Demonstration of Two Multi‐Component Target Ablation Approaches and Their Application in Combinatorial Pulsed Laser Deposition

Combinatorial pulsed laser deposition (C‐PLD) based on segmented targets has led to new possibilities in the pace of discovery of novel advanced materials with significantly reduced deposition time and material consumption. However, fabrication of established segmented targets may be complex or not even possible for certain material combinations. In this article, two alternative C‐PLD techniques based on easy‐to‐fabricate segmented targets are presented. One approach uses two semi‐circular segments A and B with a systematic adjustable lateral shift of the target rotation axis from the target center. The other approach is based on a new target design defined by an ABA‐segmentation where B is a horizontal bar between two semi‐circular segments of A. Both approaches enable growth of discrete composition libraries. The concepts and important parameters are introduced and computer simulations as function of the geometric parameters are made to yield the expected thin film compositions. As proof‐of‐concept, the techniques are employed on the transparent, semiconducting ternary alloy zinc‐tin‐oxide. The simulations are in very good agreement with the experimental data. Physical properties of films grown by the demonstrated approaches are compared with those obtained by established PLD processes.


Introduction
High-throughput screening methods for determining the physical properties of functional materials have significantly accelerated the pace of discovery of novel advanced materials.This paradigm shift is now driven primarily by high-throughput computational and machine learning algorithms, with a significant lack of experimental verification.The same is true for materials discovery through high-throughput synthesis and DOI: 10.1002/apxr.202300140characterization.However, substantial progress has been made in the development of combinatorial material synthesis by physical vapor deposition (PVD) in recent years.Due to the highly directional expansion of the source material, PVD processes are especially well suited for the creation of spatially addressable material libraries.Among them, the highly flexible and versatile pulsed laser deposition (PLD) technique has already been used to fabricate combinatorial thin films. [1,2]Most commercial PLD equipment vendors today offer a combinatorial growth option based on sequential deposition.Here, an offset PLD configuration is used and the substrate rotation is turned off during firing of some ten to some hundred laser pulses at a specific target.Then the target is changed and the substrate is rotated by 180°or by 120°t o explore a binary or ternary phase diagram, respectively.In the first case, after firing at the second target, the process is repeated until the desired layer thickness is deposited.In the latter case, after an additional 120°substrate rotation, a third target is ablated and the entire sequence is repeated.More complicated phase diagrams can be achieved by increasing the number of targets involved and decreasing the rotation angles accordingly.In all cases, a continuous composition spread (CCS) thin film material library is obtained.A major drawback of such sequential combinatorial deposition process compared to standard PLD is the very low growth rate and the long deposition time, during which the substrate is heated much longer than in standard PLD due to the numerous target changes.This can be avoided by segmented target combinatorial PLD (C-PLD) approaches described in detail in refs.[3, 4].Azimuthally segmented targets are required for multinary CCS thin film material libraries [3,5] while radially segmented targets enable discrete combinatorial material synthesis and compositional gradients in growth direction. [4,5][8] A drawback is the requirement for spatial resolution of the experiments performed.Here, laterally homogeneous thin films obtained by discrete composition synthesis offer characterization by lower spatial resolution methods at the cost of lower chemical resolution.Furthermore, optimization of electronic device properties is much easier within a discrete composition material library.However, the radially segmented PLD target approach is limited to binary phase diagrams per target, and the fabrication of such targets may be complex or not even possible for certain material combinations.
In this article, we introduce two alternative C-PLD approaches that enable discrete composition synthesis from easy-to-fabricate segmented targets using an offset PLD setup together with substrate rotation.The first approach is an azimuthally segmented target consisting of two semi-circular segments A and B where the center of target rotation can be systematically shifted toward either of the segments, and the second is an ABA-segmented target where B is a horizontal bar between two semi-circular segments of A.
First, the experimental concepts and important geometric parameters are introduced.Then, the expected thin film composition is calculated using computer simulations as a function of the geometric parameters.Finally, an experimental proof-of-concept study for the transparent, semiconducting ternary alloy zinc-tinoxide is presented, and the physical properties of films grown by the demonstrated approaches are compared with those obtained by established PLD processes.

