Parallel Diffractive Multi‐Beam Pulsed‐Laser Ablation in Liquids Toward Cost‐Effective Gram Per Hour Nanoparticle Productivity

Nanoparticles (NPs) generated by pulsed‐laser ablation in liquids (PLAL) have benefited many key applications due to their versatility, enlarged surface area, and high purity. However, scaling up NPs production represents one of the main requisites to commercialize this technology. The established upscaling strategy demands high power and repetition rate laser source with fast scanning systems, which are not widely available and costly. Herein, a cost‐effective alternative is proposed, the addition of static diffractive optical elements to achieve parallel processing through the multi‐beam PLAL (MB‐PLAL). In MB‐PLAL, the optimum repetition rate is reduced to compensate laser energy splitting, hence achieving a higher interpulse distance, reducing pulse shielding, and increasing NPs productivity. MB‐PLAL with 11 beams reached a factor 4 productivity increase for iron–nickel alloy (Fe50Ni50) NPs compared to the single‐beam setup (0.4–1.6 g h−1), and a factor 3 increase for gold (Au) NPs (0.32–0.94 g h−1). The scalability of the proposed MB‐PLAL technique setup is confirmed by Au and Fe50Ni50 NPs productivity experiments using 1, 6, and 11 beams, showing a linear increase in productivity.


Introduction
Pulsed-laser ablation in liquid (PLAL) is a versatile technique to synthesize a wide variety of colloidal nanoparticles (NPs) by ablation with a high-intensity pulsed laser of a bulk target immersed in the desired liquid. [1]When a high-intensity laser pulse (>10 9 W cm À2 ) [2] reaches the target material, the surface of the material is evaporated, forming a hot plasma plume containing ions and atoms of both the target and the liquid.Once the plasma plume collapses and the cavitation bubble is formed from the evaporation of the surrounding liquid, the ions and atoms of the bulk target are released to the liquid media due to the rapid release of vapor, [3] while larger droplets are ejected through the photomechanical spallation. [4]The process is followed by the condensation due to the rapid quenching by the liquid (evaporation-condensation mechanism [5] ).[9] The production of NPs by PLAL offers several advantages, including the synthesis of surfactant-free NPs; [10,11] the versatility of the process [12,13] that allows the generation of metallic; [14] alloyed metal, [15] oxide, [16] ceramic, [17] and organic NPs; [18] and the ability to produce NPs with complex structures and compositions that poses a challenge for standard chemical methods, [19][20][21][22][23][24] such as metastable bimetallic alloys, [25,26] or high-entropy alloy NPs. [22][33] In addition, PLAL complies with the green chemistry principles [34] compared to chemical synthesis routes as it is performed in an ambient atmosphere, requires less or no hazardous solvents and chemicals, produces less waste, and favors efficient reactant employment (atom economy). [35]espite the advantages offered by PLAL, its industrial use is still limited due to the low production rate compared to the chemical synthesis routes. [35]The productivity of PLAL ranges from several milligrams to several grams per hour depending on the experimental parameters, with the highest mass productivity at 8 g h À1 achieved for the ablation of Pt in water using a highpower laser and high scanning speed system that requires a huge initial investment. [36,37]Increasing PLAL productivity represents nowadays one of the main challenges of this technique. [2,35]everal approaches to improve and study the deciding factors I. Y. Khairani, F. Riahi, B. Gökce, C. Doñate-Buendía Chair of Materials Science and Additive Manufacturing School of Mechanical Engineering and Safety Engineering University of Wuppertal 42119 Wuppertal, Germany E-mail: goekce@uni-wuppertal.de of PLAL productivity have been discussed, [35] including target-related parameters such as geometry, [38,39] feeding method, [39,40] porosity, [41] and composition; [13,42] liquid-related parameters such as liquid dynamics, [43,44] layer thickness, [40,45] and viscosity; [46][47][48] as well as laser-and scanner-related parameters such as laser fluence, [49] pulse duration, [50,51] and scanning speed. [52,53]One of the most successful approaches up to date to increase PLAL productivity is employing a high repetition rate (MHz) and high power (hundreds of watts) picosecond laser [35] coupled with a high-speed (hundreds of meter per second) beam steering system to maximize the number of pulses ablating the target (high repetition rate) while keeping the fluence of each pulse above the threshold fluence, ideally at approximately F = e 2 ⋅ F thr , [54] where F is the fluence (on the target during the experiment) and F thr is the threshold fluence.This approach allows to maximize the inter-pulse distance on the target to avoid cavitation bubble shielding by utilizing high scanning speed. [53][59] Further increasing the repetition rate and power of the picosecond laser sources required for high productivity PLAL is limited by the technological advances in the laser manufacturing industry, finding already in the literature PLAL experiments with 3 ps, 500 W, and 10 MHz. [53]The most common laser-steering technology, galvanometer scanner, provides speeds up to 50 m s À1 , [60][61][62] but assuming a cavitation bubble size of 100 μm, [52] this speed can only avoid beam shielding for a repetition rate of 350 kHz. [53]hus, a scanning speed higher than 100 m s À1 is essential to accommodate a high-power and high-repetition-rate laser in the PLAL process.Barcikowski and coworkers utilized a polygon scanner which can reach a scanning speed of 484 m s À1 to achieve a productivity of 8 g h À1 for the ablation of Pt in water [36] and 3.8 g h À1 for the ablation of Au in water. [52,53]Nevertheless, to avoid uncontrolled beam deflection due to the edges between the rectangular mirrors of the polygon scanner, a duty cycle is required, limiting the effective laser power to 50%. [36,52,53]urthermore, the high price of the polygon scanner compared to the galvanometer scanner increases the capital investment and so the NP production cost in PLAL, [37] disrupting its prospective use in industrial applications.Consequently, even though a maximum PLAL productivity of 8 g h À1 has been achieved for Pt NPs, the specific laser source and scanning systems require a large initial investment and cannot be acquired extensively in other research labs and industrial facilities to widen the employment of PLAL for large-scale NP production.Hence, alternative approaches that can be implemented with standard commercial laser sources and galvanometric scanners should be explored to deliver pulses with energies above the ablation threshold of the material with megahertz repetition rates while achieving a sufficient interpulse distance to bypass the cavitation bubble and achieve production rates in the gram per hour scale.
In this work, we propose an approach to increase PLAL productivity by adapting the successful strategy of parallel multi-beam processing employed in laser material processing in air to PLAL.][65] The employed beamsplitting strategies distribute the laser beam into an array of M Â N spots or lines [63] with a pulse energy reduction of a factor M Â N, where M and N are natural numbers.If the initial laser pulse has enough energy to maintain the desired fluence in the individual spots, beam splitting allows the production of defined patterns on the target with a single shot instead of requiring M Â N individual laser exposures, hence highly reducing the processing time in the optimum case by a factor of M Â N. [66,67] In addition to that, beam splitting enables to operate at the optimal material processing fluence ð≈e 2 ⋅ F thr Þ for high energy and power laser sources. [58]There exist different approaches to split the laser beam, [68] including the employment of static diffractive optical elements (DOEs), spatial light modulators (SLMs), and acousto-optic or electro-optic modulators (AOM/EOM).Each of the methods has a different working principle and advantages: 1) DOEs consist of static optical elements with periodic microstructures that modify the beam's phase and amplitude, 2) SLMs dynamically modulate the beam's phase, amplitude, and/or polarization applying electrical signals to electrically anisotropic liquid-crystal molecules, 3) AOMs use a piezoelectric transducer to generate standing sound waves which modify the refractive index of a crystal, while EOMs employ variable electric voltage signals to modulate the refractive index of an electro-optic crystal.Nevertheless, SLM's diffraction exhibits an efficiency of ≈40%, [69] while AOM/EOM requires a large initial investment and can be limited by the achievable pattern size.In addition, specific SLMs and AOMs/EOMs are required for high power and repetition rate picosecond laser sources due to their damage threshold and cooling requirements, hence significantly increasing the price. [68]In this study, DOEs are chosen due to their high damage threshold; [70] high efficiency (typically in a range of 80-95%); [68,70,71] high pattern homogeneity; [63] robustness against beam parameters modification such as beam size, beam quality, and lateral displacement; [63] and lower price and easy implementation in any optical setup compared to SLMs and AOM/EOMs, allowing to easily adapt and transfer the proposed MB-PLAL system to any PLAL system available worldwide.The MB-PLAL is envisioned to provide the benefits of parallel multi-beam laser processing as higher material removal and efficient employment of high-power laser sources.In addition, an improvement of the essential factor required to increase the productivity of PLAL method is expected: the increase of the interpulse distance by reducing the repetition rate needed to achieve the optimum processing fluence, hence reducing the PLAL demand for faster scanning systems to avoid cavitation bubble pulse shielding and scale up NP production rate.This report highlights the use of MB-PLAL to successfully increase the productivity of iron nickel alloy by factor 4 (1.6 g h À1 ) through the integration of 11-beam splitter DOE.Morphology, particle size, and phase of the produced NPs were also analyzed, proving uniform and consistent properties of the generated particles as a function of beam-splitting number, rendering suitable use of MB-PLAL in the upscaling process and industrial outlook.

