1. Introduction

Owing to the extraordinary mechanical, thermal and optical properties, single crystal diamond is an excellent material for micro-optical components for high-power applications such as lasers [1], spectroscopy [2] or beam shaping [3]. However, traditional direct machining techniques such as mechanical ruling and diamond turning, or replication techniques via master fabrication, are not applicable to diamond structuring due to its hardness and chemical inertness. In contrast, patterning methods based on dry etching are commonly used for step-like micro-optics or microlenses fabrication in a variety of substrates including diamond, which allows exploiting its unique material properties.

In particular, the combination of chemical inertness, high hardness and high wear resistance, high thermal conductivity and low thermal expansion coefficient, wide optical transmission bandwidth (from deep ultra-violet to the far infrared region), low absorption and high refractive index, has spurred active research activities for the fabrication of micro-optical components such as microlens arrays (MLA) and diffraction gratings in single crystal diamond. In addition, the potential of single crystal diamond to host color centers as single photon sources for quantum information processing [4] and quantum sensing [5,6] applications has recently led to increased interest in precision micro-optical components directly structured in single crystal diamond. Among recently demonstrated micro-optical components in single crystal diamond are hemispherical convex lenses [7], double-sided lenses [8], high aspect ratio lenses [9], large focal lenses [10], concave lenses [11], hexagonal lenses [12], cylindrical and ring lenses [13], as well as immersion metalenses [14] and solid immersion lenses [15,16]. Reflection diffraction gratings [8] and trapezoidal diffraction gratings [17] have also been demonstrated. These demonstrations are mainly based on two fabrication processes, which are summarized schematically in Fig. 1. For microlens arrays, a photoresist is first spin-coated on the diamond. The resist is patterned using traditional 2D lithography into circular or hexagonal pillars. A thermal [10] or chemical [7] reflow is performed to produce surface tension induced hemispheres. The resulting structures are transferred into the diamond using dry etching, commonly with O₂/Ar [11] or Cl₂/Ar [10] chemistry. For diffraction gratings, typically a hard mask material is deposited first (such as SiO₂ [8] or Al₂O₃ [17]), in order to increase the selectivity of the etch process and to thus allow fabrication of structures with higher aspect ratios as compared to what can be achieved with only a photoresist mask. A photoresist is spin-coated and patterned with the designed grating period and duty cycle. Finally, the resulting structure is successively transferred to the hard mask and to the diamond using dry etching, with the possibility of forming trapezoidal profiles by etching without bias [17]. However, these processes present the following limitations: the attainable shapes of the lenses are limited to hemispherical profiles and the gratings can only be vertical or trapezoidal. One way to overcome these limitations resorts to greyscale lithography, where the resist is exposed with varying dose, such as demonstrated in [18]. The dose is calibrated to expose the resist only up to a defined depth. Once developed, different thicknesses of resist remain on the substrate resulting in a 3D photoresist structure. Subsequently the 3D pattern is transferred into the diamond by dry etching. Based on this fabrication process, Fresnel microlenses and blazed gratings have been demonstrated.

 

Fig. 1. Typical process flows for fabricating (a) Microlenses using photoresist reflow and pattern transfer by dry etching, and (b) Diffraction gratings using a hard mask to increase the selectivity of the etch process to allow high aspect ratio structures.

Download Full Size | PDF

We here demonstrate an alternative approach to extend the limited range of shapes that can be obtained with conventional photolithography based techniques, by using 3D laser lithography based on two-photon polymerization. This photoresist patterning method allows printing freeform 3D structures on the diamond substrate, with a lateral resolution of 200 nm, an out-of-plane resolution of 700 nm, and a build volume of several millimeters in all directions. The photoresist is applied on the diamond surface as a droplet from a pipette, without spin coating, which allows for a thick photoresist layer if required, thus permitting the fabrication of high aspect ratio (short focal length) lenses. Recent demonstrations have successfully used this method to fabricate micro-optical components on sapphire [19]. We here demonstrate the fabrication of transmission blazed (also referred to as ruled or echelette) diffraction gratings and Fresnel microlens arrays in single crystal diamond with this technique. The combination of 3D laser lithography and an optimized dry etching process with a low selectivity (1:15) yields optical elements in single crystal diamond with high surface quality.

