State-of-the-Art Materials Used in MEMS Micromirror Arrays for Photonic Applications

Abstract

This work provides an overview on micromirror arrays based on different material systems such as dielectrics, element silicon, compound semiconductors, metals, and novel 2D materials. The goal is to work out the particular strength of each material system to enable optimum performance for various applications. In particular, this review is intended to draw attention to the fact that MEMS micro-mirrors can be successful in many other material systems besides silicon. In particular, the review is intended to draw attention to two material systems that have so far been used less for MEMS micromirror arrays, that have been less researched, and of which fewer applications have been reported to date: metallic heterostructures and 2D materials. However, the main focus is on metallic MEMS micromirror arrays on glass substrates for applications like personalized light steering in buildings via active windows, energy management, active laser safety goggles, interference microscopy, and endoscopy. Finally, the different micromirror arrays are compared with respect to fabrication challenges, switching speed, number of mirrors, mirror dimensions, array sizes, miniaturization potential for individual mirrors, reliability, lifetime, and hinge methodology.

1. Introduction

Throughout the history, mirrors have played an important role. Starting from people observing reflections in water, the materials used for this purpose strongly varied with time. Limited by the knowledge and the technology, people used materials such as polished bronze, glass, and coated glass as mirrors. Nowadays, mirrors represent not only reflectors but also semi-reflective or even transparent plates. A mirror can be highly reflective in a certain spectral range and transparent in another spectral range simultaneously. The dimensions range from huge telescope mirrors of 520 m diameter for radio waves, to 40 m diameter for the visible and near-infrared spectral range, down to mini- or micro-sized membranes. By aligning and arranging many of these mini- and micro-sized mirrors together, micromirror arrays are achieved. Materials have always played a crucial role in innovation. New concepts and new designs have been triggered and enabled in many cases by new materials or a novel fabrication technology of material compounds.

Micro-opto-electro-mechanical systems (MOEMSs = optical MEMSs) have long been used as a transformative technology for the mechanical, electrical, and optical fields by approaching microscale or even nanoscale in the case of nano-electro-mechanical systems (NEMSs). For photonic applications, optical MEMS micromirrors have made and are making significant contributions to precise control of light such as switching, modulation, steering, filtering, and routing.

With the development of optical MEMS technology, mirrors and micromirror arrays have achieved applications in different spectral ranges such as in the visible, near infrared (NIR), and ultraviolet (UV) ranges. The applications range from photonic circuits to devices, such as digital projection, displays, imaging, telecommunications, sensing, etc. (see Figure 1). The performance and versatility of micromirrors are strongly influenced by their materials, design, and size via their optical, mechanical, thermal, and electrical properties. These properties, design, and size crucially impact the functionality and reliability of the micromirror. In the last 50 years, optical MEMSs have revealed a remarkable development and substantial increase in complexity; see the very informative review in [1].

Micromachined modulators for projection displays were reported in 1975 and led to research and fabrication of the development of digital micromirror devices (DMDs) by Texas Instruments [2]. There was an early period after 1960 where the focus of MEMS research was only on the use of single-crystal Si since the use of anisotropic chemical etching for micro-structurization was sufficient for many applications. Later, when precision requirements were raised, polycrystalline Si (poly-Si) technology was demonstrated and developed and overtook that of crystalline Si. In 1991, a MEMS micromirror-based pop-up structure was demonstrated by Wu et al. [3] using a poly-Si MEMS. This again accelerated MEMS free beam optics (light propagation in the air) remarkably [3,4,5] and provided amazing designs and very durable structures with high reliability and a long lifetime. With the growing list of applied MEMS materials and various fabrication technologies, the functionality and implementation of micromirrors have progressed significantly in subsequent years.

This paper reviews different materials that are used for MEMS-based micromirror arrays, each possessing distinct characteristics that influence the applicability and functionality of the final products. The goal is to distinguish the advantages and the limitations associated with different materials and to understand how material selection affects system performance. Micromirror arrays made of these material classes are finally compared with respect to parameters such as resolution, tuning ranges, switching speed, reflectivity, lifetime, capability of strong miniaturization, and reliability.

Special emphasis is also placed on the fact that transparent substrates open up completely new application options that were previously not possible when using non-transparent substrates (above all, Si). This increases the versatility of micromirror arrays and the related application diversity enormously.

The requirements for performance, miniaturization and functionalities have led researchers to investigate and develop not only Si-based micromirrors but also mirrors made of metals, dielectrics, 2D materials, and combinations of these, i.e., hybrid solutions. As shown in Figure 1, element semiconductors such as crystalline Si and polysilicon; group III/IV semiconductor compounds; dielectrics like TiO₂, ZrO₂, Si₃N₄, SiO₂; and others are preferred for optical communications, digital projection, adaptive optics, optical lithography, sensorics, spectroscopy, medical applications, smart glass, safety, security, automotive headlighting, displays, solar cell technology, and metrology.

In Section 2, dielectric DBR (distributed Bragg reflector) mirrors and applications in spectroscopy and sensorics are discussed in detail. Section 3 presents semiconductor-based MEMS micromirror arrays, such as Si and InP. Section 4 is devoted to metal-based MEMS micromirror arrays. Metallic mirrors are often investigated for the purpose of high-contrast modulators and switches, application in smart glass in buildings, personalized light steering, energy management in buildings, high-speed optical shutters, safety and security, and so on. In this section, three examples of metallic micromirror arrays are presented. Section 5 deals with the new material class of 2D materials and their potentials. Two-dimensional materials have, up to now, mainly been proposed for high-conducting materials, transistors, batteries, gas storage, and energy harvesting. However, there is much more potential, such as for mirrors. This new field will be treated elementarily. Section 6 includes a comparison of the materials and technologies with respect to micromirror sizes, number of mirrors, array sizes, switching speed, fabrication challenges, specific obstacles, material compatibilities, scalability, and reliability. Section 7 explores an outlook of MEMS micromirror array technology.

In this study, a comprehensive review of state-of-the-art, commonly employed materials is given with respect to existing and future potential micromirror array applications. By identifying highlights, breakthroughs, and challenges, the aim is to broaden the understanding of material considerations in the design and fabrication of MEMS-based micromirrors, fostering innovation in optical technology.

2. Arrays of Dielectric DBR Mirrors

A Fabry–Pérot (FP) filter consists of two DBR mirrors and a cavity in between. In most cases, DBR mirrors consist of a periodic arrangement of quarter-wave layers. The materials can be dielectric (Section 2) or made of semiconductor layers (a special case of MEMS tunable InP/air structures is presented in Section 3). A static DBR micromirror sensor array consists of an FP filter array on a photodiode CMOS (complementary metal–oxide semiconductor) or CCD (charge-coupled device) array. For each DBR, thin films are stacked with alternatively low and high optical refractive indices. The higher the refractive index contrast between the two materials, the wider the stopband. Since our DBR micromirror sensor array uses nanoimprint in the definition of all required different cavity heights, it has been named a nanospectrometer. If an FP filter array and a detector array are combined that are matching in size, we have a nanospectrometer sensor. For more information, see refs. [6,7,8,9,10]. Figure 2a shows a selection of four FP filters out of the whole array. The filter array comprises two DBRs (gray) and the cavities (orange). Each filter cavity height defines a different wavelength within all the different characteristic narrow filter lines.