Experimental Section 2.1. Material Selection, Productivity Determination, and NP Characterization
The targets employed to investigate the MB-PLAL production upscale were Au (1Â 20 Â 70 mm 3 , 99.99%, EVOCHEM Advanced Materials GmbH) and Fe 50 Ni 50 (1 Â 20 Â 70 mm 3 , 99.95%, Sindlhauser Materials GmbH).Au was employed as the reference material in PLAL productivity.On the other side, FeNi was selected as an example of a technologically relevant nanomaterial required in large amounts.[74] As green hydrogen would play a key role in decarbonization, [75] the production of this type of renewable energy through electrolysis represents a major goal to address the UNESCO sustainable development goals. [76]Hence, supplying abundant and efficient FeNi NPs catalysts to meet the demand for electrolyzers is critical.All the obtained productivity values were measured by the gravimetric method after 5 min of PLAL, weighing the target before and after ablation with an analytical balance.The splitting of the beam to generate the multiple beams was done using DOEs and 6-beam DOE and 11-beam DOE were used.The productivity measurements from the PLAL experiments using a single-beam DOE, 6-beam DOE, and 11-beam DOE were repeated three times to ensure reproducibility.The characterizations of the generated FeNi and Au colloidal NPs were performed by a high-resolution transmission electron microscopy (HR-TEM, JEOL JEM-2200FS, 200 kV, ZrO 2 /W emitter) and powder X-ray diffraction (XRD) (Bruker D8 Advance Powder Diffractometer, Bragg-Brentano geometry, CuKα radiation 1.5418 Å, 40 kV, and 40 mA).To simplify the naming, standard single-beam PLAL without any DOE will be further addressed as "standard PLAL", while the PLAL process with 1:6-and 1:11-beam-splitter DOE will be referred to as "6-beam MB-PLAL" and "11-beam MB-PLAL", respectively.