2. Methods

We design a 5 × 5 Fresnel microlens array composed of 25 individual 200 µm x 200 µm microlenses with a target focal length of 1 mm. We also design a 1 mm x 1 mm blazed diffraction grating for a maximum efficiency in the first order with a diffraction angle of 3.6°, a period of 10 µm and a blaze angle of 2.55°. All structures have a target maximum height of 445 nm and are optimized for an operating wavelength of 633 nm. The micro-optical components are fabricated on a commercially available double side polished CVD general grade single crystal diamond plate (Element Six), with dimensions of 3 mm x 3 mm x 0.3 mm, initially cleaned in hot piranha followed by a hydrofluoric acid bath. Scratches, pits and polishing lines originating from the mechanical polishing of the diamond are removed using our previously reported ion beam etching (IBE) smoothing technique [20]. During the subsequent lithography step, the diamond sample is mounted on a structured carrier wafer in order to avoid accidental damage to the objective of the lithography system during stage movement (Fig. 2). The carrier wafer comprises square recesses matching the diamond plate dimensions. After gluing the diamond in a recess, its top surface stands a few micrometers above the wafer surface. This configuration also prevents the resist to flow around the diamond excessively which can result in an uncontrolled and insufficient resist thickness on the diamond surface.

 

Fig. 2. Setup for the exposure using high resolution 3D dip-in laser lithography mode with the Nanoscribe Photonic GT. The single crystal diamond is attached in a recess etched in a silicon wafer, in order to avoid physical contact between the objective and the diamond.

Download Full Size | PDF

A droplet of photoresist (Nanoscribe IP-Dip, containing 60-80% of pentaerythritol triacrylate [21]) is deposited on the diamond surface and exposed using 3D dip-in laser lithography (Nanoscribe Photonic Professional GT). The exposure is performed in high resolution 3D dip-in laser lithography mode with a 63x magnification objective lens with a 1.4 numerical aperture. After development in PGMEA and rinsing in IPA, the structures have a maximum thickness of 6.7 µm and exhibit a visible staircase, inherent to the layer-by-layer 3D printing process (Fig. 3 (a-d)).

 

Fig. 3. (a-d) Blazed diffraction gratings and Fresnel microlenses patterned in IP-Dip photoresist on single crystal diamond surface by 3D Laser lithography. The layer-by-layer 3D printing process results in a visible staircase. (e-f) Structures obtained after the diamond etching step. The staircase is noticeably reduced due to the low selectivity (1:15) of the etching process.

Download Full Size | PDF

The photoresist structures are transferred into the diamond by reactive ion etching (STS Multiplex ICP, 800 W coil power, 300 W platen power, 5 mTorr chamber pressure, 40 sccm Cl₂ and 25 sccm Ar, etch rate 64.5 nm/min). While diamond is commonly etched using oxygen chemistry [22], we found that the IP-Dip photoresist was not etched uniformly in O₂/Ar plasma and resulted in a rough surface. In contrast, the chlorine-based etching process produces smooth surfaces and exhibits a low selectivity (1:15), which allows to smoothen out the staircase observed in the photoresist (Fig. 3 (e-f)). The staircase technique used in the 3D laser lithography results in an optical performance ceiling based on the numbers of steps. Our method not only makes use of the highest resolution available on the 3D laser lithography setup, therefore raising the numbers of steps, but also raises the resulting optical performance ceiling by smoothing these steps. Finally, in order to characterize the fabricated elements, the diamond is detached from the handling wafer and cleaned with acetone and IPA.

3. Characterization

Atomic Force Microscope (AFM) measurements of the diamond blazed diffraction gratings (Fig. 4) reveal that the surface subtends an angle of 2.48° to the top surface, deviating by less than 3% from the targeted value. The grating period is measured as 10.2 ± 0.1 µm, within 2% of the target period, while the height of the grating is measured at 310 ± 10 nm, within 30% of the target depth. A comparison of the measured profile with the design (Fig. 5) shows that sharp angled features such as the grating apexes are not accurately transferred to the diamond due to the progressive erosion of the photoresist during the dry etching. This effect could potentially be counterbalanced by adding a supplementary compensation volume to the photoresist structure. The roughness (R_a) is measured to be of less than 2 nm on a 5 µm x 5 µm measurement area.