From the top to the bottom, Figure 2b illustrates the main fabrication steps. It starts with the deposition of the bottom DBR on the detector array. Next, the liquid cavity material (orange) is deposited by spin-coating. Subsequently, the 3D cavity structure is defined using a transparent 3D stamp (light blue) that is pressed into the cavity material and hardened via UV light. After lifting the stamp, the top DBR is deposited.

Applying a single 3D nanoimprint step, 192 unequal cavity heights were defined with a single nanoimprint step to generate 192 pixels using only three different DBRs, thus combining three stopbands spectrally next to each other.

A nanospectrometer (Figure 3) was fabricated with 192 FP filters, i.e., 384 DBR micromirrors of 40 µm × 40 µm dimensions each in the lateral plane and characterized (Figure 4). The smallest possible dimensions for the visible spectral range are 8 µm × 8 µm. The system can be used for hyperspectral imaging.

Attention should now be drawn to the possibility that MEMS-tunable FP filters can also be realized on the basis of these dielectric DBR mirrors. That means that micromachined spectral tuning can be obtained in addition. The device geometries and the operation principles are similar to those explained in Section 3.2 below. Here, some examples are compiled of MEMS-tunable FP filters with dielectric DBR micromirror arrays:

SiO₂/Si₃N₄ single-membrane MEMS-tunable Fabry–Pérot filters for 1.55 µm have been published in a two-chip concept for 1.55 µm applications. A tuning range of 103 nm, a required electrostatic actuation voltage of 35 V, and a filter linewidth of 1 nm were obtained [11].

SiO₂/Si/GaInAsP single-membrane MEMS-tunable Fabry–Pérot filters for 1.55 µm in a single-chip concept have been reported, revealing a tuning range of 45 nm, a required electrostatic actuation voltage of 40 V, and a filter linewidth of 10 nm [12]. This and the concepts below are single-chip concepts that have a self-aligning mechanism and do not need complex alignment effort like in the two-chip concept.

SiO₂/TiO₂ single-membrane MEMS-tunable Fabry–Pérot filters for 1.55 µm have been presented, revealing a tuning range of 70 nm, a required electrostatic actuation voltage of 27 V, and a filter linewidth of 0.5 nm [13].

Another dielectric single-membrane MEMS-tunable Fabry–Pérot filter for 1.55 µm has been reported, revealing a tuning range of 30 nm, a required electrostatic actuation voltage of 29 V, and a filter linewidth of only 0.25 nm due to the used external resonator cavity [14].

SiO₂/Si₃N₄ single-membrane MEMS-tunable Fabry–Pérot filters for 1.55 µm have been published, including different cavities with convex, flat, and concave DBR micromirrors. By tailoring the layer stress in the individual SiO₂ and Si₃N₄ sublayers, different radii of curvature can be designed, including negative and positive radii [15].

In summary, this section demonstrates that this dielectric material class enables extremely high reflectivity in the spectral stopband range of the DBR mirrors.

3. Arrays of MEMS Semiconductor Mirrors

3.1. Silicon MEMS Micromirror Arrays

Si has outstanding mechanical properties and does not involve problems due to material stress. After removing the sacrificial layers below planar structures, they stay planar. This is not the case in III/V semiconductor MEMS technology. A further huge advantage is the compatibility of Si MEMS technology with the CMOS process. Therefore, Si MEMS can be fabricated directly on Si electronic, integrated circuits. Microelectromechanical systems (MEMS) for light processing have been designed, fabricated, and investigated for various applications.

Si micromirror array-based MEMS have been reported for N × N switches, routers, and add–drop multiplexers in wavelength division multiplexing fiber-optic communication systems [16,17,18]. In the first example, N² micromirrors are required for N input and N output channels, and the switching occurs in a 2D plane [17], as shown in Figure 5a. Since light leaving the fibers is divergent, collimating lenses, shown in blue, are required to make the light parallel. Figure 5b depicts an SEM micrograph of an 8 × 8 Si MEMS micromirror array. Figure 5c displays the design of a single mirror. Each mirror can be actuated independently of the others: A Si MEMS comb actuator is fixed to each movable translation stage. Note that 22 µm translation results in 90° rotation. An SEM micrograph of a single mirror is presented in Figure 5d, including two pushrods and the electrostatic MEMS comb actuator in front. Figure 5e describes how such specific hinges are fabricated, which look like a classical door or window hinges, with a pin, a cylindrical airgap, and a surrounding cylindrical barrel. The fabrication of such hinges has been described in refs. [3,5]. The Si MEMS fabrication process is shown in eight steps. Six lithography steps, three deposition steps of polycrystalline Si (poly-Si), and one sacrificial removal step are involved. The pin and the micromirror blade are connected behind, which is not visible in these cross-sections. This structure can be better understood in the last sub-diagram, where the sacrificial layer has been removed and the mirror and the pin have been moved by about 30° via an appropriate movement of the comb actuator. A disadvantage of this design is that not all channels can be connected to arbitrary output channels at the same time since some standing mirrors might block the path. Note that N × M switches have been developed.

Next, another N × N switch is presented that is also fabricated with Si MEMS technology, involving a beam switching in 3D, and is shown in Figure 6. Here, only 2N micromirrors are required for N input and N output channels [19], called a λ-router, as a system. Depending on how large N is (16, 36), the number of required micromirrors (N² or 2N) makes a big difference when comparing Figure 5 and Figure 6. For a large N, the λ-router (Figure 6) is by far more attractive. Also, here, a lens array (blue) is required at the input side to convert the divergent light leaving the fibers into parallel light to propagate as beams. At the output side, another lens array (blue) is required to enable the coupling of the light beams (parallel light) into the fibers. At the input side, the optical axis of each fiber is directed to the center of a corresponding mirror of the upper micromirror array. Three optical axes corresponding to three beams are shown in red at the top of Figure 6a via three red lines. At the output side, the optical axis of each fiber is directed to the center of a corresponding mirror of the lower micromirror array. Three optical axes corresponding to three beams are shown in red at the bottom of Figure 6a via three red lines. Using adequate control of the gimbal mount of each mirror (Figure 6b), the light beams can be switched from any input mirror to any output mirror. Three examples are presented in Figure 6a. Figure 6c shows an SEM micrograph of a Si MEMS micromirror array. An advantage of this design is that not all channels can be connected to arbitrary output channels at the same time.

Digital micromirror device (DMD) technology uses silicon mirrors with two stable positions, allowing fast switching in between (Figure 7). The corresponding silicon-based electronic control is located below each micromirror (each pixel) [20,21]. This system allows digital projection and can be used for beamer and programmable projections, sensing applications, and even inside optical and interference microscopes.

In adaptive optics, deformable mirrors have been developed, providing high-order correction of optical aberrations used in astronomy, laser micromachining, maskless lithography, microscopy medical technology, and optical free-space communication [22,23]. All these mirror elements are planar and electrostatically actuatable. As a side note, some adaptive optics are also based on freeform reflectors [23,24].