Single-Beam PLAL
A 10 ps neodymium-doped yttrium aluminum garnet (Nd:YAG) laser with a wavelength of 1064 nm, an average power of 120 W, a tunable repetition rate of 400-4000 kHz, a raw beam diameter (1/e 2 criteria) of 5 mm, and a beam quality of 1.11 was employed.The laser beam was directed on the Fe 50 Ni 50 (FeNi) or Au target by a galvanometer scanner coupled with an f-theta lens (focal length of 167 mm) describing an Archimedean spiral pattern (10 mm diameter) with a marking speed of 20 m s À1 .Several parameters such as repetition rate, working distance, and liquid flow rate were optimized to accommodate the highest productivity achieved in the setup.The laser parameters such as laser power, repetition rate, beam area at plane, and pulse energy are presented in Table S1, Supporting Information.The highest laser power achievable for the employed repetition rate was employed to ensure productivity maximization.Distilled water (18.2MΩ cm at 25 °C ultrapure Mili-Q water, Synergy Water Purification System) was used as the liquid source and pumped by a peristaltic pump at 500 mL min À1 .A quartz window (2 mm thickness) with an antireflective coating at laser wavelength (R < 0.25%) was employed, and the liquid layer thickness, defined as the distance between the inner side of the glass and the surface of the target, was approximately 6 mm.

MB-PLAL Process (Multiple Beams)
The splitting of the initial laser beam into 6 and 11 equivalent beams was achieved by the use of beam-splitting DOEs, a 1:6-beam-splitter DOE (HOLOEYE Photonics AG) and a 1:11-beam-splitter DOE (LIMO Lissotschenko Mikrooptik GmbH), respectively.Parameters such as repetition rate and working distance were independently optimized for each material and the number of beams produced by the DOE.The DOE was placed after the f-θ lens of the scanner to accommodate the use of a large scanning pattern, Figure 1.Placing the DOE before the scanner limits the length of the array of spots.In addition, separating the beam before the scanner may induce larger positioning uncertainty for the beams at the edge of the f-θ lens. [58]The average power of the laser before adding the DOE was measured to be approx.100 W at 400 kHz.After implementing 11-and 6beam DOEs, the power was slightly reduced to ≈98.8 and 99.3 W, respectively, suggesting only minor power losses well below 2%.The beam splitters reduced the pulse energy of each individual beam after the 6-beam and 11-beam DOEs by a factor of 6 and 11, respectively.
To explain the effect of the DOE on the MB-PLAL processing conditions, a schematic representation is shown in Figure 2. A standard PLAL without beam splitting is represented in Figure 2a, where the laser beam with a certain power P and repetition rate f rep results in a pulse energy E p as described by Equation ( 1): where P is the average laser power (W); f rep is the repetition rate (Hz); and N is the number of beams.The normalized pulse energy in the standard PLAL setup as a function of time and position on the target is depicted in the graphs in Figure 2a.As a representative example, a twofold beam splitting is illustrated, the beam from the laser with a certain repetition rate f rep and pulse energy E p was split into two beam parts, halving the pulse energy for each beam to (E p =2Þ (see Figure 2b).Hence, to obtain the same pulse energy E p as in the standard PLAL, see Figure 2a, the laser power had to be doubled ð2PÞ while keeping the same repetition rate f rep ; this way, the delivered number of pulses was twice the standard PLAL, while the pulse energy was the same.In the case where power cannot be doubled, the repetition rate needed to be halved ðf rep =2Þ to obtain, the same pulse energy Figure 2c; this way, the pulse energy for each of the beams generated after the DOE was the same as the initial standard PLAL system as shown in Figure 2a and, as a positive side effect, the spot distance on the target was increased by two times, reducing bubble shielding.
In our case, the goal is to upscale NP productivity, thus, the laser power employed in every experiment was the one that our laser source could deliver.To adjust the fluence employed and maximize productivity, the approach followed was to adjust the repetition rate.The optimum repetition rate for the standard PLAL of FeNi in water was 3000 kHz (Figure S1, Supporting Information).When the beam was split into 6 and 11 beams, in the case of 6-beam and 11-beam DOE PLAL, the repetition rate was reduced to 500 and 400 kHz, respectively, trying to reach the optimum value that would be dividing the repetition rate of the maximum productivity achieved with the standard PLAL system by 6 and 11.However, 400 kHz was the lowest repetition rate achievable in our laser system.

