Efficient Synthesis of Submicrometer‐Sized Active Pharmaceuticals by Laser Fragmentation in a Liquid‐Jet Passage Reactor with Minimum Degradation

One challenge in the development of new drug formulations is overcoming their low solubility in relevant aqueous media. Reducing the particle size of drug powders to a few hundred nanometers is a well‐known method that leads to an increase in solubility due to an elevated total surface area. However, state‐of‐the‐art comminution techniques like cryo‐milling suffer from degradation and contamination of the drugs, particularly when sub‐micrometer diameters are aspired that require long processing times. In this work, picosecond‐pulsed laser fragmentation in liquids (LFL) of dispersed drug particles in a liquid‐jet passage reactor is used as a wear‐free comminution technique using the hydrophobic oral model drugs naproxen, prednisolone, ketoconazole, and megestrol acetate. Particle size and morphology of the drug particles are characterized using scanning electron microscopy (SEM) and changes in particle size distributions upon irradiation are quantified using an analytical centrifuge. The findings highlight the superior fragmentation efficiency of the liquid‐jet passage reactor setup, with a 100 times higher fraction of submicrometer particles (SMP) of the drugs compared to the batch control, which enhances solubility and goes along with minimal chemical degradation (<1%), determined by attenuated total reflection‐Fourier transform infrared spectroscopy (ATR‐FTIR), high‐performance liquid chromatography (HPLC), and X‐ray diffraction (XRD). Moreover, the underlying predominantly photo‐mechanically induced laser fragmentation mechanisms of organic microparticles (MP) are discussed.


Introduction
High-throughput drug screening is a modern and efficient way to find new drug candidates based on receptor-drug interactions.The main drawback is that over 70% of these new drug candidates are hydrophobic. [1,2][4] Therefore, strategies to increase the solubility of these hydrophobic substances are in high demand, especially for orally administered drugs. [5,6]umerous approaches, such as precipitation of the dissolved active ingredients in a co-solvent, [7] encapsulation, [8] micellar solubilization, [9,10] complexation, [11] microfluidic synthesis [12,13] or salt formation [14] have been implemented, but may lead to negative physiological effects due to the additives used. [15]Downsizing, i.e., reducing the particle size of solids, is a well-known and established method that increases the total surface area, which in turn leads to improved dissolution properties and better oral absorption, without additional additives. [3,16]Standard comminution methods used in the pharmaceutical industry to produce drugs in the sub-micrometer size range are milling and high-pressure homogenization. [17][20] Particularly for submicrometer particles (SMP) produced via cryogenic or wet grinding, the amount of degradation products is proportional to the grinding time and becomes substantial when milling durations of hours to days are required.For example, in furosemide, 0.8% (w/w) of degradation products were detected at a grinding time of 30 min, while after 180 min, the decomposition was already 6.8% (w/w). [18]Similar amounts of degradation could also be found in wet-ground naproxen samples.After a grinding time of 240 min, the amount of detected degradation products was already at 6%-8% (w/w). [20]Furthermore, process contaminations due to abrasion of grinding media or pistons lead to additional inorganic contaminants highly undesirable in materials intended for clinical use.
In this context, the use of pulsed laser systems and the process of laser synthesis and processing of colloids has proven to be a promising comminution technique with a wide range of applications. [21]Particularly in the field of inorganic materials, such as metals, alloys, and semiconductors the technique is established and mechanistically well understood, [21][22][23] but also the transfer to organic substances has been frequently reported. [24,25]t has been established that laser processing of colloids is ideally suited for the production of high-purity organic nanoparticles as well, which are obtained in aqueous solutions and are thus directly available as stable colloids. [26]Initial attempts to transfer laser comminution to active pharmaceutical ingredients were carried out several years ago [27,28] and are still under development.In addition to drugs, the use of pulsed laser systems has also been extended to food products such as coffee or cinnamon and organic dyes, with promising results. [26,29,30]wo basic process variants are frequently described, laser ablation in liquids (LAL) and laser fragmentation in liquids (LFL).For LAL, a pressed drug target is irradiated in air or aqueous medium to obtain the resulting nano-or submicrometer-sized particles.[33][34][35] LFL, on the other hand, is based on the irradiation of a drug-microparticle (MP) suspension with a pulsed laser beam.This process has also led to successful particle size reduction down to the SMP range, as demonstrated for megestrol acetate, paclitaxel, beclomethasone dipropionate, naproxen, and fenofibrate. [27,28,36,37]Depending on irradiation time and applied laser power, an inversely proportional correlation of particle size and portion of degradation products was shown, irrespective of the laser method, laser type or drug used, leading to the formation of significant amounts of decomposition products in the case of nano-sized active substances. [37]Furthermore, better comminution success was observed at lower drug concentrations, due to the lower scattering of the incoming laser beam. [27]Previous studies using naproxen and fenofibrate as examples showed a dramatic increase in degradation of up to 20% once SMP sizes smaller than 1000 nm were reached.Further-more, ns-laser irradiation at a wavelength of 532 nm (2.5 W) lead to more degradation compared to fs-laser irradiation at 800 nm (250-400 mW). [37]When ns-lasers with a wavelength of 1064 nm (5 W) were used, only particles >1000 nm were produced.It could be shown that fs-lasers lead to more and smaller particles compared to ns-lasers with comparable wavelengths in the IR range, but a higher fragmentation efficiency (higher yield of small particles) always went along with increased chemical degradation of the drugs.In each case, 0.5 mg mL −1 of the active ingredient in water was used.To produce particle sizes of this order of magnitude, fragmentation times of 30 and 60 min were selected for 2 and 10 mL, respectively.In general, for all fragmented naproxen and fenofibrate samples, the initial normal distribution was shown to transform into a bimodal distribution with a reduced mean diameter. [37]Similar observations could be made for LFL with fs-laser at a wavelength of 800 nm using paclitaxel as an example.A sample volume of 2.5 