For DMD systems, the light is reflected on the silicon surface of the DMD system, and no light can transmit to the opaque silicon substrate. Unfortunately, this limits the applications of this very clever system. Sometime optical set-ups also would like to avoid folded beam paths due to strongly increasing alignment efforts. In Section 4.1, a smart glass based on metallic micromirror arrays is presented that does not have this disadvantage since it is based on transparent glass substrate. However, the MEMS smart glass uses passive addressing for subfield control. Regarding that point, DMD technology is advantageous since it is based on active control.

In summary, this section demonstrates that the extremely well-developed Si technology enables outstanding applications due to the two specific strengths of Si: its phantastic versatile oxide and its excellent mechanical properties.

3.2. Compound Semiconductor MEMS Micromirror Arrays

As mentioned before, the DBR micromirrors of a filter array can also be made of III/V semiconductor quarter-wave layers. If large stopbands are desired, the refractive index contrast between the two materials used in the DBR micromirrors has to be as large as possible. The largest contrast can be obtained with air used as the second material combined with InP as the semiconductor material. This provides a refractive index contrast of 3.2 − 1 = 2.2 for 1.55 µm. Figure 8a,b depicts scanning electron micrographs (SEMs) and Figure 8c shows the tuning characteristics. The top InP multiple airgap DBR micromirror is p-doped and the bottom InP multiple airgap DBR micromirror is n-doped. Together with the intrinsic GaInAs spacer layer inside the supporting posts (same thickness as the air cavity without applied voltage), it provides a pin diode that is reverse biased with different actuation voltages. Figure 8d displays an array with four MEMS FP filters. In total, they include six InP multiple airgap DBR micromirrors.

There are also a lot of MEMS devices in III/V compound semiconductors, since direct band structure is required for light-emitting devices or extreme bandgaps are required for detector devices or photovoltaic devices to address specific wavelength ranges. See also the review of III/V optical MEMS in [26].

In summary, this section demonstrates that this dielectric material class enables extremely high reflectivity in the spectral stopband range of DBR mirrors, especially if extreme refractive index contrasts are implemented via multiple airgap technology.

4. Arrays of Metallic MEMS Mirrors for Light-Intensity Modulation and Light Steering

4.1. MEMS Smart Glass for Personalized Light Steering

Metallic MEMS micromirror arrays are capable of dynamic daylight guiding by active control of reflection. The micromirrors are freestanding and fixed at one side—identical to a hinged door but indistinguishable with naked eyes from a distance farther than 20 cm. Arrays of freestanding micromirrors are mounted between the space of the panes of insulation glazing (Figure 9a,b). This space is evacuated or filled with noble gas (Ar, Kr, or Xe).

Thanks to the above-described inter gas filling, the MEMS mirrors between the glass panes are free from dust influence, damage resulting from wind, window cleaning, and unfavorable weather conditions. This also enables our MEMS smart glazing to reveal high stability and a long lifetime.

Without an applied voltage, an individual micromirror is open at a 90° angle between the mirror plane and the substrate (Figure 10b). However, the micromirror can be held in various intermediate angles between 0° and 90° by balancing the restoring force due to residual stress in the hinge and the electrostatic force by applying a tailored voltage between the mirrors and the transparent counter electrode before reaching the critical pull-in angle. In this critical position, micromirrors are closing abruptly (falling in a position that is parallel to the substrate) due to the pull-in effect. This phenomenon is typical for electrostatic actuation and is also known as the snap-in effect.

Based on the window orientation, latitude of the building, season, daytime, and user presence, a nearly unlimited number of scenarios is possible. Four of those scenarios are displayed in Figure 11: Two of the micromirror states enable daylight guiding by explicit reflection towards the ceiling in the room, one reflection scenario to a wall, and a completely closed state where the micromirrors are parallel to the windowpane due to electrostatic force to block the solar radiation. The solar radiation is reflected outside by switching all the micromirrors vertically in summer (very high solar impact) without any user in the room to keep the room cold. An automated control unit with data from an intelligent network sensing system that observes the position of persons, sun position, inside temperature, and light ambience is therefore fed to execute the functionalities according to the requirements. During summer, when there is no user in the room, all the micromirrors are placed in a closed position (parallel to the windowpane), saving huge cooling energy by minimizing the heat transfer into the room (Figure 11a). When user presence is detected during summer, some of the micromirrors (upper areas in Figure 11b) are kept open to reflect light towards the ceiling area above the user. However, the lower micromirrors are closed to keep the unoccupied room area cool, thus allowing energy saving by limiting the heat transfer into the room since only a certain number of mirrors are open.

In addition, the illuminated spot on the ceiling can be moved with the user; thus, the inside area far away from the window can also be illuminated efficiently (Figure 11b). During winter, with no user detected in the room, all the micromirrors are open to harvest energy efficiently by using all the solar radiation to keep the room warm, similar to thermal component activation, thus saving the huge amount of energy required for heating (Figure 11c). When a user is present during winter, the solar radiation is directed towards the ceiling by keeping all the micromirrors open, but with different angles using tailored actuation voltages. This saves energy needed for heating and lighting at the same time (Figure 11d).

For an overview of all our publications, see refs. [27,28,29,30,31,32,33,34]. All our obtained results are compiled in Table 5 in ref. [28]. Most of the fabricated MEMS micromirrors for smart glass applications in buildings had dimensions of 400 µm in the direction parallel to the hinge and 150 µm vertical to the hinge. Here, a brief compilation of the most important results is given.

• Typical electrostatic actuation voltages between 40 V and 60 V were measured. In a single case, an actuation voltage as low as 12 V was measured, and in some cases, 80 V, depending on details in the hinge and the metallic layer stack. Thus, a typical voltage is 50 V.
• Switching micromirror arrays from the open to the closed position takes about 0.1 s for arrays with about a sixth of a million mirrors.
• Micromirror arrays were housed in double or quadruple insulation glazing and tested at different temperatures. Between −80 °C and +120 °C, the MEMS micromirrors showed proper actuation. Beyond these extreme temperatures, measurements could not be performed due to the limitations of the used climate chamber.
• The lowest transmission measured in a closed state was 0.01%. The largest transmission measured in an open state was 73%.
• The lowest power consumption in the holding status was 0.2 mW/m² and < 0.4 mW/m² during switching.

Different rapid aging tests have been performed on our MEMS smart glass to investigate reliability and to estimate lifetime. Surviving all those different tests, a long and stable maintenance-free lifespan of 40 years for the smart glass application has been extrapolated. The following individual tests have been performed:

• Long-term electrostatic actuation at 4 kHz was performed, which corresponds to 38 billion open–closed–open cycles.
• Multiple fast temperature cycles (0 °C to +80 °C to 0 °C to +80 °C to 0 °C to +80 °C and further periodic repetitions) were performed over 225 h without any measurable damage.
• Tests under extreme UV radiation during temperature cycles were performed with double insulation glazing. In these cases, the sun spectrum was emulated by UV LED arrays with many different wavelengths. No changes in the mirror performance were measured after 1600 h of extreme UV exposure, where ten times higher irradiance values were chosen in all spectral parts of the emulated solar UV spectrum than the highest exposure values observed in big global cities.
• Mechanical vibration treatment at 3278 Hz was performed over 31,000 h, without observing mirror damage.

