Three Paradigms, One Device:
The Intellectual History of the THATTE Architecture

Why the History Matters

It is tempting to present a finished design as though it were inevitable — as though the right answer was always obvious, and the only task was to execute it. The reality of research is messier and, I think, more instructive. The THATTE device reached its final form through three distinct paradigms, each built on a different set of assumptions about what kind of device it should be. Each paradigm was wrong in a specific, identifiable way. And each failure pointed precisely at what the next attempt needed to fix.

Understanding the wrong paradigms is not just intellectual history. It explains why the final design is the way it is. The minimal two-tube structure is not elegant because elegance was the goal. It is elegant because every layer that was stripped away had a specific reason for being stripped. The device is simple because complexity was tried first and found unnecessary.

Paradigm I: The Matryoshka (March 2026)

The first device design had four concentric layers, arranged like a Russian nesting doll. From the inside out:

• Inner channel: A semiconducting (10,0) zigzag single-wall carbon nanotube
• Gate dielectric: A hexagonal boron nitride (h-BN) sheath wrapping the SWCNT
• Gate electrode: A graphene nanoribbon (GNR) wrapped around the h-BN
• Outer shield: A multi-wall carbon nanotube enclosing everything

This structure was named the Matryoshka — after the nesting dolls — and the provisional applications filed under this paradigm used codenames from the Dashavatara: MATSYA for the device structure, KURMA for the switching method, VARAHA for the fabrication process.

The logic behind it was straightforward. A transistor needs a channel, a gate dielectric, and a gate electrode. We knew we wanted carbon nanotubes. So the natural approach was to build a nanotube transistor with all three components: a semiconducting CNT channel, an h-BN dielectric (itself a layered material that can be deposited atomically), and a conducting GNR gate. The MWCNT outer shell was added to provide electrostatic shielding — protecting the device from ambient charge fluctuations that would otherwise dominate the noise floor at nanometre scales.

The choice of a (10,0) zigzag SWCNT was deliberate. Zigzag nanotubes with this chirality are semiconducting, with a bandgap that makes them suitable as FET channels. Apply a gate voltage, deplete the channel, and you have a binary switch. In principle, three carefully controlled gate voltage levels could give you three states — a DC ternary device.

What Was Right About the Matryoshka

The instinct to start from known physics was not wrong. Field-effect transistors are well- understood. Carbon nanotube FETs have been demonstrated in research labs. The h-BN dielectric is one of the best gate dielectrics available at atomic scales — it has negligible interface trap density, excellent breakdown strength, and is compatible with CVD fabrication. The GNR gate is a coherent choice given the all-carbon material system.

The Matryoshka was a thoughtful attempt to translate a proven device architecture into a nanotube material system. If the goal had been a binary nanotube transistor, it might have been a reasonable starting point.

What Was Wrong About the Matryoshka

The goal was not a binary nanotube transistor. The goal was a device that could encode three states with high fidelity. And the Matryoshka inherited the fundamental problem of every DC ternary device: three voltage levels are inherently harder to distinguish than two.

The noise margins in a DC ternary device are roughly half those of an equivalent binary device operating at the same supply voltage. With a (10,0) zigzag channel, threshold voltage variability due to local dielectric inhomogeneities, charge traps, and CNT chirality distribution in CVD-grown samples would erode those margins further. The middle state — the zero trit — would require a gate voltage precisely between the two active states, maintained against all of this noise.

Beyond the noise problem, there was a deeper structural issue: the device was trying to do something the physics did not naturally support. A semiconducting (10,0) zigzag nanotube in a gate-dielectric stack is a silicon MOSFET made of different materials. The whole logic of gate depletion, threshold voltage, and channel resistance is borrowed from CMOS. The nanotube is being forced into a paradigm designed for a completely different material.

The first device was a MOSFET with a nanotube channel.
The last device is a transmission line with a photonic gate.
Between those two sentences lies three months of physics.

Paradigm II: The Shrink-Fit (Late March 2026)

The second paradigm kept the device physics of the Matryoshka but attacked a different problem: fabrication. The four-layer structure had an assembly challenge that had been glossed over in the provisional filings. How, precisely, do you reliably position a semiconducting SWCNT inside an h-BN sheath, inside a GNR gate electrode, inside a MWCNT outer shell — at nanometre tolerances, at scale, without destroying the components?

