Design and Deposition of Ultra-Broadband Beam-Splitting Coatings

Abstract

This study aims to develop a stress-optimized ultra-broadband beam-splitting coating that integrates four spectral bands by analyzing the beam-splitting properties of coatings spanning visible to medium and long-wave infrared regions. A beam-splitting coating was deposited on a Ge substrate using ion-beam-assisted thermal evaporation, employing Ge, ZnS, and YbF₃ as coating materials. The designed coating exhibits high reflectance in the 0.5–0.8 μm and 0.9–1.7 μm wavelength bands while maintaining high transmittance in the 3–5 μm and 8–12 μm bands. The optimal deposition process for a single-layer coating was established, at a 45° incidence angle, the beam-splitting coating achieved an average reflectance (Rave) of 86.6% in the 0.9–1.7 μm band and 93.7% in the 0.9–1.7 μm band, alongside an average transmittance (Tave) of 91.36% in the 3–5 μm band and 91.3% in the 8–12 μm band. The antireflection coating achieved a single-side Tave of 98.5% in the 3–5 μm band and 97% in the 8–12 μm band. The coating uniformity exceeded 99.6%. To optimize the surface profile, a single-layer Ge coating was added to the rear surface, resulting in a root mean square deviation of less than 0.0007 μm, achieved the same precision of the surface profile successfully. The deposited beam-splitting coating possessed high surface profile precision, and successfully achieved high reflectance in the visible to short-wave infrared range and high transmittance in the medium- and long-wave infrared range. The coating demonstrated excellent adhesion, abrasion resistance, and structural integrity, with no wrinkling, cracking, or delamination.

1. Introduction

At present, the development of photoelectric technology equipment is constantly updated, driven by increasing demands for compact structures, user-friendly operation, and minimal error tolerance. In response, multispectral common optical path and common aperture technologies have emerged as essential solutions for modern information warfare, representing an inevitable trend [1,2]. The visible light, short-wave infrared, mid-wave infrared, and long-wave infrared are integrated and comprehensively applied to form an optical system of “multi-band integrated”, which is used to achieve the advanced working technical requirements of highly integrated multi-function in one machine. The beam splitter is the core component for achieving the “multi-band integrated” system, whose performance directly determines the function and quality of the system. It has extensive applications in multi-band fusion monitoring systems, electro-optical pods and other systems.

Most existing studies focus on antireflection [3,4,5,6] or beam-splitting coatings [7,8,9] within the visible to short-wave infrared and short-wave infrared to medium- and long-wave infrared bands. However, research on ultra-broadband beam-splitting coatings spanning the entire visible, short-wave infrared, medium-wave infrared, and long-wave infrared spectrum remains limited. The integration of such multi-band coatings would enable efficient separation of incoming light signals, directing reflected and transmitted beams to detectors operating in different spectral bands. This approach facilitates full-spectrum detector response. Beam-splitting coatings that reflect visible to short-wave infrared while transmitting medium- and long-wave infrared have been shown to simplify the structure of large-field-of-view, long-focal-length optical systems while significantly enhancing imaging quality [10]. However, the design and fabrication of these coatings present considerable challenges due to their broad reflection and transmission bandwidths. The required multilayer stack increases layer count and thickness errors [11,12], leading to exponentially rising layer stress and absorption. Furthermore, beam-splitting coatings must meet stringent requirements for surface profile precision, coating quality, and long-term reliability, making their production highly complex [13,14]. Future research should focus on mitigating absorption and stress-related issues to optimize spectral performance and enhance coating reliability [15,16,17], ensuring robust support for full-spectrum, high-efficiency detection in space optical systems.

Therefore, it is crucial to investigate the deposition process for a “four-band integrated” beam-splitting coating that spans the visible (0.5–0.8 μm), short-wave infrared (0.9–1.7 μm), medium-wave infrared (3–5 μm), and long-wave infrared (8–12 μm) spectral bands. Figure 1 presents the schematic diagram of the optical path for this system.