mL at a concentration of 200 μg mL −1 in poloxamer 188 solution was used, which was fragmented at laser powers of 50-400 mW for 60 min.Only above a laser power of 150 mW particles <1000 nm were formed showing 12.6% of decomposition, whereas at 400 mW already more than 23% of the drug degraded. [27]Significantly lower amounts of degradation products were detected for megestrol acetate.Depending on the volume-to-irradiation time ratio, between 1.7%-3.6%degradation products were detected, whereby larger volumes or shorter irradiation times were used compared to paclitaxel.The drug concentration was 0.5 mg mL −1 in poloxamer 188 solution in each case.The same laser parameters were used as already described for paxlitaxel.The example of megestrol acetate also confirmed that particle sizes <1000 nm are generated only with laser powers of 250 and 400 mW. [28]For beclomethasone diproprionate, an inhaled drug that ideally requires μm-scale delivery particle sizes, it was shown that corresponding to the larger particles, fewer degradation products of 2.7% maximum were generated upon fs-laser fragmentation in the IR range. [36]In order to confirm the successful downsizing of drugs, Sylvestre et al. demonstrated the increased bioavailability of fragmented megestrol acetate by means of in vivo bioavailability analyses in addition to in vitro dissolution tests.After oral administration of laserfragmented, ground, and untreated megestrol acetate control, the drug concentration in the blood of rats was monitored.Laser fragmented and conventionally ground samples were found to have significantly more drug dissolved in the blood compared to untreated samples. [28]s shown by the previous works on organic drugs LFL in a batch setup, it is highly important to improve the efficiency of the LFL-process of drug particles as the critical interplay between I) low efficiency necessitating the utilization of higher pulse numbers to obtain particles <1000 nm and II) drug degradation at a higher number of pulses/long irradiation times strongly limits its usefulness.For this purpose, a passage reactor was developed, where a liquid-jet enables selective illumination of the particle suspension, which leads to a reduction in the number of required pulses.Numerous studies on process optimization in the passage reactor have already been performed for inorganic and, to some extent, organic materials, focusing on material-specific parameters such as compound concentration, energy density, and laser wavelength. [22,25,38,39]In this work, we apply the passage reactor established for inorganic materials to organic drug suspensions.Yield and purity were determined by analytical centrifugation, scanning electron microscopy (SEM), high-performance liquid chromatography (HPLC), attenuated total reflection-Fourier transform infrared spectroscopy (ATR-FTIR), and X-ray diffraction (XRD).In comparison to previous studies on the laser processing of drugs primarily evaluating the fragmentation results qualitatively, in this work quantitative yield determinations were performed on a wide range of drug substances, in combination with in vitro solubility assays to establish a particle size-functionality correlation.Finally, fragmentation mechanisms are discussed based on calculations of stress confinement criteria at the example of naproxen particle LFL.
Figure 1 schematically shows the fragmentation setup in the passage reactor as well as the relevant readouts and analyses performed.

Results and Discussion
In initial screening studies, the influence of LFL in the passage reactor on organic drugs was tested.For this purpose, different drug concentrations (0.1 wt.%, 0.5 wt.%, and 2 wt.%) of all model drugs were added to the stabilizer solution and fragmented directly after ultrasonication in the passage reactor with different numbers of laser passages (LP).The particle size reduction, shown in Figure 2 for four model drugs, occurs independently of the particle shape and size.It is noticeable that no rounded structures are discernible.SEM images of all lasergenerated particles of the model substances indicate the presence of angular fragments with many cracks and fractured edges pointing at mechanical breaks.These findings suggest the prevalence of a mechanically dominated fragmentation mechanism in contrast to a thermally dominated one.Especially in the fragmented needle-shaped prednisolone particles in Figure 2D, these fracture edges are well visible and have been highlighted in red for better clarity.Figure 2C also shows that these cracks and fractures were not present in the untreated samples and were thus caused by laser fragmentation.Similar particle shapes and fragments were also reported by Kenth et al.Here, SEM images of water-exposed paclitaxel particles before and after ultrashort IR-laser fragmentation at 400 mW for 60 min also showed needleshaped initial particles transformed into angular fragments by fragmentation. [27]However, this observation is in contrast to the naproxen and fenofibrate particles obtained via IR-fs-LFL by Ding et al.Compared with ground samples, additional roundish particles were detected when 2 mL drug suspensions were irradiated for 30 min at 250 mW, which indicates a hybrid mechanothermal process. [37]The main difference in our work is the use of the liquid-jet passage reactor, as well as a green-ps-laser, with a 100 times higher repetition rate.The use of the flow reactor prevents the heating of the whole suspension.For all model drugs, the SEM images show that the submicrometer size range could be achieved by fragmentation.The educt particle size (LP0) of naproxen (Figure 2A) is polydisperse and ranges from 3 to 20 μm.After 100 laser passages (LP100), the fragmentation effect is clearly visible.The average particle diameter is now ≈400 nm (Figure 2B).This particle size estimated from SEM images could also be confirmed by quantitative analytical centrifugation measurements.Figure 3A,B show the corresponding hydrodynamic particle density distribution of the naproxen samples.LP100, shown in green, indicates the laser-comminuted particles, which range from 80 to 1000 nm with a maximum of ≈400 nm.In contrast, the untreated LP0 sample, shown in gray, can only be partially analyzed with a minimum diameter of 3000 nm.
To make a quantitative statement about the particle size distribution by means of an analytical centrifuge, an evaluation method was developed, which is graphically illustrated in Figure 3A,B and described in detail in the method section.The evaluation of the volume-based particle density distribution is performed by integrating the area under the curve (AUC) with a cut-off <1000 nm (Figure 3B, marked yellow) and the total AUC with a cut-off of 10 000 nm (Figure 3A, marked red) (AUC <1000 nm /AUC total ).Based on this ratio, relative information about the degree of laser-submicro-comminution can be obtained and comparability between samples is possible.