PLAL and MB-PLAL FeNi NPs Productivity
To study the influence of beam splitting on PLAL productivity and investigate the advantages and limitations of the proposed MB-PLAL configuration, we compared the mass productivity values of the standard PLAL and the MB-PLAL setups.As shown in Figure 3a, the FeNi NP productivity using the standard PLAL configuration was 0.4 g h À1 .After placing the 6-beam DOE, the repetition rate was reduced to 500 kHz (one-sixth of the repetition rate of standard PLAL) to employ a similar pulse energy value, as explained in Section 2.3.With the 6-beam splitter, we obtained a productivity jump from 0.4 to 1.24 g h À1 , which represents a factor 3 increase (Figure 3a).The 11-beam MB-PLAL, in contrast, was performed at 400 kHz due to the impossibility of further reducing the repetition rate of our laser source, hence, the pulse energy and fluence in this setup was lower than the standard PLAL and 6-beam MB-PLAL.Nevertheless, the increase in productivity is still observed, obtaining productivity as high as 1.6 g h À1 with an increasing factor of 4 compared to the standard PLAL.To the best of our knowledge, this is the first time a productivity value of 1.6 g h À1 has been reported for PLAL of Fe 50 Ni 50 in water.The power-specific productivity (Figure 3b) was calculated by dividing the mass productivity by the employed laser power.This comparison of power-specific productivities using different beam splitters is important to evaluate the power efficiency of the laser system after the DOE addition.As shown in Figure 3b, the power-specific productivities of 6-beam and 11beam MB-PLAL were increased 3.5 and 4 times compared to the standard PLAL, respectively.The increasing trend of power-specific productivity is also similar to the trend of mass productivity, which confirms the enhanced laser power delivery to the target for the MB-PLAL configuration.
The underlying reason for this improved delivery of laser power and hence productivity rise lies in the achieved interpulse distance increase from approximately 7 μm (single beam) to 50 μm (11 beams) due to the repetition rate reduction.The larger interpulse distance reduces the laser beam interaction with the cavitation bubble generated by the previous laser pulse (Table 1).The advantage of using the beam splitter is the ability to increase the interpulse distance while keeping the same number of pulses delivered to the target.In the standard PLAL system without the beam splitter (and a fixed scanning speed), one might need to reduce the repetition rate value to achieve the desired interpulse distance.However, the reduction of the number of pulses irradiating the target would highly reduce productivity.In addition, the cavitation bubble would be enlarged due to the increase of the pulse energy [77,78] resulting in the subsequent increase of the laser pulse shielding.If the repetition rate and fluence are kept constant, the compromise would be to lower the laser power, which means that the maximum outcome of the laser source is not fully utilized to achieve the highest productivity.Meanwhile, if we want to keep the same pulse energy, fluence, repetition rate, and power while increasing the interpulse distance, the scanning  speed has to be increased to at least 150 m s À1 to achieve an interpulse distance of 50 μm with 3000 kHz repetition rate (maximum FeNi NP productivity for the standard PLAL system).This high scanning speed can be achieved by a polygon scanner, but the polygon technology requires the laser shutter to be closed when the beam pathway is nearing the corner of the polygon mirrors to avoid uncontrollable beam pathways which can damage the scanner. [36]Depending on the duty cycle, the delivered laser power could be cut by half [53] and the power efficiency is therefore reduced. [36]In addition to that, the capital investment to purchase a polygon scanning system [37] is significantly greater than the price of a galvanometer system combined with the DOE.Based on these considerations, we propose the MB-PLAL system as an economical alternative to boost the productivity of PLALgenerated NPs by increasing the interpulse distance and efficiently delivering pulses at MHz repetition rates without considerable power losses or the necessity to use cutting-edge highspeed scanning systems.We further measured the productivity of the FeNi target at the same fluence and repetition rate by changing the laser power (Figure S2, Supporting Information).In this measurement, we aim to calculate the productivity increase factor as a function of the number of beam(s) at the same fluence value, as the fluences presented in Figure 3 could not be kept constant due to the laser's technical limitations not allowing to reduce the repetition rate below 400 kHz.The measurement parameters of this study are presented in Table S2, Supporting Information.From the standard PLAL to 6-beam and 11-beam MB-PLAL, the productivity increase factors are 3.6 and 6.4, respectively (Figure S2, Supporting Information).Even though the laser power is increased according to the number of beams, the productivity is not increased 6 and 11 times.The nonlinear scale-up is probably due to the energy lost as shock waves, [79] in addition to turbulences and backflow inside the ablation chamber, promoting laser shielding by the NPs, cavitation bubbles, and persistent microbubbles.Based on the linear fitting, it is possible to infer that increasing the number of beams with other DOEs could result in even higher productivity if a laser with a higher power and higher repetition rate is used.The inset in Figure S2, Supporting Information, shows the ablation area of the FeNi target after the 11-beam MB-PLAL process, confirming an increase of seven times compared to the single-beam ablation.The influence of the ablation area and spatial beam overlapping on productivity have not been evaluated; however, the 11-beam MB-PLAL shows the highest productivity with the largest ablation area, pointing out the possibility of a future further productivity increase by avoiding beam overlap.
Ensuring consistent properties, i.e., morphology, particle size, and crystalline structure, of the produced NPs in the MB-PLAL is a crucial consideration for the scalability of the process and the prospective industrial use.The particle size distribution of the generated FeNi NPs was measured to investigate the influence of MB-PLAL setup to the morphology and size of the resulting NPs, Figure 4a.Based on the TEM images in Figure 4a, FeNi NPs are formed as core-shell, which is in agreement with the structure reported in previous FeNi PLAL experiments. [19]The median values (x c ) of the log-normal fitting of all the samples are similar, 14 AE 7 nm for the standard PLAL (1 beam), 11 AE 9 nm for the 6-beam MB-PLAL, and 14 AE 13 nm for the 11-beam MB-PLAL.The x c is slightly smaller for the 6-beam MB-PLAL, but the difference is not significant and within the standard deviation.The polydispersity index (PDI), obtained from σ 2 /μ 2 where σ represents the standard deviation and μ indicates the mean value, shows a value of 0.26 for the single-beam ablation, which is smaller than the samples with beam splitting with respective values of 0.44 and 0.53 for 6-and 11-beam processing.Although there seems to be a trend of increasing PDI and standard deviation values with the increasing number of beams, a more thorough investigation should be performed to confirm this trend.Based on our current observations, as shown in the inset of Figure 4a as well as the HR-TEM images of FeNi and Au NPs generated in different number of beams and repetition rates in Figure S3, Supporting Information, we do not observe any significant changes in the morphology or particle size as a function of the number of beams.Analysis of the crystalline structure was also sought to ensure consistent phase of the produced NPs, Figure 4b.Two phases are formed in the standard 1-beam PLAL of FeNi in water: face-centered cubic (FCC) FeNi and spinel NiFe 2 O 4 .The FCC FeNi occupies the core part of the NPs, while spinel nickel ferrite (NiFe 2 O 4 ) can be found in the shell part as a result of oxidation. [19]Based on the diffractograms of FeNi NPs generated with different number of beams, Figure 4b, there is no formation of new peaks or disappearance of peaks as compared to the single-beam diffractogram, which indicates that the crystalline structure of the FeNi NPs remains constant for MB-PLAL, finding the FCC and spinel nickel ferrite composition.It can be concluded that the MB-PLAL does not influence the properties of the generated FeNi NPs and it is therefore suitable for direct PLAL NP production upscaling without influencing the composition, phase, or size of the produced NPs.