As illustrated in Figure 1, Beam Splitter 1 reflects visible (VIS) and short-wave infrared (SWIR) while transmitting medium-wave infrared (MWIR) and long-wave infrared (LWIR); Beam Splitter 2 reflects medium-wave infrared while transmitting long-wave infrared; and Beam Splitter 3 reflects visible light while transmitting short-wave infrared. Beam Splitter 1, in particular, poses significant fabrication challenges due to the broad bandwidths of both its reflection and transmission bands. This study focuses on the design and deposition process for the coating system of Beam Splitter 1.

Expected goals

•

Deposition of the beam-splitting coating on a Ge substrate.

•

Optical performance at a 45° incidence angle:

○

Average transmittance: T > 85% in the 3–5 μm and 8–12 μm bands.

○

Average reflectance: R > 85% in the 0.5–0.8 μm and 0.9–1.7 μm bands.

•

Surface profile precision maintained post-coating.

•

Coating uniformity ≥ 98%.

•

Antireflection (AR) coating performance:

○

Average transmittance: T > 95% in the 3–5 μm and 8–12 μm bands.

2. Experiment

The multi-band beam-splitting coating was deposited using a ZZS1100 box-type vacuum coating machine (Chengdu, China) equipped with an End-Hall ion source (NY, USA) and an INFICON six-position crystal monitor (Badragaz, Switzerland). This setup enables precise control of coating thickness and deposition rate, ensuring stable and reliable optical performance. Ge wafers (Φ50 mm × 2 mm) were used as the substrate material.

The transparent region of Ge is 1.7~23 μm, which possesses good mechanical properties and extremely high refractive index. The transparent region of ZnS is 0.38~14 μm, which is one of the most important membrane materials used in the visible light and infrared regions. In the visible light region, it is often combined with fluorides of low refractive index, and in the infrared region, it is often combined with semiconductor materials of high refractive index. The transparent region of YbF₃ is 0.3~12 μm, with very little absorption from the visible to infrared band and excellent mechanical properties. Therefore high-purity (99.99%) Ge, ZnS, and YbF₃, supplied by Hongrui Optics (Dongguan, China), served as the coating materials. Prior to deposition, both surfaces of the Ge substrate were cleaned with an alcohol–ether mixture (alcohol–ether = 3:1). A 100 V, 1 A ion beam was then applied to the substrate to activate and clean the surface, eliminating secondary contamination.

Ge was deposited using an electron gun, while ZnS and YbF₃ were deposited via resistive heating evaporation.

Table 1. Process parameters of three kinds of materials.

Materials	Rate (Å/S)	Temperature (°C)	Anode Voltage (V)	Cathode Current (A)
Ge	5	150	100	1
ZnS	10	150	100	1
YbF3	6	150	200	2.5

summarizes the corresponding baking temperatures, deposition rates, and ion source parameters for these materials.

The deposited Ge, ZnS, and YbF₃ coatings were all amorphous. The optical constants of the single-layer coatings were characterized using a SENTECH SER850DUV spectroscopic ellipsometer (Berlin, Germany), which provided the refractive index distribution shown in Figure 2. The Sellmeier model was used to fit the spectral ellipsometry parameters of Ge films, and the Cauchy model was used to fit the ellipsometry parameters of ZnS and YbF₃ films [18].

The reflection spectral in the visible to short-wave infrared range were measured using an Agilent Cary 7000 UV-Vis-NIR spectrophotometer (CA, USA), while the transmission spectra in the medium- and long-wave infrared range were obtained using a Nicolet iS50 infrared spectrometer (MA, USA).

At λ = 2 μm, the refractive indices of the materials were

•

Ge: 4.4;

•

ZnS: 2.26;

•

YbF₃: 1.48.

The refractive index curves of Ge, ZnS, and YbF₃ indicate stable and uniform optical properties, following the expected normal dispersion trend.

3. Results and Discussion

3.1. Design and Deposition of the Beam-Splitting Coating System

For the visible and short-wave infrared reflective coating stack, ZnS and YbF₃ were selected as the high- and low-refractive-index materials, respectively, instead of Ge, which exhibits a broad transmission range from 1.7 μm to 23 μm, and there were a large absorption in Ge/YbF₃ multilayers within the range of 770~1050 nm [19]. The medium- and long-wave infrared transmission coating stack was designed using Ge and ZnS.

