Silicon is a compelling platform for classical optical telecommunications as well as quantum communications, both through optical fibers. In general, the silicon-based integrated quantum transceivers could provide possible low-cost and high-performance secure communication networks. A reliable transfer of single photon qubits from one silicon device to another has already been shown, and the distribution of entangled states has been verified through the violation of Bell inequalities. Here, researchers first distribute the full set of entangled Bell states between two devices, and then demonstrate key missing capabilities so far, i.e., the chip-to-chip teleportation.
Schematic of the chip-to-chip entanglement distribution and teleportation of single qubits using the path-polarisation conversion technique. a, Schematic of chip A that includes a switchable router on qubit 4. A pair of MZIs choose whether the qubit is encoded into dual-rail and output via two 1d SGCs, or converted to polarisation qubits and output via the 2d SGC. The former enables the implementation of arbitrary single-qubit measurement in chip A, while the latter enables the coherent distribution of qubit 4 from chip A to chip B. b, Schematic of Bob chip, which is able to reconvert polarisation-encoded qubits to path-encoded qubits, in order to perform reconstructive projective measurements on Bob. A pair of MZIs are used for the ease of calibrating the components in chip B. c, SEM images of the 2D SGC structure fabricated on chip A and chip B. It can coherently convert the two on-chip path-encoded {|0i,|1i} states to two orthogonal polarisation modes {|Hi,|Vi} in fiber, and vise verse.
Nature Physics – Chip-to-chip quantum teleportation and multi-photon entanglement in silicon
Exploiting semiconductor fabrication techniques, natural carriers of quantum information such as atoms, electrons, and photons can be embedded in scalable integrated devices. Integrated optics provides a versatile platform for large-scale quantum information processing and transceiving with photons. Scaling up the integrated devices for quantum applications requires high-performance single-photon generation and photonic qubit-qubit entangling operations. However, previous demonstrations report major challenges in producing multiple bright, pure and identical single-photons, and entangling multiple photonic qubits with high fidelity. Another notable challenge is to noiselessly interface multiphoton sources and multiqubit operators in a single device. Here we demonstrate on-chip genuine multipartite entanglement and quantum teleportation in silicon, by coherently controlling an integrated network of microresonator nonlinear single-photon sources and linear-optic multiqubit entangling circuits. The microresonators are engineered to locally enhance the nonlinearity, producing multiple frequency uncorrelated and indistinguishable single-photons, without requiring any spectral filtering. The multiqubit states are processed in a programmable linear circuit facilitating Bell-projection and fusion-operation in a measurement-based manner. We benchmark key functionalities, such as intra-/inter-chip teleportation of quantum states, and generation of four-photon Greenberger-HorneZeilinger entangled states. The production, control, and transceiving of states are all achieved in micrometer-scale silicon chips, fabricated by complementary metal-oxide-semiconductor processes. Our work lays the groundwork for scalable on-chip multiphoton technologies for quantum computing and communication.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
You misunderstand, and trillions is based on misunderstanding. I will explain. First, I am not talking about RF switches, power and LEDs. The point is about logic and memory. There is certain budget in the world, allocated to buying CPUs, memory and logic. It cannot grow faster than the overall world economy, as other needs would be defunded. So with acceptable accuracy it is fixed size market. In that market, there is invested capital (fabs and everything for them), capital being invested (new fabs and processes), and investable capital (same for the next round). With growth (and investment return, excluding all divident-like distributions) possible only by substitution of one process or fab by another, all the money is thrown at that next round of the same, as it is fairly low-risk investment for 11-digit dollar amounts. There is no room for anything with greater risk, at all, regardless of return. At the same time, other people have no issues with investing into quantum processing – not because it is quantum, but because it is other people. They think differently: not in terms of shaving the last nanometer off the process, but in terms of multiples and orders in improvements. That is the only obstacle for big GaN chips: lack of “other people”, who are at the left half of logistical curve, not its far right corner.
I know it is already in commercial use. But what is the cost/transistor?
And then you have to sell that entire amount made to get a 22% Wright scaling learning curve (that’s the cost reduction at the end of the curve after you produce a number of units equal to the entire cumulative ever produced so far). Now do the math.
A mobile cpu with 4 billion transistors is $100. You get 400,000 transistors per cent that are 10x worse than GaN, so call it 40,000.
