of partial scattering amplitudes Aj from individual interfaces. Suppression of scattering means that the sum is equal to zero due to destructive interference of the scattered waves. As the scattering amplitudes can be considered as vectors on a complex plane, the zero sum signifies that these vectors form a closed triangle. Several examples are presented in Figs. 7(a)-7(c). As above, we consider the backscattering from a SiO2-on-Ta2O5 bi-layer at λ = 633 nm, θi = 45°, and θs = −45°.

 figure: Fig. 7

Fig. 7 The partial scattering amplitudes on a complex plane. Calculations were performed for SiO2-on-Ta2O5 bi-layers at λ = 633 nm, θi = 45° and θs = −45°. Three cases are considered: (a) bi-layer deposited at normal incidence (β = 0), with the thickness ratio Γ being equal to 0.5 or 0.6, and (b, c) two bi-layers deposited at oblique incidence with the parameters providing total backscattering suppression. The scattering patterns from the bi-layers are shown in Fig. 5(a).

Download Full Size | PDF

The case of bi-layers deposited at normal incidence (β = 0, Γ = 0.5 and 0.6) is demonstrated in Fig. 7(a). In this case, the sum of the partial amplitudes is nonzero, and the distance between the front end of the vector A1 and the origin characterizes the nonzero scattering intensity, shown in Figs. 5(a) and 5(b), curve 1. Notice that in the case of the quarter-wavelength layers (Γ = 0.5) the partial scattering amplitudes are real, while the nonzero distance between the end of A1 and the origin still persists.

Figures 7(b) and 7(c) show the behavior of the scattering amplitudes in the conditions of total backscattering suppression, which corresponds to the zero minimum in curves 2 and 3 in Figs. 5(a) and 5(b). Indeed, the partial scattering amplitudes form a closed triangle in these cases. Equation (15) thus has a simple geometrical interpretation and means that three vectors with lengths of 1, A1/A3, and A2/A3 form a triangle if Eqs. (16) are fulfilled. The last equations determine the angles between neighboring vectors necessary for the formation of a closed triangle and represent nothing but the well-known “law of cosines” for triangles.

Equations (9)-(12) allow us to design bi-layers providing total scattering suppression at any desired scattering angle. Curves 2 to 6 in Fig. 8 demonstrate this statement for negative scattering angles β, when the scattered wave propagates in the negative direction along the Z-axis in Fig. 2. For comparison, ARS from a quarter-wavelength bi-layer deposited at normal incidence is also presented (curve 1). However, the β value (in modulus) increases quickly if we want to suppress scattering in the positive direction along the X-axis. In particular, the modulus of the angle β increases to 34° if we want to suppress scattering at θs = + 20°, and it tends (in modulus) up towards an unphysical value of 90° as θs tends toward 45°, i.e., to the direction of specular reflection.

 figure: Fig. 8

Fig. 8 ARS (λ = 633 nm) in the incident plane (φs = 0) from the SiO2-on-Ta2O5 bi-layers designed to suppress scattering at different scattering angles β from 0 to −60° (curves 2 to 5). Incidence angle was set to 45° for all curves. The value of the optical thickness ratio Γ and the growth angle β are indicated in the graph. Curve 1 shows the ARS from a quarter-wavelength bi-layer deposited at normal incidence.

Download Full Size | PDF

Until now, we have analyzed the conditions of total backscattering suppression for geometry G1 in Fig. 2, in which a wave is incident onto the bi-layer from the left side. Evidently, the effect of interference suppression is useful for applications in laser gyroscopes only if it is observed at once for waves incident from both the left and the right side. For the bi-layers designed above and characterized by coincident growth angles β1 = β2, simultaneous suppression of scattering from counter-propagating waves is impossible. Figure 9(a) is an illustration of this statement. Curve G1 represents the ARS from one of the analyzed bi-layers (curve 2 in Figs. 5(a) and 5(b)) for a wave incident on the bi-layer from the left at an incidence angle θi = 45° (geometry G1 in Fig. 2). However, the effect of interference suppression disappears for a wave incident onto the bi-layer from the right side at θi = −45° (curve G2 in Fig. 9(a)). The question arises: is it possible to design a bi-layer to suppress backscattering from both counter-propagating waves at once?

 figure: Fig. 9

Fig. 9 ARS (λ = 633 nm) in the incident plane (φs = 0) from SiO2-on-Ta2O5 bi-layers for the two incident geometries indicated in Fig. 2. The beam is incident on the bi-layer from the left side (G1, θi = 45°) or right side (G2, θi = −45°). The bi-layer in (a) was designed to suppress scattering at θs = -θi = −45° for a wave incident from the left side, while the bi-layer in (b) suppresses scattering at θs = -θi for both counter-propagating waves at once. The value of the optimal thickness ratio Γ and the growth angles β are indicated in the graphs.

Download Full Size | PDF

Notice that the radiation diffraction in geometry G2 in Fig. 2 is equivalent to that in geometry G1 if we change the signs of both growth angles β1 and β2. The problem can thus be formulated in the following manner: it is necessary to design two bi-layers with the same layer thicknesses, which fully suppress backscattering (θs = -θi) at the same incidence angle θi, with the first bi-layer being characterized by the growth angles β1 and β2, while the second one is characterized by the growth angles -β1 and -β2. This means that two sets of growth angles (β1, β2) and (-β1, -β2) should obey Eq. (14) with the same values of the parameters B, C, γ1, and γ12, because the unperturbed field E0(z, θ) is the same for both geometries. Evidently, this can occur only if the phases γ1 = γ12 = 0. Indeed, in this case, the two solutions (16) of Eq. (14) have the following form

η2=2πνxd2tanβ2=±arccosB2C212Cη1=2πνxd1tanβ1=[arccosC2B212B+arccosB2C212C]
and thus, if the pair of growth angles (β1, β2) obeys Eq. (19), the second pair (-β1, -β2) obeys Eq. (19) as well.

