Page 1

ISSUE 8 / FALL SEMESTER 2019

BENDING T HE NOR M ON N A NO W IR E S P.47

F INDING WAYS T O F EED T HE W OR L D

BE T T ER R A CING C A R DE SIGN T HROUGH A N INDU S T RY PA R T NER SHIP

P.50

P.62

HARNESSING T H E A I W AV E

P.38


Innovation starts at KAUST Partnering with industry, entrepreneurs and the brightest minds to solve our shared global challenges. KAUST is an institution of enterprise and a gateway for industry in Saudi Arabia.

Learn more innovation.kaust.edu.sa


A LETTER

FROM THE VICE PRESIDENT FOR RESE ARCH

Dear Reader, In this issue of KAUST Discovery magazine, we celebrate 10 years of curiosity-driven and goal-oriented research at KAUST. From the inception of his vision for the University, the late King Abdullah recognized that universities striving for excellence depend on an atmosphere of exploration and initiative, while nurturing and protecting freedom of research, thought and discourse related to scholarly work. Inspired by the potential of this vision, 75 faculty became founding members of KAUST. They joined KAUST from across continents to pursue their research goals and form the University’s original global DNA, bridging people, ideas and traditions from around the world. As a testament to how KAUST supports its faculty, 60 percent of these founding members have stayed to make KAUST their home base for developing and expanding innovative research programs for the past 10 years. By year 10, KAUST has doubled its faculty to include more than 150 members and aims to grow that number by a further 50 percent over the next five years. In accordance with the late King’s wish for KAUST to foster collaboration and cooperation with other great research universities, KAUST faculty established partnerships with others who share the belief that tackling global challenges is essential. These partnerships are integral to the KAUST DNA. They helped to put KAUST on the map, and in some cases, they have evolved into productive partnerships that are now led by KAUST. Our article about computational scientist Matteo Parsani’s research highlights one of KAUST’s most recent partnerships with McLaren Racing. This piece tells how Matteo and his group are applying computational fluid dynamics to perfect McLaren’s racing cars’ performance. Our feature article highlights one of the new pillars of KAUST research—digital science and technology. Here, we step into the future for a glimpse of how our researchers are contributing to fundamental work in artificial intelligence and into how they are applying machine learning and artificial intelligence to

optimize their research. “Finding ways to feed the world” tells a story of how some of our plant scientists are sowing seeds to feed the ever-growing human population. From improving crop resistance to disease and parasites through to increasing plant yield, KAUST researchers are devoting substantial efforts to providing solutions to the future food challenge. This article points out an important distinction between crops that have been genetically modified and those that have undergone genome editing for breeding programs. Materials scientist Pedro Costa and his group tell us how they are using the University’s cutting-edge microscopy equipment and the cleanroom facility provided by the KAUST Core Labs to visualize and configure arrange-

“B y y e a r 1 0 , K A U S T h a s doubled its faculty to include more than 150 members and aims to grow that number by a further 50 percent over t h e n e x t f i v e y e a r s .” ments of bent nanowires with improved performance inside electronic devices. Read the article to learn about how these self-healing materials could be integrated into devices to create next-generation electronics. We look forward to continuing to share the progress of our research over the next 10 years. Keep your finger on the pulse of KAUST research by regularly visiting our website—discovery.kaust.edu.sa—and sign up to receive our fortnightly newsletter. Professor Donal Bradley Vice President for Research K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

1


KAUST & Boeing Collaborating on next-generation aerospace technologies in physics and water. Boeing is leading aerospace innovations and investing in communities worldwide. innovation.kaust.edu.sa


CONTENT S / FA LL SEMES TER 2019

ABOUT THE COVER

KAUST celebrates 10 years of research. In the next decade, our researchers will strategically harness the AI wave to accelerate innovation and technology. If you would like to update your information, send us an email at discovery@kaust.edu.sa COVER CREATED BY XAVIER PITA.

EDITORIAL COMMITTEE Pierre Magistretti Dean, Biological and Environmental Science and Engineering Division Mootaz Elnozahy Dean, Computer, Electrical and Mathematical Sciences and Engineering Division Magnus Rueping Professor, Chemical Science Physical Science and Engineering Division

MANAGING EDITOR Carolyn Unck EDITORIAL TEAM Mariyah Abualnaja Carmen Denman Virginia Unkefer ILLUSTRATIONS AND PHOTOGRAPHY Helmy Alsagaff Ivan Gromicho Anastasia Khrenova Xavier Pita

KAUST DISCOVERY is published with King Abdullah University of Science and Technology (KAUST) by the Partnership and Custom Media Unit of Nature Research, part of Springer Nature. King Abdullah University of Science and Technology (KAUST) Thuwal 239556900 – Kingdom of Saudi Arabia. Email: discovery@kaust.edu.sa

O N T HE C OV ER

47

BENDING THE NORM ON NANOWIRES FUEL COME CLEAN

50

FINDING WAYS TO FEED THE WORLD

62

BETTER RACING CAR DESIGN THROUGH AN INDUSTRY PARTNERSHIP

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

3


CONTENT S / FA LL SEMES TER 2019

C L E A N ER F U EL S

R ED S E A S C IEN C E

WAT ER B R E A K T H RO U G H S

S M A R T T EC H

25 TESTING THE BLIND SPOTS IN ARTIFICIAL INTELLIGENCE

Understanding the situations in which artificial intelligence can fail is critical for future autonomous vehicle and medical diagnostics applications.

13 RED SEA TEMPERATURES TO COOL IN COMING DECADES

6 CATALYSIS CAPTURED IN A PINCER MOVEMENT

Hydrogen-powered cars and recycling CO2 are just two of the myriad of applications that could spring from catalysts based on a new ligand platform.

8 PUTTING THE SQUEEZE ON SOOT

Understanding the effect pressure has on soot production during fuel combustion could help reduce polluting emissions.

9 READY, JET, PRINT!

Inkjet printing could produce high-efficiency organic solar cells with commercial potential.

10 BIODIESEL BY-PRODUCT HELPS FUEL COME CLEAN A compound made from the glycerol by-product of biodiesel production could promote cleaner burning in vehicle engines.

11 TURBOCHARGING THE SWITCH TO EFFICIENT ENGINES

Predicting capricious pre-ignition combustion events could enable automakers to build powerful yet more efficient engines.

4

F a l l 2 0 19

Exploring the links between natural climate cycles and the sea surface temperature of the Red Sea reveals a cooling trend during the next few decades.

14 “MARINE SKIN” DIVES DEEPER FOR BETTER MONITORING

A pioneering tagging system that monitors the movement and local environment of sea animals reaches deeper depths and higher sensitivities.

16 MANGROVE FORESTS TRAP FLOATING LITTER Marine plastic pollution accumulates in mangrove forests and is a danger to sea life.

26 TINY, FAST, ACCURATE TECHNOLOGY ON THE RADAR 17 COLORFUL SOLUTION TO A CHEMICAL INDUSTRY BOTTLENECK A graphene-oxide membrane design inspired by nature swiftly separates solvent molecules.

18 CASTING A NET FOR A SUSTAINABLE WATER NETWORK

Global water challenges addressed through improvements to desalination techniques and limits to waste.

20 COMBINING FORCES TO COMBAT E. COLI PI-7 A combined treatment may help tackle the rise of E. coli PI-7 in Saudi Arabia’s wastewater systems.

22 A DRONE ON CROP RESOURCES

Aerial imaging of plant heights could help farmers manage field crops more effectively to increase yields and preserve resources.

23 DRINKING WATER SUCKED FROM THE DUSTY DESERT AIR

An inexpensive hydrogel-based material efficiently captures moisture even from low-humidity air and then releases it on demand.

A tiny, portable radar device could allow visually impaired people, or unmanned moving devices, to detect objects in real time.

27 QUERYING BIG DATA JUST GOT UNIVERSAL

A universal query engine for big data that works across computing platforms could accelerate analytics research.

28 PROBING WATER’S SKIN Electrosprays of water cannot reliably probe the air-water interface.

30 BRIGHTER BLUE LIGHTS FOR FASTER COMMUNICATION

High-performance blue-lightemitting diodes could boost white-light, high-speed data transmission.

32 SENSITIVE ROBOTS FEEL THE STRAIN Flexible skin for soft robots, embedded with electrical nanowires, combines conductivity with sensitivity within the same material.

34 ARMBANDS DO A HEALTH CHECK WHILE YOU WORK OUT Nanotech-powered electrodes help solve the challenges of using sweat to assess biological conditions in real time.


CONTENT S / FA LL SEMES TER 2019

F E AT U R E

S C IEN C E V I S UA L IZ ED

BIOLOGICAL S YS T EM S

T R A N S L AT IN G T EC H

49 LACTATE TRIGGERS GENES THAT MODIFY BRAIN ACTIVITY

37 HARNESSING THE AI WAVE

A perfect storm is giving rise to a surge in artificial intelligence research and KAUST aims to be at its forfront.

42 CAUSAL DISENTANGLEMENT IS THE NEXT FRONTIER IN AI

Recreating the human mind’s ability to infer patterns and relationships from complex events could lead to a universal model of artificial intelligence.

43 TWEAKING TOOLS TO TRACK TWEETS OVER TIME

Computer model learns to identify Twitter users’ evolving interests by analyzing their tweets.

44 ACCELERATING THE GRAPEVINE EFFECT

Gossip is an efficient way to share information across large networks and has unexpected applications in solving other mathematical and machine-learning problems.

46 CAPTURING SPRAY FROM FLASH-BOILING LIQUID JETS

Ultrafast video capture of droplet-cloud formation should help minimize the risk of gas-leak explosions.

47 BENDING THE NORM ON NANOWIRES Arranging nanowires in bent configurations may make them less likely to fail inside electronic devices.

48 TAKING AIM AT OUTLIERS

Automatic detection of uncharacteristic data sequences could change the way data is processed and analyzed.

Study illustrates the links between brain energy metabolism and neuronal activity.

50 FINDING WAYS TO FEED THE WORLD

Genome-editing techniques have the power to transform crop yields and plant resilience to feed the growing global population.

52 SPEARGRASS RECRUITS SANDY MICROBES FOR HELP In the harsh sand of the Namib Desert, the sheaths of speargrass roots co-opt any growthpromoting bacteria they can find.

54 MINING BACTERIA FROM THE DESERT Desert bacteria protect plants from salt toxicity.

57 MOFS CAN SENSE AND SORT TROUBLESOME GASES

Fluorinated metal organic frameworks make excellent materials for selective sensing and removal of toxic gases.

58 DRIVING WATER DOWN NANOHIGHWAYS

Self-assembled channels in a polymer membrane could greatly enhance extraction of water from gases.

60 PAPER SENSORS REMOVE THE STING OF DIABETIC TESTING

Inkjet-printed device helps monitor a patient’s blood-sugar levels without painful needles.

62 BETTER RACING CAR DESIGN THROUGH AN INDUSTRY PARTNERSHIP The Formula 1 race track is the ultimate testbed for a KAUST researchers’ latest work.

64 ONE SWITCH TO RULE THEM ALL

Inkjet-printed switches make multiple frequency bands easier and cheaper to manage in wireless devices.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

5


CLEANER FUELS

Kuo-Wei Huang (second from left) and his researchers are working on a catalytic platform that has the potential to be a game changer for catalyst chemistry.

CATALYSIS CAPTURED IN A PINCER MOVEMENT

Hydrogen-powered cars and recycling CO2 are just two of the myriad of applications that could spring from catalysts based on a new ligand platform.

6

F a l l 2 0 19

When Kuo-Wei Huang and his group made a seemingly simple chemical modification to a pincer catalyst system, they did not observe the anticipated modest alteration to the catalyst’s reactivity. Instead, they discovered totally distinct catalytic behaviors. A decade of study later, the catalytic platform discovered by the Huang group looks set to spawn a new branch of catalyst chemistry, with the potential to drive revolutionary chemical transformations. Catalysts are chemical entities that assist with the making and breaking of bonds, often with exquisite selectivity. They are used across all areas of production, from commodity chemicals to pharmaceutical manufacturing, and they underpin the manufacturing of many chemicals and materials in use today.

In most catalysts, including organometallic catalysts, where each metal ion is wrapped in a carbon-based ligand, metals are the star of the show: they are the center of catalytic reactivity where bond creation takes place. The ligand might fine tune that reactivity, but it is essentially there to provide support. However, in Huang’s new catalyst family, that situation can be reversed: the ligand can be the site of catalytic activity where new bond formation takes place, and the metal is relegated to the support role. “For the last 100 years, we have been talking about how to modify the ligand to manipulate the reactivity of the metal center,” Huang says. “Now we can also use metal coordination to change the reactivity of the organic component—and maybe reach some unprecedented reactivities that cannot be achieved otherwise,” he says.


Nippy performance Huang’s new ligands fall into a class known as pincer ligands, first studied in the early 1970s, which get their name from the pincer-like grip they exert on the central metal ion. The ligand centers on a six-membered pyridine ring, notable because it possesses a special chemical feature called aromaticity, a state of electronic balance that makes it highly stable. Carbon-based side chains dangling from the pyridine ring form the pincer grip on the metal ion. That a hydrogen could be plucked from one of the CH2 groups in the side chain was a mere curiosity for three decades. Then in 2005, researchers realized that this process disturbs the aromatization of the pyridine ring and turns it into a very active catalyst. “The

central pyridine ring is very eager to regain the aromaticity,” Huang says. Most notably, the catalyst can even split water molecules into oxygen and hydrogen, a potential carbon-free future fuel. “This is a stunning reaction,” Huang says. Huang’s entry into pincer-ligand chemistry came from a seemingly small modification: he swapped the key CH 2 group for an NH group and discovered that the reactivity was completely changed. Far from decomposing water, the catalyst was unreactive, even in water heated to boiling. “That was the first big difference we observed,” Huang says. Differences such as these prompted the team to look more closely at its new pincer catalysts. Huang showed that in its reactive dearomatized form, with certain metals, the ligand’s pyridine component retained some pseudoaromaticity1. This in turn made the nitrogen arm of the pincer ligand electron-rich and highly reactive. “It is because of this special reactivity that we are no longer restricted to the reactivity of the metal center,” Huang says. In subsequent studies, the team has shown the iminic nitrogen of its pincer ligand performs like an extremely reactive form of carbon called a carbene2. “That has never been achieved before,” Huang says. For example, the team has shown they can use the catalyst to persuade carbon dioxide, a very unreactive molecule, to react with nitrogenbased molecules to make valuable chemicals3. Because of its unusual stability toward water, the team could also develop another catalyst for the decomposition of formic acid, a potential fuel, to selectively release hydrogen, which can be used to power an electric car4. A power generator developed by the team, fueled by formic acid, is going to be commercialized next year, Huang says. “No one would have envisaged that a start with the organometallic chemistry of catalyst design would lead to making electric cars,” Huang says. Huang credits KAUST’s funding model, which provides baseline

ILLUSTRATION BY XAVIER PITA.