Concept and Simulation
The two C-PLD approaches are proposed using two different target segmentations to create material libraries of discrete composition thin film samples.The first approach is an adaptation of a PLD variant based on ablation of azimuthally segmented targets introduced by Christen et al. [9] for controlling the stoichiometry of volatile potassium in KTa 1 − x Nb x O 3 .We use an azimuthally segmented target, consisting of two semicircular parts A and B. The target is rotated during deposition and the laser spot offset varies between values of 0 and R, so that the ablation on the target is homogeneous over a circular area with radius R. To achieve different compositions in the thin film library, the center of rotation is shifted horizontally by an offset . Figure 1a) illustrates such target segmentation and the homogeneously ablated target area for different offsets .In the following, this approach is denoted as ORA (Off-centered Rotation Axis) PLD.The second approach is defined by an ABA-segmentation (referred to as ABA-PLD), where B is a horizontal cuboid of width d between the two identical circular segments A (cf. Figure 1b).During deposition, the target is rotated, and since the radius of the laser track R can be controlled in situ, the path length L in each segment can be ad-justed.The laser track on the target is ring-shaped, centered and has a width t.A schematic of this target segmentation with laser ablation track is shown in Figure 1b.
By calculating the overlap between the laser ablation area and each target segment, the material composition of a deposited thin film can be determined without considering the details of plasma formation, plume propagation, nucleation, and film growth.Since this calculation assumes a priori that both segment materials have the same ablation rate, a weighting factor  A / B is introduced to account for the real material transfer rates  A and  B of the respective segment materials.This approximation is sufficient for the determination of room temperature grown thin film compositions.
The effect of geometric parameters and transfer rates on the composition A x B 1 − x is shown in Figure 2 for the two demonstrated C-PLD approaches.If not varied, the parameters have fixed default values of d = 5 mm,  A / B = 1, t = 2 mm and R = 5 mm.
Within the approach based on the azimuthally segmented target, the composition can be controlled by shifting the center of target rotation  (see Figure 2a).This method allows covering the whole composition range with only one segmented target.The radius R of the circular laser ablation area determines the gradient of the composition dependence on .Thus, larger values of R make the composition easier to be controlled, but require larger target sizes to cover the entire composition range.Changing the factor  A / B raises or lowers the values of x within the limits of With the ABA-segmented target more than 70% of the total composition range can be covered with the target and deposition parameters used for the simulations in Figure 2b.To cover the entire composition range, a second segmented target with BABsegmentation shall be used.Since this target is the inverse of the previous target, only two different targets with a single, homogeneous composition need to be fabricated and sawed to obtain the cuboid and circular segments.In general, the change in x is nearly linear for small values of R and its slope can be controlled by the width t of the laser ablation track.For larger values of R, the dependency weakens and the fraction of component B increases only gradually.For different values of the parameter d ( A / B ), the compositional dependence shifts horizontally (vertically).
The calculations for the simulation were done using a python script where the target segments and ablation areas are constructed by polygons.The overlap of target segments and ablation area was calculated using the shapely.geometrylibrary.In the following, the two depicted C-PLD approaches are employed on room temperature fabrication of zinc-tin-oxide thin film libraries as a proof-of-concept.There, zinc-oxide (ZnO) is taking the role of the previous segment material A and tin(IV)oxide (SnO 2 ) of segment material B.

