We have written several articles about Intel’s quantum effort who have provided some information about their approach. While there was the release of Intel’s Quantum Software Development Kit last year, to date, there hasn’t been a specific hardware device release. That situation has now changed with Intel’s announcement of its 12 qubit Tunnel Falls device.
Instead of building systems and setting up cloud access like other companies, Intel is working with selected government and academic institutions to provide chips so these institutions can build their own systems and research the characteristics of Intel’s quantum dot technology. The first organizations to receive Tunnel Falls are among others LPS Qubit Collaboration (LQC) at the Physical Science Laboratory in College Park, Maryland, Sandia National Laboratory, University of Rochester, and University of Wisconsin-Madison. Intel partnered with this organization to research device characteristics and will use the results for the development of the next generation of devices. We hope additional government and academic groups may be provided with these chips to assist with this research as time goes on.
Intel’s quantum technology is still in its early stages, but has great potential to scale up quickly using semiconductor-compatible quantum dot qubit technology. Intel showed that qubits built with quantum dots (also known as spin qubits) are one million times smaller than qubits built with superconducting technology. Additionally, Intel is able to manufacture devices in a state-of-the-art, high-volume, 300-millimeter wafer factory in Hillsboro, Oregon which enables them to achieve high throughput and high precision. Intel has demonstrated that they achieve a very high 95% throughput rate across wafers with each wafer housing more than 24,000 quantum devices. They also use a custom cryoprober they developed with BlueFors and Afore to quickly scan all devices on a wafer at 1.7 degrees kelvin so that they can select the best device for further investigation and installation in a system.
One additional advantage that Intel has is their expertise in materials science and relationships with equipment vendors. An example of where they leveraged this is they used the purified Silicon-28 isotope on their planar Si/SiGe heterostructure in the device as opposed to natural silicon which has a mix of other isotopes including Silicon-29 and Silicon-30. In standard semiconductors this makes no difference, but the advantage of Silicon-28, unlike other isotopes, is that it has zero nuclear spin. For quantum this makes a big difference as it allows for a significant increase in coherence time which will lower the error rate and increase the accuracy of calculations.
The device topology is a 12 qubit linear array with four sites at which the qubits can be measured. Each qubit can gate two qubits with its nearest neighbor as shown in the diagram below. This topology limits practical quantum processors and the logical next step is to create 2D spin qubit devices with qubits arranged in a lattice. While we don’t know exactly Intel’s next steps, they have shown us that they have already tapped into their next generation of devices and will make them available in 2024. The current topology is limited to practical quantum processors and the next logical step. is to create 2D spin qubit devices with qubits arranged in a lattice for more connectivity. Intel has also developed a cryoCMOS control chip called Horseridge with its partner QuTech in the Netherlands. This will allow them to closely integrate control logic or even integrate with qubit chips and will make the system more manufacturable and reliable. They don’t currently use this technology with Tunnel Falls chips, but it is something we can look forward to in the future.
Intel is taking the long view with their quantum strategy. They believed that it would be difficult to build a large quantum computing business with NISQ-level machines and that fully error-correctable, fault-tolerant processors with millions of qubits would be required for quantum computing to take off. Their technology approach is a reflection of this and they are working toward a roadmap that can scale rapidly and provision these millions of qubits. While Intel has yet to outline their long-term commercialization approach, our belief is that they may take an approach similar to the one they used with classic microprocessors. Rather than building complete systems and shipping them for on-premises use or offering quantum cloud services, Intel could continue to provide chips and other technology components as components and let OEMs build finished quantum computers and sell them or consume them. cloud. But time will tell how Intel’s strategy and the entire quantum ecosystem will develop.
Additional information about the Intel Tunnel Falls announcement is available in the newsroom article Here and a video Here. A related news release from the University of Wisconsin-Madison is available Here and other news releases from the LPS Qubit Collaboratory can be viewed Here.
June 15, 2023