4.2. MEMS Microshutter Arrays for Laser Safety Goggles

In contrast to the planarized metallic MEMS micromirror arrays presented in Section 4.1, metallic MEMS structures that are curled (rolled, i.e., non-planar) in their relaxed state, in contrast, are called microshutter arrays [35,36,37,38,39] or micro blinds [40]. Transmission modulation has been reported for these MEMS microshutter arrays for applications in smart glass [35,41], space instrumentation [36], camera shutters [38], and displays [39,41]. These MEMS microshutter arrays are curled in the open state (first state) and unrolled in the closed state by electrostatic actuation to reveal the flat condition (second state). These microshutter arrays reveal just two scenarios: completely open or completely closed. Pizzi et al. [38] developed curled microshutter arrays for car display applications. A similar system was reported using curled microshutter arrays [41], which are implemented between window planes for modulating sunlight entering rooms through windows or facades. In all these microshutter arrays, each shutter moves synchronously with the neighboring ones. In contrast, as shown in Section 4.1, our MEMS micromirror arrays enable active light steering for personalized lighting. Between the open and closed states, they allow several intermediate states. Light steering is possible since the mirrors are planarized and can be actuated into different tilt angle states.

Recently, Lamontagne et al. [42] published a laboriously reviewed article on microshutters, giving a valuable overview of that class of technology. That review also included our micromirror arrays. We updated the tables of Lamontagne et al. [42] in ref. [29] but a further update is incorporated into this review to include recent values, as presented in

Table 1. Comparison of different MEMS shutter technologies (update of ref. [42]): materials, size of single shutter elements, experimental characterization results, and potential applications [43]. The question mark (?) in the table indicates that no information is available at that position. The table includes refs. [27,29,30,31,32,33,34,35,37,38,39,42,43,44,45,46,47,48,49,50].

Research Group	Time Period	Bottom Electrode	Top Electrode	Mirror Size	Actuation	Speed To Close	Array Size	Max/Min Transmission %–Contrast	Applications, Comments
Fiat [35,36]	1999–2005	ITO	flexible metal layer	458 µm–2.4 mm	20–100 V	0.1 ms	-	20/1–20	Micro-shutter-based automotive display, IR spectrometry, low transmission
MCNC [44]	2000–2002	ITO	polyimide/Cr/Au/ polyimide	<100 µm to >200 µm	100–300 V	18 µs	5 cm2	low contrast	eyelid for protection
Kassel University [27,28,29,30,31,32,33,34]	2003–present	FTO, ITO, or Ag low-e	SiO2-SiNx/Al-Cr-Al/Al-Cr-Al-Ge	150 × 400 µm2	12–80 V	0.1 s	0.87 m2	73/0.01–7300 high contrast	sunlight steering for buildings, high contrast, low voltage
Kassel University [32,33]	2018–present	FTO, ITO	Al-Cr-Al/Al-Cr-Al	60 × 1000 µm2 100 × 1000 µm2	50–80 V	1 µs	7 cm2	77/0.01–7700 high contrast	Laser safety goggles, high contrast, high closing speed
NRC [42,45]	2005–present	SnO2, ITO, Ag low-e	Cr and others	50–300 µm2	15–25 V	40 µs	20 cm2	60/0.1–600 high contrast	high contrast, low voltage
NVMG [36,46]	2007–present	TCO	Shrinkable Polymer	≥2 mm	100–500 V	Seconds	5000 cm2	low contrast	Macro-curling shutter, commercialized
INO [39,42]	2008–2009	Al	MoCr	60 × 1000 µm2	110 V	2 ms to close, 7 ms to open	0.25 cm2	low contrast	Space instrumentation
Air Force [43]	2008	AlZnO	Ti and Au	200 × 50 µm2	-	-	-	40/1–40	Adaptive coded aperture imaging
Samsung [37,47]	2009–2011	ITO	Al-SiNx, Mo-Mo	ᴓ 2.2 mm, 36 × 1.4 mm long triangals rolled	30 V	2 ms	Iris of 0.04 cm2	?	Iris shutter for camera
KAIST [48,49]	2010–2016	ITO	Electroplated Ni	200 × 160 µm2	20-30 V	20 µs to close	small	60/ ?–?	active transparent display with TR-OLEDs
University of Tokyo [41]	2015–2016	ITO	Al-SiO2	200 × 30 µm2	38–55 V	3 ms	-	53/36–1.5	implemented on TFT
Stuttgart University [42,50]	2016–present	MoTa	MoTa on stressed SiNx	200 µm	20–60 V	-	2–225 cm2	low contrast	Transmissive display on TFT, low transmission

.

As seen in Table 1, the Fiat research group developed a micro-shutter array with a flexible metal layer for automotive displays and IR spectrometry. This was a pioneering work, but the array revealed low transmission, unfortunately. Different array sizes were fabricated, with a closing time of 0.1 ms and a contrast of T_max/T_min of 20.

Another example is from Samsung, which investigated iris shutters for cameras using a combination of metals and dielectrics. With the rolled shutter, the 2 ms speed for closing the shutter is relatively slow.

For eyelid protections, the team of MCNC investigated a microshutter array. But the contrast was also low.

For our MEMS smart micromirror glazing, we measured a minimum transmissivity T_min of 0.01% and, recently, a maximum transmissivity T_max of 73%, including the glass substrate with the transparent conductive oxide (TCO) layer. This provided a modulation contrast of T_max/T_min = 7300 between open and closed states.

There are many investigations from various groups that have been working on microshutter arrays for decades. After working for a long time on MEMS micromirror array for smart windows in building, the team from Kassel University also entered the field of microshutter arrays, with a concept for an effective fast eye protection, especially active laser safety goggles. For our MEMS microshutter arrays presented in this section, a mini mum transmissivity T_min of 0.01% and, recently, a maximum transmissivity T_max of 77%, including the glass substrate with the transparent conductive oxide (TCO) layer, wasobserved. This provided a modulation contrast of T_max/T_min = 7700 between open and closed states.

The active laser goggles are designed to protect against terroristic laser attacks against pilots, tram and bus drivers, police, and security personnel. An important application is protecting the health of the human eye against laser strikes during traffic use. It is very important to help solve this still-increasing problem, which has existed for more than a decade and has still not been solved satisfactorily. The goal is to solve this problem with active laser goggles via achieving ultrafast closing of the MEMS shutters after photodiodes have identified laser radiation.

Passive laser safety goggles deliver much less than active ones. On the market are conventional passive laser safety goggles, which are based on optical filter technology. Each type of goggle can only absorb one or a few designed wavelength ranges. The transmitted spectra are passively selected by the filter itself, which leads to a coloring effect when it is designed for visible laser wavelengths since some wavelength ranges are subtracted from the white spectrum. Another weak point of passive laser goggles, all with coatings, is that a single scratch going through the coating stack destroys the protection.