The condition γ1 = γ12 = 0 means that the ratios of the field amplitudes (or, equivalently, the partial scattering amplitudes) at different interfaces are real numbers (Eq. (13)). The simplest way to obey this condition is to use a quarter-wavelength bi-layer (Γ = 0.5) providing real-valued partial scattering amplitudes, as was demonstrated in Fig. 7(a). Then, we find the values of β1 = 27.24° and β2 = −48.72° from Eq. (19), realizing zero backscattering intensity for both counter-propagating waves. The calculated ARS for both incidence geometries are shown in Fig. 9(b).

The determined values of the growth angles are probably impractical, but we do not perform here more comprehensive analysis of how to fulfill the condition γ1 = γ12 = 0, because a bi-layer cannot be considered as a high-reflectivity mirror for laser gyroscopes. The main result of our consideration is only a demonstration of the possibility of fully suppressing backscattering from counter-propagating waves even with the use of the simplest bi-layer mirror. This statement is of importance for the subsequent analysis of high-reflectivity multilayer mirrors for laser gyroscopes, which will be described in the next paper.

At the end of this section, we note that backscattering suppression can be observed even for a single film deposited obliquely on a substrate. Indeed, if we set ε1 = ε2 = εf, which corresponds to a single film, we obtain C = 0 in Eq. (14), and the conditions of scattering suppression can be written as b = π and B = 1. The first condition unambiguously determines the growth angle β providing destructive interference of the scattered waves, while the second one requires the equality of their amplitudes. The second condition can be written as |1+r|=|t|, where r and t are the reflected and transmitted amplitudes, respectively. The last equation allows us to uniquely determine the necessary film thickness, while it is possible, if only some limitations are imposed on the values of the film and substrate dielectric constants. More detailed analysis of this case is not within the scope of this paper. For our goal, the case of a bi-layer is more interesting, because backscattering suppression can be achieved for different layer thicknesses and different growth angles, which can open new opportunities in the design of reflecting coatings. In particular, it is possible to suppress backscattering from counter-propagating waves. In addition, the conditions of backscattering suppression from high-reflectivity multilayer mirrors have the same form as Eq. (14), and their analysis, which will be performed in the next paper, is very similar to that performed above.

4. Experiment

In this section, we discuss the results of the first our experiments on the observation of interference suppression of backscattering through the oblique deposition of bi-layers, considered as the simplest example of a layered reflecting coating. First, we note the following circumstances, which should be taken into account when performing experiments.

We cannot measure the backscattered flux directed toward the incident beam (i.e. at φs = -φi), because the detector blocks the incident light within an angular interval of ± 2°. Therefore, we studied a bi-layer designed for total scattering suppression at the scattering angle φs = −35°, while the incidence angle was set to φi = 45°.

We fabricated the simplest sample holder allowing us to incline substrate at an angle α to the horizontal plane for oblique deposition of layers, while the current holder design limits the deposition angle to α < 20° and provides no way to change α during the deposition of separate layers. If we assume the validity of the so-called “tangent rule” tanα=2tanβ establishing the relation between the deposition angle α and the growth angle β, and make use of the columnar film growth model [18], we can conclude that the growth angle should be less than 10° in our experiments. However, as the authors of [18] indicated, the last equation is an approximate one even for the specific model used. Next, the authors of [20] based on the alternative model of the uniform film growth, argued that the growth angle coincides with the deposition angle, i.e., α = β. As the relation between these angles is not evident yet, it is desirable to design a bi-layer assuming the growth angle β to be significantly less than 10° to have freedom in choosing the deposition angle α. Notice that we were unable to determine the growth angle experimentally with the use of transmission electron microscopy, because the small value of β prevented visualization of the lateral interface shift with film growth.

Based on this consideration and theoretical analysis performed in the previous section, we designed a SiO2 (d1 = 79.20 nm)–on–Ta2O5 (d2 = 112.94 nm) bi-layer on a BK7 substrate assuming the same growth angle β = −5.41° for both materials. The bi-layer fully suppresses backscattering at the scattering angle θS = −35° if the incidence angle θi = 45° and the radiation wavelength λ = 633 nm. To decrease the growth angle β we chose the optical thickness of the bi-layer to be D=0.53λ exceeding slightly the value of D = 0.5λ used in Section 3 for theoretical analysis.

The bi-layers were deposited using ion beam sputtering equipment (SPECTOR-HT produced by VECCO company) onto BK7 substrates with a thickness of 1 mm and diameter of 30 mm. The deposition rates of the Ta2O5 and SiO2 layers were about 0.18 nm/s and 0.20 nm/s, respectively. The layer thickness was controlled by a broadband optical monitoring system. The accuracy of the thickness control was within 0.5%.

The surface topography of the primary substrates and the deposited coatings was analyzed with atomic force microscopy. AFM images of the substrate surface and the bi-layer fabricated with oblique deposition (the deposition angle α = 10.7°) are shown in Fig. 10(a) and Fig. 10(b), respectively. The root-mean-squared (rms) roughness of the substrate and bi-layer surfaces (σ = 3.47 nm and 3.51 nm, respectively) coincide within the measurement error. The power spectral density functions of both surfaces shown in Fig. 10(c) are practically identical as well. Thus, the AFM results support the assumption that the substrate relief was transmitted during the bi-layer growth, at least in the measurable frequency range, because the coincidence of the PSD-functions of the substrate and the bi-layer surface is the necessary (while not sufficient) condition for the roughness identity.

 figure: Fig. 10

Fig. 10 AFM image (50 × 50 μm2 scan) of (a) uncoated substrate and (b) the top surface of a bi-layer fabricated with oblique deposition (the deposition angle α = −7.7°) as well as (c) the PSD-functions of both surfaces. The straight dashed line is an extrapolation of the measured PSD-functions by the simplest fractal-like dependence PSD ~1/ν1.64.

Download Full Size | PDF

The PSD-functions correlate well with the fractal-like dependencePSD2D(ν)=K/νA, where A = 1.64 and K = 6.15·10−7 μm2.36 in the measurable range of the spatial frequencies [24]. Notice that the scattering in the direction θS = −35° is caused by the harmonic of the roughness spectrum with the spatial frequency ν2μm−1, where the total correlation of the roughness of different interfaces still persists.