Table 2. Parameters of the coating structures.

Symbol	Coating Structure	Central Wavelength/nm	Thickness/μm
S1	Ge/(0.5HM0.5H)8 0.8(0.5ML0.5M)8 0.63(0.5ML0.5M)8 0.5(0.5ML0.5M)8 0.4(0.5ML0.5M)8 0.3(0.5ML0.5M)8/Air	1800	14.02
S2	Ge/(0.5HM0.5H)9 2M 0.8(0.5ML0.5M)8 0.63(0.5ML0.5M)8 0.49(0.5ML0.5M)8 0.39(0.5ML0.5M)9 0.31(0.5ML0.5M)9/Air	1800	13.16

Remarks: H (high refractive index)—Ge; M (medium refractive index)—ZnS; L (low refractive index)—YbF₃. Superscript numbers 8 and 9 are representative of the cycle numbers.

lists the specific structures, design wavelengths, and physical thicknesses of the film layers corresponding to the film systems S₁ and S₂.

The beam-splitting coating system was designed based on a long-pass filter structure (0.5HL0.5H)^S, with the following multilayer configuration, S₁ (Figure 3), with Ge, ZnS, and YbF₃ layers measuring 716.58 nm, 5789.4 nm, and 7509.88 nm, respectively.

The design achieved an average reflectance of 98.4% in the 0.5–1.7 μm band, an average transmittance of 96.2% in the 3–5 μm band, and 96.9% in the 8–12 μm band.

The reflection curve of the deposited beam-splitting coating is shown in Figure 4, with measured results as follows: an average reflectance of 85.6% in the 0.5–0.8 μm band and 93.7% in the 0.9–1.7 μm band.

Figure 5 presents the transmittance curve for the 3–5 μm and 8–12 μm bands.

As shown in Figure 5, the double-surface average transmittance in the 3–5 μm band is 45.8%, while in the 8–12 μm band, it reaches 58.1%. Calculations indicate that the single-surface transmittances for these bands are 62.2% and 87.2%, respectively. Analysis of the medium- and long-wave infrared transmittance curve reveals two key observations:

• The overall transmittance across the entire transmission band is relatively low.
• The transmittance in the 3–5 μm band exhibits a more significant decline, with particularly low values at shorter wavelengths.

To investigate the causes of these phenomena, a thin-film composite (TFC) coating simulation was conducted to assess the impact of coating stack variations on coating thickness. The simulation results indicate that thickness variations did not alter the shapes of the actual and theoretical spectral curves, ruling out thickness errors as the primary cause of transmittance loss in the 3–12 μm band.

To further verify the cause, a reflectance test was performed on the deposited beam-splitting coating within the 3–12 μm range, as shown in Figure 6. The results demonstrate that the reflectance and transmittance trends in the 3–5 μm band are consistent, confirming that absorption is the primary factor contributing to the transmittance decline in this range.

In designing the medium- and long-wave infrared broadband antireflection coating, Ge, ZnS, and YbF₃ were selected as high-, medium-, and low-refractive-index materials, respectively. Due to its high refractive index, Ge introduces significant interface losses, which severely degrade optical performance, particularly in the infrared region. To mitigate these losses and enhance infrared transmittance, an infrared AR coating is necessary.

The designed transmittance spectrum for the 3–12 μm band is shown in Figure 7, with an average transmittance of 97.3% and a central wavelength of λ₀ = 800 nm. The multilayer coating structure is Ge/MHMHMLHLHL/Air, where H represents Ge, M represents ZnS, and L represents YbF₃. The total coating thickness is 2.21 μm, with individual layer thicknesses of 345.06 nm (Ge), 377.78 nm (ZnS), and 1489.29 nm (YbF₃).

The measured transmittance spectrum for the 3–12 μm band is shown in Figure 8, with a double-surface average transmittance of 59.6%.

One of the fundamental characteristics of optical coatings is that they are non-self-supporting heterogeneous layers deposited on a substrate. As a result, optical coatings inherently have two surfaces, and in most optical systems, multiple reflected beams from these surfaces exhibit incoherent superposition.