CLEANER FUELS

The catalyst’s reactivity is not centered on the metal ion (silver sphere), but on one of the side arms of the pincer ligand (blue sphere).

funding for faculty members, for allowing him to pursue the initial catalyst research that led to this unexpected new branch of chemistry. These examples of the catalyst’s utility published so far are just a glimpse of what is to come, he adds. “The important thing is we have created a new way for catalysts to be designed, plus set up a platform open for opportunities.” 1. Gonca̧lves, T.P. & Huang, K.-W. Metal−ligand cooperative reactivity in the (pseudo)-dearomatized PNx(P) systems: The influence of the zwitterionic form in dearomatized pincer complexes. Journal of the American Chemical Society 139, 13442−13449 (2017). 2. Li, H., Gonca̧lves, T.P., Hu, J., Zhao, Q., Gong, D., Lai, Z., Wang, Z., Zheng, J. & Huang, K.-W. A pseudodearomatized PN3P*Ni−H complex as a ligand and σ-nucleophilic catalyst. Journal of Organic Chemistry 83, 14969–14977 (2018). 3. Li, H., Gonca̧lves, T.P., Zhao, Q., Gong, D., Lai, Z., Wang, Z., Zheng, J. & Huang, K.W. Diverse catalytic reactivity of a dearomatized PN3P*–nickel hydride pincer complex towards CO2 reduction. Chemical Communications 54, 11395— 11398 (2018). 4. Eppinger, J. & Huang, K.W. Formic acid as a hydrogen energy carrier. ACS Energy Letters 2, 188−195 (2016).

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

7


CLEANER FUELS

PUTTING THE SQUEEZE ON SOOT

Understanding the effect pressure has on soot production during fuel combustion could help reduce polluting emissions. Running diesel engines and gas turbines at high pressure to boost power output and efficiency is harmful for the environment. Burning fuel at high pressure can significantly change the soot particles that are produced, William Roberts from the KAUST Clean Combustion Research Center and his team has shown. “Studying the factors affecting soot formation should lead to new ways to curb soot emissions,” says Hafiz Amin, first author of the paper. Soot is a major global pollutant, harmful to human health and the second biggest contributor to global warming after carbon dioxide. The complex substance comprises individual nanoparticles that are linked together in chainlike structures called fractal aggregates. Because the effect of soot particles on human health is highly dependent on the soot structure, the team was motivated to understand the formation processes. "The formation of the soot is the key to controlling its emissions,” Amin says. But the effect that elevated pressure has on soot production is not well understood. Roberts and his team developed an efficient method for soot sampling that provides new insights into soot formation. The team developed a way to sample, at pressures up to 10 atmospheres, the soot particles produced in a laminar flame called a counterflow diffusion flame. The counterflow burner

8

F a l l 2 0 19

produces a horizontal, circular shaped flame that is stable and uniform. “The counterflow flame allows us to control the fluid dynamics of the flame and requires less computational power for modeling,” explains Anthony Bennett, co-author of the paper and current Ph.D. student in Roberts’ team. The researchers collected soot from the flame using a fine carboncoated copper mesh held by a pneumatically powered robot arm that rapidly shoots out to collect a soot sample with minimal disturbance to the flame. Anthony Bennet (left) and Bill Roberts with the pressure vessel the group used to perform the experiments.

The team analyzed their soot samples using transmission electron microscopy and observed that elevated pressure significantly increased soot particle size. The mean particle diameter of 17.5 nanometers at 3 atmospheres of pressure rose to 47 nanometers at 10 atmospheres. “This shows the importance of the influence of pressure on soot emissions,” Bennet says. Now that they have developed an effective high-pressure experimental setup and soot-sampling procedure, the team has many more experiments planned, Amin says. “We would like to extend the investigations to a pressure of

40 atmospheres, which is relevant to the operating pressure of gas turbines,” he says. The team would also like to investigate how different fuel chemistry and residence times influence soot production at elevated pressure. “Our current experimental facilities provide the control to perform such investigations,” Amin says. Amin, H.M.F., Bennett, A. & Roberts, W.L. Determining fractal properties of soot aggregates and primary particle size distribution in counterflow flames up to 10 atm. Proceedings of the Combustion Institute 37, 11611168 (2019).


CLEANER FUELS

READY, JET, PRINT!

enhancing cell efficiency using higher performance materials. They are also investigating ways to integrate the devices in modules and with other printed electronics for self-powered autonomous sensing.

Inkjet printing could produce high-efficiency organic solar cells with commercial potential.

Corzo, D., Almasabi, K., Bihar, E., Macphee, S., Rosas-Villalva, D., Gasparini, N., Inal, S. & Baran, D. Digital inkjet printing of high-

Inkjet printing is expected to fast track the commercialization of organic solar cells. Researchers from the KAUST Solar Center have exploited this technique to generate high-efficiency solar cells at large scales. Organic photovoltaic materials could soon replace inorganic semiconductors in solarpowered devices because of their lightness, flexibility and low cost. These materials are easy to modify and process in solution, which makes them highly attractive for customization and large-scale production. In particular, customized solar cell designs can be used in conjunction with other printed electronics to power a plethora of applications, such as disposable electronics, intelligent packaging, interactive printed media and lab-on-a-chip devices. Nonfullerene acceptors are emerging materials that have helped boost the efficiency of organic solar cells close to commercialization. These components are typically blended with electron donors in a light-responsive electrochemical layer. They have proven effective for drawing the lightgenerated pairs of electrons and negatively charged holes apart and maintaining electric current when exposed to sunlight. However, scale-up and manufacturing challenges have hindered efforts to transfer these materials from the laboratory to the industrial and consumer-ready scale. To bridge this gap, Derya Baran and her colleagues have engineered inkjet-printable solar materials containing a nonfullerene acceptor and deposited these inks over large areas to produce photovoltaic cells. The resulting devices achieved efficiencies of 6 percent, which is comparable to the efficiencies of their spin-coated analogs. Ph.D. candidate Daniel Corzo explains that inkjet printing presents several advantages over traditional spin-coating and bladecoating deposition techniques, including low material consumption and rapid design changes through digital platforms. “This enables low-cost manufacturing, patterning

into complex shapes, and multilayered device fabrication without the need for multistep lithography,” he adds. The researchers optimized the printing process by tuning the viscosity and evaporation behavior of the ink to improve both how the droplets were ejected and how they interacted with the substrate surface. According to Corzo, this optimization has provided a repeatable and commercially scalable process. Baran’s team also fabricated high-efficiency turtle-shaped devices, demonstrating the potential for the process to be customized. “It is amazing that we can now fabricate solar cells with complex shapes at the push of a button, opening the door to a wide variety of applications,” Corzo says. The researchers are currently developing fully printed organic solar cells and

efficiency large-area nonfullerene organic solar cells. Advanced Materials Technologies 4, 1900040 (2019).

DANIEL CORZO PH.D. STUDENT Working in Derya Baran’s lab, Daniel Crozo is pursuing his Ph.D. degree by studying the crystallization and characterization of perovskite materials for solar cell fabrication with the aim of improving current fabrication techniques and finding new ways to manufacture optoelectronic devices.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

9


ILLUSTRATION BY XAVIER PITA.

CLEANER FUELS The researchers used computational analysis to explore the effect of differing pressures and temperatures on the combustion reactions. This allowed them to identify conditions in which glycerol carbonate will burn cleanly, while also generating the compound 3-hydroxypropanal, which is known to greatly reduce the production of soot.

“G l y c e r o l c a r b o n ate has great potential to promote cleaner Artistic representation depicting a cleaner environment within the combustion cylinder of a vehicle engine highlighting the link between clean combustion and a sustainable future.

BIODIESEL BY-PRODUCT HELPS FUEL COME CLEAN A compound made from the glycerol by-product of biodiesel production could promote cleaner burning in vehicle engines. The potential of glycerol waste from biodiesel production to be simply converted into a clean-burning sustainable fuel is being explored by scientists at KAUST’s Clean Combustion Research Center. “As one of the world’s leading teams in combustion research and innovation, we are always looking into new fuels and additives that help maximize engine efficiency and minimize environmental pollution,” says research scientist Binod Giri. Concern over diminishing fossil fuel reserves and the environmental damage caused by current fuel technologies is driving interest in more sustainable and cleaner fuels. One major avenue to reduce global dependence on fossil fuels is by producing biodiesel from waste vegetable oils and animal fats. Burning glycerol carbonate generated from the excess glycerol produced by biodiesel production offers the extra advantage of greatly reducing

10

F a l l 2 0 19

emissions of the sooty particulates that cause respiratory disease. This work on glycerol carbonate also supports Saudi Arabia’s Vision 2030 initiative, which aims to diversify the Kingdom’s economy from its current reliance on the oil industry. Giri explains that the team’s research stems from the idea that some of the hydrocarbons derived from crude oil, which contain only hydrogen and carbon, can be replaced with compounds that also contain oxygen. As well as reducing particulate emissions, burning these oxygenated compounds generates less carbon dioxide, a greenhouse gas linked to global warming. Oxygenated compounds are known to burn with near-zero soot emissions if the oxygen content is in excess of 33 percent by mass. The oxygen content of glycerol carbonate is 59 percent.

combustion as a f u e l a d d i t i v e.” “Our results show that glycerol carbonate has great potential to promote cleaner combustion as a fuel additive,” Giri concludes. Investigating the chemistry of fuels under engine-relevant extreme temperatures and pressures is a significant challenge. The KAUST researchers are collaborating with Milan Szori and Bela Viskolcz at the University of Miskolc in Hungary, which “brings together two world-leading teams with the expertise to understand the complexities of combustion chemistry,” Giri explains. Further research aims to reveal how to optimize blending glycerol carbonate with conventional fuels. “This work has the potential to change the future directions of fuel-engine research,” says Aamir Farooq who supervised this work. He and his colleagues are excited about their possible contribution to developing cleaner and greener transportation systems worldwide. Szőri, M., Giri, B. R., Wang, Z., Dawood, A. E., Viskolcz, B. & Farooq, A. Glycerol carbonate as a fuel additive for a sustainable future. Sustainable Energy & Fuels 2, 2171-2178 (2018).


CLEANER FUELS

TURBOCHARGING THE SWITCH TO EFFICIENT ENGINES Predicting capricious pre-ignition combustion events could enable automakers to build powerful yet more efficient engines. A combustion problem that has troubled engine manufacturers since the industry began still plagues the designers of fuel-efficient engines. A team at KAUST’s Clean Combustion Research Center, led by Robert Dibble, focuses on understanding pre-ignition. A single occurrence of pre-ignition—when fuel ignites prematurely in the engine cylinder to release a mistimed kick of energy—can cause major damage. Pre-ignition is sporadic and unpredictable. It is more prevalent in the downsized, turbocharged engines that car makers are now producing for improved fuel efficiency while maintaining power output. “Pre-ignition remains a bottleneck to further downsizing and improving engine efficiency,” explains Eshan Singh, a Ph.D. student in Dibble’s team. The team systematically changed engine parameters, including air intake temperature and exhaust back-pressure, thereby establishing the conditions most likely for pre-ignition. For example, the team observed more pre-ignition events with a higher exhaust back pressure: This occurs when the emptying of burnt fuel by the engine cylinder between combustion cycles is ineffective. The researchers combined all their observations to build a picture of the pre-ignition phenomenon. “Our work unravels that the pre-ignition event is divided into several steps,” Singh says. A pre-ignition event originates during the previous fuel combustion cycle, when fuel injected into the cylinder interacts with oil on the cylinder liner forming an oil-fuel droplet. At the end of each combustion cycle, the oil-fuel droplet is usually expelled from the cylinder; however, from time to time, the oil-fuel droplet remains and may trigger pre-ignition in the following cycle, the team showed. “Car makers need to break this chain of events to suppress pre-ignition,” Singh says. Stopping the fuel and oil from interacting, or completely exhausting all the burnt mixture from the engine, are two points where the event chain can be broken. Preventing the injected fuel from hitting the oil-coated cylinder liner is one approach. “We are looking at practical methods to reduce pre-ignition by changing the injection strategy,” Singh says. “We have

filed a patent that uses injection strategies to provide pre-ignition-free operation without compromising engine power.” Other routes to break the chain of events demand deeper understanding of preignition, Singh adds. "Our exciting discoveries are gratifying. Going forward, we are motivated to uncover additional controlling variables that allow for more accurate preignition predictions."

ESHAN SINGH PH.D. STUDENT Due to graduate from KAUST in the fall of 2019, Eshan is currently interning at the Combustion Research Facility in Sandia National Laboratories, California. There, he is exploring mixed-mode combustion for future internal combustion engines.

Singh, E. & Dibble, R. Mechanism triggering pre-ignition in turbo-charged engines. SAE Technical Paper No. 2019-010255 (2019).

A fuel injector can be heavily damaged by a pre-ignition event (bottom); a new injector is shown above.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

11


KAUST & Dow Delivering technologies that contribute to sustainable innovation in the Kingdom.