PLAL and MB-PLAL Au NPs Productivity
The productivity comparison of FeNi with a benchmark material such as Au is essential to help us understand the influence of the material's property on the MB-PLAL productivity value and the possibility of generally extending the results.Colloidal gold NPs are chosen due to their practical versatility in various applications, such as air and water purification, immunotherapy, cancer treatment, sensor, biomarker, and drug delivery, [80][81][82] and their high-value increase compared to their bulk counterpart. [83]The standard PLAL and MB-PLAL mass productivity of FeNi and Au NPs are compared in Figure 5.The results shown were performed at the optimized processing parameters for each material (Table S1, Supporting Information), as each material has a different ablation fluence threshold.The standard PLAL productivity of Au and FeNi is similar, 0.32 and 0.40 g h À1 , respectively.The productivity increase factors of Au from the standard PLAL to 6-beam and 11-beam MB-PLAL are found to be 2.1 and 3, which is lower than FeNi increase factors of 3.1 and 4, respectively (Figure 5).
Although usually material-dependent productivity has been reported to be linked to the trend of material density, [84] other underlying reasons might have influenced the productivity differences observed between FeNi and Au in our case.1) Larger bubble half-width of Au than FeNi, Figure S5, Supporting Information, which contributes to a larger pulse shielding.The temporal distance between two pulses in the 11-beam MB-PLAL of Au and FeNi is 2.5 μs, as the employed repetition rate was 400 kHz.At this time range, the cavitation bubble halfwidth of Au is larger than FeNi, Figure S5, Supporting Information, resulting in a larger energy shielding.2) The steady-state optical absorptance of Au at our laser wavelength (1064 nm) is 3%, [85] lower compared to the FeNi which is 35%.The higher the optical absorptance of the material at the irradiation laser wavelength, the higher the energy absorbed by the target material, [86][87][88] leading to a higher ablation volume.3) The intrinsic chemical disorder of alloy materials as FeNi [89] contributes to a stronger electron-phonon coupling and lower thermal conductivity, [89] resulting in a lower threshold fluence than the corresponding elemental materials.The HR-TEM images of FeNi and Au NPs generated by different numbers of beams are presented in Figure S3, Supporting Information, where no change in particle morphology and size could be observed.While the FeNi NPs show mostly spherical morphology, the Au NPs seem to melt forming necks between the NPs.