Which application is considered is important, e.g., for any traffic use (pilots in the cockpit, bus and tram drivers), passive laser protection is not allowed to be applied. The reasons are that passive laser goggles have (i) unacceptable low transmission during night use, when attacks mostly happen, and (ii) the pilot cannot see the correct color of displays or indication LEDs or does not see important details at all. Each single frequency range to be blocked by the passive laser goggles makes the multilayer coating more complex and less transmissive. Layer attacks against pilots have been reported for green (532 nm), red (640 and 650 nm), blue (405 nm), and others. Starting from green and adding all these colors to be suppressed, the overall transmission significantly decreases (down to 35%) and changes the color recognition very negatively. Therefore, passive laser goggles are not allowed to be used for any traffic and even less for night use. This has motivated us to develop MEMS microshutter arrays for laser safety goggles with adequate sensing and control technology that is able to block laser light of any wavelength exceeding defined eye safety levels. The active laser technology enables not only the two problems mentioned above to be solved but also to keep safety, security, and health high without penalizing them. In summary, MEMS microshutter array goggles are color neutral, have much higher transmission, and reveal much higher optical density (OD) in the closed status compared to most passive laser goggles.

Our metallic MEMS microshutter arrays in laser safety goggles are normally in a totally open state and should close on request with a high switching speed. The arrangement of these shutters in an array connected via a grid enables the shutters to move synchronously with the neighboring ones. The microshutter arrays in laser safety goggles enable the light to transmit to the glasses and reach the human eyes in normal open conditions. In case of a laser attack, they close fast and block incoming light of all wavelengths totally in the closing state, since laser light can cause serious damage to eyes. The shutters reopen after the laser radiation passes and enables free sight of the eyes.

In the default state, microshutters are open and vertically stand on the substrate, as shown in Figure 12, which displays three viewpoints. Figure 12 depicts three SEM micrographs in different perspective side views of microshutter arrays in an open state after sacrificial layer removal and a drying process. In Figure 12a, the anchor region is on the right side of the lower part of the microshutter; thus, the top part comes from the left during the standing-up process. In comparison to Section 4.1, the planarized MEMS shutter blades are much longer in the direction parallel to the hinge to increase the value of maximum light transmission. This is due to a lower number of inter-spaces between neighboring shutters. Figure 12b shows a modified perspective to make the anchor regions visible. For better understanding, the red arrow is pointing in the long direction of the anchor of a shutter. All the inter-spaces between the shutters represent opaque regions and represent the contacting grid. Figure 12c provides another perspective top view of microshutter arrays. Here, the anchor region is on the left side of the lower part of the microshutter; thus, the top part comes from the right during the standing-up process.

Figure 13 depicts the applied electrostatic actuation voltage (blue) and the light intensity transmitted through the microshutter array (orange) as a function of time. The measurement presents the highest closing speed of our microshutter arrays measured up to now. In this case, the dimensions of a single microshutter are 2000 µm × 40 µm (L_x and L_y) and the step voltage profile switches from 0 V to 80 V. Our simulations have shown that by reducing the microshutter dimensions (L_y) perpendicular to the hinge, the switching time decreases [51]. However, the height of the actuation voltage also impacts the closing time of the microshutter arrays. Very low actuation voltages and fast closing cannot be obtained at the same time [51]. A compromise must be found. The shutters could be either optimized towards ultra-fast closing accepting high voltages only or optimized towards low actuation voltages tolerating slower closing.

Different dimensions have been explored for our microshutter arrays for laser safety goggles to investigate the compromise between switching time and actuation voltage—identification of the best solution is still under investigation.

Microshutter arrays consist of thousands of microshutters in a specialized format. With high contrast and high closing speed, microshutter arrays provides a promising solution for safety usage. Depending on the detecting photodiode, microshutter arrays can be used for all wavelengths or the required spectrum as laser safety goggles. The most significant benefit of microshutter arrays for laser safety goggles is that the metallic blades reflect lasers of all wavelengths and are color neutral.

4.3. MEMS Ring Shutters for Interference Microscopy (and Endoscopy)

Utilizing the platform technology of our MEMS micromirror arrays, we developed ring-shaped shutter arrays with subfield addressing to expand potential applications towards, e.g., interference microscopy or miniaturized imaging (endoscopy) [52]. A schematic of the prototype is shown in Figure 14. In contrast to competing approaches (MEMS [47,53], optofluidic [54,55], electrochromic [56], liquid crystal elastomer [57,58]), this iris-like tunable aperture allows the actuation of different rings and even quadrants of rings, e.g., if in Figure 14 we bias contact C and contact 2 at the same time, the subfield C−2 is addressed.. Despite the concentric arrangement of shutter elements resulting in variations in shape and size, the ring-shaped array shows convincing homogeneity in the opening angle, as depicted in Figure 15a, for size parameters of angular width L_φ = 150–600 µm and radial length L_r = 60–300 µm. To match future application requirements, the design is adapted according to expected device properties [51]. Figure 15b shows an example of a promising design with an adjusted aspect ratio L_r/L_φ = 1, also leading to well-planarized shutters. Further details about design, fabrication, and characterization are the subject of a forthcoming publication.

In summary, this section demonstrates that transparent substrates enable many new applications that were previously not possible when using non-transparent substrates, increasing the versatility of micromirror arrays and the related application diversity enormously. Note that MEMS smart glass based on metallic micromirror arrays has the potential to save huge amounts of energy.

5. The Potential of 2D Materials for Photonic Applications

This section will review the new and thrilling class of 2D materials. Many different types have been found and investigated after the field was opened with the study of graphene. MXenes are 2D inorganic compound, consisting of atomically thin layers of transition metal nitrides, carbonitrides, or carbides. Fabrication of and investigations on MXenes were first published in 2012. Interestingly, among many applications (see Figure 16), reflective properties to be used in micromirrors can also be found.

The excellent electro-optical tunability [59], high reflectivity [60], better broadband performance [61], lightweight and flexible nature [62], and compatibility for integration with other materials make 2D materials a favorable choice for several photonic devices [63]. This includes photodetectors [64], photovoltaic devices [65,66], light-emitting diodes [67], and sensors [68]. Like other photonic devices, integrating 2D materials into NEMS and MEMS mirrors holds vast possibilities for diverse applications, particularly within optoelectronics and sensing [59]. Van der Waals (vdW) heterostructures—laterally stitched layers of 2D materials [69,70]—especially exhibit remarkable optical characteristics yet to be fully exploited.

Papadakis et al. demonstrated an ultralight Angstrom-scale perfect optical mirror using a graphene-based layered structure [71]. Employing ab initio techniques and optical transfer matrix computations, the authors reported that the graphene-based heterostructures exhibit prolonged electron relaxation duration, leading to enhanced optical characteristics compared to noble metals like gold and silver, specifically in the mid-infrared spectrum. These materials exhibit reflectivity exceeding 99.7% at a fraction of noble metals’ weight. Graphene–boron nitride (G-hBN) vdW heterostructures have also been investigated, and it was observed that there is a significant improvement in the plasmonic mode confinement and quality factors compared to the conventional metals Ag/Au. The photonic properties of the reported vdW heterostructures can also be actively tuned via chemical and electrostatic doping. The nearly perfect reflectors can be potentially used in designing compact waveguides, in engineering emission for back reflectors in solar cells, and for macroscopic applications such as aircraft.

Furthermore, owing to the unique and flexible optical properties of monolayer transition metal dichalcogenides (TMDCs) [72,73], these materials are receiving widespread attention for various photonic devices, including micro-/nanomirrors. Zeytinoglu et al. [74] have demonstrated that a monolayer TMDC can be an ideal mirror with an atomic-scale thickness.