The answer that emerged was thermal shrink-fit. Multi-wall carbon nanotubes have a known tendency to contract radially when annealed under specific conditions. If the MWCNT outer shell is grown with a slightly oversized inner diameter and the inner assembly is inserted before the anneal, the shrinkage brings the layers into intimate contact. The van der Waals forces between layers do the rest — they lock the assembly once contact is established.

This was a genuine fabrication insight. Thermal shrink-fit is a well-established technique in macroscopic engineering; applying it at the nanotube scale was a non-trivial concept with real patent value. Provisional P3 (VARAHA) was updated to incorporate it.

What the Shrink-Fit Got Right

The shrink-fit approach solved the assembly problem in principle. It gave a plausible route to putting the inner components inside the outer shell without mechanical manipulation of individual nanometre-scale objects. It was also compatible with CVD-based batch fabrication, which matters enormously for any eventual manufacturing process.

More importantly, thinking carefully about fabrication forced a confrontation with the material system. When you design a process flow for a four-layer structure, you have to ask at each step: does this material actually behave the way you need it to? What are the tolerances? What are the failure modes? These are good questions, and asking them revealed something uncomfortable.

The Question the Shrink-Fit Could Not Answer

The shrink-fit paradigm still used a (10,0) zigzag semiconducting channel. It still depended on DC gate control. And when the fabrication analysis became concrete enough to estimate actual performance numbers, the noise margin problem of the Matryoshka was still there.

Worse, a new problem had appeared. The shrink-fit assembly required precise diameter matching between the inner assembly and the outer MWCNT. CNTs grown by CVD do not come with controlled chirality — you get a distribution. Selecting or sorting for exact (10,0) zigzag tubes in bulk is an unsolved manufacturing problem. The device concept was still solid in principle, but the path to manufacturable yield was unclear.

The shrink-fit era had improved the fabrication story without changing the fundamental physics story. The device was still a nanotube FET.

The Break: Questioning the Tube Type

The transition from the Shrink-Fit to the current paradigm began not with a new idea but with a question that should have been asked earlier: why a semiconducting nanotube?

The answer, when examined, was: because a MOSFET needs a semiconducting channel. But the premise was the problem. A MOSFET needs a semiconducting channel because the MOSFET operating principle requires a gate to create and destroy a conducting channel between source and drain. No semiconductor, no depletion, no switching.

But what if the switching mechanism did not depend on creating and destroying a channel? What if the channel were always present — permanently conducting — and the trit encoding came from something else entirely?

A metallic carbon nanotube is always conducting. Apply an AC voltage to it and current flows. The (8,8) armchair SWCNT is one of the most well-characterised metallic nanotubes: two conducting channels, near-perfect electron-hole symmetry, essentially no bandgap. Electrons pass through it ballistically, without scattering, over distances of hundreds of nanometres.

The band structure of an (8,8) armchair nanotube is exactly symmetric about the Fermi level. This means that if you apply a positive voltage across it, you get a current in one direction. Apply a negative voltage of equal magnitude, and you get exactly the same current in the opposite direction. Not approximately equal — exactly equal, by symmetry. The relationship is I(+V) = −I(−V) to precision limited only by the measurement.

Three states. Positive AC phase. Zero (no signal). Negative AC phase. Without any gate.

This was the insight that broke the paradigm. The trit encoding was already present in the physics of AC transport on a metallic nanotube. The gate — and everything built around the gate: the h-BN dielectric, the GNR electrode, the careful DC voltage levels — was not solving a physics problem. It was importing a solution to a problem that did not exist.

Paradigm III: The Coaxial Transmission Line (April 2026)

Once the DC gate paradigm was abandoned, the structure simplified dramatically.

No gate means no gate dielectric. Drop the h-BN layer. No gate dielectric means no gate electrode. Drop the GNR. The four-layer Matryoshka became a two-layer coaxial structure: an inner (8,8) armchair SWCNT inside an outer multi-wall carbon nanotube.

But the outer MWCNT was not merely a passive shield in the new paradigm. It turned out to have a specific role that the Matryoshka had never exploited.

The inter-wall quantum coupling between the SWCNT and the MWCNT suppresses the SWCNT’s conductance. Without any photon, the two-tube system has the inner tube partially coupled to the outer shell, and that coupling pulls electron probability away from the SWCNT conducting channels. The conductance drops from the ideal ballistic value.