To investigate the effects of incoherent reflection from the rear surface of an absorption-free optical coating, the impact of this reflection was analyzed, as illustrated in Figure 9. Here, surface (a) represents the coated surface, while surface (b) is the uncoated surface.

When light undergoes multiple reflections at the front and rear surfaces, the total reflected intensity is expressed as [20],

$$R=R_a^{+}+{\displaystyle \raisebox{1ex}{$T_a^{+}{R}_b^{-}{T}_a^{-}$}\!\left/ \!\raisebox{-1ex}{$\left(1-R_a^{-}{R}_b^{+}\right)$}\right.}={\displaystyle \raisebox{1ex}{$\left[R_a^{+}+R_b^{+}(T_a^{+}{T}_a^{-}-R_a^{+}{R}_a^{-})\right]$}\!\left/ \!\raisebox{-1ex}{$\left(1-R_a^{-}{R}_b^{+}\right)$}\right.}$$

(1)

In the absence of absorption

$$R_a^{+}=R_a^{-}=R_a, T_a^{+}=T_a^{-}=T_a, R_a+T_a=1$$

(2)

which simplifies to

$$R={\displaystyle \raisebox{1ex}{$\left(R_a+R_b-2R_a{R}_b\right)$}\!\left/ \!\raisebox{-1ex}{$\left(1-R_a{R}_b\right)$}\right.}$$

(3)

For transmittance, the total transmitted intensity is given by

$$T=T_a^{+}{T}_b^{+}+T_a^{+}{T}_b^{+}{R}_b^{+}{R}_a^{-}\left[1+R_a^{-}{R}_b^{+}+{\left(R_a^{-}{R}_b^{+}\right)}^2+\dots \right]$$

(4)

$$T={\displaystyle \raisebox{1ex}{$T_a{T}_b$}\!\left/ \!\raisebox{-1ex}{$\left(1-R_a{R}_b\right)$}\right.}$$

(5)

$${\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$T$}\right.}={\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$T_a$}\right.}+{\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$T_b$}\right.}-1$$

(6)

The double-surface transmittance of the Ge substrate, measured using a spectrophotometer, was 46.5%, while the single-surface transmittance, calculated using Equation (6), was 63.5%. The double-surface average transmittance of the AR coating in the 3–12 μm band was 59.6%, yielding a single-surface transmittance of 90.7%. However, the overall transmittance remains relatively low, indicating that the coating still exhibits significant absorption.

3.2. Investigation of Absorption Issues

(1)

Decomposition of ZnS at high temperatures

During the deposition of the beam-splitting coating, YbF₃ required a high ion source energy, causing the substrate temperature to rise above 220 °C. When the system switched to ZnS deposition, the temperature remained above 200 °C with minimal cooling. According to the literature, ZnS begins to decompose at temperatures exceeding 150 °C [21], leading to increased film absorption. To verify ZnS decomposition, single-layer ZnS films were deposited at 200 °C (thickness: 1.5 μm) and 250 °C (thickness: 4.03 μm). Their transmittance curves were compared to that of a Ge substrate, as shown in Figure 10.

The 1.5 μm ZnS film at 200 °C showed a transmittance dip 1.3% lower than the Ge substrate. The 4.03 μm ZnS film at 250 °C exhibited a 1.7% lower transmittance. Both ZnS films demonstrated consistent transmittance declines across the 3–12 μm band, confirming temperature-induced absorption. This confirms that ZnS decomposition at high temperatures contributes to film absorption. However, it is not the primary cause of the significant transmittance drop in the 3–5 μm band during beam-splitting coating deposition.

(2)

Sputtering contamination from the ion source

The significant absorption observed in the medium- and long-wave infrared transmission band was further investigated from a process perspective. A suspected cause was contamination from the stainless steel ion source baffle, which, under high-energy ion bombardment, could release impurities such as C, Fe, and Cr. These contaminants may be sputtered onto the coating layers, leading to increased spectral absorption.