Results
All thin films of the three material libraries prepared by the two demonstrated C-PLD approaches were analyzed by EDX to determine their chemical composition.The experimental results are shown in Figure 3 together with the calculated values from the computer simulations.The value x refers to the Zn cation concentration in Zn x Sn 1 − x O.A sample prepared by the ORA method at an offset of  = 0 mm was used to determine the weighting factor  ZnO ∕ SnO 2 .Since for this case both segment materials are equally ablated but the chemical composition in the resulting thin film is not equally distributed, the weighting factor including the binary transfer rates  ZnO and  SnO 2 can be de-termined from the EDX measurement data.For zinc-tin-oxide a factor of  ZnO ∕ SnO 2 = 1.75 was derived.
In Figure 3a results of the threefold ABA-segmented target are depicted as a function of the laser track radius R for d = 6 mm.The path length in segments A L A of the laser track increases with R and thus the Zn cation fraction increases accordingly.The predicted non-linear increase of L A with R is very well confirmed by the experimental data points.All deviations are within the experimental error.Figure 3b shows experimental and simulated data for the threefold BAB-segmented target.The simulated data are not the same as those in Figure 3a with the Zn and Sn assignments reversed, since the different transfer rates of ZnO and SnO 2 were taken into account.Similar to the ABA-segmentation, the experimental data correspond to the simulated data for d = 6 mm with the exception of the data points of R = 3 mm and R = 5 mm.Since the employed PLD chamber is also used for hightemperature depositions the underlying mechanics have to be slightly loose in order to account for material expansion.However, at room temperature it may happen that the actual radial laser offset R differs from the adjusted value, which might explain these deviations.The compositions of the material library created with the twofold azimuthal target with fixed radial laser offset of R = 4 mm is shown in Figure 3c as a function of the target rotation axis offset .Again, the simulated and experimental data are in agreement within the experimental error.Expectedly, the achievable compositions cover the entire phase diagram, which is also possible by using an ABA-and BAB-segmented target.
X-ray diffractograms of the thin films deposited from the twofold segmented target are exemplarily shown in Figure 4. Except for the highest Zn content (in principle corresponding to binary ZnO) all thin films are X-ray amorphous.This is also true for the other two material libraries and for zinc-tin-oxide samples prepared by conventional PLD (marked by an asterisk).For x = 0.99 ZnO (002) and (103) reflections are observed.[12][13][14] Some thin films show small peaks at 37.8 °and 64.1 °, which are assigned to ohmic gold contacts on the sample corners used for the electrical transport measurements.
The surface topography of selected samples of different Zn cation content is shown in Figure 5 together with the derived surface root-mean-square roughness r RMS .We did not observe any differences for the two different C-PLD methods presented in this letter.In general, smooth surfaces with r RMS ⩽ 4 nm are observed for low and medium Zn contents.However, within this range there is no tendency observable.For high Zn cation content (x = 0.99) the surface roughness abruptly increases to r RMS = 9 nm where the sample is no longer amorphous but polycrystalline (cf. Figure 4).
LSM scans of zinc-tin-oxide samples of selected rotation axis offsets  are depicted in Figure 6 together with a figure displaying the distribution of droplet density for different offsets .Droplet density is highest for a rotation axis offset of  = 0 mm with decreasing density toward the sides.On all samples prepared by the ABA-and ORA-PLD methods droplets are present.After a few PLD processes using the segmented targets an increased ablation at the interface between target segments was observed, which was then stated as the source of droplet formation.
The electrical transport properties of all samples in the three material libraries were characterized by room temperature Hall effect measurements.The data are summarized in Figure 7. Regardless of the C-PLD method used, the resistivity  is lowest for the highest Sn contents with  ≈ 8 × 10 −5 Ωm.With increasing Zn content the resistivity increases exponentially up to x ≈ 0.5.For higher Zn contents a saturation of the resistivity is visible and for x ≈ 0.83 a maximum of  ≈ 10 Ωm is observed.A spline function applied to the data points (dashed line in Figure 7) has its maximum of  ≈ 5 Ωm at x ≈ 0.75.A further increase of the Zn content leads to a decrease of  and for the highest Zn content of x = 0.99 a resistivity of  ≈ 0.5 Ωm is observed.Since the carrier mobility does not systematically depend on the alloy composition and is in a rather small range between 0.2 and 13.0 cm 2 V −1 s −1 (cf. Figure S1, Supporting Information), the carrier density has a similar but reversed dependence as the resistivity.The highest free electron density n = 1 × 10 20 cm −3 is observed for the lowest Zn content and decreases exponentially with increasing x up to x ≈ 0.4, at which point the decrease begins to saturate.The lowfree carrier concentration n ≈ 7 × 10 14 cm −3 is observed for x ≈ 0.83.A spline function applied to the data points (dashed line in the Figure 7) has its minimum of n ≈ 3 × 10 15 cm −3 at x ≈ 0.75.[17][18] According to ref. [16] an oxygen pressure of 0.03 mbar during the deposition of zinc-tin-oxide leads to the highest conductivity, and most literature data concerning the deposition of zinc-tin-oxide thin films are based on this pressure and therefore comparability is guaranteed, which is why this pressure was chosen for this work.