To evaluate the efficiency of comminution processes in the laser passage reactor, naproxen samples were additionally fragmented at the same pulse number in a control batch process.For this purpose, an analogous sample volume of the same drug con- centration of 0.1 wt.% was used and the fragmentation time was adjusted according to the pulse number.Fragmentation was performed with the same lens setup.In a stirred glass vessel, the laser beam was focused into the center of the suspension.After ≈333, 833, and 1665 s, which corresponded to a pulse number of 0.33*10 8 , 0.83*10 8 , and 1.67*10 8 , equivalent to 20, 50, and 100 passages in the liquid-jet reactor, samples were taken, and the hydrodynamic particle size distribution was determined using the evaluation by analytical centrifugation described previously.The fraction of particles <1000 nm was plotted against the pas-sage and pulse numbers for the batch and liquid-jet processes, respectively (Figure 3C).The graph shows that with an increasing number of passages in the liquid-jet setup, the fraction of particles <1000 nm significantly increases.The size reduction effect, which occurs after 20 passages, is not particularly strong, with ≈10% of the particles <1000 nm, whereas after 50 passages 88% of the particles are already below a diameter of 1000 nm.After 100 passages, 100% of the particles were below the limit of 1000 nm.
The fragmentation effect resulting from the batch setup can only be shown when the graph is enlarged.Only for a fragmentation time of 1665 s, which corresponds to 1.67*10 8 pulses, a weak comminution effect was observable, and therefore the portion of SMP was less than 1%.In comparison to previous works, which also used a batch process and reported successful particle comminution, significantly lower fragmentation durations or larger sample volumes were chosen in this work to guarantee a fair comparison of liquid-jet and batch setup.Thus, it can be clearly demonstrated that by using the passage reactor setup and the associated better illumination of the thin liquid-jet, particle size reduction occurs at lower pulse counts and consequently with a strongly elevated efficiency.It should be noted that the presence of smaller particles with diameters <<100 nm cannot be fully excluded, as they could not be detected by analytical centrifugation at the used sedimentation protocol.SEM analysis, on the other hand, also showed no pronounced fractions of smaller particles, which seems to indicate that this size fraction is either not formed at all or at a low yield (or coalesce quickly to larger particles).
This laser-comminution success in the liquid-jet setup was tested on a wide range of model drugs and the influence of drug concentration was analyzed.Figure 4A-D shows that the same comminution trend is evident for all four model drugs.With increasing drug concentrations from 0.1 wt.% to 2 wt.%, the com-minution success decreases due to attenuation of laser energy by scattering of the particles.Ding et al. also reported attenuation of laser energy at higher drug concentrations, resulting in a lower nanoparticle production yield. [37]hereas ketoconazole (Figure 4B) and prednisolone (Figure 4C) still exhibit very good particle fragmentation yields of ≈80% at a concentration of 0.5 wt.%, megestrol acetate (Figure 4D) only achieves values of 66% and naproxen (Figure 4A) of 20%.A further increase to 2 wt.% leads to a reduction in the laser-comminution efficiency for all four model substances, a trend well documented in the literature and associated with energy loss due to intensity attenuation along the beam path. [27,37]However, the concentration dependency of the comminution efficiency seems to be drug-specific, a phenomenon not associated with attenuation effects alone, as attenuation of the laser beam (power loss) measured in the liquid-jet reactor for different drug concentrations was similar (<10% difference) for the model drugs prednisolone and naproxen (Figure S2, Supporting Information).Figure 4 also shows that a high number of passages leads to better particle size reduction for all four drugs analyzed.
Figure 5 shows the characterization of the chemical composition of the drugs analyzed via HPLC (A) and ATR-FTIR spectroscopy (B).HPLC measurements using naproxen as an example show that, as described in previous works, there is an inversely proportional relationship between irradiation time (number of pulses/passages) and degradation. [27,28]If the number of fragmented passages increases, the amounts of degradation products also increase, but the total level of degradation is lower than reported in previous works.Despite performing 100 laser passages and thus fragmenting all naproxen particles to the <1000 nm size threshold, less than 1 wt.% of degradation products were detected.LFL batch procedures, also using naproxen and yielding similar particle sizes, were associated with portions of degradation products ranging from 2.3% to 10% when using IR-fs lasers. [37]When the number of passages was reduced to 20, 99.9% of the initial naproxen drug was recovered.The same observations could be made from the FTIR spectra (Figure 5B).The spectra of the untreated and laser-generated naproxen samples show a consistent peak pattern and shape.No changes in the fingerprint region (400-1800 cm −1 ) could be detected.Compared to FTIR analyses of previous works, no oxidation could be detected, [28,37] nevertheless, at a wave number of 3500-3600 cm −1 new broad naproxen-untypical peaks resulting from a stretching vibration of the OH bands, present in both laser-fragmented and untreated samples, could be detected, however, are atypical of dry powder samples and are therefore caused by the fact that all drug were handled and treated as aqueous suspensions.In the case of sensitive drugs, hydration occurs, however, not as a result of fragmentation. [27,40]This remarkably low amount of detected degradation products is also in accordance with the potential fragmentation mechanism of I) photomechanical fragmentation under stress confinement conditions, II) the formation of a stress gradient by inhomogeneous heating in MPs and III) evaporation and consecutive expansion of water pockets in polycrystalline drug particles.These mechanisms are summarized graphically in Figure 9 and discussed in more detail in the following chapter.Due to the dominance of mechanisms based on mechanical stress when ultrashort ps-lasers are used, the degradation temperature of the drugs is not reached, and the formation of degradation products is mostly avoided.The key, however, is that degradation is inversely proportional to the number of pulses, and in the liquid-jet passage reactor the number of incident laser pulses, required to achieve a definite particle size is lower than in the batch setup, which results in reduced degradation at high fragmentation yields.
