Making 3D-Printed Titanium Structures Float: Buoyant Metal–Polymer Open-Cell Hybrid Lattice Metamaterials

Researchers at RMIT University’s Centre for Additive Manufacturing describe the development of a world-first buoyant metal–polymer open-cell lattice metamaterial that combines a laser 3D-printed Ti-6Al-4V framework with polyurethane foam.

Metallic lattice metamaterials offer an attractive combination of low mass and architecturally tuneable mechanical and functional properties. These characteristics facilitate their use in advanced applications across biomedicine, aerospace, robotics and the automotive industry. However, one area has remained largely inaccessible: marine structures. Put simply, lattices do not float.

The challenge cannot be easily solved by reducing the bulk density below that of water. In an open-cell lattice, water enters the externally accessible pores and replaces the air. Consequently, a metallic lattice may have a nominal bulk density well below that of water (1 g/cm3) and still sink. The open volume contributes to the conventional density calculation but provides no sustained displacement once flooded.

Our research began with metallic hollow-strut lattices (HSLs), which we had previously developed to improve the load-bearing efficiency of metallic lattices. By positioning material farther from the neutral axis, hollow struts can increase their second moment of area and resist bending more efficiently than solid struts of equivalent mass. Importantly for this project, they also contain an interconnected network of internal channels.

We realised that these channels offered a way to integrate a second material without filling the large spaces between the struts. This distinction is important. In conventional polymer-infilled hybrid lattices, the polymer typically fills the external pores of the metallic lattice and converts an open-cell architecture into a closed-cell structure, sacrificing permeability and accessible surface area. Instead, we injected polyurethane (PU) only inside the hollow struts and nodes of the titanium alloy (Ti-6Al-4V) lattices. Water could still flow through the lattice, while the material sealed inside the struts and nodes could provide buoyancy.

The result is a Ti-6Al-4V+PU hybrid lattice that retains its external open-cell architecture while floating in water, making it the first reported metallic lattice structure to do so.

Designing Around Skeletal Density

Developing the architecture required us to reconsider how density should be defined for an open structure interacting with a liquid. Conventional bulk density divides total mass by the total external volume, including all open and closed voids. This is useful for many mechanical comparisons, but it can be misleading when predicting flotation because externally accessible pores become occupied by the surrounding liquid.

We therefore introduced skeletal density, an environment-independent design parameter that excludes externally accessible porosity. For the hybrid lattice, the calculation includes the mass and volume of the titanium strut walls and the foam contained within the sealed channels but excludes the open spaces through which water can flow.

The design rule is straightforward: the skeletal density of the fluid-excluding framework must be lower than the density of the surrounding liquid.

We tested this framework using Ti-6Al-4V simple-cubic HSLs comprising 4 × 4 × 4 unit cells. The HSL cell size and wall thickness were held constant at 10 mm and 0.2 mm, respectively, while the internal strut diameter was varied from 2.5 to 4.0 mm. This produced CAD-derived hybrid skeletal densities ranging from 0.75 to 1.17 g/cm3, spanning both sides of the freshwater buoyancy threshold.

The experimental skeletal densities were approximately 7.4% higher than predicted, ranging from 0.80 to 1.27 g/cm3. This difference was primarily attributed to partially fused powder and the associated increase in effective wall thickness. The printed titanium lattices were, on average, 11.5% heavier than their CAD models, although their external volumes differed by less than 1%. This illustrates why process-induced mass deviations must be incorporated into buoyancy margins.

When immersed in freshwater with a measured density of 0.997 g/cm3, every unfilled titanium HSL sank. Water entered its interconnected internal channels and displaced the trapped air. By contrast, foam-filled lattices with experimental skeletal densities below that of the liquid floated, while those with densities above it sank. The outcome aligned directly with the skeletal-density criterion.

Manufacturing The Hybrid

The Ti-6Al-4V frameworks were fabricated by laser-based powder bed fusion (PBF-LB), a 3D-printing process that uses a focused laser beam with an 80 µm spot size to selectively melt and fuse 30 µm layers of metal powder into solid objects. Producing 0.2 mm walls around interconnected internal channels is demanding because powder adhesion, dross formation and dimensional deviations become significant relative to the intended feature size.

Powder removal was particularly important. Residual powder adds non-functional mass, reduces the buoyancy margin and can obstruct polymer infiltration. In a separate sensitivity study, we printed 100 mm long hollow Ti-6Al-4V struts with internal diameters between 0.5 and 2.0 mm. The PU foam could not reliably infiltrate the 0.5 mm channel, whereas internal diameters of 0.7 mm and above allowed successful powder removal and complete foam infiltration under our processing conditions.

