Masked-assisted radial-segmented target pulsed-laser deposition: A novel method for area-selective deposition using pulsed-laser deposition

Area-selective deposition (ASD) allows the growth of functional materials on predefined positions of a substrate. Selective growth can be induced by substrate prepatterning resulting in spatial dependent nucleation, desorption, or shadowing of regions by, e.g., self-assembled monolayers and is well established for, e.g., chemical vapor and atomic layer deposition as recently reviewed.¹ The resulting feature size ranges from some 100 $μm$ down to $100nm$ or below. ASD is hardly exploited within physical vapor deposition methods. Shadowing mask approaches, including simple and cost-effective filter paper masks, patterned using $CO2$ laser scribing, or steel masks enabling sub-mm feature sizes, were reported.^2,3 The subsequent application of differently patterned masks (mask sets) during magnetron sputtering or pulsed-laser deposition (PLD) was originally used for the realization of material libraries.^4,5 For that, the sputtered or ablated target is exchanged when a mask is changed resulting in a spatially addressable material library. We note that continuous composition spread material libraries can be realized without shadow masks by cosputtering⁶ or ablation of azimuthally segmented targets.^7,8 In this letter, we introduce an ASD approach by PLD allowing, in principle, arbitrary patterning of solid state solutions with precisely controlled alloy composition on the $μ$m scale. The proposed method, combining the ablation of segmented (here radial segmentation) targets with shadow masks movable independently in both in-plane directions with a spatial precision of about $10μm$⁠, is denoted as mask-assisted radial-segmented target PLD (MARS-PLD). It allows the deposition of lateral as well as vertical (out-of-plane) compositional gradients with high spatial and chemical resolution as required, e.g., for graded refractive index layers or wavelength-selective photodetectors.^9,10

Within a conventional PLD setup, it is common to ablate homogeneous targets resulting in homogeneous thin films. Azimuthally segmented PLD targets were used to fabricate thin films with a well-defined material gradient (combinatorial PLD, C-PLD).⁸ Additionally, it is possible to mask off certain areas of the sample with adequate shadow masks.¹¹ In the technique introduced here, we combine a laterally movable mask setup with the ability to precisely control the composition of the incoming material flux by ablating radially segmented targets (compare Fig. 1).^7,8 For our processes, we control the time-averaged material flux composition by the radius of the laser track.^7,12 The radius is determined by the laser spot and the adjustable position of the center of rotation of the target. Thus, we can precisely control the material flux during PLD by virtue of selecting the radius of ablation. With our setup, it is possible to ablate any material composition from two present target segment materials at any desired and predefined lateral substrate position. That also means that we are able to combine lateral and vertical material gradients and control via the number of PLD pulses. Furthermore, multiple radially segmented targets can be mounted in the target carousel, which increases the freedom in material design.

In order to investigate and optimize the method of MARS-PLD, we selected the ternary material system of magnesium zinc oxide (Mg,Zn)O, primarily, in its wurtzite structure. Since it is already thoroughly investigated over the past decades, it is a suitable model system for demonstrating our MARS-PLD method.

The segmented targets for MARS-PLD were optimized and fabricated within our laboratories. In the case of segmented targets, the procedure is not as straightforward as with homogeneous targets since different grain sizes and different coefficients of shrinkage may lead to brittle or cracked targets after the sinter process.

In our standard process for (Mg,Zn)O, we first presinter the source powders at $T=1150°C$ for $12h$ in air. After cooling down, the powder is ball-milled using a tungsten carbide mortar for three times and 45 s. Afterward, the powder is filled into a stainless steel mold with an elliptically shaped metal tube and pressed into a target with a diameter of $d=28mm$⁠. Therefore, the respective powder is pressed by a hydraulic press with $p=4.5MPa$ for $10min$⁠. The target now has a height of approximately $h=6mm$ and is sintered after removal of the auxiliary tube for $12h$ at $T=1150°C$⁠. The segmented PLD targets were fabricated using ZnO (⁠$99.99%$⁠) and MgO (⁠$99.995%$⁠). The samples were grown from a segmented target with an elliptical inner segment, made from binary ZnO powder, and an outer segment, made from a mixture, consisting of $50at.%$ ZnO + 50 $at.%$ MgO. Figure 2 shows a radially segmented PLD target as described above after fabrication (left) and after application of approximately 3 00 000 laser pulses (right).

In order to fabricate these novel chemical composition spreads on selected substrate positions, two key components need to be implemented into the ordinary PLD setup.