Advancing science and innovation in response to the world’s most pressing challenges.

innovation.kaust.edu.sa


RED SEA SCIENCE

RED SEA TEMPERATURES TO COOL IN COMING DECADES Exploring the links between natural climate cycles and

States and the UK Met Office from the 1980s to the present with observations from NOAA’s Extended Reconstructed SST and the Met Office’s Hadley Centre Sea Ice and SST spanning a period of 140 years. They found that the high warming trend observed from 1980 to 2010 was dominated by a positive phase of the

the sea surface temperature of the Red Sea reveals a cooling trend during the next few decades. A natural climate cycle called the Atlantic multidecadal oscillation (AMO) strongly influences the surface temperature of the Red Sea, according to a study by KAUST, that will lead to a shift to a cooling period over the decades to come. The majority of E arth’s oceans warmed over the past century, with the rate of warming increasing over the last few decades. As one of warmest ocean basins on the planet, the Red Sea contains a diverse and fragile ecosystem that is particularly vulnerable to further warming. Most studies of the sea surface temperature (SST) of the Red Sea have covered short periods, where warming rates may have been slowed or intensified by natural, longer-term cycles.

“Fu t u r e s t u d i e s

How the AMO influences the SST of the Red Sea over the long term has now been investigated by George Krokos and colleagues, led by Ibrahim Hoteit. “The AMO is a permanent feature of the Earth’s climate system and is primarily associated with variations in the Atlantic conveyor belt,” explains Krokos. “It has been linked to important global climate impacts, such as the multidecadal variability of mean surface temperatures in the northern hemisphere.” To investigate the effects of the longer-term temporal variability of the AMO on the SST of the Red Sea, the team combined several sources of data. This included satellite data from the National Oceanic and Atmospheric Administration (NOAA) in the United

will increase our understanding of the Red S e a ’s r e s p o n s e to climate v a r i a b i l i t y.” natural SST oscillation; however, this is expected to shift to a cooling phase during the next few decades as the AMO enters a negative phase. “Our analysis showed that the effects of natural climatic variability, manifested in the Red Sea as low-frequency oscillations that follow the AMO, are dominant over the identified long-term trends,” says Krokos. The work shows that the long-term oscillation related to AMO has modulated the warming of the Red Sea over the past three decades and, having reached its peak over the past decade, is projected to enter a negative phase in the upcoming years. “Future studies using modern oceanographic tools, including atmospheric and ocean models, will increase our understanding of the Red Sea’s response to climate variability,” says Krokos. Krokos G., Papadopoulos, V. P., Sofianos, S. S., Ombao, E., Dybczak, P. & Hoteit, I. Natural climate oscillations may counteract Red Sea warming over the coming decades. Geophysical Research

Ibrahim Hoteit (left) and George Krokos discuss the results of their simulation.

Letters 46, 081397 (2019).

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

13


RED SEA SCIENCE

“MARINE SKIN” DIVES DEEPER FOR BETTER MONITORING A pioneering tagging system that monitors the movement and local environment of sea animals reaches deeper depths and higher sensitivities.

A new and greatly improved version of an electronic tag, called “Marine Skin”, used for monitoring marine animals could revolutionize our ability to study sea life and its natural environment, say KAUST researchers. Marine Skin is a thin, flexible, lightweight polymerbased material with integrated electronics that can track an animal’s movement and diving behavior and the health of the surrounding marine environment. Early versions of the sensors, reported previously1, proved their worth when glued onto the swimming crab, Portunus pelagicus. The latest and much more robust version can operate at unprecedented depths and can also be attached2 to an animal using a noninvasive bracelet or jacket. This can, when necessary, avoid the need for any glues that might harm an animal’s sensitive skin. “The system can now operate down to a depth of 2 kilometers, which has never been achieved before by anyone,” says Ph.D. student Sohail Shaikh of the KAUST team. The sensitivity of the monitoring electronics has

14

F a l l 2 0 19

also been enhanced by up to 15 times. The data collected reveals a tagged animal’s depth and the temperature and salinity of the surrounding water. Further development is planned to incorporate additional environmental sensing capabilities, such as measuring oxygen and carbon dioxide levels and precise geolocation tracking. Shaikh reports that a major challenge in developing the enhancements was to make the system sufficiently robust to tolerate operating at much greater depths. The researchers also managed to reduce the size down to half that of the previous version. Tests also showed improved performance, flexibility and durability when the skin was attached to different fish, including sea bass, sea bream and small goldfish. Lab tests in highly saline Red Sea water also demonstrated integrity through a full month’s immersion and 10,000 extreme bending cycles. "Marine Skin is a unique and groundbreaking innovation in wearable technology for marine animals,” says Muhammad Mustafa Hussain, whose


“T h e s y s t e m c a n now operate down to a depth of 2 kilometers, which has never been achieved before b y a n y o n e .”

research group has developed the system in collaboration with Carlos Duarte's group. Hussain adds that Marine Skin outperforms all existing alternatives, while emphasizing that ongoing development work will continue to enhance the sensing capacities, overall performance, reliability and affordability. 1. Nassar, J.M., Khan, S.M., Velling, S.J., Diaz-Gaxiola, A., Shaikh, S.F., Geraldi, N.R., Torres Sevilla, G.A., Duarte, C. & Hussain, M.M. Compliant lightweight non-invasive standalone “Marine Skin” tagging system, npj Flexible Electronics 2, 13 (2018). 2. Shaikh, S.F., Mazo-Mantilla, H.F., Qaiser, N., Khan, S.M., Nassar, J.M., Geraldi, N.R., Duarte, C.M. & Hussain, M.M. Non-invasive featherlight wearable compliant “Marine Skin”: Standalone multisensory system for deep-sea environmental monitoring. Small 15, 1804385 (2019).

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

15

ILLUSTRATION BY XAVIER PITA

RED SEA SCIENCE


RED SEA SCIENCE

MANGROVE FORESTS TRAP FLOATING LITTER Marine plastic pollution accumulates in mangrove forests and is a danger to sea life. Mangrove forests on the coasts of Saudi Arabia act as litter traps, accumulating plastic debris from the marine environment. The study offers an explanation for the fate of missing marine plastic litter and highlights the threat it poses to coastal ecosystems. “Of all the plastic discarded in the marine environment globally, only 1 percent is found floating in surface waters. That means that 99 percent of the plastic is elsewhere, but yet we don’t know where exactly,” says Cecilia Martin of KAUST’s Red Sea Research Center. In previous work, Martin and others in Carlos Duarte’s research group found relatively low levels of plastic litter in the Red Sea1. Next, to identify the location of this missing litter, the team used an unmanned aerial vehicle to scour the beaches2. Now, together with Hanan Almahasheer, Martin has surveyed litter in mangrove forests along the Red Sea and Arabian Gulf3. They recorded the type and location of the litter, as well the weight of some items and various features of the sites, such as the distance to the nearest coastal city and to intense marine traffic, the density of the trees, and how far the litter was from the In surveying the mangrove forests, the researchers found one litter item every square meter or two. The litter consisted mostly of small plastic items.

16

F a l l 2 0 19

shore or the sea. They found one litter item every square meter or two, with greater density along the Arabian Gulf than the Red Sea. Plastic made up more than 90 percent of the litter: the most common were small items, such as bottles, bags, lengths of rope and food wrappings. High-density groves had more litter, and the aerial roots acted like a sieve, capturing large plastic debris from the water and causing mangroves to accumulate more plastic than bare beaches. The team’s analysis showed that the density of debris depends on the distance to major maritime traffic routes rather than land-based factors, such as the distance to the nearest city. Traffic further from the coast caused less litter accumulation, but only up to a certain distance. Unexpectedly, litter density began to increase again once the traffic was more than 15 kilometers away, which the researchers propose is due to currents transporting the litter to mangroves. As well as harming the mangroves,

debris could be ingested by other marine organisms that use the forests as a nursery. Chemicals associated with the debris, such as industrial additives or pollutants absorbed by the plastic, could also damage these ecosystems. The team is now checking whether microplastics are building up in mangrove sediment, which could be a major sink for plastic marine pollution and explain the fate of some of the plastic that is not accounted for. These findings reinforce the need to reduce plastic consumption and properly deal with plastic waste in order to preserve these important ecosystems. 1. Martí, E., Martin, C., Cózar, A., & Duarte, C. M. Low abundance of plastic fragments in the surface waters of the Red Sea. Frontiers in Marine Science 4, 333 (2017). 2. Martin, C., Parkes, S., Zhang, Q., Zhang, X, McCabe, M.F. & Duarte, C. M. Use of unmanned aerial vehicles for efficient beach litter monitoring. Marine Pollution Bulletin 131, 662–673 (2018). 3. Martin, C., Almahasheer, H., Duarte, C.M. Mangrove forests as traps for marine litter. Environmental Pollution 247, 499-508 (2019). CECILIA MARTIN PH.D STUDENT In Carlos Duarte’s group, Cecilia is pursuing her degree by assessing the fate of plastic pollution in the Red Sea. She focuses on micro plastic abundance and interaction with ecosystems and biota.


WATER BRE A K THROUGHS

COLORFUL SOLUTION TO A CHEMICAL INDUSTRY BOTTLENECK A graphene-oxide membrane design inspired by nature swiftly separates solvent molecules.

The nanoscale water channels that nature has evolved to rapidly shuttle water molecules into and out of cells could inspire new materials to clean up chemical and pharmaceutical production. KAUST researchers have tailored the structure of graphene-oxide layers to mimic the hourglass shape of these biological channels, creating ultrathin membranes to rapidly separate chemical mixtures. “In making pharmaceuticals and other chemicals, separating mixtures of organic molecules is an essential and tedious task,” says Shaofei Wang, postdoctoral researcher in Suzana Nuñes lab at KAUST. One option to make these chemical separations faster and more efficient is through selectively permeable membranes, which feature tailored nanoscale channels that separate molecules by size. But these membranes typically suffer from a compromise known as the permeancerejection tradeoff. This means that narrow channels may effectively separate the different-sized molecules, but they also have an unacceptably low flow of solvent through the membrane, and vice versa— they flow fast enough but perform poorly at separation.

Nuñes, Wang and the team have taken inspiration from nature to overcome this limitation. Aquaporins have an hourglassshaped channel: wide at each end and narrow at the hydrophobic middle section. This structure combines high solvent permeance with high selectivity. Improving on nature, the team has created channels that widen and narrow in a synthetic membrane. The membrane is made from flakes of a two-dimensional carbon nanomaterial called graphene oxide. The flakes are combined into sheets several layers thick with graphene oxide. Organic solvent molecules are small enough to pass through the narrow channels between the flakes to cross the membrane, but organic molecules dissolved in the solvent are too large to take the same path. The molecules can therefore be separated from the solvent. To boost solvent flow without compromising selectivity, the team introduced spacers between the graphene-oxide layers to widen sections of the channel, mimicking the aquaporin structure. The spacers were formed by adding a silicon-based molecule into

the channels that, when treated with sodium hydroxide, reacted in situ to form silicon-dioxide nanoparticles. “The hydrophilic nanoparticles locally widen the interlayer channels to enhance the solvent permeance,” Wang explains. When the team tested the membrane’s performance with solutions of organic dyes, they found that it rejected at least 90 percent of dye molecules above a threshold size of 1.5 nanometers. Incorporating the nanoparticles enhanced solvent permeance 10-fold, without impairing selectivity. The team also found that there was enhanced membrane strength and longevity when chemical cross-links formed between the grapheneoxide sheets and the nanoparticles. “The next step will be to formulate the nanoparticle graphene-oxide material into hollow-fiber membranes suitable for industrial applications,” Nuñes says. Wang, S., Mahalingam, D., Sutisna, B. & Nunes, S.P. 2D-dual-spacing channel membranes for high performance organic solvent nanofiltration. Journal of Materials Chemistry 7, 11673-11682 (2019).

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

17


WATER BRE A K THROUGHS

CASTING A NET FOR A SUSTAINABLE WATER NETWORK Global water challenges addressed through improvements to desalination techniques and limits to waste.

Freshwater resources are becoming increasingly valuable and worryingly scarce as they come under pressure from climate change and global population growth. KAUST’s Water Desalination and Reuse Center (WDRC), led by Hans Vrouwenvelder, aims to find innovative solutions for the global challenge of a sustainable infrastructure for water security. Within this remit, Vrouwenvelder and his colleagues are optimizing existing technologies and developing novel water-treatment methods to transform the water cycle. The Center focusses on three flagship research themes: waste to resource, water security and greener desalination. “Several of our research projects focus on recovering water, energy, minerals and nutrients from wastewater that would

18

F a l l 2 0 19

otherwise be lost,” says Vrouwenvelder. For example, Peiying Hong and her team aim to demonstrate that by combining treatments, wastewater itself can be made into a safe, valuable water resource. Wastewater has considerable potential value as a water resource, especially in arid regions, like Saudi Arabia. “Our Kingdom partners, both in industry and government, are rapidly implementing the results of our Center’s work,” notes Vrouwenvelder. “They are requesting our assistance to anticipate upcoming changes in environmental legislation, too.” Because agriculture consumes around 80 percent of freshwater supplies in Saudi Arabia, largely taken from unsustainable underground aquifers, food production and food security in the Kingdom are intricately linked to water security. Matthew McCabe and his team are monitoring agricultural water use and collecting satellite, UAV and ground data to determine more efficient use of water and improve crop yields. There are two main methods of desalination: thermal desalination, which involves heating salt water until the fresh water evaporates and is collected, and membrane desalination, in which salt water is pushed through filters to separate the salt from water molecules. There are opportunities to make each method greener and more efficient, but both methods require significant energy, often fueled by oil. “All current desalination processes in the Kingdom use oil, directly or indirectly,” says Vrouwenvelder. “At present, there are limited opportunities for solar-powered desalination because the energy requirements are high and the technologies are difficult to scale up.” KAUST’s innovative start-up company MEDAD has taken steps to improve the efficiency of thermal desalination by creating a hybrid process. “MEDAD has added a stage at the end of the traditional process, which involves the adsorption of water molecules


WATER BRE A K THROUGHS

“T h i s m e a n s w e c a n wa t e r u s i n g t h e s a m e a m o u n t o f e n e rg y.” into a silica gel, lowering the condensation temperature” says Vrouwenvelder. “This means we can extract more fresh water using the same amount of energy.” Membrane desalination comes with its own added challenges. While salt is removed to yield desalinated water, the membrane through which it passes accumulates microorganisms that ultimately compromise the efficiency of the membrane and the quality of the desalinated water. This is Vrouwenvelder’s area of specialization. “We are tackling the growth of bacterial films, or biofilms, on desalination membranes, pipes and cooling towers because they affect cost, quality, reliability and safety of water across the network,” he explains. “At KAUST we have the benefit of testing our work on our desalination and wastewater treatment pilot plants,” says Vrouwenvelder. “Under the greener desalination theme, researchers are working to reduce associated environmental impacts by improving efficiency, reducing or eliminating the use of chemicals, recovering resources used during the process, and refining and optimizing membrane technologies.” However, all desalination processes can trigger further environmental problems. When leftover brine is pumped back into the ocean, it can lead to the development of highly saline, dense layers of water that rest

on the seafloor, impeding oxygen flux and killing marine life. “In addition, brines can contain concentrated nutrients and toxic chemicals that are added during the desalination process,” says Carlos Duarte, marine biologist at KAUST’s Red Sea Research Center. “WDRC researchers are seeking solutions to this—brines either need to be recycled, cleaned or diluted prior to being discharged.” To find solutions, the WDRC and The Red Sea Development Company, along with other partners in the Kingdom, launched the “Brains for Brine Challenge.” This challenge aims to bring the innovation required to provide potable water to a luxury resort being created by the Red Sea Development Company on an archipelago of islands in the northern section of the sea. “We called for fresh ideas and new perspectives from young scientists, engineers and innovators from different fields to address the challenge of managing brine in an environmentally sustainable manner,” says Martin Stahl at The Red Sea Development Company. “The selected proposals will not only eliminate discharge to the sea, but also provide a commercially viable, ecological solution for managing recovered salts.” “Central to the Red Sea Project is a commitment to set new standards in sustainable development and to protect, preserve and enhance the stunning natural environment that makes the destination so unique,” says Stahl. “One initiative is to achieve zero discharge, 100 percent wastewater reuse, and no brine discharge into the lagoon environment—this is where Brains for Brine will have a true impact.”