Thus, these findings clearly demonstrate the suitability of LFL of drugs in the passage reactor as a novel, wear-free comminution method.When all MPs were completely transformed into SMPs via LFL, less than 1% of the drug was degraded, which is significantly less than established methods as well as previous laser processes.Conventional milling methods produce 6 wt.% to 8 wt.% of degradation products at similar final particle sizes. [18,20]o evaluate changes in the physical parameters of the drugs after LFL, solubility assays were developed to demonstrate a correlation between size reduction and dissolution properties.Naproxen suspensions with a concentration of ≈90 mg L −1 were covered by dialysis membranes.The membrane and surrounding medium contained 0.01 M PBS buffer, pH 7.4 that was tempered to 37 °C.The concentration of dissolved drug outside the membrane was analyzed over time by UV-Vis extinction spectroscopy.In each case, the untreated drug particles were compared with the laser-generated drug particles produced after LFL with 100 passages (Figure 7A).
Already at the beginning as well as over the entire time period of the dissolution tests, the laser-treated samples show higher concentrations of dissolved naproxen.The concentration curves of both samples (untreated and laser-treated) differs minimally; the laser-treated samples show a faster increase in concentration with a steeper slope, whereas a flatter curve is observed for the untreated samples.Both dissolution curves reach a saturation concentration after ≈24 h, which is higher in the laser-treated sample.Neither the untreated LP0 samples nor the laser-generated drug particles dissolved completely.However, in the laser-fragmented LP100 samples, more than 40% of total naproxen dissolved after 32 h, while in the untreated samples only 27% of naproxen mass dissolved in the same period.This is an increase in the concentration of dissolved naproxen by over 50% after 32 h in the laser-treated sample.For both samples, a dissolution equilibrium between dissolved and undissolved drug is established, which is shifted toward the dissolved form for the laser-processed drugs.The dissolution behavior and the associated potential bioavailability/-activity were improved by the laser process.This solubility improvement can be explained by different phenomena.The reduction in particle size increases the specific surface area of the drug, i.e., more surface molecules are available, which means that more active/soluble surface sites are present in the particles, which shifts the equilibrium toward the dissolved drug.As already described by Ding et al., even a size reduction of the naproxen particles to the submicron scale is sufficient to obtain significantly better dissolution properties. [37]evertheless, the improved dissolution behavior of the laserprocessed drugs may also have been induced by a change in the crystal structure of the drug or a chemical modification of the drug itself.Based on Figure 7A, it is not possible to fully differentiate between these phenomena.However, the evaluation of the XRD analyses, which can be seen in Figure 6, shows that the crystalline structure of the laser-generated particles remains unchanged compared to the untreated samples.The crystallite size of the untreated and laser-treated drug particles calculated by the Scherrer equation (example insets Figure 6) are in a uniform nanometer-size range (40-50 nm) for the polycrystalline particles and do not change by the use of lasers.Even though laser-induced subtle changes to the surface chemistry of the drug particles, affecting solubility, cannot be fully ruled out based on the data, an effect derived from the higher volume-specific particle surface area due to size reduction is the most probable effect causing the higher portion of dissolved drug in the laser-irradiated samples.
Figure 6.XRD pattern of untreated (LP0) and laser-generated (LP100) naproxen-particles.3][44] Table 1.Dissolution parameters of the untreated and laser-generated particles.To gain deeper insight into the dissolution mechanism and specifically the dissolution kinetics/dissolution rates, the Weibull fit typical of dissolution and release experiments [45,46] was applied to the experimental data.This general distribution function is a 2-parameter equation that can be adapted to a wide range of different drugs due to a scale () and shape factor (). Experimental data can be reliably described by an exponential curve, which asymptotically approaches a plateau value, the saturation concentration, over the entire time period and, depending on the shape factor, considers the real curve shape.The Weibull fit gives the accumulated amount of drug in solution at time t (m t ) relative to the saturation amount (m ∞ ).By extending the scale ( = k 1 /(1-h)) and shape factors ( = 1-h) by the parameters k 1 and h, further statements can be made about the dissolution kinetics.The h-value indicates the progression of dissolution rates, with values <1 reflecting heterogeneous dissolution conditions and values equal to 1 reflecting homogeneous dissolution conditions.By converting the k 1 parameter to the kinetic constant K 1 of the drug (K 1 = k 1 /(1-h)), direct conclusions can be made about the rate of dissolution.In Table 1, the parameters have been summarized for the untreated and laser-generated samples, respectively.The h-values of both curves are below 1 and thus represent heterogeneous dissolution rates.In Figure 7A,B, this monotonic decrease in dissolution rates can also be confirmed.
Both graphs show an initially high dissolution rate, which asymptotically approaches 0 as the concentration increases and reaches saturation.For the calculation of all parameters, the final value determined in each case after 32 h was assumed to be m ∞ .
Corresponding to the faster dissolution rate of the lasergenerated drugs already discussed in Figure 7A, a higher K 1 value could also be calculated compared to the untreated drugs.Nevertheless, it can be seen that although the K 1 kinetic constant of the untreated drugs has a lower value, it is in the same range as the constant of the laser-treated drugs.Laser fragmentation enables the generation of drug particles on a submicrometer scale, which, in addition to the faster dissolution rate, also leads to a higher saturation concentration (according to Ostwald-Freundlich). Accordingly, a superposition of two solubility-determining effects can be observed.Thus, the closely related rate constants are due to the different final concentrations of the dissolved drug.
From these data, it can be concluded that drug fragmentation leads to an increased dissolution rate, and, in addition, the dissolution equilibrium is shifted to the dissolved form, resulting in a higher saturation concentration.This suggests that the macroscopic dissolution kinetics are determined by the specific surface area.Hence, the diffusion of the drug molecules from the bulk of the particle to the solution seems to represent the ratedetermining step in the tested release setup.

Proposed LFL-Mechanisms for Organic MPs
The exact mechanism of laser fragmentation of organic materials is unclear, therefore an interplay of three different processes may occur, explained and differentiated in the following:  Calculations of stress confinement criteria for sphere-equivalent naproxen particles; A) thermal and acoustic relaxation times as a function of the smallest dimension of heated volume (particle diameter or optical penetration depth dependent on particle size); B) estimated maximum pulse duration as a function of the smallest dimension of heated volume (particle diameter or optical penetration depth dependent on particle size) for different organic materials (with different absorption coefficients (μ a )) required for the stress confinement.In the micrometer size range, the 10 ps pulse duration applied in our experiments will always fulfill the stress confinement condition (maximum pulse duration of 490 ps for sizes of 1 μm or larger).