To verify this, an experiment was conducted using three B270 glass substrates (Φ30 × 1 mm), labeled #1, #2, and #3, each exposed to different ion source energy levels for one hour. The first substrate was subjected to 280 V and 8 A, the second to 250 V and 5 A, and the third to 200 V and 4 A. A visual inspection revealed that substrates exhibited progressive blackening at higher ion source energies, with #1 being the darkest and #3 the least affected. This blackening suggests increased absorption due to contamination from sputtered C, Fe, and Cr impurities. To confirm the presence of these elements, X-ray photoelectron spectroscopy (XPS) was performed on all three substrates. The corresponding spectra are shown in Figure 11a–c, while

Table 3. Sputtering components of ion source baffle.

Substrate	Ion Source Energy	C Atomic Composition Percentages	Fe Atomic Composition Percentages	Cr Atomic Composition Percentages
#1	280 V, 8 A	38.21	10.51	3.99
#2	250 V, 5 A	35.36	9.26	8.13
#3	200 V, 4 A	35.16	2.83	3.68

lists the atomic composition percentages. The results indicate that the primary contaminants include C, O, Si, Fe, and Cr, with C, Fe, and Cr being the major contributors to coating absorption and substrate blackening.

To address this issue, the stainless steel baffle was replaced with a high-purity graphite baffle, which offers excellent heat and chemical resistance along with superior thermal conductivity. Due to its dense crystalline structure, graphite can withstand higher ion source energy, thereby minimizing sputtered contaminants and effectively reducing coating absorption.

(3)

Cross-contamination between coating materials

During the deposition of the beam-splitting coating, Ge was vapor-deposited using an electron gun, while ZnS and YbF₃ were deposited through resistive evaporation. ZnS, being a sublimable material, tends to accumulate on the cover above the evaporation boats, as well as on the baffle plate and dividers between the boats. When the process switches from ZnS to YbF₃, the higher evaporation current required for YbF₃ can cause ZnS to resublimate and redeposit onto the YbF₃ material, leading to contamination. Additionally, because the infrared beam-splitting coating consists of thick layers, prolonged evaporation increases the accumulation of material on the cover above the evaporation boats. Over time, these deposits may fall through small holes in the covers, contaminating the ZnS or YbF₃ materials inside the boats.

Due to equipment limitations, modifying the layout of the evaporation boats is not feasible. The most effective solution to minimize cross-contamination between ZnS and YbF₃ is to increase the height of the divider between their respective evaporation boats.

(4)

High absorption of Ge at high temperatures

To investigate whether the decline in transmittance in the 3–5 μm band was caused by Ge absorption, single-layer Ge films were vapor-deposited on ZnS substrates under the conditions listed in

Table 4. Different processes on three substrates.

Number	Process
#1	No ion source, temperature 250 °C
#2	Ion source, temperature 250 °C
#3	Ion source, temperature 150 °C

. These samples, labeled #1, #2, and #3, had their transmittance spectra measured, as shown in Figure 12. Ge, having a small bandgap, excites more free carriers as temperature increases, leading to higher absorption. Consequently, with rising temperature, the peak transmittance of Ge decreases [22,23].

As shown in Figure 12, a significant drop in transmittance was observed for Ge films deposited under all three conditions across the 3–12 μm band, with a more pronounced decline in the 3–5 μm range. This confirms that Ge films exhibit significant absorption. To rule out any influence from the ZnS substrate itself, its transmittance was first measured before placing it in a vacuum chamber and baking it at 250 °C for two hours. A second transmittance measurement was then conducted, as shown in Figure 13. The comparison reveals that after baking, the overall transmittance of the ZnS substrate slightly decreased; however, no significant transmittance drop was observed in the 3–5 μm band. This confirms that the decline shown in Figure 5 was caused by Ge film absorption rather than any inherent property of the substrate.

Comprehensive analysis indicates that the significant absorption across the entire transmission band, as well as the pronounced transmittance decline in the 3–5 μm range, results from the combined effects of four major factors: the decomposition of ZnS at high temperatures, sputtering contamination from the ion source, cross-contamination between coating materials, and the high absorption of Ge at elevated temperatures.