A GaN power switch (e.g. LGM1210 and competitors) is $4 for a 10,000 transistor power switch. You get 4 transitors per cent.
Both will be in your new mobile phone.
Now calculate the total sales required to get the Wright scaling to match for a GaN general purpose 400 million transitor CPU.
Yeah, trillions.
Oh, I guess I should not have explicitly mentioned GaAs in my post and instead used the more generic term computronium.
GaN is a recent addition, GaAs has been around for decades, which does not change my point on silicon dominance in the least. You have not shown the derivation of the astronomical costs, because there is none. Epitaxial GaN-on-Si is already commercialised in power. Scaling down such transistors and improving the process does not cost trillions.
You sound disingenous though. The R&D costs to get say a Gallium Nitride (better than GaAs) chip to maybe (not guaranteed) 10 million transistors would be like 2 or 3 trillion dollars (let alone 10 billion like current silicon ones).
So yeah it’s “just money” but it’s a meaningless statement. Nobody will spend 2 trillion on a 50% attempt.
Microelectronics industry overcame far more difficult problems that wafer fragility. Countless times. The only obstacle for overcoming has been the money: if no money in overcoming, then no overcoming. All money went into overcoming in mass produced silicon; now everything is silicon, though at once unbelievable level of perfection.
I’m guessing you don’t work in the chip industry. I do.
You’re confusing cause and effect here.
GaAs wafers are extremely fragile.
That follows from the difference in investment and production volumes. How many fabs, in the world, work with 300mm GaAs wafers? None, that is how many. Such wafers do not even exits. 1″ and 2″ wafers is what is available for InP. That is like silicon in 70’s, when a chip cost tens or hundreds of dollars, which means hundreds or thousands in today’s dollars. That is my whole point: silicon consumed everything, and now everything is silicon. That is as much “non-future” as future may be.
GaAs vs silicon isn’t about price suppression due to technological first to market.
Gallium is 17 parts per million of the Earth’s crust. Silicon is 28% or 280,000 parts per million.
Duh.
You forgot to mention enormity of the price difference. It is 1000x more expensive to use GaAs than silicon.
This sounds like quantum bullshit. Are they claiming faster then light transport of information by quantum teleportation? If so, definitely bullshit.
I read Victor as saying that eventually it will be so expensive to push silicon any further that it will be cheaper to develop another material and so we’ll see new stuff on offer.
Circuit to circuit or chip to chip wiring replaced by teleportation sounds impressive. Like always, the proof will be in the pudding…
It is already prohibitive. New fabs (under 10nm) are so expensive and full of exclusivity, that only one or two can economically exist in the world. When it gets to 2 or 3nm, there can be only one. End of the road. Try finding a higher level of concentration and dependency in any industry, not even such a critically vital as microelectronics. I could not.
Now consider this. You have one of the two (TMSC and Samsung), or just one fab in the whole world, and everyone comes to you. No one can compete with you, as there is no economic sense in that. Why would you innovate? Why even bother looking at anything else, if silicon is the foundation of your very well established monopoly?
An interesting example of one technology suppressing others by getting there the firstest with the mostest. But at some point the R&D costs for getting silicon to perform new tricks is going to become prohibitive and they will be force to look into these other materials.
Let’s not confuse “compelling” with “cheap”.
Silicon is about as good for photonics as cheese is for drill bits. It took an awful lot of trickery to make half-decent photonic components in silicon, while the really compelling photonics materials (such as GaAs) never made it into mass production. Silicon is the reason we do not have GaAs, InP and other fascinating III-V materials in mass-produced microelectronics, not even SiC and GaN outside their niches: no memory, no processors, no FPGAs, no ASICs, and no photonics (LEDs and such is off topic). Silicon everything. Clocked at 474MHz in 1993, Cray-3 supercomputer was made with GaAs chips, and now it is a long forgotten story. In the same 1993, Intel made the “revolutionary” Pentium processor clocked at revolutionary 60MHz – eight times slower, because silicon. The term “compelling” belongs to the III-V semiconductors, especially in photonics applications. Silicon is the “el cheapo” way to the”el cheapo” future, because industry is heavily invested into silicon fabs and business is better without revolutions.
Remember graphene? Other 2D materials, like MoS2? There were also vacuum nanoscale FETs, which no one remembers now, while they could come handy with quantum and photonic work. Nope, silicon.