Saudi Arabia’s Red Sea Project will include more than 90 untouched islands, and environmental preservation of these pristine ecosystems is the cornerstone of the ambitious project.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

19

RED SEA DEVELOPMENT COMPANY

ex t ra c t m o r e f r e s h


WATER BRE A K THROUGHS

COMBINING FORCES TO COMBAT E. COLI PI-7

A combined treatment may help tackle the rise of E. coli PI-7 in Saudi Arabia’s wastewater systems.

Peiying Hong (left), Nada Al-Jassim (right) and colleagues seek answers to the growing problem of antibioticresistant bacteria found in wastewater.

Solar irradiation treatment, combined with naturally occurring viral components, could help destroy a highly antibiotic-resistant form of Escherichia coli (E. coli PI-7) that has been identified in Jeddah’s wastewater network. Antibiotic resistance continues to develop rapidly in bacteria, which have evolving genes that can confer resistance through generations. One of these genes is the New Delhi metallo-β-lactamase gene, blaNDM, which E. coli PI-7 carries. Crucially, blaNDM confers resistance to antibiotics that are considered to be our last line of defense against certain diseases. This gene, and others like it, are found in increasing numbers in wastewater systems, which is a real concern for public health. Bacteriophages are naturally occurring viruses that infiltrate and destroy bacteria by breaking down their cell membranes. As such, they

20

F a l l 2 0 19

represent a largely untapped source of agents that could help tackle antibiotic-resistant pathogens. “Our previous study showed that solar irradiation helped partly destroy E. coli PI-7, but not eradicate it completely1,” says Nada Al-Jassim, former student of Peiying Hong. “Under solar irradiation, PI-7 cells can enhance their cellular wall and DNA repair mechanisms to protect themselves and survive. However, the mechanisms preventing phage infection were downregulated, suggesting a weakness in the bacteria that we could exploit.” The team isolated seven bacteriophages from wastewater that specifically targeted PI-7—this was crucial to ensure the safety of the treatment in the wider environment. “We selected three phages that remained stable under solar irradiation long enough to act


WATER BRE A K THROUGHS on PI-7,” says Al-Jassim. “The phages slowed E. coli PI-7’s growth, downregulated cell wall functions and DNA repair, and limited the destruction of reactive oxygen species within the bacteria. In other words, the phages appear to actively exploit the bacteria’s own defense mechanisms2.” PI-7 cells treated with phages persisted for a shorter time under solar irradiation than untreated cells, suggesting that the phages increase PI-7’s sensitivity to sunlight. However, the team does not know how the combined treatment will fare in real wastewater treatment systems. For a start, coping with large volumes of water and maintaining phage concentrations under those conditions could prove challenging. More pressing, however, Hong explains, is that “Regardless of the disinfection strategies, these techniques will cause the bacterium to release antibiotic resistance genes into the environment, where they could be taken up by other bacteria. We are keen to examine this in greater depth.”

1.Al-Jassim, N., Mantilla-Calderon, D., Wang, T. & Hong, P.-Y. Inactivation of a virulent wastewater Escherichia coli and non-virulent commensal Escherichia coli DSM1103 strains and their gene expression upon solar irradiation. Environmental Science and Technology 51, 3649-3659 (2017). 2.Al-Jassim, N., Mantilla-Calderon, D., Scarascia, G. & Hong, P. Bacteriophages to sensitize a pathogenic New Delhi Metallo β-Lactamasepositive Escherichia coli to solar disinfection. Environmental Science and Technology 52, 14331–14341 (2018).

NADA AL-JASSIM ALUMNA Nada completed her Ph.D. under the supervision of Peiying Hong in 2017. Her Ph.D. work focused on mitigating antibiotic resistance threats arising from the use of treated wastewater. She is the first Ph.D. student to graduate through the KAUST Gifted Student Program.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

21


WATER BRE A K THROUGHS

A DRONE ON CROP RESOURCES Aerial imaging of plant heights could help farmers manage field crops more effectively to increase yields and preserve resources. Monitoring the growth patterns of crop plants provides farmers with a strong indication of potential yield, allowing them to tweak crop management to boost production. Researchers have now demonstrated that using fixed-wing unmanned aerial vehicles (UAVs) to collect data on vegetation height throughout the crop growth cycle provides a low-cost, simple way of monitoring plant health on a farm-wide scale. “Farmers routinely have problems identifying areas in their fields that need attention: They cannot see what is happening in a field of 2-meter-high corn that is 800 meters in diameter,” says KAUST Ph.D. student Matteo Ziliani, who worked on the project with colleagues under the supervision of Matthew McCabe. “Consequently, farmers often uniformly fertilize and water fields, regardless of which areas need more or less water. UAVs could

help achieve effective precision agriculture, saving money and resources and growing better quality crops.” Ziliani set out to show that the imaging data collected by UAVs is as useful as data retrieved by more expensive LiDAR laser scanning. To do this, the team took multiple images of the same 50-hectare maize field at different points throughout the growing season—one set of color images taken by UAV and the other by ground-based LiDAR. The researchers used each image set to build three-dimensional canopy models and to compare how closely the UAV images replicated the accurate LiDAR data. “We can create three-dimensional models of objects using sets of two-dimensional images taken from different angles in the same way as the human brain processes our binocular vision,” says Ziliani. The digital field maps generated using UAV data provided an accurate representation of

crop height over time, comparable to those created using LiDAR. Ziliani acknowledges that further work is needed, for example, the computational power required to process the data is currently too large to be processed by an average desktop computer. Also, when the plants were very young, the UAV struggled to pick up their structural details. This could be resolved by flying at lower levels, but this limits the area covered in the available flight time, which is in turn linked to battery life. Indeed, battery life will need to be improved if UAV technology is to meet its potential in monitoring crops across larger regions, searching for pockets of productive and unproductive land, or examining storm damage. Ziliani, M.G., Parkes, S.D., Hoteit, I. & McCabe, M.F. Intra-season crop height variability at commercial farm scales using a fixed-wing UAV. Remote Sensing 10, 2007 (2018).

A typical irrigation boom that heavily waters a 50-hectare field.

22

F a l l 2 0 19


WATER BRE A K THROUGHS complicated,” he says. To overcome the problem, the researchers incorporated the salt into a polymer called a hydrogel, which can hold a large volume of water while remaining a solid. They also added a small amount of carbon nanotubes, 0.42 percent by weight, to ensure that the captured water vapor could be released. Carbon nanotubes very efficiently absorb sunlight and convert the captured energy into heat.

Renyuan Li (left) and Peng Wang make a hydrogel in the lab.

“T h e h y d r o g e l’s most notable aspects are its high performance and low c o s t .”

DRINKING WATER SUCKED FROM THE DUSTY DESERT AIR An inexpensive hydrogel-based material efficiently

captures moisture even from low-humidity air and then releases it on demand. A simple device that can capture its own weight in water from fresh air and then release that water when warmed by sunlight could provide a secure new source of drinking water in remote arid regions, suggests new research from KAUST. Globally, Earth’s air contains almost 13 trillion tons of water, a vast renewable reservoir of clean drinking water. Trials of many materials and devices developed to tap this water source have shown each to be either too inefficient, expensive or complex for practical use. A prototype device developed by Peng Wang from the Water Desalination and Reuse Center and his team could finally change that.

At the heart of the device is the cheap, stable, nontoxic salt, calcium chloride. This deliquescent salt has such a high affinity for water that it will absorb so much vapor from the surrounding air that eventually a pool of liquid forms, says Renyuan Li, a Ph.D. student in Wang’s team. “The deliquescent salt can dissolve itself by absorbing moisture from air,” he says. Calcium chloride has great water-harvesting potential, but the fact that it turns from a solid to a salty liquid after absorbing water has been a major hurdle for its use as a water-capture device, says Li. “Systems that use liquid sorbents are very

The team incorporated 35 grams of this material into a simple prototype device. Left outside overnight, it captured 37 grams of water on a night when the relative humidity was around 60 percent. The following day, after 2.5 hours of natural sunlight irradiation, most of the sorbed water was released and collected inside the device. “The hydrogel’s most notable aspects are its high performance and low cost,” says Li. If the prototype were scaled up to produce 3 liters of water per day—the minimum water requirement for an adult—the material cost of the adsorbent hydrogel would be as low as half a cent per day. The next step will be to fine tune the absorbent hydrogel so that it releases harvested water continuously rather than in batches, Wang says. Li, R., Shi, Y., Alsaedi, M., Wu, M., Shi, L. & Wang, P. Hybrid hydrogel with high water vapor harvesting capacity for deployable solar-driven atmospheric water generator. Environmental Science and Technology 52, 11367-11377 (2018).

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

23


KAUST & Aramco

Collaborating on innovative and original research to advance technology in Saudi Arabia and the world.

Putting energy to work and enabling opportunities that revolutionize the world. innovation.kaust.edu.sa


TESTING THE BLIND SPOTS IN ARTIFICIAL INTELLIGENCE Understanding the situations in which artificial intelligence can fail is critical for future autonomous vehicle and medical diagnostics applications.

applications . It is much more likely that semantic or contextual scenarios occur that the networks have not been trained on.” For example, rather than being faced with a perturbed image of an object, an autonomous driving system is more likely to be faced with an object orientation or a lighting scenario that it has not learned or encountered before; as a result, the system may not recognize what could be another vehicle or nearby pedestrian. This is the type of failure that has occurred in a number of high-profile incidents involving self-driving vehicles. “Our approach involves training another deep network to challenge the deep learning model. This will allow us to discover the weaknesses and strengths of the AI model that will eventually be used in applications,” says Hamdi.

“A I a n d d e e p Deep-trained artificial intelligence models can make mistakes if they encounter scenes that they do not recognize, such as an object in an orientation, color or lighting that conflicts with the datasets used to train the model.

By investigating the robustness of deep learning models using a context-based approach, KAUST researchers have developed a means to predict situations in which artificial intelligence might fail. Artificial intelligence (AI) is becoming increasingly common as a technology that helps automated systems make better and more adaptive decisions. AI is an algorithm that allows a system to learn from its environment and available inputs. In advanced applications, such as self-driving cars, AI is trained using an approach called deep learning, which relies solely on large volumes of sensor data without human involvement. However, these machine learning systems can fail when the training data misleads the decision-making process. In Bernard Ghanem’s team, Abdullah Hamdi and Matthias Müller have been researching the limitations of AI and deep neural networks for safety-critical applications, such as self-driving cars. “AI and deep learning are very powerful, but the technology can fail in rare edge cases, which are likely to eventually happen in real-world scenarios when these deep learning models are used in our daily

lives,” explains Hamdi. For many applications, an AI failure is merely an inconvenience as long as the AI works as expected most of the time. However, for applications like self-driving cars and medical diagnosis, in which failure can result in death or catastrophe, repeated unexpected failures are not tolerated. “Deep learning is successful in a wide range of tasks but no one knows exactly why it works and when it fails,” says Hamdi. “Because deep learning models will be used more and more in the future, we should not put blind faith in those that are not well understood and that can put lives at risk.” The team has shown how these very powerful AI tools can fail badly in seemingly trivial scenarios. They have developed a method to analyze these scenarios and establish risk maps that identify these risks in real applications. “AI models are commonly tested for robustness using pixel perturbations, that is, adding noise to an image to try to fool the deep-trained network,” says Hamdi. “However, artificial pixel perturbations are actually unlikely to occur in real-life

learning are very p owe r f u l , b u t the technology c a n f a i l i n ra r e e d g e c a s e s .” The researchers expect their AI approach to contribute to the development of robust AI models and the verification of deep learning technologies before they are applied. 1. Hamdi, A. & Ghanem, B. Towards analyzing semantic robustness of deep neural networks. Preprint arXiv:1904.04621 (2019). 2. Hamdi, A., Müller, M. & Ghanem, B. SADA: Semantic adversarial diagnostic attacks for autonomous applications. Preprint arXiv:1812.02132 (2018).

ABDULLAH HAMDI PH.D. STUDENT In Bernard Ghanem’s research group, Abdullah is investigating deep learning for activity recognition, circulant structure matrices for tracking and correlation filters.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

25

METAMORWORKS/ SHUTTERSTOCK.COM

SMART TECH


SMART TECH

TINY, FAST, ACCURATE TECHNOLOGY ON THE RADAR A tiny, portable radar device could allow visually impaired people, or unmanned moving devices, to detect objects in real time.