The authors directed a laser field (both collimated and coherent) towards a monolayer of transition-metal dichalcogenides (TMDCs), as shown in Figure 16. They observed a two-mode squeezed state (TMSS) due to the exciton–exciton interactions in addition to the transmitted and reflected coherent fields. The destructive interference between the field directly transmitted, and the field produced by the TMDC excitons suppressed transmission and amplified reflection. In the case of the perfect coupling between the incidence lines and excitons generated through TMDCs, the system will behave like a perfect mirror.

Van de Groep et al. [75] have presented the role of excitonic resonances in atomically thin optical lenses. The article also illustrates electrical gating techniques to turn exciton resonance on and off by enhancing focusing efficiency using the excitons in optical lenses. A tungsten diselenide (WSe₂) monolayer has been used to synthesize the tunable optical element using chemical vapor deposition (CVD) techniques on a sapphire substrate. A monolayer graphene is transferred to the top of WSe₂ to improve the DC surface conductivity, achieve uniform gating on the WSe₂, and enhance long-term stability.

Ma et al. [76] studied the Reststrahlen effect in hexagonal boron nitride across different thicknesses experimentally and theoretically. The Reststrahlen effect is vital in designing atomically thin NEMS mirrors that predominantly use 2D materials.

Figure 16. Design setup for a TMD monolayer-based optical mirror with excellent reflectivity. Reproduced with permission [74].

Figure 16. Design setup for a TMD monolayer-based optical mirror with excellent reflectivity. Reproduced with permission [74].

A heuristic time-domain model and a Fourier-transform infrared spectrometer have been used to depict coherent radiative decay and estimate absolute reflectance. The article reported that the high reflectivity plateau evolves into a single peak in the optically thin limit. Two distinct regimes emerge at this thickness: a strong-response regime dominated by coherent radiative decay and a weak-response regime dominated by damping. The researchers also developed a simple 2D model that can be applied to a wide range of thin media.

The optoelectronic behavior of 2D materials can be modulated extensively using different engineering approaches, such as making vdW heterostructures, creating defects with the induction of strain and electric fields, surface passivation, etc. An exciting study has been presented by Patrick Back et al. [77], where the researchers designed electrically controlled, atomically thin, resonant mirrors based on TMDC monolayers. In this work, the researchers employed an hBN passivated monolayer of molybdenum diselenide (MoSe₂) placed on a 285 nm thick SiO2/Si substrate. Platinum (Pt) electrodes contacted the MoSe₂ monolayer, while Si was a back gate. When monochromatic laser light was tuned with the exciton transitions in MoSe₂, a significant reflection was observed at the cryogenic temperature. After normalizing the reflected intensity using the measured reflectance of the bare substrate and the metal electrodes, the mirror reflected up to 85% of the incident light.

In a similar study, Daniel et al. [78] have reported that the TMDC materials can be atomically thin mirrors for light confinement. The article illustrated an experimentally verified new technique for achieving nanometer-thick planar optical cavities with inherent chiral properties using two atomically thin TMDC mirrors as the primary photonic components. The device has been designed so that two MoSe₂ monolayer mirrors are sandwiched within the layers of h-BN encapsulation, with a total h-BN thickness of 240 nm. The monolayers are symmetrically positioned at a distance of 60 nm from the top and bottom of the vdW heterostructures, respectively. The schematic representation for the design set-up is shown in Figure 17. The TMM (transfer matrix method) simulations carefully predicted the design parameters. It was observed that the electromagnetic mode was generated in the system from the effective optical excitation and recombination of excitons in the two TMD mirrors, in contrast to the traditional Fabry–Pérot interferometric cavities. Due to the excitonic nature of the cavity’s mirrors, spin-polarized cavity modes, which are bifurcated due to the valley Zeeman effect under an external magnetic field with adjustable energies, were observed. The standing optical mode was also observed at a wavelength of λ = 746 nm, and the FDTD simulation further confirmed that the quality factor for this mode is about Q ≈ 1060 at the resonance wavelength. Additionally, the authors reported an excitonic saturation of the optical mode based on pump power, displaying its resilient adjustability through temperatures up to 100 K.

Rui and his co-workers [79] have suggested a new approach for light–matter interaction at the atomic scale. The article demonstrated the development of a subradiant optical mirror using the single-structured atomic layer. The authors observed the subradiant response from the mirror, which was generated due to the interference of the scattered field. The response can be tuned by tuning the interatomic distance and the external field. Using spatially resolved spectroscopic measurements, the researchers claimed that the array can act as a high-quality mirror formed by a single monolayer of a few hundred atoms.

We aim at functionalizing our MEMS metallic micromirror smart glass (Section 4.1) with the fascinating properties of these new 2D materials. This will be the subject of forthcoming publications.

In summary, this section shows how the use of 2D materials can also contribute a great deal to micromirrors in the future and again enhances the versatility.

6. Comparison of the Different Designs and Technologies

6.1. Dimensions, Complexity, Speed, Miniaturization, and Reliability

The micromirror technologies selected for this review paper are very different in materials, design, performance, and application. The selected materials, design complexity, and technical parameters are strongly related to the requirements of the application and the price that can be tolerated in that field.

Table 2. Comparison of different MEMS shutters: material, number of micromirrors, array size, size of single shutter elements, reflectivity, switching time, miniaturization potential, reliability. The table includes refs. [6,18,19,26,27,28,29,30,31,32,33,34,80,81,82,83,84,85,86,87].

No.	Materials and Type of Arrays	Number of Micromirrors	Array Size	Dimensions of A Single Mirror	Reflectivity %, Wavelength Range In nm	Switching Time	Miniaturization Potential of a Single Mirror	Reliability, Lifetimes
1	Dielectric DBRs [6]	384 mirrors	90 mm2	40 µm × 40 µm	99.8 for 400–1700 nm ***	-	6 µm × 6 µm for VIS	high, very long for static DBRs
							13 µm × 13 µm for NIR
2	Si micromirrors with gold, dielectric coating, N × N switch [80,81]	64 (8 × 8)	100 mm2	Ø = 400 µm		500 µs mirror itself
		36 (6 × 6)
						560 µs with springs
3	Si micromirrors λ-router [16,19]	256 (16 × 16)	2.5 cm2	Ø = 500 µm		<5 ms		18 billion actuation cycles, long
		1296 (36 × 36)	≈36 cm2
4	Si micromirrors DMD [82,83,84,85,86,87] *	1024 × 768	≈14 × 10.5 mm2	13.68 µm	89 for 420−680 nm 88 for 363−420 nm	12.5 µs		>500 million actuation cycles per mirror, >20 years **
		1920 × 1080	≈20.7 × 11.7 mm2	10.8 µm		12.5 µs
		2560 × 1600	≈19.4 × 12. mm2	7.56 µm (± 12°)		16 µs
		3840 × 2160	≈17.3 × 9.7 mm2	9 µm (± 14.5°)		8 µs
		1920 × 1080	≈ 10.4 × 5.8 mm2	5.4 µm (±17°)		10 µs
5	InP multiple airgap micromirrors [26]	306	70 mm2	Various Ø=10 µm–80 µm	99.995 for a DBR micromirror with 4.5 InP/air periods ***	Simulated <10 µs	Membrane Ø = 13 µm for λ = 1.55 µm
6	Metal micromirror arrays for smart glass [27,28,29,30,31,32,33,34]	>1,200,000	29 × 29 cm2	400 µm × 150 µm	≈90 at 600 nm	≈0.1 s	not recommended for smart glass in buildings	31,000 h vibration tests, 37 billion actuation cycles
		>12,600,000	126 × 69 cm2
7	Metal microshutter arrays for safety goggles [32,33]	up to 10,000	elliptical main axis 3 cm and 2.5 cm	1000 µm × 100 µm and many other dimensions	≈90 at 600 nm	1 µs	20 µm
8	Metal microshutter arrays for ring shutters	3000–10,000	Ø = 18 mm	Lφ = 150–600 µm Lr = 60–300 µm (various dimensions)	≈ 90 at 600 nm	≈10 µs interpolated	not required for interference microscopy

