The Sealed
Machine

The world of his first patent was already rigidly canonical. By the mid-1980s the implantable pulse generator had a fixed anatomy: a titanium can laser-welded from stamped halves; a feedthrough crossing the hermetic wall through a glass-to-metal seal; and inside, the lithium-iodine battery that had stretched device life from two years to ten. Above that hardware raged the era’s glamorous contest — rate response, the race to let a pacemaker sense exertion — and nearly everyone wanted in.

Ruben entered beneath that race, not in it. His first filing, in November 1993, could not be more foundational: the pacemaker’s body itself. Where the standard “clam shell” demanded weld-ring protection and separate plastic cups to position every component, US5535097A reshapes the enclosure so the parts locate themselves — a stepped interior seat for battery and circuit, wire bonds borrowed from the chip industry, one lid closed once. Fewer pieces, fewer welds, fewer ways to fail. A hundred later inventions would cite it.

The same instinct runs through the rest of the decade. When Medtronic’s own devices moved from piezo crystals listening for footfall to true accelerometers, Ruben didn’t bet on which sensor would win the clinical argument — he made the winning answer buildable. US5885471A is the beam-stop that kept a fragile piezoelectric cantilever from snapping when the device was dropped; with 203 citations it is the most-cited work in the corpus. US6216537B1 stood the sensing axis upright in a surface-mount package so the part could ride a standard assembly line.

Unglamorous work, on its face. But it is the kind everything else stands on — and it set the pattern of the whole career: while others argued about how the machine should think, he made sure it could be built, sealed, and trusted for a decade in warm salt water.

Around 2010 the comfortable shrink-schedule broke against a wall made of volume. A pacemaker’s logic follows Moore’s law; its battery and hermetic package do not. Energy is set by chemistry and geometry, so shrinking the device tenfold makes the can wall, the weld seam, and the feedthrough header a punishing fraction of a one-cubic-centimeter implant. Medtronic’s leadless Micra — a 0.8 cc, two-gram pill dropped straight into the right ventricle — reached a first human in December 2013. Someone had to make a package that small even thinkable.

The technique that could was wafer-level packaging — sealing thousands of devices in parallel on the wafer, imported from consumer electronics. It came with a fatal catch: its classic bonding steps run at 400–1100 °C, and would cook any battery sealed inside. This is the seam Ruben’s mid-career work runs straight down. US8666505B2 — priority October 2010, three years before Micra’s first implant — folds control module and energy source into one wafer-scale cavity sealed at no more than 200 °C, explicitly naming leadless form factors. Its continuation, US9318400B2, carries the corpus’s highest measured influence: 151 citing patent families.

And the laser weld — heat put exactly where the joint is and nowhere else — dissolves the cook-the-battery dilemma entirely. The priority dates show him writing the transition down while the race was still forming.

Every problem Ruben attacked had a dominant incumbent solution — and he repeatedly declined the simplest robust one for the harder option demanding more sealing, more power, more control. The pattern is the signature. Feedthroughs belonged to co-fired ceramic and glass-to-metal seals, proven across tens of millions of devices — but both apply heat, late, to a part that must therefore be empty. His laser-bonded feedthrough line trades batch-oven throughput for the ability to seal a package already full of electronics and a live battery.

In pressure sensing, CardioMEMS chose a fully passive resonator — no battery, no wire crossing the body — and won a large clinic. Ruben’s US8424388B2 keeps an amplifier awake inside, buying precision at the exact cost the passive design was built to avoid. In rate response, his accelerometer bet durability over physiologic fidelity — a trade the field ultimately hedged with dual-sensor blends. And against the fixed and magnetically programmable valves that own the hydrocephalus market, US11701503B2 pushes toward a MEMS closed-loop actuator — re-importing every electronics-in-fluid problem the rest of his corpus exists to solve.

The choices were not stubbornness; the rulebook rewards them. Where “hermetic” is a measured number and every change to a Class III seal is a change to the safety case, a laser weld you can cross-section and leak-test beats a cheaper adhesive with a vaguer aging story. The regulations select for techniques that can be characterized to death — precisely the clean, individually testable joins he builds. When the standard leak test physically broke down at small scale, he answered with US7902851B2: test structures built into the bond. The standard couldn’t follow implants down in scale, so he built the test into the seal.

Given an easy road and a hard one, he kept choosing the hard one — active over passive, closed-loop over fixed, seal-it-full over seal-it-empty — because the hard road was more capable, and because he had spent his career building the tools that made it survivable. His packaging craft is the down payment that lets him afford the ambitious sensor.

Then the industry’s center of gravity moved again — from stimulating the body to reading it. Continuous glucose monitoring, barely a commercial category in 2010, became med-tech’s growth engine, and a chronic sensor’s accuracy turned out to be set less by the transducer than by the tiny hermetic room built around it. The frontier had wandered onto Ruben’s home ground, and it converged on his materials from three directions at once.

Joining: in 2018 the glass giant Schott acquired Primoceler, pioneer of heat-free laser glass micro-bonding aimed explicitly at implants — the same fire-through-the-transparent-layer physics his corpus had practiced for a decade. Ruben’s own answer is co-assigned with the other glass giant: the Corning–Medtronic kinetically limited nano-scale diffusion bond, US10124559B2 — sapphire laser-bonded to titanium in a joint under a thousand nanometers thick, its named applications reaching vacuum windows, photonics, and spacecraft.

Power: the microwatt-for-decades problem revived Medtronic’s own plutonium-pacemaker heritage by the modern betavoltaic route — and while startups made the 2024 news cycle, US10096393B2 and US11881325B2 had claimed a 3D-collector converter across tritium, nickel-63, and other beta fuels with priority back in 2014. Sensing itself: the pending stacked-analyte line fabricates mutually incompatible electrodes — glucose, ketones, potassium, creatinine — separately and stacks them: a metabolic panel as an implant.

The newest grant closes the loop. US12610487B2, granted April 2026, seals a device on a transparent substrate — for laser bonding through it — to a power-source housing, with a pressure-only interconnect: capsule, seal, optic, interconnect in one geometry, reached by removing the solder joint a chronic fluid-immersed sensor would eventually lose. Reading a body from inside it means weighing a feather in an earthquake; the winner is whoever builds the quietest, smallest, longest-lived sealed room around the sensor. That room is what he had been building all along.

Each study follows one line of work from first filing to latest grant — the original drawing sheets set as plates, real photographs in the dark rooms, the physics drawn as working instruments, and every claim tiered to its source.