Radar technology has been used for decades in aviation, defense and speed-camera technology. Now, a team at KAUST, in collaboration with scientists at the VTT Technical Research Center of Finland, have created a compact, low-cost radar with potential applications in healthcare and personal security. Radar provides detailed information about the size, distance and speed of moving objects. However, for closerange applications, the transmitted radio waves must have short wavelengths to pick up as much detail as possible about their immediate environment. Such sensors could help visually impaired people, and unmanned moving devices, to “see” by translating radar reflections into useful information. “Current radar modules are large and bulky. They also miss key details because they operate using long radio wavelengths,” says Seifallah Jardak, who worked on the project under the supervision of Sajid Ahmed and Mohamed-Slim Alouini from KAUST along with Tero Kiuru and Mikko Metso from VTT. “We wanted to develop a low-power, portable radar. Colleagues at VTT brought the necessary experience in millimeter-wave and hardware design, while I focused on the signal processing side and developed modular radar software,” explains Jardak. The earliest prototype performed a single scan every 2 seconds, making it difficult to acquire enough input data. Jardak optimized the signal processing modules and improved the performance to eight scans per second, providing better real-time monitoring. The device design incorporates a frequency-modulated continuous wave radar. This means that the radar produces

26

F a l l 2 0 19

“To l i m i t t h e size of our system, we chose an operating frequency of 2 4 g i g a h e r t z .”

continuous pulses of millimeter-wavelength radio waves that have a frequency that varies during each pulse. The small wavelength means that the time taken for pulses to reach an object and reflect back, and therefore the distance to the object, are calculated accurately. “To limit the size of our system, we chose an operating frequency of 24 gigahertz. This enabled us to reduce the size of the microstrip antenna,” says Jardak. “Our design also has one transmitting and two receiving antennae, meaning it can better estimate the angular location of a target.” The device fits into a 10-centimeter box, weighs less than 150 grams and is powered by a 5 volt battery. Initial trials suggest that the device is capable of target detection, speed estimation and tracking at ranges of up to 12 meters. The team even used it to detect whether a person was breathing when sitting in a chair. “Our prototype may also be useful for unmanned robotic and quadcopter applications where a collision avoidance system is required,” adds Jardak. Jardak, S., Kiuru, T., Metso, M., S. Ahmed, S. & Alouini, M.-S. Low-cost compact FMCW radar: Hardware and software implementation. IEEE Aerospace & Electronic Systems Magazine 34, 36-44 (2019).

The new 5-volt device is compact: it fits into a 10-centimeter box, weighs less than 150 grams and is powered by a 5-volt battery.


Fuad Jamour, Panos Kalnis and Yanzhao Chen (left to right) are building systems and algorithms for processing and analyzing very large datasets.

QUERYING BIG DATA JUST GOT UNIVERSAL A universal query engine for big data that works across

“T h e f i r s t research graphquery engine

computing platforms could accelerate analytics research. To solve one of the key obstacles in big data science, KAUST researchers have created a framework for searching very large datasets that runs easily on different computing architectures. Their achievement allows researchers to concentrate on advancing the query engine itself, rather than painstakingly coding for specific computing platforms. Big data is one of the most promising, yet challenging aspects of today’s information-heavy world. While the huge and ever-expanding sets of information, such as online-collected data or genetic information, could hold powerful insights for science and humanity, processing and interrogating all this data require highly sophisticated techniques. Many different approaches to querying big data have been explored. But one of the most powerful and computationally effective is based on analyzing data with a subject-predicate-object “triplestore” structure of the form (e.g., apple, is a, fruit). This structure lends itself to being treated like a graph with edges and vertices, and this characteristic has been used to code query engines for specific computing architectures for maximum efficiency. However, such architecture-specific approaches cannot be readily ported to different platforms, limiting the opportunities for innovation

and advancement in analytics. “Modern computing systems provide diverse platforms and accelerators, and programming them can be intimidating and time consuming,” say Fuad Jamour and Yanzhao Chen, Ph.D. candidates in Panos Kalnis’s group in KAUST’s Extreme Computing Research Center. “Our research group focuses on building systems and algorithms for processing and analyzing very large datasets. This research addresses the desire to write a program once and then use it across different platforms.” Rather than the previously used graphtraversal or exhaustive-relational indexing approaches, the group queried triplestore data by using an applied mathematical approach called sparse matrix algebra. “Our paper describes the first research graph-query engine with matrix algebra at its core to address the issue of portability,” says Jamour. “Most existing graph-query engines are designed for single computers or small distributed-memory systems. And porting existing engines to large distributed-memory systems, like supercomputers, involves significant engineering effort. Our sparse matrix algebra scheme can be used to build scalable, portable and efficient graph-query engines.” The team’s experiments on large-scale

with matrix algebra at its core to address the issue of p o r t a b i l i t y.” real and synthetic datasets achieved performance comparable with, or better than, existing specialized approaches for complex queries. Their scheme also has the capacity to scale up to very large computing infrastructures handling datasets of up to 512 billion triples. “These ideas can facilitate building analytics components in graph databases with cutting-edge performance, which is currently in high demand,” says Chen. Jamour, F., Chen, Y., Abdelaziz, I. & Kalnis, P. Matrix algebra framework for portable, scalable and efficient query engines for RDF graphs. Fourteenth EuroSysConference 2019 (EuroSys ’19), March 25–28, 2019, Dresden, Germany. ACM, New York, NY, USA.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

27


SMART TECH

PROBING WATER’S SKIN

Electrosprays of water cannot reliably probe the air-water interface.

First author of the study, Adair Gallo, uses a high-voltage device to create electrosprays of water.

From the wind-whipped surface of the open ocean to trillions of tiny water drops in clouds, the air-water interface—water’s skin— is the site for crucial natural processes, including ocean-atmosphere exchange and cloud acidification. The air-water interface has even been postulated as the cradle of life. However, factors, such as its subnanometer size and dynamic nature, render the investigation of interfacial water a daunting task. In recent years, researchers have investigated the air-water interface using electrosprays of water: fine sprays produced by applying greater than 5000 volts to water solutions passing through a metallic capillary. Traditionally, electrosprays have been used to study ions in the gas phase. Using electrosprays, researchers have suggested that the surface of mildly acidic water behaves as a highly reactive superacid. But the debate remains whether the air-water interface at mild pH can really behave as a superacid. Motivated to untangle this mystery, a team of KAUST researchers led by Himanshu Mishra has employed complementary techniques to disentangle purely interfacial effects from electrospray-specific effects. The researchers investigated the reactivity of isoprene—a volatile molecule released by plants experiencing heat stress—at the

28

F a l l 2 0 19


water interface. “We compared two scenarios: electrosprays of water interacting with isoprene gas and vigorously stirred mixtures of water and isoprene in closed vials,” explains Adair Gallo Jr., a Ph.D. student in Mishra’s team and lead author of the study. The team looked for the formation of short isoprene chains called oligomers. “Intriguingly, isoprene spontaneously formed oligomers in electrosprays under acidic and basic conditions, and even in the absence of water,” Gallo says. No oligomerization products were detected when acidified water was vigorously stirred with isoprene for hours. But when the same organic phase was electrosprayed, oligomers formed. The findings unequivocally establish that the oligomerization took place exclusively in electrosprays. Computer simulations carried out by Adriano Sanchez, a postdoctoral scholar in Mishra’s team, gave molecular-scale insights into the results. “We found that oligomer formation was only possible on

gas-phase clusters comprising not more than three water molecules and an excess proton,” Sanchez says. Collectively, the team’s results demonstrated that electrosprays present highly energetic gas-phase pathways for chemical reactions to occur that are impossible at natural air-water interfaces. “Electrosprays should therefore be complemented with surface-specific techniques and computer simulations to avoid incorrect conclusions when studying interfacial processes,” Mishra says. “I have been thinking about this problem for over six years, and now, thanks to my team, we have managed to disentangle purely interfacial effects from artifacts” says Mishra. Gallo Jr, A., Farinha, A. S. F., Dinis, M., Emwas, A.H., Santana, A., Neilsen, R.J., Goddard, W.A. & Mishra, H. The chemical reactions in electrosprays of water do not always correspond to those at the pristine air– water interface. Chemical Science 10, 2566-2577 (2019).

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

29

ILLUSTRATION BY IVAN GROMICHO.

SMART TECH


SMART TECH

BRIGHTER BLUE LIGHTS FOR FASTER COMMUNICATION High-performance blue-light-emitting diodes could boost white-light, high-speed data transmission.

Ultrabright diodes could bring white-light-mediated data transmission up to speed: KAUST researchers have developed high-performance, blue-light-emitting diodes useful for both solid-state lighting and visiblelight communication. These superluminescent diodes (SLDs) could become primary light sources and make data communication available at the flick of a light

30

F a l l 2 0 19

switch, such as on the internet and in audio and video entertainment. Light-emitting diodes have superior durability and ability to produce white light with minimal heat generation and energy loss compared to incandescent lightbulbs. They are becoming ubiquitous in everyday lighting applications ranging from general lighting to automotive headlamps and medical devices. They have also shown potential for wireless data transmission, but problems in achieving the required high-speed transmission rates have stalled implementation efforts and prompted researchers to turn to laser diodes. But, despite their higher efficiencies and data transmission rates, laserbased devices present the risk of eye injuries and a high speckle concentration, which are unwanted in solid-state lighting. To address these shortcomings, a team led by Boon Ooi has now created SLDs on galliumnitride semiconductor structures. “These diodes bring the advantages of light-emitting and laser diodes without showing any of their drawbacks,� explain Ph.D. students Abdullah Alatawi and Jorge

Ph.D. student Jorge Holguin-Lerma (left) and Abdullah Alatawi investigate the optical properties of superluminescent diodes.


SMART TECH Holguin-Lerma. The devices achieve exceptional light output power of 474 milliwatts and a data rate of 1.45 gigabits per second. At the heart of the SLDs, the researchers used an indium-gallium-nitride semiconductor layer, which becomes optically active under applied electric voltage and generates light via a process combining spontaneous and stimulated emission. They sandwiched

1,000+

Students from over 74 countries

“T h e s e d i o d e s b r i n g the advantages of light-emitting and laser diodes without showing any of their d r a w b a c k s .” this layer between several positive and negative semiconductor layers. The active layer presents multiple quantum wells—potential wells narrowly split into numerous discrete energy levels, giving rise to the high number of states and, consequently, broad spectral emission of the diode when electrically produced electrons and holes recombine to create light. When coupled with a commercially available luminescent plate, the blue SLD produces intense white light with a color rendering index of 85.1 and a correlated color temperature of 3392 Kelvin. This is consistent with its ability to accurately reproduce all colors present in natural white light and its suitability for solid-state lighting. “Altogether, these results prove the great potential of SLDs,” says Holguin-Lerma. These blue SLDs could directly replace light emitters in current technologies, such as optical coherence tomography (a noninvasive imaging retinal test), picoprojection (tiny projectors) and sensing. The researchers are now developing SLDs displaying different emission colors and greater performance. Alatawi, A.A., Holguin-Lerma, J.A., Kang, C.H., Shen, C., Subedi, R.C., Albadri, M.A., Alyamani, A.Y., Ng, T.K. & Ooi, B.S. High-power blue superluminescent diode

Join an international community dedicated to science & technology innovation.

APPLY NOW!

kaust.edu.sa

for high CRI lighting and high-speed visible light communication. Optics Express 26, 26355 (2018).

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

31


SMART TECH

SENSITIVE ROBOTS FEEL THE STRAIN Flexible skin for soft robots, embedded with electrical nanowires, combines conductivity with sensitivity within the same material.

An artificial soft skin imbued with flexible electronics could enhance the way robots sense and interact with their surroundings, KAUST researchers have shown. The team has discovered how to program electrical conductivity and strain sensing into a single material embedded in a stretchy polymer skin. The discovery could also have applications in wearable electronic devices. When an animal stretches a limb, a network of nerves and sensors within the skin provides feedback that help it orient the limb in space and interact with its surroundings. Embedding a network of strain sensors and connective wiring into a flexible artificial skin would give soft robots similar sensory feedback, helping them autonomously navigate their environment, says Gilles Lubineau, who led the research. Until now, researchers have used different materials for the sensing and conductive wiring components, adding cost and complexity to the fabrication process, explains Ragesh Chellattoan, a Ph.D. student in Lubineau’s team. “Our objective is to get both sensing and wiring connectivity

32

F a l l 2 0 19

in the one material,” he says. The team developed an artificial material comprising a flexible polymer embedded with silver nanowires. Individually, each nanowire is conductive, but high resistance at the junctions between them limits overall conductivity through the material. The resistance increases markedly when the material is flexed and the nanowires are pulled apart such that the nanowire network acts like a strain sensor. But that behavior can be altered, the team showed. Applying a DC voltage made the nanowire network very hot at the points of high resistance, where the nanowires meet. This localized heating acts to weld neighboring nanowires together, forming a highly conductive, firmly bonded network that is impervious to stretching and flexing. “Electrical welding joins thousands of junctions in the network within 30 seconds,” Chellattoan says. Changing how the current is introduced controls which parts become conductive. The researchers created a stretchy skin for a


SMART TECH Researchers coated the leg of an action figure in a strainsensing artificial skin to demonstrate the material in action.

“E l e c t r i c a l welding joins thousands of junctions in the network within 3 0 s e c o n d s .” toy action figure to demonstrate their material. They coated one of the figure’s legs with the artificial skin and then applied DC voltage only to the leg’s left side before flexing the leg at the knee and observing what happened. On the right side, the nanowire network acted as a strain sensor that could detect leg position as the figure’s knee was bent and straightened; the left side showed high conductivity regardless of leg position. The next step, Chellattoan says, is to gain greater control over where nanowire welds form. This would enable researchers to draw precise conductive patterns into the artificial skin. Chellattoan, R., Lube, V. & Lubineau, G. Toward programmable materials for wearable electronics: Electrical welding turns sensors into conductors. Advanced Electronic Materials 5, 1800273 (2019).