First, the position of the laser spot on the target needs to be adjustable in order to precisely control the radius of the laser track. In our setup, this is accomplished by moving the rotating target relative to the laser spot since the latter is focused on a fixed position. The target can be moved laterally with respect to the laser spot with an accuracy of approximately 200 $μm$⁠. Second, for the control of the lateral position of the mask, we use two linear manipulators with an accuracy of $6μm$ and a travel range of $Δy≥300mm$ and $Δz=24mm$ (compare Fig. 1). Due to the facile use of shadow masks, the only limiting factor for geometry and shape of the material distribution on the substrate is the fabrication of masks with appropriate small notches and close proximity to the substrate. For the PLD-process itself, we use an LPX305 krypton fluoride excimer laser, which operates at a wavelength of $λ=248nm$⁠. The laser has an energy density of $2Jcm−2$ with a pulse length of 25 ns at the surface of the target. The distance between the target and substrate is 90 mm, and the lateral offset is 10 mm relative to their centers.

In order to understand the influence of physical and mechanical process parameters during MARS-PLD on the quality of the thin films, we fabricated several different samples and will now present results on lateral material gradients based on (Mg,Zn)O. The thin films were deposited on $10×10mm2$ a-sapphire substrates at an oxygen partial pressure of $p(O2)=0.02mbar$ and a temperature of $TG=660°C$⁠. The process itself consists of 20 individual steps. With every step, the radial position of the laser spot on the target was increased by $400μm$ and the lateral position of the mask, which has a slit width of $dS=500μm$⁠, was changed by $250μm$⁠. With these parameters, we obtained a compositional gradient across 5 mm of which a photographic image is shown in Fig. 3. The thin films are transparent and have a slight variation in color due to different layer thicknesses, which can be explained by different growth rates of magnesium- and zinc oxide. We note that the same number of pulses was used for each lateral position, not adjusting for different growth rates. The thickness of the thin film was determined by spectroscopic ellipsometry to 20 nm at the Mg-rich side and 60 nm at the Zn-rich side.

Energy dispersive x-ray spectroscopy (EDX) measurements have been conducted with an electron beam spot size of $100×140μm2$ with a pitch of $250μm$ between data points. The results are shown in Fig. 4. We observe a nearly linear increase in the magnesium content with increasing the position up to 80%, whereby at around 50%, a phase boundary between the wurtzite and cubic phase¹³ is present. In the wurtzite regime, we observe two linear regions. One, within the first millimeter with an increase of $8%mm−1$ magnesium content and second, for the subsequent 2.5 mm with an increase in the $17%mm−1$ magnesium content. This steep slope for the lateral material composition gradient for (Mg,Zn)O is much higher than reported from other papers.¹² 

The pseudo-epsilon 2 $⟨ϵ2⟩$ spectra (similar to the absorption coefficient) determined every $200μm$ of the 5 mm wide graded layer by spectroscopic ellipsometry (at $60°$ angle of incidence) is shown in Fig. 5. We find a systematic shift of the absorption edge toward higher photon energy with an increasing magnesium content. Due to the small layer thickness of 40 nm or below, a modeling of the spectra was not possible.¹⁴ Nevertheless, we can observe a defined edge up to a magnesium concentration of around 40%. For higher concentrations, the respective spectra flatten and show a strong shift toward higher energies. In this region, the phase transition from the hexagonal to the cubic phase of (Mg,Zn)O is present.

Investigations of structural properties by x-ray diffraction (XRD) could not be acquired as a function of composition, since a necessary divergent slit used for a line scan reduces the intensity to a degree, that the resulting spectrum is within the noise level. Nevertheless, a measurement averaging over the entire material gradient provides quantitative insight into the crystal structure. In Fig. 6, a $2Θ−ω$ scan of a 5 mm (Mg,Zn)O material gradient is shown. The (00.2) reflection of wurtzite (Mg,Zn)O exhibits a wide peak because of the varying material composition. This leads to a broadening of the (Mg,Zn)O (00.2) reflection toward higher angles. Further, the reflexes at around $36.3°$ can be assigned to the (111) orientation of cubic (Mg,Zn)O.¹⁵ 

Here, we demonstrate another promising feature of MARS-PLD. We make use of the option to independently control both the in-plane mask movement relative to the substrate and the composition. This means that we can, in principle, deposit any desired material composition at any point within the lateral range of mask movement limited only by the used target materials. Not only can we deposit different material compositions next to each other in order to create lateral material gradients, it is also possible to deposit them on top of each other, which leads to vertical gradients. In Fig. 7, examples of area-selective deposition of material gradients are illustrated. The image shows a combination of lateral and vertical chemical composition gradients, a separated vertical gradient as well as a widespread barlike geometry. All structures can be deposited within a single process using only one heterogeneous target.