Discussion
We introduced two alternative C-PLD methods that can be used for fast and efficient creation of material libraries covering entire ternary phase diagrams for rapid exploration of material properties.Both methods allow deposition of discrete composition material libraries, as well as samples with vertical composition gradients.However, vertical composition gradients by the ORA method require in situ control of the  parameter, which is not currently implemented in our chamber.Vertical gradients by the ABA-PLD approach are feasible in any PLD chamber that allows target scanning.The methods can be easily implemented in any conventional PLD chamber.In addition, because there is no need to change targets during the deposition process, there is no loss of deposition rate.The fabrication process of targets for both approaches is fairly simple.For each approach only two single binary targets have to be fabricated independently and subsequently sawed in the respective shapes and assembled in a target holder.This allows materials with incompatible sintering recipes to be explored by C-PLD.Furthermore, with such simple geometric shapes, radial target segmentation as proposed in refs.[4] and [5] can be avoided if necessary.
Simulations of the material composition in the resulting thin film show excellent agreement with EDX measurement data of samples fabricated with the demonstrated C-PLD methods at   Overview of charge carrier concentration n and resistivity  in dependence on Zn cation content of thin films prepared by ABA-and ORA-PLD methods and data points from amorphous zinc-tin-oxide thin films prepared by conventional PLD. [15]om temperature.Hence, they can be used to predict the composition for designated target and deposition parameters and thus thin films with specific composition can be fabricated in a targeted manner on demand.][21][22] In this work, the main focus is on the methodology, and thus for electrical characterization contact material and contact properties were not optimized leading to a scattering in resistivity and charge carrier concentration observable in Figure 7.
As seen in Figure 6 the occurrence of droplets on the thin film surface originating from the interface between segments has to be addressed.The more often a laser pulse is incident to the interface during deposition the more droplets eventually appear on the thin film surface (cf. Figure 6f)).Similar effect was observed for an azimuthally segmented ZnO-SnO 2 target where both target segments were fabricated independently [15] (cf. Figure S2, Supporting Information).Corresponding LSM scans together with an LSM scan of a sample prepared by the ORA method can be found in the Supporting Information part.High droplet density can be avoided by mixing the different segment materials, filling the material powders in respective geometric forms, removing the forms and sintering the target materials at once (as in ref. [15]).However, this procedure annihilates the approach of independent target fabrication as presented in this letter.Instead, frequent grinding and polishing of the target surface can also drastically reduce droplet density, which will be systematically investigated in upcoming experiments.
After a few deposition processes utilizing the ORA-PLD method with different rotation axis offsets  pit formation occurs, since for different offsets the ablation areas might overlap.Further ablation is possible but the composition of such samples can no longer be accurately predicted because plume propagation and size of ablated area differ from those of a flat target.Similar applies for the ABA-PLD method where radial laser track grooves form.This undesirable occurrence can again be easily circumvented by grinding and repolishing of the target surface, as already suggested for the reduction of droplet density.
Both C-PLD methods were proven to be viable for the creation of thin film composition libraries but each approach has its benefits and drawbacks.For the ORA method only one segmented target is needed to cover the whole composition range while for the ABA-PLD method two segmented targets are required.The semicircular target segments are easier to fabricate than the circular and cuboid segments since no close attention has to be paid on the precise width of the cuboid during sawing.However, for the ORA method an additional target holder with adjustable mounting has to be manufactured.Depending on the needs of accuracy of composition the ABA-PLD method might be the method of choice since for larger radial laser offsets R the change in composition is only gradual while the change of composition with the ORA approach is comparably steep.Applying a high laser pulse frequency together with a rotation axis offset far different from zero might lead to layered film growth of alternating segment material A and B because in this case during one rotation cycle one target segment is ablated considerably more before ablation of the other segment begins.Hence, laser pulse frequency needs to be adjusted accordingly (and not commensurate to target rotation rate) in order to ensure a sufficient point density on the laser track but at the same time provide a homogeneous distribution throughout the deposited layer on the thin film. [23]For the ABA-PLD method this effect is less significant since ablation of the different segments changes more frequently.