3.3. Optimization of the Coating System and Process

To enhance the spectral performance, the original multilayer coating structure S₁ was slightly adjusted and optimized to S₂(Figure 14), with individual layer thicknesses of 848.35 nm for Ge, 5817.3 nm for ZnS, and 6492.5 nm for YbF₃.

Figure 14. The design spectral curve of beam-splitting coating after optimized.

Figure 14. The design spectral curve of beam-splitting coating after optimized.

This design achieved an average reflectance of 98.72% in the 0.5–1.7 μm band, an average transmittance of 96.22% in the 3–5 μm band, and an average transmittance of 96.27% in the 8–12 μm band.

To improve adhesion between layers, reduce film stress, and minimize absorption, the baking temperature for all three materials was set to 100 °C.

The beam-splitting and AR coatings were deposited on a Φ50 mm × 2 mm Ge substrate, and its transmittance curve in the 3–12 μm band is shown in Figure 15.

Using an infrared spectrophotometer, the initial double-surface transmittance of the uncoated Φ50 mm Ge substrate was measured as

•

Tave@3–5 μm = 45.5%;

•

Tave@8–12 μm = 46.9%.

Applying Equation (6), the calculated single-surface transmittances were as follows:

•

Tave@3–5 μm = 62.6%;

•

Tave@8–12 μm = 63.8%.

The measured transmittance curve in the 3–12 μm band after coating the front surface with a beam-splitting coating of #3 substrate is shown in Figure 16. The average transmittance is 59.7% in the 3–5 μm band and 61.3% in the 8–12 μm band.

Figure 17 presents the measured transmittance curves in the 3–12 μm and 8–12 μm bands after applying a medium and long-wave infrared AR coating to the rear surface. The transmission spectrum test results, after coating the front surface with a beam-splitting coating and the rear surface with an AR coating, indicate that at a 45° incidence angle, the average transmittance is 91.4% in the 3–5 μm band and 91.3% in the 8–12 μm band.

The measured reflectance curves in the 0.5–0.8 μm and 0.9–1.7 μm bands are shown in Figure 18. The corresponding reflectance spectrum test results show an average reflectance of 86.6% in the 0.5–0.8 μm band and 93.7% in the 0.9–1.7 μm band.

Figure 19 displays the transmission curve of the AR coating applied to the reverse side within the 3–12 μm band. The double-surface transmittance values are Tave@3–5 μm = 62% and Tave@8–12 μm = 62.5%. Using Equation (6), the single-surface transmittance values for the AR coating are calculated as Tave@3–5 μm = 98.5% and Tave@8–12 μm = 97%.

Figure 20 and Figure 21 illustrate the transmittance measurements taken at three points along the diameter of the Φ50 mm Ge wafer. Using the lowest point in the spectrum as a reference, the wavelengths corresponding to the curves are 5.885 μm for curve 1, 5.91 μm for curve 2, and 5.89 μm for curve 3. The calculated coating uniformity exceeds 99.6%.

3.4. Non-Optical Properties of the Coatings

Beyond optical performance, durability and environmental stability are essential evaluation criteria for infrared coatings [24]. To assess these properties, the deposited beam-splitting coating samples (with an AR coating on their rear surfaces) underwent abrasion resistance, adhesion, and damp-heat tests in accordance with the standard [25].

Adhesion and abrasion resistance tests were conducted on a Φ50 × 2 mm sample. Figure 22 presents the surface conditions of the coating before and after the 3 M tape test. No detachment or visible damage was observed under reflected light, confirming strong adhesion between the coating and the substrate. Figure 23 illustrates the surface conditions before and after rubbing the front and rear surfaces with absorbent cotton gauze. No scratches or detachment were detected, demonstrating excellent abrasion resistance.

Figure 24 shows the coating surfaces before and after the damp-heat test. Following exposure, the coating exhibited no wrinkling, cracking, or detachment, confirming high environmental stability.

3.5. Surface Profile Correction

The precision of the beam splitter’s surface profile primarily depends on the initial surface accuracy and the degree of deformation induced by the residual stress of the deposited coating. Even with a high-precision substrate, residual coating stress can significantly degrade surface profile accuracy, ultimately affecting the system’s transmission performance. Future studies should focus on analyzing coating stress and substrate deformation mechanisms to develop a correction model, ensuring that the surface profile remains stable before and after coating.