Notice that observation of the backscattering suppression effect would be easier, if the contribution of uncorrelated intrinsic roughness into the scattering pattern is smaller. Therefore, assuming the intrinsic roughness to be independent of the substrate one, we used rather rough substrates (rms roughness ~3.5 nm) for bi-layers deposition in our first experiment. We can say that the smoother the substrate is, the lower the correlation between the substrate and interface roughness becomes. This fact was shown experimentally in [11,12] and was explained as a result of a competition between the substrate replication effect and the material grain size.

First, we briefly analyze the effect of adverse factors influencing the backscattering suppression. The calculated ARS in the incident plane from the designed bi-layer is shown in Figs. 11(a) and 11(b), curve 2; the experimental PSD-function was used for this calculation. For comparison, curve 1 shows the ARS from a similar bi-layer that was fabricated by deposition at normal incidence. Curves 3 and 4 in Fig. 11(a) demonstrate the variation of the scattering pattern assuming an error in the growth angle β of ± 0.6 degrees. The cumulative effect of ± 0.7 nm errors in the thickness of both layers is illustrated by Fig. 11(b). While the errors result in the disappearance of total scattering suppression, the minimum in the ARS curves still persists, with the scattering intensity at θS = −35° being two orders of magnitude lower than that from the bi-layer deposited at normal incidence. Therefore, we consider the indicated errors in the growth angle and the layer thickness as acceptable for observation of interference scattering suppression. Such an error in the layer thickness is well controllable in the experiment, while the error in the growth angle escapes direct experimental detection and thus constitutes a crucial factor hindering the observation of the effect.

 figure: Fig. 11

Fig. 11 The effect of errors in (a) the growth angle and (b) the layer thickness on the scattering suppression from a SiO2/Ta2O5 bi-layer. Curve 1 is the ARS from a bi-layer deposited at normal incidence, while curve 2 is the ARS from a bi-layer optimized for total scattering suppression at the scattering angle θS = −35° (θi = 45°, λ = 633 nm). Curves 3 and 4 demonstrate the effect on the ARS curve of either (a) ± 0.6° error in the growth angle β or (b) ± 0.7 nm error in the thickness Δd of both layers (b).

Download Full Size | PDF

The effect of inaccuracies in the refractive indexes of the materials used when designing bi-layers is illustrated by Figs. 12(a)-12(c). Curve 1 is the ARS from the bi-layer deposited at normal incidence, while curve 2 is the ARS from the bi-layer optimized for total scattering suppression at the scattering angle θS = −35° (θi = 45°, λ = 633 nm). Curves 3 and 4 demonstrate the effect on the ARS curve of negative and positive deviations in the refractive index of (a) SiO2, (b) Ta2O5, and (c) substrate material. The values of the deviations indicated in the graphs are the maximum possible values that still provide suppression of the backscattering by two orders of magnitude compared to the bi-layer fabricated with normal incidence deposition.

 figure: Fig. 12

Fig. 12 The effect of inaccuracies in the refractive indexes of materials used when designing bi-layers on the scattering suppression from a SiO2/Ta2O5 bi-layer. Curve 1 is the ARS from a bi-layer deposited at normal incidence, while curve 2 is the ARS from a bi-layer optimized for total scattering suppression at the scattering angle θS = −35° (θi = 45°, λ = 633 nm). Curves 3 and 4 demonstrate the effect on the ARS curve of negative and positive deviations in the refraction index of (a) SiO2, (b) Ta2O5, and (c) the substrate material.

Download Full Size | PDF

The calculations showed that the effect of backscattering suppression is mostly sensitive to variations in the refractive index of the SiO2 layer. While the position of the minimum in the ARS curve remains almost the same while varying the refractive indexes of the SiO2 and Ta2O5 layers, changing the substrate refractive index results in a pronounced shift of the minimum. The values of the refractive index were determined by analyzing experimental reflectance and transmittance data measured from the single films. We estimated the accuracy of the refractive indexes to be within ± 0.5% and can thus conclude that the refractive indexes are controllable in experiments with a reasonable accuracy. The indexes used for designing bi-layers and presented in Section 3 are the averaged values obtained for several films of slightly different thickness.

There is one more crucial factor that can influence the scattering suppression and is poorly controlled in experiments. This factor is the scattering from uncorrelated interfacial roughness, because when the partial scattering intensities, rather than the scattering amplitudes, are added, interference suppression in the scattered waves is out of the question. Film growth is always associated with the development of short-scale intrinsic film roughness uncorrelated with the substrate roughness. Then, assuming the total PSD-function of each interface to be a sum of correlated and uncorrelated parts, we calculated the ARS curves shown in Fig. 13. Curves 1 and 2 represent scattering from the bi-layers deposited at normal and oblique incidence, respectively, with the effect of uncorrelated roughness being neglected and the experimental PSD-function being used for the calculations. To estimate the effect of short-scale uncorrelated roughness, we suppose that its PSD-function can be determined by Eq. (17), where the Hörst parameter h = 0.5 and the rms roughness σun = 0.15 nm. Calculations were performed for two values of the correlation length ξun = 50 nm (curve 3) and 100 nm (curve 4).

 figure: Fig. 13

Fig. 13 The effect of small-scale uncorrelated roughness on backscattering suppression from a SiO2/Ta2O5 bi-layer. Curve 1 is the ARS from the bi-layer deposited at normal incidence, while curve 2 is the ARS from the bi-layer with the skewed roughness optimized for total scattering suppression at the scattering angle θS = −35° (θi = 45°, λ = 633 nm), not taking the uncorrelated roughness into account. Curves 3 and 4 are the ARS from the bi-layer with the skewed roughness in the presence of small-scale uncorrelated roughness with σun = 0.15 nm and correlation lengths ξun = 50 nm (3) and 100 nm (4). Curve 5 demonstrates the contribution of uncorrelated roughness to the total ARS. Curve 6 is the same as curve 1, but assuming the interface roughness to be fully uncorrelated.