Economical Perspective of MB-PLAL for High-Throughput NPs
The definition of productivity in PLAL does not solely revolve around mass productivity, but also power-specific productivity and investment-specific productivity.The power-specific productivity tells us the efficiency of the available laser power used for PLAL.It is important to make this distinction because high productivity values do not always imply high power-specific productivity.The laser power can be partially wasted due to the cavitation bubble pulse shielding, laser-liquid interaction, or the beam steering method such as in the case of polygon scanners. [35]We have discussed in the previous section the powerspecific productivity of FeNi and how it follows a similar trend as mass productivity.A similar trend is also observed for Au (Figure 6a), although the values are smaller than FeNi due to the aforementioned reasons.As we are aiming to scale up PLAL productivity by proposing the MB-PLAL configuration suitable for any PLAL setup, the value of power-specific productivity can give a hint of how efficiently the MB-PLAL can be implemented in other labs worldwide to upgrade their PLAL system to produce larger amounts of NPs.
However, another definition of productivity is needed to evaluate NP productivity as a function of capital investment, which is related to the price to procure the instruments and units required to start ablation.Capital investment budgeting is a vital part of management policy formulation because it correlates with many business factors such as growth, expansion, budget diversification, modernization, and long-term planning. [90]The value of investment-specific productivity tells us the hourly produced amount of NPs for every 1000 € of capital investment poured into the PLAL system.The higher the value, the more efficient the production is in terms of the required investment.Here, we define the PLAL system as the combination of the laser system, the optical table, the scanning system (including the software and the f-θ lens), the water pumping system, and the DOE if applicable.From the results presented in Figure 6b, the investmentspecific productivity of Au and FeNi with 11-beam MB-PLAL are 4.6 and 7.9 mg (hk€) À1 , respectively.A linear increasing trend, similar to the mass-and power-specific productivities, is also observed for the investment-specific productivity.This is because the price of the beam-splitter DOE is marginal compared to the laser and scanning systems, thus, the DOE does not significantly increase the total capital investment of the PLAL system.Hence, another advantage of MB-PLAL is the possibility to increase productivity without significantly increasing capital investment.
An intriguing question arises, whether the performance of the MB-PLAL system is more suitable and beneficial for industry compared to the world-record PLAL productivity system proposed by Streubel et al. [52] Here, we present Table 2, where we compare our experimental parameters and productivity results of Au PLAL to ref. [52].The first and the most striking comparison is the scanning speed.A high-speed polygon scanner with a speed of 484 m s À1 is employed in ref. [52] employed to achieve an interpulse distance of 48 μm at a repetition rate of 10.1 MHz.Meanwhile, our beam-splitting system could achieve an interpulse distance of 50 μm using a galvanometer scanner with a scanning speed of 20 m s À1 , which is 24 times slower than the polygon scanner. [52]The number of delivered pulses onto the target's surface is also another point worth comparing.In our case, we reduced the repetition rate to the lowest value in our system to compensate for the pulse energy splitting by the DOE and achieved a 50 μm interpulse distance.Nevertheless, we could still deliver 4.4 Â 10 6 pulses s À1 due to the splitting by the 11-beam DOE.In addition, as the pulse energy is reduced in the MB-PLAL setup, we can avoid the formation of a large cavitation bubble. [91]Although the number of delivered pulses, 4.4 Â 10 6 pulses s À1 , is still lower than in ref. [52] (10 7 pulses s À1 ), we were able to deliver the equivalent number of pulses to a laser operating in the megahertz regime even using a lower repetition rate of 400 kHz.The number of delivered pulses by the 11-beam MB-PLAL system at 400 kHz also exceeds the maximum number of pulses achievable with our laser in the standard PLAL configuration, operating at 4 MHz.Hence, beam splitting is advantageous for a laser system with high pulse energy but a low repetition rate, as the number of delivered pulses can be adjusted externally by selecting the optimum DOE.
Another point we would like to emphasize is power-specific productivity.The power-specific productivity is trimmed down to 7.6 mg hW À1 , which is lower than our value at 9.5 mg hW À1 even when we only used an average power of 100 W. Lastly, the investment-specific productivity of our system reached almost a similar value at 4.6 mg (hk€) À1 despite requiring only one-third Table 2. Comparison of experimental parameters and the resulting mass-, power-, and investment-specific productivities from ref. [52] and our experiment for the ablation of Au in water.The initial investment covers the price of purchasing the laser, the scanning system (including the f-θ lens), the optical table, the pumping system, and the DOE (if applicable).

