Direct sound printing

Photo- and thermo-activated reactions are dominant in Additive Manufacturing (AM) processes for polymerization or melting/deposition of polymers. However, ultrasound activated sonochemical reactions present a unique way to generate hotspots in cavitation bubbles with extraordinary high temperature and pressure along with high heating and cooling rates which are out of reach for the current AM technologies. Here, we demonstrate 3D printing of structures using acoustic cavitation produced directly by focused ultrasound which creates sonochemical reactions in highly localized cavitation regions. Complex geometries with zero to varying porosities and 280 μm feature size are printed by our method, Direct Sound Printing (DSP), in a heat curing thermoset, Poly(dimethylsiloxane) that cannot be printed directly so far by any method. Sonochemiluminescnce, high speed imaging and process characterization experiments of DSP and potential applications such as remote distance printing are presented. Our method establishes an alternative route in AM using ultrasound as the energy source.


The origin of DSP in SCL experiments
The SCL experimental setups are shown schematically in Supplementary Figs. 1a, d and g. In all setups, SCL patterns, captured using a digital single-lens reflex (DSLR) camera, constitute general five regions, I to V. Region I is the most focused reactive location which resembles the laser beam spot in SLA. Ultrasound is transmitted directly to the luminol solution in Supplementary Fig. 2a. We captured the SCL patterns for varying electrical power ( Supplementary Fig. 2) and duty cycle, DC ( Supplementary Fig. 3). Electrical pulse period is set to 2.5 ms in the present paper therefore DC 100% and 30% means 2.5 ms and 0.75 ms active pulse duration, respectively. Supplementary Figs. 2b and c show axial SCL distribution spectrum (ASDS) diagrams, which are the normalized color intensity (CI) of the captured pictures along z-axis, for varying power and DC, respectively, corresponding to Supplementary Figs. 2 and 3. Supplementary Figs. 1b and c show that in wide range of powers and DCs, region I remains chemically active as CI level indicates. If we could harness the reactivity of this region to drive radical polymerization via sonochemical route, this region would solidify the liquid printing material locally. However, the solidified material would move due to acoustic streaming along streamlines and consequently nothing would remain at the desired printing locations (pixels of the desired part). If the solidified material is made to stick to a platform, then this problem would be resolved as the streaming forces will be less than adhesive or bonding strength. We conducted tests with the presence of a platform with varying height, h,  Supplementary Figs. 1e and f, respectively. Reducing the platform height disturbs the SCL patterns, however, region I remains present and active until the platform height is less than ~ 50 mm (the geometric focal is located at 54.5 mm with respect to the transducer face). The luminol solution is contained in a chamber, as shown in Supplementary Fig. 1g, to investigate the effect of ultrasound transmission from degassed deionized water through a barrier (the chamber shell) on the ASDS diagrams for varying h. Figs. 2h and i show that despite compaction of the five SCL active regions in the chamber, region I still could be identified. Supplementary Figs. 6 and 7 show the SCL patterns for this setup. The idea of DSP method emerges from the observation of these tests and taking advantage of presence of region I which is localized and chemically reactive as well as useful for printing. If we could harness the reactivity the region I on the platform (the platform is kept at region I) in the chamber filled with the resin, a 3D object could be created pixel by pixel by moving the transducer or the platform along a designated path.

Tuning microstructure between transparent and porous structure in DSP
The viscosities of the base and the curing agent are 5100 cP 1 and 110 cP 2 , respectively. The viscosity of the mixture of 10:1 (10 base and 1 curing agent) is 3500 cP 1 . The curing agent has lower viscosity and adding more curing agent, assuming keeping the base weight constant, leads to less viscous mixture and vise versa. For example, 13:1 mixture is more viscous than 10:1 mixture. In DSP, mixing ratio 13:1 for Sylgard 184 works as a borderline ratio, higher than 13:1 (14:1, 15:1 and so on) results in transparent structure while ratios lower than 13:1 (12:1, 11:1 and so on) lead to porous structures. The reason behind this phenomenon can be explained by the bubble rupture at the collapse phase. The chemically active bubbles undergo collapse phase (the chemically active phase of bubble oscillation in the acoustic field when temperature and pressure inside bubble reach their maximum). The cavitation bubbles which undergo collapse are also called inertial cavitation (IC) bubbles. The word "inertial" refers to the fact that the inertia of the surrounding liquid flows toward a bubble during the bubble collapse 3 . The IC bubbles upon rupture due to collapse create chemically inactive bubbles called "daughter bubbles" 4 which are then dissolved in the medium. In DSP, created IC bubbles undergo collapse. During collapse, two mechanisms occur simultaneously: curing the surrounding medium of the bubbles with extraordinary fast rate and the birth of daughter bubbles due to the IC bubble collapses. If the viscosity of the fluid were lower, the creation of the daughter bubbles would get easier. Therefore, in maxing ratio 10:1 in comparison with 13:1, more daughter bubbles are created and since the surrounding medium is being cured fast, some daughter bubbles can not be dissolved to the medium and stay in the cured medium and creates the pores in the structures. In higher ratios (such as 14:1, 15:1 and so on), the creation of daughter bubbles are prevented and the bubbles after collapse are disappeared in the medium due to the inertial flow of the medium toward inside of the collapsing bubbles.