Download Full Size | PDF

For the case of a small correlation length, the denominator in Eq. (17) is practically equal to unity at the spatial frequency ν ~2 μm−1 corresponding to the zero minimum in curve 2, so that the effect of uncorrelated roughness is determined by the product of σun and ξun. Figure 13 demonstrates that the scattering suppression is clearly observed if this product does not exceed 15 nm2 (curve 4). Then, the value of the PSD-function of uncorrelated roughness is only 3.4*10−11 μm4 at ν ~2 μm−1, i.e., almost four orders of magnitude lower than the experimental PSD-functions shown in Fig. 10(c). Therefore, Fig. 10(c) shows that there is no possibility of recognizing the presence or absence of small-scale uncorrelated roughness, which thus represents an uncontrollable factor hindering the observation of scattering suppression.

Curve 5 in Fig. 13 demonstrates the contribution of uncorrelated roughness (at ξun = 100 nm) to the total ARS (curve 4). The angular distribution of the scattering from small-scale roughness is almost isotropic, because the correlation length is much smaller than the radiation wavelength. The value of the scattering intensity from uncorrelated roughness at θS = −35° is only two orders of magnitude less than that from the sample deposited at normal incidence (curve 1), while the PSD-function of uncorrelated roughness is almost four orders of magnitude smaller.

The fact is that the partial suppression of scattering due to interference effects occurs even for the bi-layer deposited at normal incidence. Indeed, curve 6 was calculated for a bi-layer with uncorrelated roughness, while the PSD-function of each interface is the same as that of curve 1; it is shown in Fig. 10(c). In this case, the scattering intensity is enhanced by almost two orders of magnitude and the difference between curves 5 and 6 reaches four orders of magnitude at θS = −35° in total agreement with the PSD-functions ratio. A geometrical interpretation of this effect is illustrated by Fig. 7(a). In the case of uncorrelated roughnessARS~|Aj|2, while for fully conformal roughness ARS~|Aj|2. The second sum, i.e., the distance between the end of vector A3 and the origin, is essentially smaller.

Until now, we have analyzed s-to-s scattering only, while the effect of s-to-p scattering is one more uncontrollable factor. However, calculations showed that the s-to-p scattering intensity is extremely low (ARS ~10−20 srad−1 in the range of the scattering angles of interest to us), and thus can be neglected.

In addition, we can mention at least two other poorly controlled factors influencing the scattering suppression. They are the scattering from the volume defects of different natures and the scattering from dust particles on the sample surface. However, their analysis is too complex to be discussed in the present paper.

Thus, the preliminary analysis has demonstrated that the major factors influencing the scattering suppression are either well controllable in experiments or can be neglected. The exception is represented by an inaccuracy in the growth angle, which is uncontrollable experimentally. Therefore, we fabricated and studied a set of bi-layers at different deposition angles α, and thus different growth angles β, to experimentally obtain the optimal value providing the best scattering suppression at the angle θS = −35°. The deposition angle α was set to α = 7.7, 10.6, and 13.5 degrees as well as α = 0 (normal incidence deposition) for comparison. A 3D small-angle laser scatterometer (ALBATROSS-TT [25]) was used for the scattering analysis. All experimental ARS curves shown in Figs. 14-15 were measured in the incident plane, and the incident radiation was s-polarized.

 figure: Fig. 14

Fig. 14 The experimental ARS curves (1 to 3) from a set of bi-layers measured in the incident plane. The bi-layers were fabricated at the different deposition angles indicated in the graph. The dashed curves 4 and 5 are results of ARS calculation from the bi-layers with growth angles β = 0 (normal incidence deposition) and β = −5.41° providing total scattering suppression at θS = −35°. Curve 6 is the same as curve 5, but with the noise added. The experimental PSD-function shown in Fig. 10(c) was used when calculating. The specular reflected beam of considerably higher intensity is not shown in the figure. The experimental ARS curves were interpolated within the angular interval [-47°,-43°], where the scattering could not be measured.

Download Full Size | PDF

 figure: Fig. 15

Fig. 15 Experimental ARS curves (1 to 3) from the same bi-layer fabricated by oblique deposition (α = 7.7°). The scattering was measured in the incident plane, but at different angles ψ between the radiation incident plane and the XZ-plane in Fig. 2, which is the incident plane of particle flux during deposition. An angle of ψ = 0 or 180° corresponds to geometry 1 or 2 in Fig. 2, while ψ = 90° indicates radiation incident perpendicular to the XZ-plane. Curve 4 is the ARS from a bi-layer deposited at normal incidence. The dashed curves 5 and 6 are the results of ARS calculations from the bi-layers deposited at growth angles of β = 0 (normal incidence deposition) and β = −5.41° providing total scattering suppression at θS = −35°. Curve 7 is the same as curve 6, but with the noise added. The specular reflected beam of considerably higher intensity is not shown in the figure. The experimental ARS curves were interpolated within the angular interval of [-47°,-43°], where the scattering could not be measured.

Download Full Size | PDF

Curve 1 in Fig. 14 is the measured ARS from the bi-layer deposited at normal incidence, while curve 4 is a theoretical prediction based on the PSD-function found experimentally and shown in Fig. 10(c). Curve 2 presents the experimental ARS from the sample deposited at an oblique incidence angle α = 10.6°, so that the growth angle β is close to the optimal one, assuming the validity of the tangent rule. We see that the scattering in the prescribed direction decreases, but only by a factor of 2. The ARS from the sample deposited at larger angle α = 13.5° lies between curves 1 and 2. However, decreasing the deposition angle down to 7.7° resulted in a sharp suppression of the scattering (curve 3) by a factor of about 30 at the scattering angle θs = −35°.