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

33


SMART TECH

Youngjiu Lei (left) and Husam Alshareef demonstrate the ability of their patch to detect biomarkers in sweat.

ARMBANDS DO A HEALTH CHECK WHILE YOU WORK OUT Nanotech-powered electrodes help solve the challenges of using sweat to assess biological conditions in real time.

Next-generation fitness sensors could give deeper insights into human health through noninvasive testing of bodily fluids. Researchers at KAUST have used a graphene-like material called MXene to develop a stretchy patch that makes it easier to analyze sweat for critical biomarkers. Human perspiration contains trace amounts of organic molecules that can act as measurable health indicators—glucose fluctuations, for example, may point to blood sugar problems, while high levels of

34

F a l l 2 0 19

lactic acid could signal oxygen deficiencies. To detect these molecules, researchers are developing flexible prototypes that sit on the skin and direct sweat toward special enzyme-coated electrodes. The specific nature of enzyme-substrate binding enables these sensors to electrically detect very low concentrations of target compounds. One obstacle with enzyme biosensors, however, is their relatively short lifetimes. “Even though human skin is quite soft, it can delaminate the enzyme layer right off the biosensor,” says Ph.D. student Yongjiu Lei. Lei and his colleagues in Husam Alshareef ’s group have now developed a wearable system that can handle the rigors of skin contact and deliver improved biomarker detection. Their device runs on a thin, flat ceramic known as MXene, which resembles graphene but contains a mixture of carbon and titanium atoms. The metallic conductivity, hydrophilic nature and low toxicity of this 2D material make it an ideal platform for enzyme sensors, according to recent studies. The team attached tiny dye nanoparticles to MXene flakes to boost its sensitivity to hydrogen peroxide, the main by-product of enzyme-catalyzed reactions in sweat. Then, they encapsulated the flakes in mechanically tough carbon nanotube fibers and transferred the composite onto a membrane designed to draw


ILLUSTRATION BY XAVIER PITA.

SMART TECH

1,762+ Alumni

The team has developed a light-weight and stretchy patch that can analyze sweat for critical biomarkers during a workout.

sweat through without pooling. A final coating of glucose or lactose-oxidase enzymes completed the electrode assembly. The new electrodes could be repeatedly swapped in or out of a stretchy polymer patch that both absorbs sweat and transmits the measured signals of hydrogen peroxide to an external source, such as a smartphone. When the team placed the biosensor into an armband

“P a t c h t h a t b o t h absorbs sweat and transmits the measured signals of hydrogen peroxide to an e x t e r n a l s o u r c e .” worn by volunteers riding stationary bicycles, they saw lactose concentrations in sweat rise and fall in correlation with workout intensities. “We are working with KAUST and international collaborators under the umbrella of the Sensors Initiative to integrate tiny electrical generators into the patch,” says Alshareef, who led the project. “This will enable the patch to create its own power for personalized health monitoring.” Lei, Y., Zhao, W., Zhang, Y., Jiang, Q., He, J.-H., Baeumner, A.J., Wolfbeis, O., Wang, Z.L., Salama, K.N., Alshareef, H.N. A MXene-based wearable biosensor

Sow the seeds of your future at KAUST.

APPLY NOW!

kaust.edu.sa

system for high-performance in vitro perspiration analysis. Small 15, 1901190 (2019).​

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

35


GET SOCIAL WITH KAUST

follow us on social media

(screen shot of kaust other social media page)

KAUSTofficial @kaust_news kaustedu kaustofficial KAUST

discovery.kaust.edu.sa


ILLUSTRATION BY XAVIER PITA.

HARNESSING THE AI WAVE

This illustration depicts the process by which a deeplearning model is trained on a protein database to predict RNA-binding specificity on a protein of interest. This information tells researchers about the function of the protein and the RNA.


FE AT URE

A perfect storm is giving rise to a surge in artificial intelligence research and KAUST aims to be at its forefront. The quest to design machines with humanlike intelligence has a long history. The term artificial intelligence (AI) was coined in the 1950s, but realizing its expected potential has not been straightforward. Scientists describe two AI winters, in the 1970s and 90s, when commercial and scientific activities in the field declined significantly. Then, from 2010, a new subset of AI, called machine learning, and an even more specialized field, called deep learning, burst onto the research scene, bringing AI closer to the realm of almost limitless possibility. “It’s like a perfect AI storm,” explains KAUST president, Tony Chan. “We have very inexpensive and universally available data-collection devices, like cell phones. We also have fast, powerful hardware and very clever researchers with interesting models and fast algorithms. When everything works together, and you get a killer app, like face recognition, you suddenly see results.” Facial recognition, as when social media websites suggest you tag a person detected in an image you’ve uploaded, is just one example of what can be achieved by training machine-learning algorithms. KAUST researchers are already involved in many aspects of theoretical and applied

38

F a l l 2 0 19

artificial intelligence research. Chan wants to see even more. Theoretical AI A good place to start is with experienced AI researchers. Peter Richtarik joined KAUST in 2017, the same year Google released its federated learning platform, a paradigm-shifting approach to machine learning. Richtarik’s group worked with colleagues at Google to design the platform, which enables developers to design apps that can learn from the private data stored on mobile devices without it having to be uploaded to the cloud. Richtarik is now using his experience to design new and improved algorithms that can train machine-learning and deep-learning models. “I work on the theoretical foundations of machine learning, but my work has many applications,” he explains. One of Richtarik’s 50 active machinelearning-related research projects focuses on distributed machine learning with compression. Training machine-learning models requires huge amounts of data located on tens to even thousands of computers. Communication among them can be very slow, acting as a key bottleneck in the training process. Richtarik and his team develop new training algorithms that work, even when part of the data is discarded, to reduce the amount communicated, a process called lossy compression. “All of the algorithms I design also use randomization. The methods flip coins at every step to make fast and simple decisions during training. This enables them

to learn much faster,” explains Richtarik. “Without randomization, it wouldn’t be possible to have deep learning, machine learning or artificial intelligence—without it we couldn’t train the huge models now used in industry. KAUST colleague Marco Canini works to improve the efficiency of processing huge amounts of complex data through thousands of computers. He and his team collaborated with academic and commercial partners to design a system to speed up the processing per second of training samples by a machine-learning model by 300 percent. He is also studying machinelearning models to develop a tool that can help users understand what they do and how to solve problems when they arise. Training more sophisticated machinelearning models with larger amounts of data is a major trend in the current era of deep learning. Yet, the lack of thorough characterization of the uncertainty of these models limits this technology from becoming more ubiquitous and applicable in life-critical decision-making applications, such as self-driving cars, self-flying drones/planes and automated medical diagnosis. Many deep models can reach very high average performance— super-human in some cases—in many challenging AI tasks, yet they fail in very trivial scenarios: scenarios so trivial that the average human would rarely get it wrong. Such scenarios could be behind the mishaps and fatal accidents that have occurred recently. Bernard Ghanem and his team at


KAUST are developing the theoretical tools and frameworks to predict and assess these failure cases, thus avoiding the usual need to run exhaustive, expensive, and time-consuming tests. This research provides robustness certificates to deeplearning models. These can serve as verifiable guarantees of performance. They can also be automated and cheap mechanisms to expose scenarios where such a model fails, such as specific weather/road/lighting conditions for a self-driving car or noise level in a biomedical image. This line of work will make deep-learning models more robust, their failure cases more predictable, and their deployment in the field less prone to accidents. Finding the top picks for you Xiangliang Zhang worked on her first artificial intelligence problem as an undergraduate student at Xi’an Jiaotong University in China. Her aim was to determine who was using a computer based on the way they type on a keyboard: was it the computer’s owner or someone who stole the password? Her focus has broadened since then, and now at KAUST she is developing algorithms that can identify patterns in complicated, streaming, largescale datasets. For example, she and her team are trying to teach machines to read and interpret the nuanced language use— known as natural language processing— in people’s social media posts to identify their evolving interests and recommend new places to visit or people to meet. They are also using mobile phone data to track and understand the underlying patterns

in people’s movements to improve recommendations for getting to know people with similar interests, habits or preferences. Zhang’s team is also developing programs that can recommend relevant research studies for scientists to incorporate in their work. Solving biological problems with AI Innovation often happens at the margins of technology, and the research of Xin Gao and Jesper Tegnér lies at the intersection between computer science and biology. In addition to his work on developing machine-learning theories, methodologies and algorithms, Gao applies machinelearning techniques to solve problems in biology and biomedicine. His group, for example, designed a computational approach that integrates data on existing FDA-approved drugs with information on their known protein targets to help researchers discover new disease targets for these drugs—drug repositioning. It could save pharmaceutical companies years and billions of dollars in drug development, he says. Gao’s group also developed a model that can forecast the clinical problems someone might face based on their medical history and genotype. Additionally, the group has designed a collection of deep-learning models that can decipher the hidden information in the DNA code that regulates gene expression. Jesper Tegnér and his team are designing techniques that decode the molecular programs found in cells and living systems with the ultimate aim of not only

identifying patterns in these complicated datasets but also of deciphering the causal rules that generate these patterns. This work could lead to the development of new tools with generic applications for designing intelligent machines. Tegnér and his team have proven a mathematical proof of principle for their approach. They are now tailoring it to single-cell genomics analysis, as well as extending it to design truly intelligent machines that can detect causes and ask reasoning questions, like “what if?” Such questions may lead to ethical concerns about AI. For example, would you trust an AI doctor to diagnose you? What about the element of common sense, which is crucial when it comes to human health? Can AI systems learn a physician’s common sense? Regarding such questions, Chan warns, “As a society we have to be very careful to balance between fostering innovation and over-regulating.” Where does KAUST fit in the future of AI? “As a relatively small university of science and technology we are mapping out what KAUST is in the best position to do. We want to focus our efforts on using AI and machine learning to help us do better science and engineering. Our plan is to use AI and machine learning to move Saudi Arabia beyond oil and into innovation and technology,” adds Chan. “For example, our partnerships with large local technological companies, like Saudi Aramco, are likely to drive our focus. For certain, we need to be strategic about what we do to contribute.”

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

39

DMITRIYRAZINKOV/ SHUTTERSTOCK.COM

FE AT URE


RECOGNIZING PARASITIC PLANTS

OPTIMIZING ENGINES

Machine learning methods can help to quantify the effects of chemicals that act to stimulate or inhibit the germination of parasitic seeds.

Applications of machine learning can be used to predict fuel properties and optimize engine design for greener vehicles.

germinating striga seeds

INTERPRETING DATA FROM SENSORS Application of AI tools to data collected from wearable sensors could remove the need for human intervention in response to some health conditions.

CLASSIFYING FISH SPECIES Hours of scientists’ time can be saved by using machine learning tools to identify coral and fish species.

DETECTING LARGE-SCALE ACTIVITY AI for tools can be used to detect human activity and for recognition. This information is valuable for evaluating risk in crowded situations. crossing road

Chaetodon semilarvatus

speaking on the phone


ESTIMATING CROP FEATURES

THE RISE OF AI

Machine learning and AI can be applied to image analysis, for automatic feature extraction, and to classification, to detect changes in land-cover use.

Forecasts suggest that by 2020 each person will have more than six network-connected devices. There is already a flood of data entering a central data-storage system at essentially no cost. Meanwhile, supercomputers are becoming more powerful and better equipped to handle this ever-increasing data at lower computational cost. Cheap and available data combined with super-powerful processing units explain the current rise of AI. Artificial Inteligence Programs that can sense, reason, act and adapt

Machine Learning

Deep Learning

Algorithms improve their performance with training

Multilayered neural networks that learn from vast amounts of data

AI

ML

ANALYZING FLUID SIMULATION DL AI application to computational fluid dynamics provides more accurate aerodynamic simulations and facilitates the exploitation of the full capabilities of current- and next-generation supercomputers.

1950

1960

1970

1980

1990

2000

2010

2020

Large-scale data Tools

Performance and efficency

Mathematical foundation Hardware/Software

UNDERSTANDING THE UNDERGROUND Deep-learning tools are useful for seismic imaging and inversion techniques, which are helpful for detecting features underground.

Aplication and adoption by industry

THE FUTURE OF AI Supply and demand have always driven product development, and the expansion of AI will be no different. Its development will be driven by industry powerhouses that will benefit from more targeted efforts and reduced losses. It will also be pushed by consumer demand for programs and devices that improve the quality of life. From targeted healthcare to appliances that respond from remote locations, AI promises to make life easier and more comfortable.

Xavier Pita, Research Publications Services (RPS), KAUST 3D models: NASA, CADNAV, Cinema4D Sources, RPS Sources: Bernard Ghanem


CAUSAL DISENTANGLEMENT IS THE NEXT FRONTIER IN AI Recreating the human mind’s ability to infer patterns and relationships from complex events could lead to a universal model of artificial intelligence. A major challenge for artificial intelligence (AI) is having the ability to see past superficial phenomena to guess at the underlying causal processes. However, new research has yielded a novel approach that moves beyond superficial pattern detection. Humans have an extraordinarily refined sense of intuition or inference that gives us the insight, for example, to understand that a purple apple could be a red apple illuminated with blue light. This sense is so highly developed in humans that we are also inclined to see patterns and relationships where none exist, giving rise to our propensity for superstition. This type of insight is such a challenge to codify in AI that researchers are still working out where to start; however, it represents one of the most fundamental differences between natural and machine thought. Five years ago, a collaboration between KAUSTaffiliated researchers Hector Zenil and Jesper Tegnér, along with Narsis Kiani and Allan Zea from Sweden‘s Karolinska Institutet, began adapting algorithmic information theory to network and systems biology in order to address fundamental problems in genomics and molecular circuits. That collaboration led to the development of an algorithmic approach to inferring causal processes that could form the basis of a universal model of AI. “Machine learning and AI are becoming ubiquitous in industry, science and society,” says Tegnér. “Despite recent progress, we are still far from achieving general-purpose machine intelligence with the capacity for reasoning and learning across different tasks. Part of the challenge is to move beyond superficial pattern detection toward techniques enabling the discovery of the underlying causal mechanisms producing the patterns.” This causal disentanglement, however, becomes very challenging when several different processes are intertwined, as is often the case in molecular and genomic data. “Our work identifies the parts of the data that are causally related, taking out the spurious correlations, and then identifies the different causal

42

F a l l 2 0 19

Using algorithmic information theory, KAUST researchers have developed an approach for inferring the causal processes that give rise to a complex observed interaction.

mechanisms involved in producing the observed data,” says Tegnér. The method is based on a well-defined mathematical concept of algorithmic information probability as the basis for an optimal inference machine. The main difference from previous approaches, however, is the switch from an observer-centric view of the problem to an objective analysis of the phenomena based on deviations from randomness. “We use algorithmic complexity to isolate several interacting programs and then search for the set of programs that could generate the observations,” says Tegnér. The team demonstrated their method by applying it to the interacting outputs of multiple computer codes. The algorithm finds the shortest combination of programs that could construct the convoluted output string of 1s and 0s. “This technique can equip current machinelearning methods with advanced complementary abilities to better deal with abstraction, inference and concepts, such as cause and effect, that other methods, including deep learning, cannot currently handle,” says Zenil. Zenil, H., Kiani, N., Zea, A. & Tegnér, J. Causal deconvolution by algorithmic generative models. Nature Machine Intelligence 1, 58–66 (2019).