Linear acoustics in high speed imaging tests
In the high speed imaging experiments, the ultrasound reflection from the platform is taking into account using linear acoustics. Assuming infinitesimal variation of density during the isentropic propagation of the focused ultrasound waves, the linear acoustic theory can be used. The Acoustics Module of COMSOL Multiphysics 5.4 software was used to investigate the linear behavior of the ultrasound wave propagation inside the build chamber and the locations of focal regions considering the reflection. The mass and momentum conservation equations and energy equation are the employed governing equations for driving the linear wave equation in viscous medium as where ω is the angular frequency (rad/s), p is the spatial pressure (Pa), cc is the complex-valued speed of sound (m/s) expressed as cc=ω/k where k = ω/c-iα is the complex wave number in which c and α are the speed of sound and plane wave attenuation function respectively. The plane wave attenuation function, α (Np/m), in power law form is used as α= α0(f/f0) n where f, α0 and n are transducer driving frequency (MHz) and the attenuation of material at f0= 1 MHz, and a constant [ ] 0,2 n∈ . ρc represents the complex-valued density defined as ρc=ρ(c/cc) 2 where the ρ (kg/m 3 ) is the fluid density.

Nonlinear acoustics for bubble dynamics
Time dependent acoustic pressure at UAMR locations are predicted using non-linear acoustics for the later use in bubble dynamics. HITU_Simulator v2.0 5 is used to predict the propagated wave pressure inside the build chamber at UAMR. The software implements the wide-angle parabolic approximation of the Westervelt equation (wide-angle Khokhlov-Zabolotkaya-Kuznetsov, WAKZK) to calculate the pressure. This approximation results in one-way equation that takes into account spatial distribution of pressure of each harmonic, beam diffraction, interference effects and power-law frequency-dependence as 6 where p is the time-dependent pressure field in (Pa) and p(ω) is its frequency-domain representation, c sound speed (m/s), t time (s), β dimensionless nonlinear parameter, ρ density (kg/m 3 ), ω angular frequency (rad/s), 2 ∇ is the Laplacian in cylindrical coordinate (cm -2 ), α(ω) is the attenuation function (cm -1 ) which follows the power law form presented earlier, and is the convolution of α and p . WAKZK neglects reflection and scattering which are crucial for predicting the locations of high pressure regions which create UAMRs. Therefore in our high speed imaging tests, linear acoustic is used instead. Material related parameters for the multilayer medium domain for linear wave and WAKZK are as follows for water c=1486.6 m/s, ρ= 998 kg/m 3 , α0= 0.217 dB/m, n =2, β= 3.5; for PDMS c=1020 m/s, ρ= 965 kg/m 3 , α0= 147.66 dB/m, n =2, β= 4.5 and for the polystyrene platform c=2400 m/s, ρ= 1060 kg/m 3 , α0= 7.49 dB/m, n =2, β= 1. The thickness of the polystyrene barrier and platform of the build chamber is 1mm.

Bubble dynamics at UAMR
Since the printing material is the mixture of the monomer and the curing agent, the modified Keller-Miksis model for mixtures 7 where and R is the radius of the bubble, R0 is the initial bubble radius, γ is the adiabatic index, Ps is the static pressure and Pv is the vapor pressure (can be neglected 8 ). ce, ρe, μe and σe are the equivalent sound speed, density dynamic viscosity, and surface tension of the mixture, respectively. P(t) in Eq. 4 is the external pressure imposed by focused ultrasound and calculated via WAKZK (Eq. 3). In Eq. 3, the effects of bubble-bubble/particle interaction and gravity are neglected and it is assumed that the bubbles size are much larger than the size of particles suspended in the liquid. The input parameters for PDMS (10:1 mixing ratio) are ce =1020 m/s, ρe = 965 kg/m 3 , μe= 3.5 Pa.s. Bubble dynamics of different printing conditions are shown in Supplementary Fig. 13d and h and Supplementary Fig. 14i. Due to high viscosity of PDMS, R/R0 is small (<1.045) and the concentration of the bubble size distribution is around R0. The analyses of the bubble size distribution confirm this finding as the porosity size distribution are found out to be localized between 1μm to 5 μm.