Next, we note that the background noise-equivalent ARS level was about (5-6)·10−8 srad−1 in our experiment. The noise was measured in the incident plane using the same procedure as in the ARS measurements, but in the absence of a sample. Figure 14 demonstrates that just the noise of the measurements limits the minimal scattering value in the ARS of curve 3. Hence, we cannot exclude the possibility that the actual ARS minimum is deeper (and maybe even significantly deeper) than the measured one. However, more accurate measurements require a much lower level of instrumental noise to prove or disprove this assumption. Another possibility is to study ARS from the bi-layers deposited onto considerably rougher substrates. Then, all the ARS curves will be proportionally enhanced, while the noise level will remain the same.

We did not perform experiments with smaller steps in the deposition angle near the value of α = 7.7°, as the instrumental noise does not permit us to distinguish the optimal α providing the deepest scattering minimum.

Notice that the difference between experimental curve 3 and predicted curve 5 is not so pronounced if we take into account the averaged noise value. Indeed, curve 6 is the same as curve 5, but with the noise added.

To verify that the observed scattering suppression is not a casual fact, we measured ARS from the same sample deposited at α = 7.7°, but turned through an angle of 90° and 180° in the XY plane when measuring scattering. The results are presented in Fig. 15. Curve 1, the same as curve 3 in Fig. 14, is the measured ARS demonstrating suppression of scattering. The curve was measured in geometry G1 shown in Fig. 2, i.e., the incident plane of the light beam coincides with the incident plane of the deposited particles. Curves 2 and 3 are the ARS from the same sample, while the sample was turned through an angle of 90° and 180° in the XY plane, so that curve 3 was measured in geometry G2 in Fig. 2, and the light beam incident plane was perpendicular to the XZ plane for curve 2. The conditions of interference scattering suppression are invalid in both these cases, and the scattering patterns proved to be close to that of the normally deposited sample (curve 4), as was demonstrated earlier in [21].

Thus, we can state that the suppression of backscattering in the desired direction due to interference was experimentally observed. The destructive interference of the waves scattered from different interfaces was achieved through the oblique deposition of the bi-layer. The experimental demonstration of the backscattering suppression is an evidence of almost ideal transmittance of the substrate relief during the bi-layer growth.

5. Conclusion

In the present paper we analyzed the problem of light backscattering suppression in the direction toward the incident beam after reflection from a rough layered coating. We specifically considered the case of s-to-s scattering, which is of especial interest for laser gyroscope physics.

First, we proved theoretically that, in the case of fully conformal roughness replicated perpendicular to the substrate surface during deposition, the backscattering intensity could only be reduced through a decrease in the reflectivity. Therefore, it is impossible to design high-reflectivity multilayer mirrors of any type (periodic or aperiodic, having abrupt or smooth interfaces, consisting of two or more materials) suppressing, even slightly, light backscattering, if the interfacial roughnesses are conformal.

Next, we theoretically demonstrated the possibility of entirely suppressing light backscattering from a rough layered coating fabricated by oblique deposition, so that the substrate relief is replicated at a certain angle to the substrate surface. Scattering suppression is caused by the destructive interference of waves scattered by different interfaces. In the present paper we limited our consideration to the simplest case of a bi-layer deposited at oblique incidence. The necessary conditions of the scattering suppression were formulated.

We proved the possibility of entirely suppressing the backscattering from waves incident onto the reflector from opposite sides, i.e., for the scattering geometries G1 and G2 shown in Fig. 2, which is of special importance for laser gyroscopes.

We analyzed the effect of different adverse factors on the backscattering suppression. We demonstrated that the major factors influencing the scattering suppression are either reasonably controllable in experiments or can be neglected. The exception was inaccuracy in the growth angle, which was uncontrollable experimentally, because a too small growth angle (~5°) prevented visualization of the lateral interface shift with transmittance electron microscopy.

Interference suppression of light backscattering was experimentally observed. We fabricated and studied a set of SiO2-on-Ta2O5 bi-layers fabricated at different deposition angles. The bi-layers were designed to observe total backscattering suppression of radiation (633 nm wavelength, 45° incidence angle) at the scattering angle of θs = −35°. This value of θs was chosen because we were unable to measure scattering in the direction toward the incident beam because the detector blocked it.

We experimentally observed the suppression of the scattering from the bi-layer fabricated at a deposition angle of 7.7° by a factor of about 30 compared with that from the sample deposited at normal incidence. The observed value of the scattering suppression was limited by the background noise of the measurements. It is most likely that the actual minimum of the ARS curve was deeper than the measured one.

We consider the present paper as the first step to the final goal of our work: development of high-reflectivity multilayer mirrors with suppressed backscattering from both counter-propagating waves in laser gyroscopes. Results of that study will be published in the next paper.

Funding

National Natural Science Foundation of China (NSFC) (61522506, 61621001); NSAF (U1630124); and the Russian Ministry of Science and Higher Education within the State assignment FSRC «Crystallography and Photonics» RAS.

References

1. V. V. Yashchuk, L. Samoylova, and I. V. Kozhevnikov, “Specification of x-ray mirrors in terms of system performance: new twist to an old plot,” Opt. Eng. 54(2), 025108 (2015). [CrossRef]  

2. M. Wen, I. V. Kozhevnikov, F. Siewert, A. V. Buzmakov, C. Xie, Q. Huang, Z. Wang, L. Samoylova, and H. Sinn, “Effect of the surface roughness on X-ray absorption by mirrors operating at extremely small grazing angles,” Opt. Express 26(16), 21003–21018 (2018). [CrossRef]   [PubMed]  

3. R. X. Adhikari, “Gravitational radiation detection with laser interferometry,” Rev. Mod. Phys. 86(1), 121–151 (2014). [CrossRef]  

4. V. E. Asadchikov, I. V. Kozhevnikov, Yu. S. Krivonosov, R. Mercier, T. H. Metzger, C. Morawe, and E. Ziegler, “Application of x-ray scattering technique to the study of supersmooth surfaces,” Nucl. Instrum. Meth. A 530(3), 575–595 (2004). [CrossRef]  