After powder evacuation, a commercially available expandable polyurethane precursor was injected into the connected channels. Its low viscosity enabled it to move through the internal network before expansion, while its final density of approximately 0.08–0.11 g/cm3 added little mass.

Cross-sectional microscopy showed continuous contact between the PU foam and the internal surfaces of the as-built Ti-6Al-4V HSL, with no visible interfacial cavities or delamination. The foam developed a heterogeneous closed-cell morphology, with most pores ranging from approximately 10 to 200 μm. This closed-cell internal structure is central to the material’s damage-tolerant buoyancy.

Mechanical Performance Without A Large Mass Penalty

The Ti-6Al-4V framework remained the primary load-bearing component. Introducing the foam increased bulk density by only 5.9–7.3%, while producing relatively modest changes in yield strength, peak strength and elastic modulus during compression testing. Across the hybrid designs, the average yield strength was 10.8 ± 0.3 MPa, the peak strength was 12.4 ± 0.3 MPa, and the elastic modulus was 567.9 ± 10.6 MPa.

The lowest-density hybrid had a bulk density of approximately 0.27 g/cm3 and a yield strength of 10.3 MPa. At the same density, the estimated density-scaled yield strengths of high-density polyethylene (HDPE) and 316L stainless steel (SS316L) were approximately 5.5 and 6.9 MPa, respectively. The hybrid therefore achieved almost twice the estimated density-scaled yield strength of HDPE and approximately 50% greater density-scaled yield strength than SS316L.

This performance reflects a useful decoupling of functions: the titanium architecture carries structural loads, while the polyurethane facilitates damage-resilient buoyancy by limiting flooding. The infill can therefore add functionality without replacing the lattice topology or substantially changing its governing deformation mechanism.

Experimental compression testing and finite-element simulations showed that failure initiated around the foam-entry openings in the hollow nodes. These geometrically compliant and comparatively porous regions developed high von Mises stresses, followed by node ovalisation, cracking and propagation along an approximately 45° fracture plane. The agreement between simulations and experiments provides a basis for future node optimisation.

Buoyancy After Structural Failure

The most striking result was the material’s ability to remain afloat after severe mechanical damage. We interrupted compression tests at several stages: after the specimens reached peak compressive strength; at the onset of nodal fracture; following the formation of a continuous fracture across an entire layer; and near densification. At each point, we returned the specimens to water.

The lattices continued to float even after an entire structural layer had fractured. A conventional sealed hollow structure could flood rapidly after a crack creates an opening for water ingress. In our hybrid lattice, the polyurethane acts as a distributed buoyancy medium: damage at one location does not expose the entire internal volume to flooding because the foam contains many small, closed cells.

The specimens sank only near densification, after undergoing approximately 33–37% compressive strain. At this stage, structural collapse reduced the displaced volume sufficiently to raise the corrected skeletal density to approximately 1.15–1.25 g/cm3. The dominant cause of buoyancy loss was therefore compaction rather than catastrophic water infiltration.

Longer-term testing also produced encouraging results. Samples remained afloat in freshwater for more than two months, and two samples have remained afloat on my desk for nearly six months. During a separate two-week exposure to natural seawater from Port Phillip Bay, floating samples lost only 0.15 ± 0.03% of their mass and exhibited reductions of less than 1% in yield and peak compressive strength.

Finally, we fabricated a marine-buoy demonstrator measuring 100 mm in height and 85 mm in width. Under simulated turbulent-flow conditions using natural seawater, it maintained stable flotation while rotating by up to 45°, without an external sealed casing or auxiliary flotation device.

The underlying skeletal-density framework is scale-independent, but its implementation is not. Larger structures will require reliable powder evacuation, uniform foam distribution and quality assurance across increasingly complex channels. Multiple injection ports, vacuum-assisted infiltration, industrial computed tomography (CT) and large-format PBF-LB will be important development pathways.

More broadly, this work demonstrates that the hollow interior of a metallic lattice can become a functional design space. Polyurethane provides damage-resilient buoyancy, but future infills could introduce thermal regulation, damping, sensing or electromagnetic functionality while preserving an externally open architecture.

This study, “Breaking the Surface: Buoyant Metal–Polymer Open-Cell Hybrid Lattice Metamaterials,” by Dr Jordan Noronha, Joey Tallon, Raad Omar, Jason Dash, Prof Andrey Molotnikov, Prof Martin Leary, Prof Milan Brandt and Prof Ma Qian, is published in Advanced Materials: https://doi.org/10.1002/adma.74641