Effects of environmental conditions, pressure and temperature, on DSP
Environmental conditions such as hydrostatic pressure or temperature could affect the printing resolution as well because the cavitation can be affected by these conditions. We investigated the effect on temperature by cooling the printing material ( Supplementary Fig. 11) and we investigated the size of the printed spot on the platform. In the ambient temperature, 80 W power and 2.15 MHz leads to 1.3 mm spot diameter. However, by decreasing the temperature to -5 o C at the printing spot using thermoelectric coolers, the size of the spot reduced to 0.9 mm in diameter. In another setup, we investigated the effect of the static pressure on the spot size by applying external pressure to the printing medium ( Supplementary Fig. 12). The spot size is reduced by increasing the static pressure using the hydraulic syringe. The gauge static pressure of 0, 0.9 kPa and 1.8 kPa resulted in 1.3 mm, 0.8 mm and 0.5 mm spot sizes, respectively. Although, as seen in these experiments, increasing the internal pressure and decreasing the temperature of the printing material could reach to a finer resolution, manipulating ultrasound frequency and power seem to be an easier way of tuning the resolution practically.

Synthesis and patterning application of DSP
In another application, DSP can be used for selective integration of functionalities such as electrical, optical, etc. to desired objects. Here we demonstrate simultaneous synthesis and patterning of nano particles due to localized chemical activity of UAMR leading localized and selective patterning and making gold nano particle-PDMS composite. In this application, focused ultrasound waves induce almost instantaneous reduction of Gold (III) chloride trihydrate (HAuCl4.3H2O) with reducing agent present in PDMS to produce nano particles embedded in PDMS compared to the conventional heat assisted chemical systhesis process that takes hours. In addition, the instantaneous and localized ultrasound induced chemical synthesis of gold nano particles is used for fabricating a Localized Surface Plasmonic Resonance (LSPR) integrated mircrofluidic bio-sensing chip where DSP synthesizes and patterns gold nano islands (AuNIs) on a PDMS substrate.
Plasmonic property of AuNIs is used to detect extracellular vesicles or exosomes 9 . Exosomes or extracellular vesicles are nano sized particles containing mRNAs, microRNAs and lipids from their origin cells and have a critical role in cell to cell communication. A common method for exosome isolation and quantification is ultracentrifugation which is not practical for clinical settings 10 . In an alternative method 10 , a specially synthesized polypetide (Vn96) is used to capture and quantify exosomes as shows in Supplementary Fig. 19a. A protocol (see Methods) to attached Vn96s to a PDMS substrate is illustrated in Supplementary Fig. 18. AuNIs are conventionally created on a glass or PDMS substrate by thermal convection and annealing process 9 . Here, in case of DSP, the desired pattern of AuNIs compatible with the designed micro chip is printed on the PDMS substrate directly as shown in Supplementary Fig. 19b where gold ions are reduced by polymer's cross-linking agent at UAMR instantly (Supplementary Movie 11). Supplementary Figs. 19c-e show a spiral, a maple leaf and a filled gear pattern of AuNIs which demonstrate the flexibility of DPS in printing customized patterns for bio-sensing purposes ( Supplementary Movies 12 and 13). The designed microchip with microfluidic channels (500 μm by 150 μm) and chambers (Dia. 5 mm) is illustrated in Supplementary Fig. 19f where the localized surface plasmon resonance (LSPR) band of AuNIs is measured by a spectrophotometer. Supplementary Fig. 19g shows the fabricated micro chip while exosomes are passed through the chip. A plasmonic wavelength shift (6 nm) towards longer wavelength, due to the interaction between the immobilized Vn96 on gold nano islands and exosomes from breast cancer cells MCF7, is detected as shown in Supplementary Fig. 19h. The patterned gold nano islands on the PDMS substrate create a surface and sub-surface gold nanoparticle-polymer composite locally. The diffusion depth, dp, of the nanoparticles are about a few micrometer by conventional heat and annealing methods 11 . However, X-ray photoelectron spectroscopy (XPS) measurement of the printed patterns shows that using DSP, gold nano particles could diffuse a few hundreds of micrometers from the surface of the polymer as shown Supplementary Fig. 19i.