Impact case study database

The research of the Microelectronics Group (MG) is mainly funded by the EPSRC. EPSRC emphasizes the impact of research on industry and society. MG shares this value and has used industry relevance as a core criterion when selecting research topics. The projects were created by consulting industrial partners and all test samples used in the projects were supplied by partners. To increase the impact, MG chose to publish their results at industry-centric conferences, such as IEDM, where leading companies, such as Intel, announce their latest breakthroughs. The knowledge gained through this industry-centric research approach enables the MG to provide a service to the industry in terms of:

(W1) The predicative modelling

Early works from some research groups verified their device-ageing models by showing that the models can fit test data well. This, however, did not deliver the original mission of the modelling: to predict device and circuit performance where test data does not exist. MG determined to take up this challenge and qualified their model by its capability to predict device performance under real use conditions [R3]. This has attracted the attention of industry and the model has been used to:

(I) Qualify new processes and materials by using the model to predict device lifetime:

Ever since the invention of integrated circuits in 1958, the intensive competition has driven the industry to develop a new process every 1.5~2 years. Before one process can make its commercial debut, the industry standard requires a device lifetime of 10 years. This requires prediction, as it is impractical to test devices for that long. The MG’s predictive model has been used to qualify new processes [S1, S2 (Section 5)].

S1 is a partner of the EPSRC-funded projects and supplied some of the test samples used by the projects. The research results were disseminated directly to it through progress meetings. Two members of MG joined S1 to work on process qualification: Dr MB Zahid in 2008 and Dr B Tang in 2016.

S2 is an international chip foundry. After MG presented the AG model at the 2013 IEEE International Electron Devices Meeting (IEDM), S2 invited members of MG to present the model to the company’s test engineers in 2014. After the presentation, the company decided to adopt the model. This involves changing test procedure and data analysis to improve the accuracy of device lifetime estimation. The company’s lawyers did not allow the company to provide a financial figure on the benefit of the model to the company.

In addition to S1 and S2, the model has received an increasingly wide attention and members of MG have been invited and funded to deliver tutorials to train test engineers on how to implement the model. Over 500 engineers have taken part in the training so far. The latest tutorial was delivered at the 26th IEEE International Symposium on the Physical and Failure Analysis (IPFA) of Integrated Circuits, Qingdao, July 2019, with over 100 attendees [S3].

(II) Product Design Kit (PDK)

The modern microelectronic industry has some companies, such as ARM, specializing in design and others, such as TSMC, focusing on fabrication. The bridge between them is PDK. The ‘First Silicon Success’ relies on the accuracy of models used in PDK and the discrepancy between models and silicon performance is a major challenge to the industry. The research and models of MG have been used to develop reliability- and variability-aware PDKs [S4, S5]. These PDKs allow designers to take the reliability and time-dependent device-to-device variation into account when verifying their circuits, which in turn not only increases the ‘First Silicon Success’, but also optimizes the performance of designed circuits in term of power, speed, and cost.

Semiwise [S4] is a start-up, while S5 is a prime international supplier of Electron Design Automation (EDA) software. S5 is a partner in the EPSRC projects [G4, G6] and at the end of project G4, the post-doctoral researcher, Dr M Duan, joined S5 in 2018 to work on variability-aware PDK. The collaboration with S4 and S5 is further strengthened through the award of the joint EPSRC project (“Realistic fault modelling to enable optimization of low power IoT and Cognitive fault-tolerant computing systems”, EP/T026022/1, £487k) in 2020.

(W2) Embedding test techniques into commercial instruments

Based on commercial instruments, MG developed a number of advanced characterisation techniques that require technique-specific programs and hardware setups. MG realised that these programs and setups were not available to other users of the same instruments, as MG were invited by a project partner to implement the technique for their laboratory [S1].

After this experience, MG contacted the instrument supplier and explored embedding these techniques into their instruments as a module. The company then invited MG to visit their research and development centre in Ohio, USA, followed by the company visiting Liverpool John Moores University. The module has been jointly developed and embedded into the supplier’s most advanced semiconductor parameter analyser, KEITHLEY 4200, and shipped to the customers. MG worked together with the supplier to prepare the module’s user manual. In addition, MG was invited and funded to train test engineers at the workshops organized by the supplier [S6]. This collaboration started in 2012 and is on-going, as more techniques have been added to the module when they are developed by MG.