 

Fig. 4. (a) AFM measurements of the diamond blazed diffraction gratings and of (b) the diamond Fresnel microlenses. (c) Detail of a 5 µm x 5 µm area for roughness measurement. The surface shown has a roughness of 1.45 nm (Ra) and 1.76 nm (Rms)

Download Full Size | PDF

 

Fig. 5. Profiles of (a) the blazed diffraction gratings and of (b) the Fresnel microlenses measured with AFM and compared to the designed profile. The two structures exhibit a very high accuracy in period as compared to the target dimensions (2% deviation), while the height deviates by as high as 30% from the target height due to photoresist erosion during the etching.

Download Full Size | PDF

The blazed gratings were characterized by measuring the power diffracted into each of the -4 to + 4 diffraction orders (Fig. 6), using a laser with a 633 nm wavelength, an iris to reduce the beam diameter and a power meter. The relative diffraction efficiency is obtained by dividing the power measured in a given order by the total power measured in all the -4 to + 4 orders. The measurement reveals that 66% of the outgoing light is diffracted in the first diffraction order, closely matching the value obtained by simulating the diffraction efficiency of the measured profile of the gratings, using the Reticolo software [23] (Rigorous Coupled Wave Analysis for gratings with Matlab interface) (Fig. 7).

 

Fig. 6. Setups for measuring (a) the blazed diffraction gratings relative efficiency and (b) the Fresnel microlenses focal lengths. The relative efficiency was measured by dividing the power diffracted in a given order by the sum of the powers diffracted in the orders from -4 to + 4. The microlenses focal length is directly given by the stage displacement when focusing first on the lenses and then on the focal point of the lenses.

Download Full Size | PDF

 

Fig. 7. Measured and simulated relative diffraction efficiency of the blazed diffraction gratings. The gratings show a measured maximum relative efficiency of 66% in the first diffraction order. While the measurements closely match the simulated values, we attribute the discrepancies (and notably the higher measured than simulated values in some orders) to the difference between the gratings profile used in the simulation (i.e. a software repetition of a randomly chosen, single AFM measured period of the grating) and the actual gratings profile that slightly fluctuates from one period to another and that can be of overall better optical quality than the single grating period used for the simulation.

Download Full Size | PDF

The Fresnel microlenses’ focal lengths were measured using a transmission mode light microscope with an illumination wavelength of 633 nm. The focal length is extracted by the stage displacement when focusing first on the lenses base and then on their focal point. The average measured focal length was 960 ± 10 µm, which deviates by 4% from the targeted value. Finally, the optical efficiency of a single Fresnel microlens was obtained by placing a 200 µm x 200 µm pinhole between a light source and the unpatterned side of the diamond. The light intensity on the patterned side was measured in a zone with no lens on the diamond surface plane, and on the focal plane of a zone with a lens. The ratio of the two values results in an optical efficiency of 48%.

4. Conclusions

In summary, we demonstrate a method to fabricate continuous relief micro-optical components in single crystal diamond with an improved optical performance using 3D laser lithography, allowing to extend the range of achievable shapes compared to conventional processes. We used a low selectivity etch process that smooths out the staircases produced by the 3D laser lithography, raising the performance ceiling of the technique. The fabricated diffraction gratings and Fresnel microlenses present surface roughness R_a below 2 nm, as well as excellent lateral dimensional accuracy with less than 0.3 µm deviation from design. Surface quality is expected to be further improved using optical grade diamond substrates. The demonstrated fabrication process enables the fabrication of a wide variety of single crystal diamond structures that are not achievable using standard lithography techniques, such as micro-prism arrays or free-form optical elements that can be adapted to different wavelengths by varying the photoresist thickness. Based on the surface quality and dimensional accuracy, precision micro-optical components can be fabricated in single crystal diamond for high-power laser, spectroscopy or beam shaping applications.