Summary
Two CPLD variants based on the ablation of segmented targets with simple segments were described in detail.Both methods were used to create discrete composition zinc-tin-oxide material libraries whose physical properties were compared with thin films grown by conventional PLD for benchmarking.The timeaveraged composition of the plasma plume is controlled for both methods over the radius of the ablation track or area, respectively.In ABA-PLD, two semi-circular pieces of material A are joined with a rectangular centerpiece of material B to form a target.For small ablation track radii, thin films are rich in material B, and the amount of material B systematically decreases with increasing ablation track radius R.This was demonstrated by numerical modeling and by experimental deposition of zinc-tin-oxide as a function of R. For ORA-PLD an azimuthally segmented target consisting of two semi-circles is ablated.The center of rotation of the target can be shifted, which is used to adjust the ratio of ablation area in the two segments and with that the resulting thin film composition.Pros and cons of both methods were discussed in detail.Finally, structural, morphological and electrical trans-port properties sets of zinc-tin-oxide thin films were characterized as function of the cation composition.The results obtained coincide with properties of zinc-tin-oxde thin films realized by conventional PLD which illustrates the suitability of the methods.We note that both methods, but especially ABA-PLD, can be used to realize compositional gradients in the growth direction and provide a tool for strain and bandgap engineering, refractive index gradients, polarization doping, and more.

Experimental Section
All zinc-tin-oxide thin films were deposited on 10 × 10 mm 2 Corning Eagle XG glass substrates utilizing a large-area offset PLD setup. [24]In this setup the distance between target and substrate was 9.5 cm and an offset of 8 mm was chosen (the offset is the lateral distance between the center of the plasma plume and the substrate center).[27][28] A KrF excimer laser (Lambda Physik,  = 248 nm) with pulse frequency of 15 Hz and a fluence of approx. 2 J cm −2 at the target surface and a total pulse number of 40000 were used.The samples were deposited at room temperature and in oxygen atmosphere at a chamber pressure of 0.03 mbar leading to thin film thicknesses in the range of 0.5 to 1.1 μm.The substrate was rotated during the deposition process in order to achieve thin films of homogeneous thickness.
ZnO and SnO 2 source powders purchased from Alpha Aesar with purities of 99.99% and 99.9%, respectively, were used for the fabrication of segmented, ceramic targets.In total, four targets were fabricated for the deposition of zinc-tin-oxide thin films, two binary ZnO and two binary SnO 2 targets.ZnO powder was dried at 105 °C for 2 h, cold pressed in a circular mold and sintered at 1150 °C for 12 h in air.SnO 2 powder was cold pressed in a circular mold, sintered at 1200 °C for 24 h in air, ground, cold pressed, and sintered again at 1350 °C for 72 h in air.After the sintering process, the targets were cut into the respective shapes and then assembled in a target holder.For the ORA-PLD approach a target holder with adjustable mounting was manufactured allowing to manually vary the offset of center of rotation .
The chemical composition of thin films was measured using energydispersive X-ray spectroscopy (EDX).For this a Nova NanoLab 200 by FEI company was utilized.On each sample three measurements, each at different positions on the thin film, were conducted and averaged.For each of the investigated samples the difference between the three measurements were within the error bar of the experiment.For zinc and tin the L-edge was measured.Crystal structure was investigated by X-ray diffraction (XRD) measurements using a Philip's X'Pert diffractometer with 2Θ- scans in the range of 10 − 80 °.A Park System XE-150 atomic force microscope (AFM), as well as a laser scanning microscope (LSM) consisting of a Keyence VKX200 K control unit combined with an X210 microscope were used to obtain information on surface morphology.The LSM was equipped with a 100 W halogen lamp as light source and a CCD camera for capturing optical microscopic images.Measurements were conducted with a laser of wavelength of 408 nm and power of 0.95 mW.Charge carrier density (n) and Hall mobility of carriers (μ) were determined by Hall effect measurements at 0.43 T at room temperature.Resistivity () was measured using a four-probe van der Pauw setup.Beforehand, ohmic gold contacts were sputtered through a mask on the corners of each sample in argon atmosphere at a pressure of 0.02 mbar.