5. K. Yamauchi, H. Mimura, K. Inagaki, and Y. Mori, “Figuring with sub-nanometer-level accuracy by numerically controlled elastic emission machining,” Rev. Sci. Instrum. 73(11), 4028–4033 (2002). [CrossRef]  

6. M. Störmer, H. Gabrisch, C. Horstmann, U. Heidorn, F. Hertlein, J. Wiesmann, F. Siewert, and A. Rack, “Ultra-precision fabrication of 500 mm long and laterally graded Ru/C multilayer mirrors for X-ray light sources,” Rev. Sci. Instrum. 87(5), 051804 (2016). [CrossRef]   [PubMed]  

7. R. B. Hurst, N. Rabeendran, K. U. Schreiber, and J. P. R. Wells, “Correction of backscatter-induced systematic errors in ring laser gyroscopes,” Appl. Opt. 53(31), 7610–7618 (2014). [CrossRef]   [PubMed]  

8. A. G. Kuznetsov, A. V. Molchanov, M. V. Chirkin, and E. A. Izmailov, “Precise laser gyroscope for autonomous inertial navigation,” Quantum Electron. 45(1), 78–88 (2015). [CrossRef]  

9. T. J. Kippenberg, S. M. Spillane, and K. J. Vahala, “Modal coupling in traveling-wave resonators,” Opt. Lett. 27(19), 1669–1671 (2002). [CrossRef]   [PubMed]  

10. J. H. Apfel, “Optical coating design with reduced electric field intensity,” Appl. Opt. 16(7), 1880–1885 (1977). [CrossRef]   [PubMed]  

11. C. Amra, J. H. Apfel, and E. Pelletier, “Role of interface correlation in light scattering by a multilayer,” Appl. Opt. 31(16), 3134–3151 (1992). [CrossRef]   [PubMed]  

12. C. Amra, “From light scattering to the microstructure of thin-film multilayers,” Appl. Opt. 32(28), 5481–5491 (1993). [CrossRef]   [PubMed]  

13. C. Amra, “Light scattering from multilayer optics. II. Application to experiment,” J. Opt. Soc. Am. A 11(1), 211–226 (1994). [CrossRef]  

14. C. Amra, G. Albrand, and P. Roche, “Theory and application of antiscattering single layers: antiscattering antireflection coatings,” Appl. Opt. 25(16), 2695–2702 (1986). [CrossRef]   [PubMed]  

15. A. Yu Karabekov, I. V. Kozhevnikov, and V. E. Fedyukovich, “Interference suppression of X-ray scattering under total external reflection from rough surfaces,” J. XRay Sci. Technol. 4(1), 37–43 (1993). [CrossRef]   [PubMed]  

16. J. Zhang, H. Wu, H. Jiao, S. Schröder, M. Trost, Z. Wang, and X. Cheng, “Reducing light scattering in high-reflection coatings through destructive interference at fully correlated interfaces,” Opt. Lett. 42(23), 5046–5049 (2017). [CrossRef]   [PubMed]  

17. A. G. Dirks and H. J. Leamy, “Columnar microstructure in vapor-deposited thin films,” Thin Solid Films 47(3), 219–233 (1977). [CrossRef]  

18. S. Lichter and J. Chen, “Model for columnar microstructure of thin solid films,” Phys. Rev. Lett. 56(13), 1396–1399 (1986). [CrossRef]   [PubMed]  

19. C. Amra, “Light scattering from multilayer optics. I. Tools of investigation,” J. Opt. Soc. Am. A 11(1), 197–210 (1994). [CrossRef]  

20. E. M. Gullikson and D. G. Stearns, “Asymmetric extreme ultraviolet scattering from sputter-deposited multilayers,” Phys. Rev. B Condens. Matter Mater. Phys. 59(20), 13273–13277 (1999). [CrossRef]  

21. M. Trost, T. Herffurth, S. Schröder, A. Duparré, and A. Tünnermann, “Scattering reduction through oblique multilayer deposition,” Appl. Opt. 53(4), A197–A204 (2014). [CrossRef]   [PubMed]  

22. I. V. Kozhevnikov and M. V. Pyatakhin, “Use of DWBA and perturbation theory in x-ray control of the surface roughness,” J. XRay Sci. Technol. 8(4), 253–275 (1998).

23. J. M. Elson, Theory and Software for Light Scattering from Multilayer Optical Components with Interfacial Roughness (Naval Air Warfare Center, China Lake, CA, 1992).

24. J. E. Harvey, N. Choi, S. Schroeder, and A. Duparré, “Total integrated scatter from surfaces with arbitrary roughness, correlation widths, and incident angles,” Opt. Eng. 51(1), 013402 (2012). [CrossRef]  

25. A. von Finck, M. Hauptvogel, and A. Duparré, “Instrument for close-to-process light scatter measurements of thin film coatings and substrates,” Appl. Opt. 50(9), C321–C328 (2011). [CrossRef]   [PubMed]  

Cited By

Optica participates in Crossref's Cited-By Linking service. Citing articles from Optica Publishing Group journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (15)