A suitable correction method involves incorporating a buffer layer to mitigate stress-induced deformation. This can be achieved by introducing a layer with opposite stress between the substrate and coating to counteract stress effects or by adding a layer with the same stress on the rear surface of the substrate to reduce or neutralize deformation [26,27].

In this study, the Ge substrate was coated with a beam-splitting coating on the front surface and an AR coating on the rear surface. The thickness difference between these coatings was 10.95 μm, which increased the substrate’s optical power, leading to deformation and reduced surface profile accuracy. As shown in Figure 25, TFC software (version 3.5) simulations indicate that incorporating a single Ge layer between the Ge substrate and the AR coating on the rear surface has minimal impact on the transmission spectra. Consequently, a single Ge layer was introduced at this interface to enhance surface profile stability.

Figure 26 compares the surface profiles of substrates 3 and 8 before and after the deposition of the beam-splitting coating, with root mean square (RMS) differences listed in

Table 5. RMS deviation before and after the deposition of the beam-splitting coating.

Number	RMS/μm (Substrate)	RMS/μm (After sp Coating)	RMS/μm (Deviation)
#3	0.099	0.166	0.066
#8	0.026	0.083	0.056

. The laser test wavelength is 632.8 nm.

The Ge layer in the beam-splitting coating is 848.35 nm thick. For the initial surface profile correction, a 200 nm Ge layer was deposited on the rear surfaces of substrates 3 and 8 under the following conditions: baking temperature of 100 °C, deposition rate of 5 Å/s, and ion source energy of 100 V/A.

Figure 27 presents the corrected surface profiles after this adjustment, with the corresponding RMS values before and after coating listed in

Table 6. RMS deviation after the deposition of 200 nm Ge.

Number	RMS/μm (Before)	RMS/μm (After)	RMS/μm (Deviation)
#3	0.166	0.156	−0.010
#8	0.082	0.073	−0.009

.

Further correction was performed by depositing a 650 nm Ge layer on the rear surfaces of the beam-splitting coatings on substrates 3 and 8. As shown in Figure 28, the RMS value of substrate 3 improved from 0.156 μm to 0.122 μm, approaching the uncoated value of 0.099 μm, resulting in an RMS difference of 0.023 μm. Similarly, the RMS value of substrate 8 improved from 0.073 μm to 0.044 μm, nearing the uncoated value of 0.026 μm, with an RMS difference of 0.017 μm.

The substrate 8 with smallest RMS deviation under the current state was selected to prepare the antireflection film within the 3–12 μm on the rear surface of the beam-splitting coating, the surface profile after deposition was shown in Figure 29. The RMS value of substrate 8 improved from 0.044 μm to 0.0259 μm, nearing the uncoated value of 0.0266 μm, with an RMS difference of 0.0007 μm. The precision of the surface profile remained unchanged after coating.

4. Conclusions

An ultra-broadband beam-splitting coating, based on a long-pass filter structure, was designed and deposited on a Ge substrate, covering four spectral bands: visible, short-wave infrared, medium-wave infrared, and long-wave infrared. At a 45° incidence angle, the coating achieved an average reflectance of 86.6% in the 0.5–0.8 μm band and 93.7% in the 0.9–1.7 μm band. After applying an AR coating to the rear surface, the overall average transmittance was 91.4% in the 3–5 μm band and 91.3% in the 8–12 μm band. The single-surface AR coating transmittance reached 98.5% in the 3–5 μm band, and 97% in the 8–12 μm band. In addition to excellent optical performance, the coating demonstrated strong adhesion to the substrate, superior abrasion resistance, and high environmental stability, with no signs of wrinkling, cracking, or parting after damp-heat testing. Furthermore, surface profile correction effectively minimized substrate deformation caused by residual coating stress. After deposition, the corrected surface profile achieved an RMS deviation of less than 0.0007 μm compared to the uncoated substrate and achieved the same precision of the surface profile after coating.