49] Thermal damage is induced by a temperature increase in the particle leading to surface evaporation at low fluence and phase explosions, accompanied by the formation of vapor bubbles, that cause the particles to burst at higher fluence. [51]In the case of mechanical damage, the photoacoustic shockwaves, generated by thermoelastic pressure, lead to tensile and compressive stress, with the stress being confined in a finite region.In case the amplitude of the shockwave exceeds the tensile strength of the (drug) material mechanical breakage occurs, [47,49,50] as calculated in Figure 8 for our particle and laser parameters with a graphical mechanistic summary in Figure 9.Both damage mecha-nisms may differ dependent on whether the educt MP is homogeneously or inhomogeneously heated, which depends on the optical penetration depth of the specific material, as shown in the sketch in Figure 8A.Homogeneous heating is achieved for particle sizes smaller or equal to the optical penetration depth.However, as soon as the particle size exceeds the optical penetration depth, only a fraction of the particle facing the incident laser beam is heated.
Here, the size and the mechanical and optical properties of the absorbing material, as well as the used laser parameters (fluence and pulse duration), influence the corresponding processes, [48] and the thermal and acoustic relaxation times (t th and t ac ) are the main criteria. [47,49]The acoustic relaxation time corresponds to the time required to dissipate the stresses formed within the particle, while the thermal relaxation time is defined as the time required to achieve a significant reduction (equilibration) of temperature in the region of the particle where the energy of the laser was deposited.If the pulse duration is smaller than these relaxation times, this results in a maximum temperature and pressure increase in the particle.It should be mentioned that photoacoustic relaxation is significantly shorter than thermal relaxation, as also shown in Figure 8A using naproxen particles as an example, so thermal confinement conditions are almost always fulfilled while mechanical stress confinement is more critically affected by the laser pulse duration.Particles can thus suffer thermal and mechanical interactions as soon as the condition t pulse <t ac is fulfilled (stress confinement criterion). [47]However, in case the tensile strength of the particle material is already exceeded by the rapid tensile stress amplitude of the photoacoustic response, a stress-induced fracture can occur even without a significant temperature increase of the whole particle. [47]ince in our experiments, a pulse duration of 10 ps was used, which is below the acoustic relaxation time for the particle range calculated in Figure 8A, a maximum pressure increase occurs within the particles, and the conditions of stress confinement are fulfilled.It should be noted that the smallest dimension of the heated volume was used for all calculations of relaxation times in Figure 8A.This is the particle diameter for particles with dimensions smaller than the optical penetration depth (homogeneous heating) and the optical penetration depth for larger particles (inhomogeneous heating).This is why plateau values of t th and t ac are reached for large particles with diameters above the optical penetration depth.
Furthermore, we calculated the maximum pulse duration under that stress confinement is still possible, a criterion that critically depends on the absorption coefficient (μ a ), reciprocal to the optical penetration depth (1/μ a ) (Figure 8B).Here, we assume a constant speed of sound in the materials, which is a fair assumption for organic particles (e. g. various drugs, polymers, etc.). [52,53]With an decreasing particle size, the applied pulse duration must also be reduced to remain within the stress confinement regime.For the model substance naproxen used in this work (with an absorption coefficient of 950 cm −1 and a sound velocity of 2041 m s −1 ), the pulse duration must be below 5 ns for particles 10-1000 μm in size, whereas a maximum of 490 ps can be used for particles with a particle size of 1 μm.
The resulting maximum temperature increase on the surface of the heated region of a sphere-equivalent naproxen particle due to laser deposition can be estimated to be 116 °C (assuming a homogeneously heated region within a particle and thermal confinement, an absorption coefficient of 950 cm −1 , an effective fluence of 197 mJ cm −2 , a particle density of 1266 kg m −3 , [54] and a specific heat capacity of 1273 J (kgK) −1 [54] (for calculations see Equation S7 and Table S2, Supporting Information), which is slightly below the typical melting temperature of bulk naproxen (152 °C).This clearly indicates that the predominant fragmentation effect cannot be solely based on thermal effects and contributions from photoacoustic waves are probable.For the fragmentation mechanism of MP, this means that photoacoustic effects are definitely relevant and only one laser pulse in the appropriate fluence range is sufficient for particle fragmentation if a high enough tensile stress amplitude can be generated to exceed the tensile strength of the particulate material. [48]Recently, pump-probe studies on single IrO 2 particles confirmed that a single laser pulse is enough to fragment a particle, and highlighted the contribution of stress-mediated processes to the laser fragmentation mechanism. [50]he tensile stress generated within the particles depends, among other things, on their shape and symmetry.In experiments with liquid spheres and gelatin cylinders, Paltauf et al. described that particularly in particles with high symmetry, spherical collapsing waves lead to very high tensile stress amplitudes with a very short tensile peak.For asymmetric orientations with a long orientation in the z-direction, this results in a longer tensile phase duration.Since the integral over time of the pressure curve P(t) must be equal to zero, a longer tensile peak consequently leads to a lower amplitude.The highest tensile stress could be observed in each case in the center of a sphere, as well as on the axis of cylinders, with maximum values occurring in symmetrical particle shapes.It should be noted that high tensile peaks can still occur in asymmetric particles, albeit without simultaneous focusing of the tensile waves propagating from the surface to the center of the particle, due to the difference in the extent of the path length, which may be still sufficient to exceed the tensile strength of the material.Analogous to Paltauf's publication, [47] the laser irradiation in our work is along the short axis (hitting the needle from the side) of symmetry of the needle shape because laminar flow conditions in the liquid jet reactor result in a vertical orientation of the asymmetric particles. [55]As asymmetric particles like needle-shaped prednisolone consequently undergo longer tensile stress phases with lower tensile stress amplitudes not arriving simultaneously at the center of the particle, only partial breakage and crack formation could be observed.The comparatively symmetrical particles in Figure 2 (except C+D), on the other hand, would be exposed to shorter tensile stress phases with higher amplitudes, concentrated at the center of the sphere, which would result in full fragmentation and formation of smaller particles. [56]altauf et al. also reported strong cavitations inside organic liquid and gelatin samples, but they could not detect any disintegration due to their high surface tension.In contrast, drug particles are much more brittle, which facilitates fragmentation by pressure waves. [56]n addition, especially for large particles whose diameter exceeds the optical penetration depth of the material (>1/μ a ≈ 10 μm for naproxen), simple inhomogeneous heating can lead to a quasi-stationary stress gradient which, if the tensile strength of the material is exceeded, also leads to fracture of these large particles, outlined in Figure 9C,D.In the case of needle-shaped particles as in Figure 2C,D, stress-induced fracture is also fundamentally favored, due to the asymmetric shape.Accordingly, based on the SEM images, it is not possible to clearly distinguish whether the fractures and cracks of the drug particles observed in Figure 2D were caused by exceeding the tensile strength due to shock wave generation in the stress confinement regime or by inhomogeneous heating within the particles.