Fig. 1
Fig. 1 The calculated ARS (at λ = 633 nm) in the incident plane (φs = 0) from two high-reflectivity Ta2O5/SiO2 multilayer mirrors: HR45 and LSHR45 designed to suppress scattering into the back hemisphere. Here and below the ARS was calculated with the computer program described in [23], Elson et al.
Fig. 2
Fig. 2 Sketch of the bi-layer fabricated by oblique deposition of materials.
Fig. 3
Fig. 3 Green lines limit the area of the parameters B and C, where the necessary conditions of backscattering suppression in Eq. (15) are fulfilled. Red curves EF show the change of these parameters with the variation of the thickness ratio Γ from zero to unity according to Eq. (18). Calculations were performed for (a) SiO2-on-Ta2O5 and (b) Ta2O5-on-SiO2 bi-layers deposited onto BK7 substrates. The radiation wavelength, incidence angle, and scattering angle corresponding to total scattering suppression are set to 633 nm, 45°, and −45°, respectively.
Fig. 4
Fig. 4 Two solutions of Eq. (12) found with varying thickness ratio Γ are indicated by colored solid curves. The straight dashed line corresponds to the optimal practical case β1 = β2. Calculations were performed for (a) SiO2-on-Ta2O5 and (b) Ta2O5-on-SiO2 bi-layers deposited onto BK7 substrates. Radiation wavelength, incidence angle, and scattering angle corresponding to total scattering suppression are set to 633 nm, 45 degrees, and −45 degrees, respectively.
Fig. 5
Fig. 5 ARS in the incident plane (φs = 0) from the bi-layers with the (a) SiO2 or (b) Ta2O5 layer placed on the top. The radiation wavelength, incidence angle, and scattering angle corresponding to total scattering suppression in the incident plane (φs = 0) are set to 633 nm, 45°, and −45°, respectively. The value of the optimal thickness ratio Γ and the growth angle β are indicated in the graphs. Curve 1 shows ARS from a quarter-wavelength bi-layer deposited at normal incidence.
Fig. 6
Fig. 6 The calculated 2D ARS from SiO2/Ta2O5/sub bi-layers. The scattering intensity is presented on a logarithmic scale. Calculations were performed for a quarter-wavelength bi-layer grown perpendicular to the substrate surface (β = 0, Γ = 0.5) and a bi-layer deposited at an oblique incident angle (β = −10.11°, Γ = 0.2844).
Fig. 7
Fig. 7 The partial scattering amplitudes on a complex plane. Calculations were performed for SiO2-on-Ta2O5 bi-layers at λ = 633 nm, θi = 45° and θs = −45°. Three cases are considered: (a) bi-layer deposited at normal incidence (β = 0), with the thickness ratio Γ being equal to 0.5 or 0.6, and (b, c) two bi-layers deposited at oblique incidence with the parameters providing total backscattering suppression. The scattering patterns from the bi-layers are shown in Fig. 5(a).
Fig. 8
Fig. 8 ARS (λ = 633 nm) in the incident plane (φs = 0) from the SiO2-on-Ta2O5 bi-layers designed to suppress scattering at different scattering angles β from 0 to −60° (curves 2 to 5). Incidence angle was set to 45° for all curves. The value of the optical thickness ratio Γ and the growth angle β are indicated in the graph. Curve 1 shows the ARS from a quarter-wavelength bi-layer deposited at normal incidence.
Fig. 9
Fig. 9 ARS (λ = 633 nm) in the incident plane (φs = 0) from SiO2-on-Ta2O5 bi-layers for the two incident geometries indicated in Fig. 2. The beam is incident on the bi-layer from the left side (G1, θi = 45°) or right side (G2, θi = −45°). The bi-layer in (a) was designed to suppress scattering at θs = -θi = −45° for a wave incident from the left side, while the bi-layer in (b) suppresses scattering at θs = -θi for both counter-propagating waves at once. The value of the optimal thickness ratio Γ and the growth angles β are indicated in the graphs.
Fig. 10
Fig. 10 AFM image (50 × 50 μm2 scan) of (a) uncoated substrate and (b) the top surface of a bi-layer fabricated with oblique deposition (the deposition angle α = −7.7°) as well as (c) the PSD-functions of both surfaces. The straight dashed line is an extrapolation of the measured PSD-functions by the simplest fractal-like dependence PSD ~1/ν1.64.
Fig. 11
Fig. 11 The effect of errors in (a) the growth angle and (b) the layer thickness on the scattering suppression from a SiO2/Ta2O5 bi-layer. Curve 1 is the ARS from a bi-layer deposited at normal incidence, while curve 2 is the ARS from a bi-layer optimized for total scattering suppression at the scattering angle θS = −35° (θi = 45°, λ = 633 nm). Curves 3 and 4 demonstrate the effect on the ARS curve of either (a) ± 0.6° error in the growth angle β or (b) ± 0.7 nm error in the thickness Δd of both layers (b).
Fig. 12
Fig. 12 The effect of inaccuracies in the refractive indexes of materials used when designing bi-layers on the scattering suppression from a SiO2/Ta2O5 bi-layer. Curve 1 is the ARS from a bi-layer deposited at normal incidence, while curve 2 is the ARS from a bi-layer optimized for total scattering suppression at the scattering angle θS = −35° (θi = 45°, λ = 633 nm). Curves 3 and 4 demonstrate the effect on the ARS curve of negative and positive deviations in the refraction index of (a) SiO2, (b) Ta2O5, and (c) the substrate material.
Fig. 13
Fig. 13 The effect of small-scale uncorrelated roughness on backscattering suppression from a SiO2/Ta2O5 bi-layer. Curve 1 is the ARS from the bi-layer deposited at normal incidence, while curve 2 is the ARS from the bi-layer with the skewed roughness optimized for total scattering suppression at the scattering angle θS = −35° (θi = 45°, λ = 633 nm), not taking the uncorrelated roughness into account. Curves 3 and 4 are the ARS from the bi-layer with the skewed roughness in the presence of small-scale uncorrelated roughness with σun = 0.15 nm and correlation lengths ξun = 50 nm (3) and 100 nm (4). Curve 5 demonstrates the contribution of uncorrelated roughness to the total ARS. Curve 6 is the same as curve 1, but assuming the interface roughness to be fully uncorrelated.
Fig. 14
Fig. 14 The experimental ARS curves (1 to 3) from a set of bi-layers measured in the incident plane. The bi-layers were fabricated at the different deposition angles indicated in the graph. The dashed curves 4 and 5 are results of ARS calculation from the bi-layers with growth angles β = 0 (normal incidence deposition) and β = −5.41° providing total scattering suppression at θS = −35°. Curve 6 is the same as curve 5, but with the noise added. The experimental PSD-function shown in Fig. 10(c) was used when calculating. The specular reflected beam of considerably higher intensity is not shown in the figure. The experimental ARS curves were interpolated within the angular interval [-47°,-43°], where the scattering could not be measured.
Fig. 15
Fig. 15 Experimental ARS curves (1 to 3) from the same bi-layer fabricated by oblique deposition (α = 7.7°). The scattering was measured in the incident plane, but at different angles ψ between the radiation incident plane and the XZ-plane in Fig. 2, which is the incident plane of particle flux during deposition. An angle of ψ = 0 or 180° corresponds to geometry 1 or 2 in Fig. 2, while ψ = 90° indicates radiation incident perpendicular to the XZ-plane. Curve 4 is the ARS from a bi-layer deposited at normal incidence. The dashed curves 5 and 6 are the results of ARS calculations from the bi-layers deposited at growth angles of β = 0 (normal incidence deposition) and β = −5.41° providing total scattering suppression at θS = −35°. Curve 7 is the same as curve 6, but with the noise added. The specular reflected beam of considerably higher intensity is not shown in the figure. The experimental ARS curves were interpolated within the angular interval of [-47°,-43°], where the scattering could not be measured.