A photon absorbed by the MWCNT changes this. The absorbed photon excites electrons in the MWCNT into higher energy states, shifting its energy levels away from the SWCNT’s. The coupling weakens — not because anything physical was inserted between the tubes, but because the MWCNT’s quantum states are no longer in resonance with the SWCNT’s. The SWCNT recovers its conductance. Current can flow.

The photon does not switch the device in the MOSFET sense. There is no gate voltage, no depletion, no threshold. The photon modulates the coupling between two tubes that are always in contact. It is a subtler mechanism than a gate, and a more powerful one.

The trit encoding in the final paradigm works as follows: the AC polarity determines the current direction (trit sign); the photon determines whether the SWCNT conductance is high (illuminated) or suppressed (dark). No photon and no AC signal gives the zero state — true absence of signal, not a constructed voltage midpoint.

Fabrication: Metallocene CVD

The shift to a two-tube metallic structure also changed the fabrication approach. The shrink-fit process had been designed around a four-layer assembly. A two-layer coaxial structure does not need it.

The fabrication route in the final paradigm is based on metallocene CVD — chemical vapour deposition using organometallic precursors. Ferrocene and related metallocenes have been used since the early 2000s as both the catalyst source and the carbon feedstock in CNT growth. A single precursor decomposes to provide both the catalyst nanoparticles that nucleate CNT growth and the carbon atoms that form the tube walls.

The (8,8)@(13,13) coaxial structure is a specific double-wall carbon nanotube (DWCNT). DWCNTs are naturally occurring products of certain CVD conditions — they form when a second nanotube nucleates inside an already-growing tube, or when conditions favour coaxial rather than parallel growth. Selecting for the (8,8) inner tube and (13,13) outer tube requires chirality control, which remains an active area of CNT synthesis research. But the fabrication route exists. It does not require multi-step assembly of disparate materials.

This is a significant simplification. The Matryoshka and Shrink-Fit paradigms required building a heterostructure from at least three different materials (SWCNT, h-BN, GNR) in sequence. The Coaxial paradigm requires growing one kind of structure: a double-wall carbon nanotube with specific chirality. Fewer materials, fewer interfaces, fewer process steps, fewer failure modes.

The Simulation That Settled It

A paradigm shift in theory is not the same as a paradigm shift confirmed by numbers. The NEGF (Non-Equilibrium Green’s Function) quantum transport simulation, completed on 2 April 2026, provided the numbers.

The simulation computed the transmission function of the exact (8,8)@(13,13) DWCNT geometry using the Kwant framework. Three operating conditions. Three trit states. The results:

The signal-to-noise ratio between active states and the zero state exceeded 2,000. The current symmetry ratio was 1.0000 — not a model assumption but a consequence of the armchair band structure. An independent GW+BSE optical calculation placed the MWCNT E₂₂ gate transition at 1678 nm and refined the initial design estimate of 710 nm for the SWCNT M₁₁ drive transition to 693 nm (a 2.4% correction).

The Matryoshka had never been simulated to this level of rigour. The noise margin problem was known theoretically but never quantified for the specific structure. The NEGF simulation of the Coaxial device provided something the first two paradigms never had: a precise, first-principles confirmation that the device does exactly what it is claimed to do.

What the Three Paradigms Reveal

Looking back at the three paradigms, the trajectory is clear: each transition removed something that was present for the wrong reason.

The Matryoshka had a gate because transistors have gates. When the need for a gate was questioned — specifically, whether a device that encodes trits as AC polarities rather than DC voltage levels even needs a gate — the gate went away. And with it went the h-BN dielectric, the GNR electrode, and two of the four layers.

The Shrink-Fit era had a complex assembly process because a four-layer structure needed one. When the four layers became two, the assembly complexity dissolved. Metallocene CVD can grow a DWCNT in a single process step.

The (10,0) zigzag channel was there because MOSFETs use semiconducting channels. When the MOSFET paradigm was abandoned, the zigzag channel was abandoned with it. The (8,8) armchair metallic tube is a better choice for AC transmission-line operation in every dimension: lower resistance, perfect symmetry, no threshold voltage to control, no depletion region to manage.

None of these simplifications were obvious at the start. The Matryoshka was not a naive design — it was the product of careful thinking within a particular set of assumptions. The assumptions were wrong, but identifying why they were wrong required working through the implications of each one until the failure became visible.

This is not an unusual arc for research. The simplest form of a correct idea is rarely the first form. The first form is the simplest thing you can build given what you currently believe. The final form is the simplest thing that the physics actually requires.