Parameters
Streubel et al. [52] This work of the capital investment compared to ref. [52].The price difference comes mostly from the laser and scanning system, as a custom high-power picosecond laser (500 W) and a high-speed polygon scanner are employed in ref. [52].A lower initial capital investment, in addition to the commercially available laser and DOE employed in this work, could increase the likelihood for the industry to invest and utilize the MB-PLAL configuration especially for small to medium-sized companies, and also to facilitate access to PLAL to research institutions and universities requiring a versatile NP production technique.
With an aim to upscale and introduce MB-PLAL to the industry, we present the economical perspective of Au NPs production using MB-PLAL.We performed a cost-benefit analysis related to the labor and electricity costs that are needed to produce 1 kg of Au NPs (Table 3) and the daily and yearly production rate (Table 4).The calculations of labor and electricity costs to produce Au NPs are based on the highest Au productivity achieved in our laser system.Several assumptions are considered, such as the gross salary assumption of the working staff at 21 € h À1 and the labor time which considers 2 h daily initial set up and adjustment for every 8 h ablation.The electricity consumption is measured for the whole PLAL system, which is approximately 3 kWh.
Based on the calculation, using the 11-beam MB-PLAL, the total labor working hours are reduced by one-third compared to the standard PLAL, which saves approximately 54 000 € of labor cost for every 1 kg of Au NPs produced.In addition, the electricity consumption could be trimmed down by 66% and it is possible to save around 2500 € per 1 kg of Au NPs when using the 11-beam splitter.As also calculated by Jendrzej et al. in their cost comparison to produce 1 g of Au NPs, [83] the energy cost (electricity) to produce the Au NPs is much smaller compared to the labor cost to operate the laser.Hence, the main advantage of using the MB-PLAL is the labor cost reduction.
Meanwhile, for the time-based production rate, as shown in Table 4, we focus on the daily and annual production as well as the percentage increase of production when using the DOE compared to the standard PLAL.By assuming the 16 h daily ablation time and 350 days of annual working days, the percentage increase of the daily and annual production rate using the 11-beam MB-PLAL compared to the standard PLAL is almost 200%, where the daily production is increased from 3 to 8 g per day, and the annual production is increased from 640 to 1800 g per year.The most important point based on these calculations is that we can achieve this jump in production and cost reduction just by increasing the capital investment by 1.3% to acquire the 11-beam splitting DOE.As for the case of the 6-beam DOE, the price is more affordable and the increase in capital investment is only 0.5%.The minuscule addition of capital investment and the low implementation time of the DOE in the PLAL setup are insignificant compared to the benefits achieved by the proposed MB-PLAL system.
Further calculations of the capital expenditure (CAPEX), operational expenditure (OPEX), and production cost per kilogram of the NPs have been done as both are crucial factors in deciding to start a new investment.In the CAPEX calculation, we consider the equipment needed to produce colloidal NPs.For the OPEX calculation, we measured the electricity consumption of the devices and the operational cost, which includes the labor cost to operate and maintain the system.Please note that for these calculations, we do not include the expenditure related to the land acquisition, building and infrastructure establishment, characterization of the produced NPs, and other parameters that are commonly included.Our focus is on the PLAL setup.Hence, the calculated values of CAPEX, OPEX, and production cost per kilogram are still less than the actual expenditure.
The calculated CAPEX of the 11-beam MB-PLAL is 267,000.00€ with 25% of contingency.Meanwhile, the total OPEX is 95,500.00€ consisted of total direct cost of 39,300.00€ (electricity and material cost) and total indirect cost (operational and services) of 56,200.00€.The calculated production cost per kilogram of Au NPs is approximately 22,400.00€.As for the standard PLAL and 6-beam MB-PLAL, the CAPEX is only slightly different from the 11-beam MB-PLAL as the DOE is the only changed element and the price is only approximately 1% of the total CAPEX.The CAPEX of standard PLAL is 263,900.00€ while 6-beam MB-PLAL is at 265,200.00 €.The OPEX of the standard PLAL is the lowest as fewer gold targets are required in Assuming daily initial setup and manual adjustment of the working distance (+2 h for every 8 h).b) Estimation of the average gross salary of a chemical laboratory assistant of 21 € h À1 .c) Measured value of electricity consumption of the PLAL system of 3 kWh.Assuming daily ablation time of 16 h.b) Assuming 1 year production is equal to 350 days.
production.Nevertheless, production cost is the highest at 52,200.00 € kg À1 .The OPEX of 6-beam MB-PLAL is approximately 81,400.00€ and the production cost is 26,900.00€ kg À1 of Au NPs.Based on this calculation, 11-beam MB-PLAL offers the lowest production cost per kilogram of Au NPs, which is half of the standard PLAL.
Comparison of the PLAL and chemical procedure has been thoroughly investigated by Jendrzej et al. [83] The break-even point of the PLAL method compared to the chemical reduction method to produce Au NPs is at NPs productivity of 550 mg h À1 . [83]ence, 11-beam and 6-beam MB-PLAL already surpassed the cost effectiveness of reduction method and are more economically viable.An interesting discussion of the sustainability footprint of PLAL technique compared to the chemical reduction method can be found in recent publication by Havelka et al., [92] emphasizing the importance to shift the NPs production method to PLAL as a green chemistry method.