Nevertheless, this mechanism cannot fully explain the fact that almost no degradation products were generated during LFL since thermal and photomechanical effects are partially coupled (even if photomechanical effects probably dominate our experiment).Hence, a third mechanism may contribute to the fragmentation of the drug particles, based on internal pressure-driven fragmentation/ablation, generated by the evaporation and expansion of residual water pockets in polycrystalline educt MP during laser penetration (Figure 9E,F). [57]Tabetah et al. showed the strong influence of residual water during laser irradiation in simulations with lysozyme targets.Even very minor water contents of 5 wt.%-10 wt.% in the material lowered the required ablation threshold fluence by more than a factor of 2. The rapid evaporation of the water pockets and the associated expansion and pressure generation leads to bursting of the surface of the lysozyme target, by disrupting the interconnected viscous network, resulting in the release of lysozyme molecules.Remarkably, this release mechanism resulted in fully intact molecules, with no intramolecular fractures and thus minimal formation of degradation products. [57]Especially for polycrystalline drug particles dispersed in an aqueous environment, the presence of small water pockets in the educt MP crystal structure is likely.In addition, the previously calculated surface temperature of 116 °C of the naproxen particles during fragmentation (using Equation S7) indicates that the expansion and evaporation of present water residues is conceivable, as temperatures above the expansion and evaporation temperature of water are reached.Based on this, contributions from a water pocket expansion mechanism are proba-ble for drug MP fragmentation and could explain the low content of degradation products found in our study.
However, the limited homogeneity of the irradiation conditions in the current experimental setup makes it highly unlikely that all educt particles will be hit with sufficient laser fluence in one passage.For once, inhomogeneity of the irradiation fluence could be caused by the Gaussian beam profile of the laser, which could mean that particles in the center of the beam are fragmented due to the suitable fluence regime, while those along the edge may not receive sufficient energy.Second, extinction of the laser light along the beam path (Lambert-Beer law) may cause fluence gradients and a reduction in fragmentation probability through the irradiated liquid-jet. [38,58]The fact that higher drug concentrations and the correlated elevated turbidity with increasing concentrations (Figure S2, Supporting Information) strongly reduces fragmentation efficiency suggests that shielding effects along the beam path impact the fragmentation process.Third, laminar flow conditions prevail in the nozzle of the passage reactor, which leads to an inhomogeneous distribution of pulses per volume element.On average, 2.0 pulses per unit volume are introduced per passage.Particles located in the liquid volume elements close to the wall of the cylindrical nozzle have a much higher residence time than particles in the center of the flow field.These location-specific residence time distribution consequently lead to a pulse number that deviates from the average, while particles in the center of the flow are hit by fewer pulses (≈1.0 PPV) than those at the edges (≈4.0 PPV).
These previously discussed fluence and pulse number inhomogeneities suggest that an interplay between both effects (number of pulses and laser fluence) is responsible for the observed phenomena since both a certain threshold fluence and, depending on that, a minimum number of pulses are required to induce a successful fragmentation event. [48]Multipulse irradiation causes an accumulation of laser-induced damage leading to material ejection even at lower fluence regimes, with sufficiently high fluences leading to successful fracture with just one pulse. [48]However, differentiation between the two processes is not possible with the cylindrical jet setup used in this study but would require an alternative flat-jet setup, where a superior control of fluence and number of pulses is possible, a design already described for the fragmentation of gold nanoparticles. [58]

Conclusion
To improve the dissolution properties of solid drugs, particle size reduction is used by default to achieve better solubility and thus better oral absorption by increasing the specific surface area.Though it is already known to allow drug particle downsizing by contamination-free LFL, the low efficiency (many pulses needed) of batch setups and on the other hand the chemical degradation of the drug with high pulse numbers made utilization of this technique impractical in real-life applications.To address this problem, we introduce the passage reactor concept as a 100 times more efficient setup for drug particle laser fragmentation, which also reduced the problem of drug degradation to a negligible <1%.These results pave the way toward the applicability of laserbased comminution techniques in pharmaceutical research and industrial formulation development, as the modern passage reactor setup can solve the problems of low productivity as well as impaired product quality by degradation due to a boosted fragmentation efficiency.Furthermore, a functionality assay verified that increased solubility and thereby bioavailability of the drug was achievable with this method, proofing the superior functionality of the laser-processed drug.In addition, we gained mechanistic insight into the LFL process of organic molecules, postulating a simultaneous contribution of multiple fragmentation mechanisms based on the stress confinement criterion, a degradation-free ablation enabled by residual water, as well as the formation of an internal quasi-stationary stress gradient within MP, prone to stimulate further fundamental research in this area.Here, particularly a newly developed flat-jet passage reactor, which was demonstrated to enable a more homogeneous illumination and more precise pulse per volume control in metal [58] or oxide [59] nanoparticle pulsed laser excitation, could be a useful asset to further clarify the fragmentation mechanism.