Figure 1 .
Figure 1.a) Azimuthally segmented target with segment materials A and B for three different offsets of center of rotation  at fixed radius R of laser ablation area.b) ABA-segmented target with horizontal cuboid of width d and with ring-like laser ablation area of radius R and width t.

Figure 2 .
Figure 2. Composition x of A x B 1 − x for a) the azimuthally segmented target as a function of the shift of center of rotation , the radius of the laser ablation area R and the weighting factor  A / B , and for b) the ABA-segmented target approach as a function of the radius of the laser ablation track R, the width of the cuboid d, the weighting factor  A / B and width of laser ablation track t.

Figure 3 .
Figure 3. EDX measurements of samples prepared by a) ABA-, b) inverted ABA-and c) ORA-PLD method.Solid lines show simulated Zn and Sn cation contents in (a)+(b) for d = 6 mm, t = 2 mm in dependence on the radial laser offset R and in c) for R = 4 mm in dependence on the rotation axis offset .For both methods a weighting factor of  ZnO ∕ SnO 2 = 1.75 was applied.

Figure 4 .
Figure 4. XRD measurements of samples of different Zn cation contents fabricated with the ORA-PLD method and conventional PLD (marked by the asterisks, taken from ref. [15]).For low and medium Zn content all samples are X-ray amorphous.For high Zn cation content (x = 0.99, in principle binary ZnO) the sample is polycrystalline since ZnO (002) and (103) reflections are observed.The peak at 41.5 °likely originates from the sample holder.All other reflections above the noise floor stem from ohmic Au contacts used for the measurement of electrical properties.

Figure 5 .
Figure 5. a−d) AFM scans of samples fabricated by ABA-and ORA-PLD for selected Zn cation contents.e) r RMS in dependence on Zn cation content.The thin films exhibit a surface roughness r RMS of less than 4 nm except for x = 0.99 with r RMS = 9 nm where the sample is no longer amorphous but polycrystalline.

Figure 6 .
Figure 6.a−e) LSM measurements of samples fabricated by ORA-PLD for selected offsets .All LSM pictures have dimensions of 73 × 73 μm 2 .f) Droplet density of zinc-tin-oxide thin films fabricated with different offsets .Droplet density is highest for  = 0 mm and decreases toward the sides.

Figure 7 .
Figure 7.Overview of charge carrier concentration n and resistivity  in dependence on Zn cation content of thin films prepared by ABA-and ORA-PLD methods and data points from amorphous zinc-tin-oxide thin films prepared by conventional PLD.[15]