Conclusion
MB-PLAL has been shown as a feasible and practical technique to increase NP productivity by only integrating a static DOE into the PLAL system, improving the efficiency and economic viability of colloidal NPs production.The employed 6-beam and 11-beam DOE increased the productivity of FeNi PLAL in water by factors 3 and 4, respectively, with a maximum FeNi NPs productivity of 1.6 g h À1 .The energy splitting into multiple beams achieved by the DOEs has been shown to require the reduction of the repetition rate by a factor equal to the number of beams generated to keep the same optimum processing fluence per beam produced by the DOE compared to the single-beam PLAL.This fact allows us to increase the interpulse distance from 7 to 50 μm without the use of expensive high-speed scanners, reducing the beam shielding by the cavitation bubble, while the number of pulses delivered to the target is not affected.In the case of reducing the repetition rate without the use of the DOEs, the high pulse energy promotes nonlinear interactions with the liquid and induces optical breakdown, lowering the productivity due to the extra energy losses.The proposed MB-PLAL system thus enables us to employ the optimal pulse energy and fluence to achieve the highest productivity while increasing the interpulse distance without modifying the scanning parameters.This is further confirmed by the trend observed for increasing power-specific productivity as well as mass productivity, indicating that the MB-PLAL system causes a reduction of the factors affecting the energy delivery to the target as pulse shielding due to the cavitation bubble.
The properties of the produced NPs do not change for MB-PLAL compared to single-beam PLAL, based on the HR-TEM, XRD, and UV-vis results, indicating that the MB-PLAL with repetition rate compensation to keep the same processing fluence as single-beam PLAL does not modify the ablation mechanism or chemical processes during the NPs generation.The only parameter modified from the standard PLAL to the MB-PLAL is the interpulse distance (due to the repetition rate compensation), while the pulse energy, delivered number of pulses to the target's surface, and pulse width are kept approximately the same.The increase of the interpulse distance reduces cavitation bubble shielding, thus increasing NP productivity.Meanwhile, pulse energy and the number of delivered pulses in our setup are kept approximately constant, which is only possible due to the use of the DOE and repetition rate compensation.Hence, the properties of the generated NPs in the MB-PLAL are not modified compared to the single-beam PLAL system.
A comparative study with Au productivity was also performed to investigate the material's influence on the MB-PLAL system and the possibility of extending the MB-PLAL technique to any material processable by PLAL.Productivity increase factors of 2.1 and 3 were observed for the 6-beam and 11-beam MB-PLAL of Au in water, respectively, with a maximum productivity value of 0.94 g h À1 for the 11-beam MB-PLAL.The lower productivity increase factor of Au compared to FeNi can be caused by the typically larger cavitation bubble observed for Au.Hence, even for the achieved larger interpulse distance, pulse shielding from the cavitation bubble is expected to further influence Au NP production.In addition, the lower optical absorptance of Au at the 1064 nm employed laser wavelength results in lower energy absorbed by the target material.Further material properties affecting productivity are Au's higher threshold fluence, the weaker electron-phonon coupling, and higher conductivity of Au compared to FeNi, leading to energy thermal dissipation.
With the aim to introduce the MB-PLAL system as a costeffective NP production system, the MB-PLAL system proposed is compared with the current most productive PLAL system worldwide based on a high-speed polygon scanner configuration proposed by Streubel et al. [52,53] With a laser power five times lower than the polygon-PLAL system, higher power-specific productivity was achieved for the MB-PLAL system, 9.50 mg (hW) À1 .Meanwhile, the MB-PLAL allowed to reach a similar interpulse distance, 50 μm, using a typical galvanometer scanner with a speed of 20 m s À1 , which is 24 times slower than the polygon scanner employed by Streubel et al. [52,53] A further benefit of the MB-PLAL system lies in the lower initial capital investment, which is approximately one-third of the polygon-PLAL configuration, easing the access of PLAL to research institutions and industries that require colloidal NPs.
A comparison of MB-PLAL and PLAL employing the same laser source allows us to conclude that MB-PLAL can achieve an annual production of approximately 1800 g of colloidal Au NPs, which is almost a 200% production increase compared with the PLAL system.A decreased labor cost and electricity consumption of 54 000 € and 20 000 kWh are expected for every 1000 g of Au NPs produced with the 11-beam MB-PLAL compared to the PLAL using the same laser source and scanning system.Prospectively, the proposed MB-PLAL system can be integrated into the current high-productivity PLAL systems available worldwide to further boost achievable productivity with a minimum cost and low experimental effort.Furthermore, the employment of the MB-PLAL system with higher laser power systems with repetition rates in the MHz range or higher would allow to employ beam splitters to generate a larger number of beams.Integration of such systems with a faster scanning speed will increase the interpulse distance, further lowering pulse shielding and reaching even higher production rates, enhancing the efficiency and economic viability of PLAL for industrial applications.

Figure 1 .
Figure 1.a) Schematic illustrations of the MB-PLAL process in a flow chamber with a 1:11-beam-splitter DOE and b) real setup of the experimental procedure with top and side views.Note that the beam splitter DOE is placed close to the f-θ lens to avoid the focused beam damaging the DOE.

Figure 2 .
Figure 2. Schematic diagram of the PLAL process to illustrate the relation between pulse energy E p and repetition rate f rep in a) a standard PLAL without beam splitting, b) an MB-PLAL setup generating two beams without repetition rate compensation (using the same repetition rate f rep as the standard PLAL), and c) an MB-PLAL setup generating two beams and with repetition rate compensation (repetition rate value is reduced by a factor of 2 ðf rep =2Þaccording to the number of beams generated, leading to a two times larger spot spacing on the target).Please note that the normalized pulsed energy of the undiffracted beam in (c) is twice (2E p ) of the normalized pulsed energy of diffracted beam (E p ).The term "a.u." is an abbreviation from "arbitrary unit," indicating a comparative unit for a relative quantification.

Figure 3 .
Figure 3. a) FeNi NP productivity in water values at the optimized parameters.b) Productivity and power-specific productivity comparison as a function of the number of beams.

Table 1 .
The influence of the proposed repetition rate compensation in MB-PLAL depending on the number of generated beams by the DOE and the effect on the interpulse distances.

Figure 4 .
Figure 4. a) Particle size distribution and b) XRD diffractogram of standard PLAL (1 beam) and MB-PLAL (6 beams and 11 beams) of FeNi NPs.The measurements of particle size in (a) were done using ImageJ software of more than 400 particles.The insets in (a) show the HR-TEM images of FeNi NPs generated using different numbers of beams.

Figure 5 .
Figure 5. Mass productivity comparison between FeNi and Au ablation in water for PLAL and MB-PLAL NP production experiments.

Figure 6 .
Figure 6.a) Investment-specific productivity and b) power-specific productivity of FeNi and Au NPs for PLAL and MB-PLAL as a function of the number of beams.The shown error bars define the standard deviations calculated from three measurements.

Table 4 .
Calculation of the annual production rate of Au NPs using different numbers of beam(s).Please note that the electricity and labor costs are the same as the installment of DOE does not change the production costs.DOE Productivity [g h À1 ] Daily production [g] a) Annual production [g] b)

Table 3 .
Calculation of labor and electricity cost to produce 1 kg of Au NPs using different numbers of beam(s).