Preparation of Drug Suspensions: Suspensions of all model drugs with different concentrations (0.1 wt.%, 0.5 wt.%, and 2 wt.%) were produced in a previously prepared aqueous stabilizer solution of 1 wt.% Tween 80 and 0.1 wt.% HEC in deionized water.After drug addition, the suspensions were each pre-treated three times for 5 s with an ultrasonic-sonotrode (Hielscher Ultrasonics GmbH, Teltow, Germany) and used immediately for laser fragmentation.
For the dissolution experiments, 0.01 wt.% SDS was used instead of the previous stabilizers Tween 80 and HEC at a naproxen concentration of 0.1 wt.%.Immediately before laser fragmentation, the suspension was also redispersed for 5 s using an ultrasonic sonotrode.
Pulsed Laser Fragmentation in Liquids: For LFL, the TruMicro 5000 (TRUMPF SE + Co. KG, Ditzingen, Germany) with a wavelength of 515 nm, 30 W average power, a repetition rate of 100 kHz, 10 ps pulse duration, and a pulse energy of 300 μJ was used.The applied fragmentation setup was the free liquid-jet previously established for inorganic materials, [22,23,38,60] as shown in Figure 1A.In this setup, a plano-convex cylindrical lens with a focal length of 100 mm was used, which focuses the laser precisely on the liquid-jet containing the drug particles to obtain improved illumination compared to the batch process and achieve the highest possible incident fluence of ≈428 mJ cm −2 .By opening the valve on the liquid-jet, the particle dispersion passes through the laser beam.This fragmentation process was repeated for 20, 50, and 100 cycles, respectively.
For comparison, additional batch experiments were performed in a beaker which was stirred using a magnetic stirrer and irradiated horizontally.The laser was focused by a lens with a focal length of 100 mm in the middle of the beaker.Identical sample volume, drug concentration, laser parameters as well as the total number of pulses were used in both, the passage reactor setup and the batch setup (Table S1, Supporting Information).
Qualitative Particle Size and Morphology Analysis: In order to qualitatively determine the hydrodynamic particle shape and size, SEM images of all model substances were acquired.For this purpose, the stabilizers had to be washed out of the drug suspension by ultracentrifugation (Beckman Coulter GmbH, Optima MAX-TL, Krefeld, Germany).The supernatant was discarded and refilled with deionized water, the pellet was redispersed, and the procedure was repeated (two cycles at 186 000 x g for 152 min).The washed-out drug suspension was added onto Si sample slides by spin coating, dried, and sputtered (80% Au + 20% Pd).All untreated and laserfragmented samples were recorded using the Apreo S LoVac (Thermo Fisher GmbH, Kandel, Germany).
To determine the crystallite size, XRD measurements of the dried drug powders were performed using Malvern Panalytical Empyrean (Malvern Panalytical GmbH, Kassel, Germany) with Cu K radiation ( = 1.5405980Å) and PIXel 3D Detector with 255 channels.XRD patterns were recorded with a step size of 0.04°2  with 240 s/step (step time) between 10°and 95°2 .The subsequent calculation of the crystallite sizes was performed using the Scherrer equation.
Quantitative Particle Size Analysis by Analytical Centrifugation: For quantitative particle size analysis, the dispersion analyzer LUMiSizer 651 (LUM GmbH, Berlin, Germany) was used, which determines the particle size distribution via sedimentation profiles based on ISO 13318-2. [61]ntreated and laser-generated model drug suspensions of all concentrations were externally tempered to 7 °C and redispersed in an ultrasonic bath directly before measurement.Particle size analysis was performed at 7 °C and a velocity ramp from 300 to 4000 x g was used.Sedimentation profiles were detected at a wavelength of 410 nm.For the evaluation, a method was developed to allow the comparability of different model drugs and drug concentrations related to the comminution success, as shown in Figure 3A,B.By integrating the volume-based particle density distribution (q 3 (x)), the AUC for particles <1000 nm (Figure 3B, marked yellow) could be determined and normalized to the total AUC of all particles with a cut-off diameter of 10 000 nm (AUC <1000 nm /AUC total ) (Figure 3A, marked red), to ensure comparability between samples.From this quotient, a quantitative statement can be made about the relative portion of particles comminuted down to the submicron level.Please note that based on the sedimentation protocols used (based on a defined particle mass concentration and sedimentation ramp), the LUMiSizer can quantitatively characterize particle suspensions in the range of 100-10 000 nm, size fractions outside of this range cannot be reliably quantified.
Characterization of Drug Degradation: Drug composition was determined via ATR-FTIR and HPLC analysis.For ATR-FTIR spectroscopy, the untreated and laser-generated drug suspensions were dried, with the stabilizer solution subtracted as background before measurement.For each sample, 10 scans were recorded in the range of 400-4000 cm −1 with a resolution of 4 cm −1 using an ALPHA Platinum FTIR spectrometer with ATR sample module (Bruker Corporation, Billerica, USA).In order to determine the quantitative amounts of initial drug compound and degradation products formed during laser fragmentation, additional HPLC measurements were performed using an LC-system with an UV detector (Knauer Wissenschaftliche Geräte GmbH, Berlin, Germany) at a wavelength of 230 nm.The analysis was based on the method of the European Pharmacopoeia. [62]A 100 mm Nucleosil-C18 column with an inner diameter of 4 mm and particle sizes of 3 μm were used (Macherey-Nagel GmbH & Co. KG, Düren, Germany).The mobile phase is composed of 42 vol.%acetonitrile and 58 vol.% of a 1.36 g L −1 solution of potassium dihydrogen phosphate previously adjusted to pH 2.0 with phosphoric acid.Drug suspensions with a concentration of 0.1 wt.% and acetonitrile were mixed in a ratio of 3:2 to dissolve the drug particles.The dissolved samples were filtered through 0.2 μm syringe filters.The injected volume was 10 μL.The flow rate of the column at 50 °C was adjusted to 1.5 mL min −1 .Before sampling, a calibration line of the initial drug compound was recorded by plotting the AUC of the chromatogram against the respective concentration (Figure S3, Supporting Information).The amount of drug and degradation products of the laser-generated samples were also calculated by determining the AUC and the corresponding concentration of the calibration line.Any additional peaks that appeared in the fragmented samples and not in the untreated control were assumed to be degradation products.