* Number of micromirrors and array size varies with the required resolution, listed numbers are typical-used number. ** the real value might be higher, most probably. But this value can be found in the literature. *** see Figure S1 in the supplementary material.

provides a comparison of switching speed, number of mirrors, mirror dimensions, array sizes, miniaturization potential for individual mirrors, reliability, and lifetime.

According to Table 2, the complexity of micromirror arrays is high, and the number of mirrors per array is from 36 to 64 mirrors for the λ-router, nearly 400 for DBR micromirror arrays for spectroscopy, over 8,000,000 for DMD applications, and up to 12,000,000 for metal micromirror arrays for smart glass applications.

DBR mirrors show by far the highest reflectivity. According to the supplementary material of this review, dielectric SiO₂/Si₃N₄ DBRs reveal a reflectivity of a maximum of 99.9%. The highest reflectivity of 99.995% can be achieved by InP multiple airgap micromirrors with 4.5 InP/air periods. The reflectivity of metal micromirrors is about 90%. In comparison, pure silicon micromirrors have limited reflectivity. However, it has been improved by coatings with metals [88] or dielectrics.

Another noticeable point is the switching speed of the MEMS mirrors. The switching time of metal micromirrors for smart glass is about 0.1 s, which is sufficient for personalized lighting since it is related to maximum human walking speed. For laser safety goggles, a value of <10 µs is mandatory. For metallic microshutter arrays for laser safety goggles, shutter closing times of down to 1 µs have been recorded up to now experimentally.

As seen from Table 2, miniaturization of the micromirrors is not mission impossible. Multiple factors contribute to this, for example, materials, fabrication technologies, application requirements, etc.

6.2. Hinges of Micromirrors: Geometry, Methodology, and Fabrication Challenges

In case MEMSs are involved, these technologies are also very different with respect to the hinge geometry and hinge methodology. The following numbering is related to the numbering in Table 2 (first column).

• In the DBR micromirror arrays in the example chosen above, the FP-based nanospectrometer includes no hinges since they are static.
  If very high reflectivity >99.5% is required, the number of periods must be larger than 14.5, and the optical absorption inside the two dielectric layers must be <20 cm⁻¹, which is a challenge. If the SiO₂/Si₃N₄ DBRs and a reflectivity of 99.9% are required, a low absorption of even 10 cm⁻¹ has to be achieved, which is an extreme challenge for deposition machines. The minimum required number of periods is 15.5 for that. In such a case, it would be better to use a dielectric material system with much higher refractive index contrast such as SiO₂/ZrO₂ or SiO₂/TiO₂. For more details, see the Supplementary Materials (Figure S1).
  The potential minimum space requirement for a spectrometer with static dielectric DBR micromirrors might be 0.07 mm². This is estimated for a wavelength range λ = 400–800 nm and a single micromirror pixel size of 10 × 10 μm² and TiO₂/SiO₂ dielectric stacks.
  The fabrication challenges lay in the requirements for the multilayer deposition to keep the grating period as well as the composition highly constant throughout the whole DBR stack. This requires precision engineering. Using the nanoimprint over several DBR mirror stacks (to combine spectrally different stopbands) [6], this technology is scalable.
• The hinges of the N² micromirrors used in the fiber-optic N×N switches are the only ones that are built like a classical door or window hinge, with a pin, a cylindrical airgap, and a surrounding cylindrical barrel. The fabrication requires three sub-steps, amorphous silicon is used for all remaining parts, and SiO₂ is used as the sacrificial layer material. The technological process is displayed in Figure 5e. The fabrication process is similar to that described in refs. [3,4,5,89]. One of the challenges is to reduce friction inside the hinge and to overcome the problems of diverging light beams. In addition, it should be noted that this technology does not allowall arbitrary combinations of simultaneous switching processes from input to output.
• The gimbal mount of the 2N micromirrors of the λ-router uses four cantilevers as torsion blades for each mirror. The challenge is to implement the required angle accuracies. Precision engineering is required here.
• Each micromirror of a DMD array uses two cantilevers as torsion blades and two landing tips to stylize a distinct orientation and position. In the projection or dump mode, each mirror plane is defined precisely by three points.
  One of the multiple challenges that the engineers had to solve was filling the square central hole existing at the beginning.
• The InP multiple airgap micromirrors use four, three, or two cantilevers for each membrane. One end of each cantilever is connected to a solid supporting post, and the other end is deflected vertically to the substrate plane. Four suspensions show the largest stability, three suspensions show the best compromise between stability and wider tunability, and two suspensions reveal the widest tunability but are unstable and can undergo unwanted tilt.

The fabrication challenges in this technology are layer stresses, buckling, and membrane deformations out of the flat planes. To obtain very high reflectivity and wide stopbands is not a large challenge in the InP multiple airgap DBR mirror. Using only 3.5 periods of InP–air already provides 99.99%. A reflectivity of 99.995% is obtained with 4.5 periods. The saturation at 99.996% is due to residual absorption. Note that these are the lowest possible absorption values obtained in state-of-the-art high-quality undoped InP. For these values, abrupt air/InP and planar membranes are considered. These very high reflectivity values are excellent and require very few periods, which is a big advantage. For more details, see the Supplementary Materials (Figure S1).

The potential minimum space requirement for a spectrometer with MEMS-tunable InP multiple airgap DBR micromirrors might be 0.05 mm². This is estimated for a wavelength range of λ = 1100–1900 nm, a single membrane diameter of 10 µm, and a potential λ_InP/4 or λ_InP/4 configuration in the InP/air DBR stacks, revealing a spectral stopband width of >1300 nm, enough to cover the above-mentioned 800 nm wavelength span.

Here, the challenges are to control unwanted membrane bending due to internal material stress. This is unexpected, since InP on InP should be lattice matched. But it originates from, e.g., arsenic carry-over in metal–organic–chemical–vapor phase–epitaxy from the GaInAs sacrificial layers. That means that due to the growth direction, GaInAs/InP and InP/GaInAs interfaces are not equivalent, resulting in asymmetries in the composition of the membranes. Specific obstacles in marked penetration might be that the device yield is limited by residual stress influence.

6.

The metallic MEMS micromirror in arrays for smart glass applications in building facades or windows uses blades as hinges that are rolled up and unrolled. The mirror part is planarized and can be tilted in the desired angles via electrostatic actuation. Here, the challenges are to control material stress in the metallic sublayer stress to enable precise hinge bending angles and to allow planarization. In addition, upscaling is challenging.