Equations (19)

Equations on this page are rendered with MathJax. Learn more.

ARS( θ s , φ s , θ i )= 1 Q i d Q s dΩ = |A( θ s , φ s , θ i ) | 2 Scos θ i .
A( θ s , φ s , θ i )= k 2 4π e i( q 0 q ) ρ Δε( r ) E 0 (z, θ i ) E 0 (z, θ s ) d 3 r ,Δε( r )=ε( r ) ε 0 (z), q 0 =k( sin θ i ,0 ), q =k( sin θ s cos φ s ,sin θ s sin φ s ), ρ =( x,y ),k=2π/λ.
E 0 (z,θ)+ k 2 ( ε 0 (z) sin 2 θ ) E 0 (z,θ)=0, E 0 (z,θ)= e ikzcosθ +r(θ) e ikzcosθ , E 0 (z+,θ)=t(θ) e ikz ε sub sin 2 θ .
ARS( θ s , φ s , θ i )= k 4 (4π) 2 cos θ 0 | E 0 ( z, θ i ) E 0 ( z, θ s ) ε 0 (z)dz | 2 PSD( ν ), PSD( ν )= ζ( ρ )ζ(0) exp( 2iπ ρ ν ) d 2 ρ ,2π ν = q q 0 .
S + E 0 2 ε 0 dz = lim L 1,2 { ( E 0 2 ε 0 ) z= L 1 z= L 2 L 1 L 2 ε 0 d E 0 2 dz dz }.
ε 0 d E 0 2 dz = sin 2 θ i d E 0 2 dz 1 k 2 d dz ( d E 0 dz ) 2 .
S= lim L 1,2 { ( ε 0 sin 2 θ i ) E 0 2 + 1 k 2 ( d E 0 dz ) 2 } z= L 1 z= L 2 , = 1 k 2 lim L 1,2 { E 0 d 2 E 0 d z 2 + ( d E 0 dz ) 2 } z= L 1 z= L 2 .
ARS( θ i ,π, θ i )= k 4 π 2 cos 3 θ i |r( θ i ) | 2 PSD( 2ksin θ i ).
ζ 2 ( ρ )= ζ sub ( ρ n x d 2 tan β 2 ), ζ 1 ( ρ )= ζ sub ( ρ n x d 2 tan β 2 n x d 1 tan β 1 ).
ζ 2 F ( ν )= ζ sub F ( ν ) e i η 2 , ζ 1 F ( ν )= ζ sub F ( ν ) e i( η 1 + η 2 ) , η j 2π ν x d j tan β j .
A( θ s , θ i )= k 2 4π [ ( ε 1 ε 0 ) E 0 ( z 1 , θ 0 ) E 0 ( z 1 ,θ) e i( η 1 + η 2 ) + ( ε 2 ε 1 ) E 0 ( z 2 , θ 0 ) E 0 ( z 2 ,θ) e i η 2 +( ε 3 ε 2 ) E 0 ( z 3 , θ 0 ) E 0 ( z 3 ,θ) ] ζ sub F ( ν ), η j 2π/λ( sin θ i sin θ s ) d j tan β j .
ARS( θ s , θ i )= k 4 (4π) 2 cos θ 0 | ( ε 1 ε 0 ) E 0 ( z 1 , θ 0 ) E 0 ( z 1 ,θ) e i( η 1 + η 2 ) +( ε 2 ε 1 ) E 0 ( z 2 , θ 0 ) E 0 ( z 2 ,θ) e i η 2 +( ε 3 ε 2 ) E 0 ( z 3 , θ 0 ) E 0 ( z 3 ,θ)| 2 PS D sub ( ν ).
A( θ s , θ i )~1+ ε 2 ε 1 ε 3 ε 2 E 0 ( z 2 , θ i ) E 0 ( z 2 , θ s ) E 0 ( z 3 , θ i ) E 0 ( z 3 , θ s ) Cexp(i γ 2 ) e i η 2 + ε 1 ε 0 ε 3 ε 2 E 0 ( z 1 , θ i ) E 0 ( z 1 , θ s ) E 0 ( z 3 , θ i ) E 0 ( z 3 , θ s ) Bexp(i γ 12 ) e i( η 1 + η 2 ) .
1+С e iс +B e ib =0.
|1B|C1+B.
c η 2 + γ 2 =±arccos B 2 C 2 1 2C , b η 1 + η 2 + γ 12 =arccos C 2 B 2 1 2B .
PSD sub (ν)= σ 2 ξ 2 h π ( 1+ ξ 2 ν 2 ) 1+h .
D 1 = d 1 ε 1 sin 2 θ i = λ 2 Γ, D 2 = d 2 ε 2 sin 2 θ i = λ 2 (1Γ).
η 2 =2π ν x d 2 tan β 2 =±arccos B 2 C 2 1 2C η 1 =2π ν x d 1 tan β 1 =[ arccos C 2 B 2 1 2B +arccos B 2 C 2 1 2C ]
Select as filters


Select Topics Cancel
© Copyright 2024 | Optica Publishing Group. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.