In Vitro Solubility Assays: To verify increased solubility in the laserirradiated sub-micrometer drug particles, in vitro dissolution tests were performed using untreated and laser-treated naproxen suspensions as a model substance (0.1 wt.% naproxen, LP100, corresponding to the best comminution parameters).To exclude interactions and blocking at the membrane of naproxen and the previously used high-viscosity long-chain stabilizers, the drug suspension was fragmented in an aqueous solution containing 0.01 wt.% of the short-chain surfactant sodium dodecyl sulfate (SDS).
For the dissolution experiments, 5 mL of this drug suspension containing 5 mg of the drug and 5 mL of 0.01 M PBS buffer, pH 7.4, was added to a dialysis membrane with a MWCO of 0.5-1 kDa, and the membrane was sealed.The MWCO was chosen to guarantee that only the solubilized drug would pass through the membrane.The sealed membrane was placed in a dialysis vessel containing 45 mL of 0.01 M PBS, pH 7.4, continuously tempered to 37 °C, and mixed using incubation vibratory shakers (VWR International GmbH, Darmstadt, Germany).900 μL of sample was taken at specific time intervals, which was then refilled with the same amount of PBS buffer.
All dissolution experiments were limited to a duration of 32 h.UV-vis extinction spectroscopy (Evolution 201, Thermo Scientific, Waltham, US) was used to analyze the concentration of the dissolved drug.For UV-vis extinction spectroscopy, a calibration line of the initial drug in acetonitrile and 0.01 M PBS buffer, pH 7.4, at a volume ratio of 1:9 was prepared for concentration determination (Figure S6, Supporting Information).To ensure that samples taken from the dialysis vessel and calibration lines were determined under identical conditions, every 900 μL of the samples was also mixed with 100 μL of acetonitrile prior to the measurement.The peak maximum at 330 nm, typical for naproxen, was chosen for quantification of the drug concentration.
Statistical Analysis: The bar charts show the mean values of several independent measurements (n).The error bars indicate the standard deviation.Asterisks indicate the significance of differences between LP0 and LP20, LP20 and LP50, LP50, and LP100, and the difference between LP0 and LP100 determined by one-way analysis of variance (1-way ANOVA) with a significance level  of 0.05.P values ≤0.05 are summarized with one asterisk, P values ≤0.01 are summarized with two asterisks, and P values ≤0.001 are given with three asterisks.Values greater than 0.05 were marked as not significant ("ns" in the graphs).

Figure 1 .
Figure 1.Schematic illustration of the LFL process; A) setup of liquid-jet passage reactor with educt drug MP and fragmented SMP); B) overview of the relevant readouts and used analytical methods.Analytical centrifuge and SEM were used to determine particle size distributions, HPLC, ATR-FTIR and XRD determined purity.Solubility assays provided proof of function.

Figure 3 .
Figure 3. Particle size distribution of untreated (LP0) and laser-generated (LP20, LP50, LP100) model substance; density distribution q 3 (x) of 0.1 wt.% naproxen determined by analytical centrifugation at different passage numbers, A) (total) area under the curve (AUC) with a cut-off at 10 000 nm exemplarily marked in red for the LP50 case; B) AUC for particles <1000 nm exemplarily marked in yellow for the LP50 sample; C) comparison of the efficiency of passage reactor and batch-setup at drug loadings of 0.1 wt.% naproxen for different laser passages.The bar charts show the mean values of three independent measurements (n = 3).The error bars indicate the standard deviation.Asterisks indicate the significance of differences between LP0 and LP20, LP20 and LP50, LP50 and LP100, and the difference between LP0 and LP100 determined by one-way analysis of variance (1-way ANOVA) with a significance level  of 0.05.

Figure 4 .
Figure 4. Laser fragmentation yield: Influence of drug concentration on LFL comminution success for A) naproxen, B) ketoconazole, C) prednisolone, and D) megestrol acetate.The bar charts show the mean values of three independent measurements (n = 3).The error bars indicate the standard deviation.Asterisks indicate significance of differences between LP0 and LP20, LP20 and LP50, LP50 and LP100, and the difference between LP0 and LP100 determined by one-way analysis of variance (1-way ANOVA) with a significance level  of 0.05.

Figure 5 .
Figure 5. Characterization of drug degradation using A) the AUC of HPLC chromatograms and B) ATR-FTIR spectra of the untreated (LP0) and lasergenerated (LP100) water-exposed naproxen particles.

Figure 8 .
Figure 8. Calculations of stress confinement criteria for sphere-equivalent naproxen particles; A) thermal and acoustic relaxation times as a function of the smallest dimension of heated volume (particle diameter or optical penetration depth dependent on particle size); B) estimated maximum pulse duration as a function of the smallest dimension of heated volume (particle diameter or optical penetration depth dependent on particle size) for different organic materials (with different absorption coefficients (μ a )) required for the stress confinement.In the micrometer size range, the 10 ps pulse duration applied in our experiments will always fulfill the stress confinement condition (maximum pulse duration of 490 ps for sizes of 1 μm or larger).

Figure 9 .
Figure 9. Potential mechanisms for LFL of organic MP: A) and B) photomechanical fragmentation mechanism under stress confinement conditions; C) and D) crack formation by thermally induced stress gradient due to inhomogeneous heating, and E) and F) degradation-free ablation enabled by evaporation and expansion of residual water inside the drug particle, leading to degradation-free downsizing.