Since 2017, over 4,000 engineers have taken part in the training. The Company could not tell MG how many customers bought the instrument because of this test module. What they can say is that, at September 2019, 3145 copies of the module were shipped together with the instrument to world-wide customers [S6]. To continue the collaboration and add more techniques to the module, the Company offered a PhD studentship to MG [S7].

(W3) Contributing to development of new technologies

As transistors are downscaled to nano-meters, the cost of developing future technologies becomes so expensive that even the sector’s largest companies, such as Intel, cannot afford to do everything on their own. To share the cost, the industry has formed a consortium, based at IMEC, Belgium, whose members include Intel (US), ARM (UK), Micron (US), Samsung (Korea), Toshiba (Japan) and TSMC (Taiwan). Each company has its own assignees based at IMEC. Every six months, there is a partner technical week (PTW) to review the progress of the R&D work.

The cost of fabricating the industry-relevant nano-meter transistors is well beyond the reach of MG. To contribute to this world-wide effort to develop future generation technologies that impact the daily life of society, MG’s strategy has been to collaborate with this industrial consortium and to focus on the qualification of new processes, materials and devices sourced from the consortium.

To strengthen the link with the consortium, MG has one researcher based at IMEC, who takes part in the daily R&D activities of the consortium. This offers a direct bridge between MG and the consortium. The issues to be addressed and the test samples needed by MG are co-identified with the researchers at IMEC. The results are disseminated directly to the consortium through MG’s researcher at IMEC. The MG’s works have impacted the Consortium’s development work in terms of material selection, structure evaluation, and performance optimization [S8]. For example, one joint paper published at 2019 IEDM conference is entitled “Endurance improvement of more than five orders in GexSe1-x OTS selectors by using a novel refreshing program scheme” [R4].

The value of this collaboration has been recognized by IMEC and IMEC has supported the research at LJMU through a rolling contract of Euro100k p.a. for over 20 years [G7]. To the best of MG’s knowledge, LJMU is the only UK university that IMEC has supported continuously in this way for over two decades. LJMU is also the only UK university that has been regularly invited to give presentations to the Consortium at its ‘partner technical week’ meetings. It is a privilege for MG to work together with THE world leading consortium at the forefront of developing new technology. This has allowed MG to punch well above its weight. The collaboration goes from strength to strength, initially in the area of logic devices and now also in memory devices.

(W4) Defect-based True Random Number Generator and its commercialization

The three key issues identified for IoT are security, cost, and power consumption. Random Numbers are required in cryptography and authentication and their generator is an essential component of security systems. The algorithm-based software generators used in some security systems are not truly random and are vulnerable to attack. The True Random Number Generator (TRNG) harvests the randomness of natural phenomena in hardware without using an algorithm.

MG has gained an in-depth knowledge of defects in transistors, whose charging-discharging produces random telegraph noise. Although noise is unwanted for normal electronic circuits, MG innovatively harvested its randomness to design and build a TRNG, which is 10 times faster than the TRNG proposed by early works [R6]. The events comprising this impact are outlined in chronological order below:

• MG submitted a paper on its TRNG to the 2018 Symposium on VLSI technology. The conference organizer not only accepted the paper [R6], but also invited and funded MG to demonstrate the TRNG to the conference attendees, along with the demonstrations by Intel Corporation and Panasonic Corporation [S9].

• MG applied for, and was awarded a grant by Innovate UK to develop a proto-type product for its TRNG [G8].

• MG was invited to demonstrate its proto-type product at the 2018 International Security Expo in London. A number of companies expressed their interest in the product.

• This led to the award of USA patent (Attorney-Docket-Number: P34923US1).

• Crypta Labs and Secure Technologies Ltd are bidding for the IP for security application in their products [S10].

In summary, to maximize the significance of the impact of its research, MG has chosen to work together with world leaders by providing better models (W1), new test techniques (W2), material/structure optimization (W3), and new product design (W4). This collaboration also led to a wide reach of its impact. For example, by embedding the new technique into the instruments of a prime supplier, it reached the supplier’s world-wide customers. The sustainability of MG’s approach has been demonstrated by the continuous collaboration and by the partnership in the EPSRC projects with these leading organizations over the last two decades.