7.

The metallic MEMS microshutter in arrays for safety goggles uses blades as hinges that are rolled up and unrolled. The shutter part is planarized and actuated electrostatically. Here, to obtain different hinge orientations and high area filling at the same time with no interference and planarity is very challenging.

8.

The metallic MEMS micromirrors for ring shutters in interference microscopes and endoscopes use blades as hinges that are rolled up and unrolled. The mirror part for the ring shutter is planarized and can be tilted in the desired angles via electrostatic actuation.

In summary, the challenges of Nos. 5–8 are in the planarization of the mirror/shutter plane and the definition of distinct elevation angles of the mirror/shutter plane implemented by the rolling status of the hinges in unactuated mode. This requires precision engineering. No material incompatibilities have been observed. No extraordinary specific obstacles can be seen.

7. Outlook

In the past, very successfully, Si-based MEMS micromirror arrays have offered new ways of switching optical signals in optical communication with high complexity. Above all, Si-based DMD MEMS arrays for digital projection have shown a major impact in penetrating beamer technology (digital light processors). But DMD technology also caused a revolution in bio-imaging systems and led to outstanding economic developments. DMD arrays proceed from digital beamer projection and spatial light modulators into the huge medical field via biomedical instruments. Note that DMD technology for biomedical applications [90] is also highly interesting for optical diffraction tomography, interference microscopy, structured illumination microscopy, quantitative phase microscopy, tomographic phase microscopy, ophthalmoscopy, and others. Also for optical spectroscopy, digital holography, Raman spectroscopy, bio-printing, lithography, surface plasmon (polariton) resonance techniques, and others, DMD methodology is very attractive.

A massive economic penetration was already happening. Since metallic MEMS micromirror arrays on glass substrates can be scaled up potentially to much larger sizes as Si-based MEMSs, a huge impact on economy and society is likely. Note that Si MEMSs are limited to Si wafer sizes. Since the main focus of our review is on metallic MEMS micromirror arrays, more real-world use case studies are added to provide a better understanding of potential impact and benefits in society and economy. Metallic MEMS micromirror arrays in double-pane modules can be used for renewable energy technologies in light-concentrating systems such as heliostat mirror parks, dish collectors, trough collectors, linear concentrator solar power plants, and others. These metallic MEMS arrays can also be used in smart headlights of cars and other mobile systems. Staying with cars, they are also highly attractive for novel laser radar systems. This review presented applications in laser safety goggles against terroristic attacks, as well as in ring shutter arrays for interference microscopy and endoscopy.

The biggest significance, however, we see in smart glass applications in buildings. Buildings are responsible for 40% of the primary energy consumption and 36% of the total CO₂ emissions. There is a huge potential to decrease energetic consumption for lighting and heating/cooling by substituting the 85% of inefficient glazing areas in Europe and North American buildings with energy-efficient smart-glazing windows. In Asia, South America, and Africa, similar numbers or even larger percentages could apply. At the moment, the Chinese government forces public buildings and industry to achieve much better insulation in building windows and facades. The metallic MEMS micromirror technology for smart building facades, which we describe in this review, reflects instead of absorbs the incoming sunlight according to user actions, sun positions, and daytime-/season-variable requirements, providing a personalized light steering inside the free solar heat in winter and overheating prevention in summer permanently with healthy natural daylight, leading to huge energy savings (up to 35%) and massive CO₂ reduction (up to 30%), as well as savings of 10% steel and concrete in high-rise buildings. Two unique safety features enable vision in case of electricity failure and trendsetting digitalization for smart homes. The technology can be used in new and old buildings and could play a key role in smart green cities with huge economic, social, and environmental benefits.

The use cases for metallic MEMS micromirror smart glass are not only limited to personalized light steering and energy management in buildings but are also for security and safety. The MEMS smart glass can be used as black/white displays to make the life of elderly people safer. Our metallic MEMS smart glass integrates a twofold safety feature: (1) In case of power failure, the window will be transparent and not blacked out/obscured as with other smart glass technologies. This will allow, in case of accidents (e.g., fires), intervisibility and communication with people in need inside the building and emergency forces outside the building. (2) Thanks to the operation with movement sensors, our product can raise the alarm in case of the collapse of a person in the room. In case a person accidently drops and stays in the same position for more than 5 min, our system identifies (knowing that that position is not the bed) an emergency case, alerts rescue services and close relatives, and displays in all the windows “I need help”—readable from outside. This can be especially helpful in environments such as hospitals, nursing homes, and elderly people’s apartments, considering that 73% of elderly people’s accidents/falls happen inside their apartments or inside residences.

Materials always have driven innovation and novel applications. The authors posit a careful statement: Due to limitations arising from Si wafer sizes, Si MEMSs might not really reach very-large-scale sizes. This is not the case if we withdraw from Si substrates and proceed towards MEMSs on glass panes. In ref. [34], Ge is demonstrated as a nm-scale coating for switchable color effects on huge glass facades, and 2D materials come into consideration in Section 5 of this paper. The company Nanoscale Glasstec GmbH in Germany [28,91] is working towards new applications using metallic MEMS smart glass and is upscaling glass substrates now to a few m². Very interesting advancements in materials and fabrication techniques such as nanoimprints and 3D imprints are going on now worldwide, and this will definitively drive innovation in this field.

8. Conclusions

The diversity of materials enabled the versatility of micromirror arrays with respect to their various optical applications. DMD mirrors exhibit exceptionally high reflectivity. InP multiple-airgap DBRs combined with specifically designed MEMS suspension hinges allow very large spectral tuning of Fabry–Pérot filters, high miniaturization potential, small required space, high resolution in spectrometers, and high flexibility. These characteristics of micromirror arrays aim for applications in optical communication and strongly miniaturized high-resolution spectrometers and sensors. Si has outstanding mechanical properties and Si MEMSs are compatible with CMOS technology, allowing integrated electronic circuits including complex optical MEMS arrays. The controlled flexibility makes them particularly appealing for applications such as digital projection, adaptive optics, and optical switching, where ultra-precise mechanical moments are imperative. Compound semiconductors offer the advantage of integrating LEDs, lasers, and photodiodes for the NIR range but with reduced complexity. This review also focuses on fabrication challenges and miniaturization potentials and requirements.

The dimensions and the array size of metallic MEMS micromirrors or microshutters exhibit strong variations based on their intended applications. Therefore, they have widespread applications in daily optics, such as smart window glass, personalized daylight steering, heat management in buildings, active laser safety goggles, ring shutters for endoscopy, and others.

Owing to the extraordinary optoelectronic properties of 2D materials, extensive attention of researchers has been drawn to designing atomically thin optical mirrors. Recently, many interesting excitonic phenomena such as the Reststrahlen effect and the optical Kerr effect (OKE) have been observed in atomically thin 2D layers, and exceptional tunability and flexible characteristics demonstrated by various 2D materials such as MXenes, TMDCs, and van der Waals heterostructures accelerates the research interests in photonic devices, including micro-/nanomirrors. Atomically thin optical mirrors can potentially be used for designing the back reflectors in solar cells, compact waveguides, macroscopic applications such as aircraft, and various sensing applications. Including these attractive properties in MEMS micromirror technology would be a worthwhile goal.