Nevertheless, in the following, we illustrate a material gradient over a predefined two-dimensional pattern. Therefore, we used a shadow mask with a $1×1mm2$ notch, which allows us, in the case of a $10×10mm2$ sample, to deposit at 100 individual spots. In Fig. 8, we show EDX measurements of such a sample with a (Mg,Zn)O graded pattern of the letters “U” and “L” referring to “Universität Leipzig” grown at $TG=650°C$ and $p(O2)=0.02mbar$⁠. The material that forms the two letters is deposited on a thin buffer layer of binary ZnO in order to have a background signal as a reference. Since this layer also contributes to the EDX measurements, we expect slightly higher values for the Mg-content of the (Mg,Zn)O thin film than depicted in the color scale. We note that this ASD PLD method would, in principle, also be possible by repeatedly exchanging two individual homogeneous targets. However, in order to accomplish a similar thin film quality, a target change for every five pulses is necessary. This would lead to a total deposition time between 50 and 65 h depending on the estimated time for a target change and does not seem practical. Another method would include 20 individual targets, each with a distinct and homogeneous magnesium content. This would require opening the chamber multiple times and is thus not a viable alternative.

In this work, we report on a novel area-selective deposition method for pulsed-laser deposition for deposition of varying material compositions. We use radially segmented targets and a freely moving shadow mask in order to exemplary demonstrate material composition gradients. As a model system, we use (Mg,Zn)O. We achieved a linear increase of the magnesium content with a slope of $17%mm−1$⁠. Further experiments by spectroscopic ellipsometry show a corresponding systematic increase of the absorption edge with an increase in the magnesium content and a phase transition to cubic Mg_$x$Zn_$1−x$O at $x≅0.45$⁠.

Further, we want to highlight the ability to deposit a selected material composition at any desired 1D, 2D, and even 3D pattern. Additionally, there is no restriction in the usage of shadow masks, which allows for masks with multiple notches or a single spotlike hole to trace a predefined pattern. On the latter approach, we present first quantitative results as we fabricated a 2D (Mg,Zn)O pattern forming the letters “UL.” The process of optimizing the method of MARS-PLD also brought up certain difficulties. We note that radial-segmented targets of high quality are the basis for likewise high quality material gradients. We can report on difficulties to produce flat, gapless and, in the best case, ceramic targets with a well-defined separation of the two compact material segments. We see further need for optimizing this process. This includes a procedure for a consistent border between the materials in order to guarantee thin films of high surface quality. Further, it is not trivial to fabricate the segmented target with accuracy to meet predictions of already mentioned simulations. As a result, every target needs to be individually adjusted and every ablation process configured independently. Another important aspect is the varying size of ablation areas. These result from differently sized laser tracks, which the laser spot describes during the PLD. Assuming that the pulse number for every step is constant, a corrugation of the target surface is likely to emerge. This might change the desired material composition and thus lead to nonideal material gradients. However, a flat target surface might be regained by a careful grinding procedure or by well adjusted sequences of laser pulses. All the latter aspects are important to guarantee a sufficient reproducibility and are currently thoroughly investigated.

The authors want to thank Monika Hahn for the preparation of PLD targets, Holger Hochmuth for assisting in modifications at the PLD chamber, Lukas Trefflich and Gabriele Benndorf for conducting the measurements by spectroscopic ellipsometry. This work was supported by the Federal Ministry of Education and Research in the framework of project “Ultrakompaktes Spektrometer - UltraSPEC2” (No. 03VP08180).

The authors have no conflicts to disclose.

Laurenz Thyen: Data curation (lead); Investigation (equal); Visualization (equal); Writing – original draft (equal). Daniel Splith: Data curation (supporting); Software (lead); Visualization (supporting); Writing – review & editing (equal). Max Kneiß: Formal analysis (lead); Methodology (equal); Supervision (equal); Writing – review & editing (equal). Marius Grundmann: Conceptualization (lead); Funding acquisition (lead); Methodology (lead); Project administration (lead); Supervision (lead); Writing – review & editing (equal). Holger von Wenckstern: Conceptualization (lead); Supervision (equal); Validation (equal); Writing – original draft (equal); Writing – review & editing (lead).

The data that support the findings of this study are available from the corresponding author upon reasonable request.