ILLUSTRATION BY XAVIER PITA.

FE AT URE


FE AT URE

TWEAKING TOOLS TO TRACK TWEETS OVER TIME Computer model learns to identify Twitter users’

The researchers improved the model’s output by developing and incorporating a streaming keyword diversification component, which can identify closely related keywords and remove redundant entries from the top keyword list. The resulting model can capture a diverse range of interests for each user and adapt to their evolving interests over time. When the team tested their model on a set of tweets, it was a significant improvement on previous approaches, Zhang says. “Our model significantly outperforms many state-of-the-art user-profiling models.” The team has already produced a new iteration of their embedding model approach, she adds, in which user-user relationships are also captured to begin to identify interests that users have in common. “The next model will be more advanced and build dynamic co-embedding vectors that capture user-user social proximity and user-attribute relevance simultaneously,” Zhang says.

evolving interests by analyzing their tweets.

Your social media posts reveal a lot about you. KAUST researchers have developed a dynamic computational model that can analyze tweets to identify Twitter users’ interests and track changes over time. “Understanding the evolution of users’ interests means we can group them accordingly and recommend friends, news, events and other services,” says Xiangliang Zhang who led the research. Creating computer models that can identify a person’s evolving interests from their social media posts is a multifaceted problem. The first challenge is to understand the meaning of the posted text, a research area known as natural language processing (NLP). “The objective of NLP is to make computers as intelligent as human beings in understanding language,” Zhang says. “It is one of the most challenging tasks of AI,” she adds. Rule-based NLP models have not been very successful at interpreting the nuance of language in the way that humans use words in diverse and creative ways, such that the meaning of words can often be highly dependent on context. One alternative approach is to apply machine learning to represent words in a semantic space— where semantically related words for example, Paris, Beijing and Riyadh—are mapped closely together. To identify Twitter users’ interests by analyzing their tweets, the key challenge is to characterize individual users by their most important keywords. Zhang and her team have created an embedding model in which words and users are handled together. “We created a dynamic-user and word-embedding model that can jointly and dynamically learn user and word representations in the same semantic space,” Zhang says.

“O u r m o d e l significantly outperforms many stateof-the-art user-profiling

Liang, S., Zhang. X., Ren, Z., & Kanoulas, E.

m o d e l s .”

Twitter. Proceedings of the 24th ACM SIGKDD

Dynamic embeddings for user profiling in International Conference on Knowledge Discovery & Data Mining, 1764-1773 (2018).

Xiangliang maps out some of the theory behind her computational models.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

43


Gossip is an efficient way to share information across large networks and has unexpected applications in solving other mathematical and machine-learning problems. By looking at classical gossip algorithms from a novel perspective, Peter Richtarik has found a way to significantly speed up gossip-based information sharing, and in the process, he discovered new applications for this efficient mathematical approach. Gossip involves the sharing of information among individuals in a network and can be applied mathematically in both human social networks and data networks, such as distributed sensors. “A network is a collection of nodes, each connected to other nodes via links,” explains Richtarik. “In social networks, for instance, individuals are connected to others via friendship links. In sensor networks, sensors could be connected if they are close enough to communicate over a wireless connection.” In many real-world applications, it is often useful to perform calculations based on the data stored by all nodes across a

44

F a l l 2 0 19

network, such as computing the average of the private data stored by each node, which is known because the average consensus problem. However, because communication is limited to direct links between nodes, in practice, this is very challenging. “The idea of gossip algorithms is to perform this calculation by pairwise communication between randomly selected friends, and to repeat this process until all individuals learn the result,” says Richtarik. “This mimics the way gossip works among humans. It’s surprising that it is possible to show mathematically that this simple communication strategy can solve a global network-wide problem.” In collaboration with Nicolas Loizou from the University of Edinburgh in Scotland, Richtarik has been studying randomized gossip algorithms and their connections to other branches of mathematics and

computer science. Their theoretical study revealed a deep connection between randomized gossip algorithms and a branch of mathematics called linear algebra, which involves solving systems of many equations with many unknowns. They also established a direct deep link with one of the most famous algorithms in machine learning, the stochastic gradient descent method, which is used to train the deep-learning models employed in almost all industrial applications. These insights helped the researchers develop new and much faster gossip protocols. “We were able to develop an accelerated gossip algorithm that needs many fewer gossip rounds to reach the average consensus value,” says Richtarik. “Our method needs only the square root of the number of rounds needed for a classical gossip algorithm, that’s 100 rounds instead of 10,000. We proved this mathematically and observed this acceleration in practice as well.” Lizou, N. & Richtarik, P. Accelerated gossip via stochastic heavy ball method. 56th Annual Allerton Conference on Communication, Control, and Computing 2-5 October 2018 IEEE, 927-934.

NESTIGN / ALAMY STOCK PHOTO

ACCELERATING THE GRAPEVINE EFFECT


KAUST & McLaren Pushing the limits of science and engineering on the track to improve the performance of Formula 1 race cars.

Driven by a relentless desire to innovate, McLaren leads the way in the development of groundbreaking racing technologies.

innovation.kaust.edu.sa


SCIENCE VISUALIZED

CAPTURING SPRAY FROM FLASH-BOILING LIQUID JETS

Ultrafast video capture of droplet-cloud formation should help minimize the risk of gas-leak explosions. Using an ultrahigh-speed video camera operating at speeds of 5 million frames per second, KAUST researchers have captured footage of the fastest-expanding bubbles ever recorded. The bubbles formed when a highly compressed liquid fuel escaped through a tiny nozzle to create a fine jet of superheated liquid that boils instantly. Footage of this bubble-forming phenomenon, known as flash boiling, should improve our understanding of gas compression processes and the explosion potential of compressed gas tanks or pipeline leaks. “This research reveals the responsible physics behind the rapid phase change (liquid to gas) that many industries rely on for processes, such as cooling and refrigeration, power generation and thermal water desalination” says Ph.D. student Tariq Alghamdi. The study was inspired by a gas explosion in a Mexican maternity hospital in 2015, caused by a leaking gas storage tank, says postdoc José Federico Hernández-Sánchez, who carried out the research with Alghamdi in Sigurdur Thoroddsen’s lab. “The explosion in the hospital happened far from the leak, which is very common in these types of accidents because the explosive clouds are unpredictable and may move through windows and doors until finding a spark,” Hernández-Sánchez says. Understanding the explosive clouds formed by flash-boiling fuels has been difficult until now: flash boiling occurs too fast for capture by traditional high-speed cameras, which typically operate at up to 12,500 frames per second. “In most reports about flash boiling, every picture of a spray shows blurred motion,” Hernández-Sánchez says. The blur masks behavior with potentially significant implications, Alghamdi and Hernández Sánchez have now shown. “It took us three months to design, construct and optimize the experimental setup,” says Alghamdi, who built the setup. The team used a camera capable of capturing 5 million frames per second, fitted with a microscopic lens, which they aimed at a jet of compressed liquid being released into a vacuum chamber. The camera captured sharp images of flash boiling, which the team could use to accurately measure the ultrafast phenomena taking place. “We measured bubbles expanding at up to 140 meters per second. I believe that high-speed expansion has never been reported at this rate for a bubble,” Hernández-Sánchez says. More significantly, the team could also observe what happened once the bubbles burst. “The bubbles that disrupt a flash-boiling jet have remained unseen due to their extremely rapid speed of

46

F a l l 2 0 19

“M a n y n u m e r i c s i m u l a tions have attempted to predict the propagation o f e x p l o s i v e c l o u d s .” expansion. We visualized them and observed that after the expansion, the jet breaks into a polydispersed spray,” says Hernández-Sánchez. The explosive cloud of tiny droplets spreads almost 360 degrees around the end of the nozzle—a much larger spray angle than previously suspected. “Many numeric simulations have attempted to predict the propagation of explosive clouds,” Hernández-Sánchez says. But they have all failed because they missed the speeds of propagation of the gas, the real direction where the fuel is going, and the polydispersity of the spray. In our study, we address all these missing ingredients.” Al-Ghamdi, T., Thoroddsen, S. T., & Hernández-Sánchez, J. F. Ultra-high speed visualization of a flash boiling jet in a low-pressure environment. International Journal of Multiphase Flow 110, 238-255 (2019).

TARIQ ALGHAMDI PH.D. STUDENT Tariq grew up near KAUST in the city of Tabuk. He completed his M.S. at KAUST and is now studying under Sigurdur Thoroddsen for his Ph.D. Tariq is interested in studying the dynamics of laser-induced cavitation and the breakup instability of microliquid jets.


SCIENCE VISUALIZED

BENDING THE NORM ON NANOWIRES

Arranging nanowires in bent configurations may make them less likely to fail inside electronic devices. The team suspended silver nanowires from platinum electrodes over their custom-made TEM chips.

These TEM chips are patterned onto a 4-inch silicon wafer, which comprises hundreds of chips with through holes for TEM imaging and metal pads for electrical connections.

Nitin Batra (left), Pedro Costa (right) and Ahad Syed load a sample in the TEM.

New methods of arranging silver nanowires make them more durable, shows a study by KAUST. These nanowires form flexible, transparent conductive layers that can be used for improved solar cells, strain sensors and next-generation mobile phones. Applying nanotechnology in electronic devices requires rigorous testing of individual tiny components to ensure that they will stand up to use. Silver nanowires show great promise as connectors that could be arranged in flexible, neartransparent meshes for touchscreens or solar cells, but it is unclear how they will respond to prolonged stresses from bending and carrying current. Testing the bulk properties of a large sample of nanoparticles is easy, but not completely revelatory. However, adopting transmission electron microscopy (TEM) makes it possible to examine individual nanoparticles. Ph.D. student Nitin Batra and his supervisor Pedro Costa are at the forefront of developing new TEM techniques. This has allowed them to study single silver nanowires in detail1. “A major part of our work has been designing and fabricating sample platform prototypes (or ‘chips’) for TEM, which allow us to characterize and manipulate nanomaterials with an unsurpassed spatial resolution,” says Batra. To improve on expensive commercially available chips that contain a very fragile membrane to support nanoparticles, Batra and Costa, with help from Ahad Syed, an operational lead of the Nanofabrication Core Lab at KAUST, have now submitted

for patenting2 their own robust, reusable chips that do not require a membrane. The researchers suspended silver nanowires from platinum electrodes over their custom-made TEM chips and applied a range of voltages until the nanowires failed due to heating by the electrical current. They found that straight nanowires tended to snap when they reached a certain high current density, at points determined by local structural defects. More interesting behavior was seen when the nanowires were bent from the beginning. These samples tended to buckle instead of snapping at high voltage and exhibited an ability to “self-heal” because they remained held together by the carbon coating on the outside of the wires. Some nanowires even exhibited resonant vibrations, like the harmonics on a guitar string, before they failed. “Many devices are expected to undergo repeated bending and twisting by the end user, which means that it is not realistic to limit the study of the electrical response of silver nanowires to straight configurations,” says Batra. “Our results suggest that the failure rate of such devices could be minimized by using bent nanowires instead of straight ones. The self-healing capability could effectively delay the breakdown of the circuit.” 1. Batra, N.M., Syed, A. & Costa, P.M.F.J. Current-induced restructuring in bent silver nanowires. Nanoscale 11, 3606-3618 (2019). 2. US provisional patent 62/677,333. Patentees: Nitin M. Batra, Pedro M.F.J. Costa. Title “Membraneless platform for correlated analysis of nanomaterials” (2019).

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

47


SCIENCE VISUALIZED

TAKING AIM AT OUTLIERS

Automatic detection of uncharacteristic data sequences could change the way data is processed and analyzed.

“We we r e a b l e t o a c h i ev e g o o d o u t l i e r detection while accounting for the n a t u ra l c o r r e l a t i o n i n

48

F a l l 2 0 19

t h e d a t a .” a sudden jump. Building on previous work, Huang and Sun developed a metric for characterizing shape in a data series statistically, and used it as a basis for shapeoutlier detection. “We decompose the variation in the data into two components corresponding to magnitude and shape,” says Huang. “The challenge, however, was how to detect shape outliers as accurately as possible since there are many different types of outliers. In our study we tried out a wide range of outlier models and were able to achieve good outlier detection while accounting for the natural correlation in the data.” One direct benefit of this work is through preprocessing real-world datasets to remove erroneous observations and thus improve the accuracy of statistical analyses. However, as Huang points out, the approach could also be used to pick out unusual sequences for further investigation. “Sometimes, the outliers themselves can be more interesting than the regular observations,” Huang says. Huang, H. & Sun, Y. A decomposition of total variation depth for understanding functional outliers. Technometrics (2019).

RTIMAGES / ALAMY STOCK PHOTO

By deriving a measure of the characteristic way that a data series varies over time, KAUST statisticians have found a powerful solution to the problem of detecting outliers in monitoring data. An important first step in statistical processing is to remove potentially abnormal values in the data. A faulty temperature sensor, for example, could be generating data that might significantly alter the average temperature and range calculated for a meteorological station, or trigger a shutdown in a manufacturing process. Usually, the detection of such outliers is based on some threshold or rule, and it is comparatively easy for a computer to use this rule to highlight erroneous data. But what if the data is within the normal range but behaving erratically? What if the temperature reading suddenly drops by just a few degrees due to a sensor fault, or fluctuates incorrectly? These types of “shape” outliers are easy enough for a human to spot, but tremendously difficult to detect using an automatic algorithm. Huang Huang and Ying Sun encountered this problem through their research on ways to clean, analyze and present complex environmental data. “There is no universal definition of shape outliers because too many types of shape outliers exist, which makes them more difficult to detect,” explains Huang. Shape outliers can be regarded as data originating from any source process that leads to a different pattern, such as a different frequency of oscillation or


BIOLOGICAL SYSTEMS

LACTATE TRIGGERS GENES THAT MODIFY BRAIN ACTIVITY Study illustrates the links between brain energy metabolism and neuronal activity. A genome-wide study led by Dean Pierre Magistretti sheds light on the mechanisms through which lactate regulates long-term memory formation and neuroprotection. The breakdown of sugar in non-neuronal brain cells, called astrocytes, produces lactate, which gets shuttled to neurons as a source of energy. This

Hanan Mamood (left), Michael Margineanu (right) and colleagues have identified genes that may assist with developing novel therapeutic targets for neurodegenerative diseases.

lactate not only supports the energy demands of neurons, but also rapidly and transiently activates multiple genes that modulate neuronal activity and regulate brain function. Previous studies have shown that lactate stimulates the expression of genes encoding proteins involved in neuronal activity by signaling through N-methylD-aspartate (NMDA) receptors. Magistretti’s latest study reveals the extent to which lactate modifies gene expression in cortical neurons and also points to the mechanisms through which lactate modulates brain function. The team’s genome-wide analysis of gene transcription revealed that lactate exposure triggers the expression of more than 400 genes, some of which are crucial for modifying brain structure and function in response to internal and external influences, known as neuroplasticity. “We found that lactate stimulates synaptic activitydependent genes in the short term and genes involved in regulating neuronal excitability in the long term,” explains the first author of the paper Michael Margineanu, a KAUST Ph.D. student. After only an hour of exposure to lactate, 113 genes were differentially expressed compared with controls. Among them were genes known to mediate the NMDA-receptor-dependent response to neuronal

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

49


BIOLOGICAL SYSTEMS activity and those known to be involved in the mitogen-activated protein kinase (MAPK) signaling pathway that regulates neuronal survival. Selectively blocking NMDA receptors with the inhibitor MK-801 confirmed that 41 percent of the genes modulated by lactate after 1 hour were activated in an NMDA-receptor dependent manner. The study also revealed that nicotinamide adenine dinucleotide (NADH), a by-product of the metabolic processing of lactate to pyruvate, was able to regulate the expression of more than 60 percent of the genes that were modulated by lactate after 1 hour. NADH is crucial for the production of adenosine triphosphate (ATP), the major energy currency of the cell. This finding highlights the influence of the cells’ energy production process on neuronal gene expression and signaling. In addition to identifying genes that are stimulated in a non-NMDA-receptor dependent manner, the study also revealed upregulation in the expression of genes involved in neuronal excitability after 6 hours of lactate exposure. “The genes that we have identified could contribute to the development of novel therapeutic targets for neurodegenerative diseases in which brain energy metabolism is altered, such as Alzheimer’s disease,” concludes Margineanu. Margineanu, M.B., Mahmood, H., Fiumelli, H. & Magistretti, P.J. L-lactate regulates the expression of plasticity and neuroprotection genes in cortical neurons: a transcriptome analysis. Frontiers in Molecular Neuroscience 11, 375 (2018).

MICHAEL MARGINEANU PH.D. STUDENT Michael joined Pierre Magistretti’s group at KAUST in 2013 from Jacobs University in Germany to complete his Master’s degree. After graduating in 2015, he continued on with his Ph.D. aimed at investigating the effects of lactate on cellular signaling in brain cortical neurons using genome-wide transcriptome and bioimaging studies.

50

F a l l 2 0 19

FINDING WAYS TO FEED THE WORLD

Genome-editing techniques have the power to transform crop yields and plant resilience to feed the growing global population.

Genome-editing techniques have the power to transform crop yields and plant resilience to feed the growing global population. By 2050, global food production will have to almost double to feed the predicted population of 9 billion people. Yet achieving food security for the future is plagued with challenges, from increasing pressures of global warming and shifting climatic belts, to a lack of viable agricultural land and the significant burdens on freshwater resources. Plant scientists at KAUST are searching for ways to improve crop resilience, boost plant stress resistance, and combat emerging diseases. By showcasing the latest genome-editing technologies, the team is highlighting ways of securing enough nutritious, highquality food to feed the world’s growing population. “Essentially, we must work from the bottom-up to alleviate


BIOLOGICAL SYSTEMS poverty and undernourishment,” says Mark Tester, a plant scientist with a strong interest in developing salt-tolerant crops. “Studies have shown that, if the incomes of smallholder farmers are increased, poverty reduces rapidly. It is vital that farmers have access to crop plants that will provide the best possible yield within their given environment. This is feasible using safe and affordable genome-editing techniques.” Farmers have been genetically selecting crop plants for thousands of years, choosing superior-looking plants (based on their appearance or phenotype) for breeding. From the early 20th century, following breakthroughs in understanding of genetic inheritance, plant breeders have deliberately cross-bred crop cultivars to make improvements. However, phenotypic selection is timeconsuming and often expensive. “Since the 1990s, phenotypic selection has been aided by DNA sequence tags that are associated with favorable traits. However, this has issues because if a trait does not exist within the gene pool for a particular species, then it can’t be selected,” says Simon Krattinger, an expert in cereal genetics and plant-pathogen interactions. Other techniques include genetic modification (GM), which involves inducing random mutations in plants or incorporating genes from other species (transgenics). The idea of GM foods continues to raise safety fears in the general public, particularly in those regions that have not yet had to rely on such methods to feed their populations. Thanks to the GM stigma, it may be difficult to persuade people that genome editing is safe.

“Genome editing is a targeted, precise method of tweaking a natural plant to be more resilient and produce higher yields, without the need for incorporating foreign DNA or antibiotic resistance from bacterial genes—it is an entirely different process from GM,” says Magdy Mahfouz, who is heavily involved in developing and refining that genome-editing techniques. “These exciting techniques mean that it is now feasible to introduce specific DNA changes into a crop species and induce traits that did not exist before.” Mahfouz hopes to accelerate and expand the scope of nextgeneration plant genome engineering, with a specific focus on nutritional cereal crops and plant responses to abiotic stresses. A key editing technique Mahfouz is working with is CRISPR— molecular scissors that allow scientists to snip out and replace specific fragments within DNA. His team recently developed a CRISPR platform that allows them to efficiently engineer traits of agricultural value across diverse crop species. Their primary goal is to breed crops that perform well under climate-related stresses. “We also want to unlock the potential of wild plants, and we are working on CRISPR-guided domestication of wild plants that are tolerant of hostile environments, including arid regions and saline soils,” says Mahfouz. Krattinger and coworkers recently used CRISPR to introduce disease resistance into rice. With 10-15 percent of the global harvest lost each year to disease, inducing resistance is hugely valuable. Salim Al-Babili and his team have long used transgenics to

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

51


BIOLOGICAL SYSTEMS enhance crop nutrition by transferring genes across species and they are excited to explore the more refined approaches offered by genomic editing. “We successfully increased provitamin

“We’r e i d e n t i f y i n g t a rg e t g e n e s t h a t can be edited to increase p e a r l m i l l e t ’s resistance to t h e r o o t p a ra s i t e S t r i g a .” A content in crops like Golden Rice using transgenic approaches. However, genome editing opens up unprecedented possibilities for crop improvement,” says Al-Babili. “Now, we’re identifying target genes that can be edited to increase pearl millet’s resistance to the root parasite Striga, which causes enormous losses in African cereal production.” However, Krattinger notes, while disease resistance is relatively simple to tweak—because it is controlled by only a few genes—more complex traits, such as overall yield or drought tolerance, controlled by hundreds of genes working together, have a far more challenging editing process. KAUST teams will continue trialing and refining techniques for complex traits in the coming years. With climate change shifting pathogenic strongholds and presenting new agricultural challenges, it is time to encourage people to embrace these technologies. Education across all nations is vital to secure our future food supplies, notes Tester. Zaidi, S.S., Vanderschuren, H., Qaim, M., Mahfouz, M.M., Kohli, A., Monsoor, S. & Tester, M. New plant breeding technologies for food security. Science 363, 1390-1391 (2019).

52

F a l l 2 0 19

SPEARGRASS RECRUITS SANDY MICROBES FOR HELP In the harsh sand of the Namib Desert, the sheaths of speargrass roots co-opt any growthpromoting bacteria they can find. Sticky, sandy sheaths surrounding the roots of three speargrass species growing in the Namib Desert recruit whatever growth-promoting bacteria are available in the surrounding sand. This is contrary to the more specialized root sheaths of plants

growing in resource-rich soils, where different plant species recruit different types of bacteria. The research led by KAUST has implications for predicting how well plants with root sheaths, including some food crops, will


BIOLOGICAL SYSTEMS desert speargrass species would randomly recruit whatever bacteria available in the surrounding sands. The team analyzed the microbial content of the soil and the roots of three different species of desert speargrass growing on the top, middle and bottom of a single dune in the Namib Desert. “It was important for us to analyze the plants growing in a restricted area to nullify the variability determined by the type of soil and the climate,” explains study coauthor, Maria Mosqueira. The team found that the rhizosheaths of all three species acted as a hot spot for relatively similar growth-promoting bacteria and fungi. Their finding indicates that the types of microbes present in the rhizosheath-root systems of the three kinds of speargrass were not driven by plant species, but rather by microbial availability in the surrounding sand. “We are now trying to define which microbes are commonly associated with the rhizosheath and to understand their roles in favoring the fitness of their hosts,” says the study’s first author, Ramona Marasco. “The selection and isolation of such microorganisms can be a step toward evaluating their protective capacity on agricultural crops, such as wheat or barley,” she says. Ramona Marasco views the rhizosheath of speargrass through a stereomicroscope (Leica DFC295).

Marasco, R., Mosqueira, M. J., Fusi, M., Ramond, J-B., Merlino, G., Booth, J.M., Maggs-Kölling, G., Cowan,

adapt to changing environmental conditions and stress. The rhizosheath is an adaptive feature developed by some plant species that strengthens the contact between plant roots and the surrounding soil to improve nutrient and water uptake. Root hairs, fungal filaments and a sticky material formed by roots and microorganisms cause sand or soil granules to aggregate and form a sheath of soil encasing the entire root system. Desert speargrass species are among the types of plants to develop this adaptive trait and, until now, few studies have been conducted on this

species’ rhizosheath-root system. In resource-rich soils, where plants have abundant choice, different species are known to recruit specific types of bacteria to their roots. These growth-promoting bacteria facilitate the availability of nutrients, such as nitrogen, iron, and phosphorous, to the plants, and they can also confer disease resistance. KAUST bioscientist Daniele Daffonchio and colleagues predicted that a plant’s species would have a minimal impact on bacterial recruitment in the resource-poor sands of the Namib Desert and that different

D., Daffonchio, D. Rhizosheath microbial community assembly of sympatric desert speargrasses is independent of the plant host. Microbiome 6, 215 (2018).

RAMONA MARASCO RESEARCH SCIENTIST In Daniele Daffonchio’s group, Ramona is studying plant-microbe interactions in extreme environments to understand the ecological role of microbial communities that naturally associate with plants under stressful conditions.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

53


BIOLOGICAL SYSTEMS

MINING BACTERIA FROM THE DESERT Desert bacteria protect plants from salt toxicity.

Bacteria isolated from the Saudi desert have demonstrated plant-growth-promoting properties that could make them useful as biofertilizers. “The vast majority of deserts, especially in Saudi Arabia, have never been explored for agricultural potential,” says doctoral student, Abdul Aziz Eida, of KAUST’s Desert Agriculture Initiative. “Many people think deserts are sterile and inhospitable to any form of life. But there are many plants able to grow and survive in the harsh conditions found there. We believe that one of the key factors enabling plants to survive in such environments is their association with microbes in the soil.”

54

F a l l 2 0 19

Eida is part of a team working on the Darwin21 project led by microbiologist Heribert Hirt. The researchers in this team study desert bacteria for their potential to promote plant growth in stressed conditions, such as those facing drought, salinity, extreme temperatures or nutrient deficiency. The team collected soil samples from sites in two desert regions in Saudi Arabia: Jizan, located on the southern Red Sea coast, and Al Wahbah Crater, part of the extinct Harrat volcanic chain in western Saudi Arabia. They also took root samples from four types of plants and examined the samples for their bacterial content. They found large numbers and diverse kinds of bacteria in the desert soil, but their number and diversity were smaller in the soil attached to the plants’ roots, a zone known as the rhizosphere, and even smaller in the endosphere, within the roots1. A significant number of the bacteria isolated from the plants’ endospheres were shown to have growthpromoting traits. The researchers introduced some of these bacteria, under laboratory conditions, to saltstressed soil surrounding the roots of Arabidopsis thaliana (thale cress), which is often used to study plant development. They found the bacteria conferred


BIOLOGICAL SYSTEMS

The habitat of the remote Al Wahbah crater, located in western Saudi Arabia, is characterized by low humidity, high evaporation rates and limited rainfall.

salt-stress tolerance to the plant, promoting its ability to grow. “This was particularly exciting as it indicated that these bacteria might also promote salinity-stress tolerance in other plants, like wheat, barley or cucumbers,” says Eida. Next the team investigated what was happening at the molecular level to improve plant growth when these bacteria were added to salt-stressed soil surrounding their roots. They found that the presence of the bacteria affected the expression of certain genes involved in the transport of sodium and potassium ions2. Sodium is toxic to plants, while potassium is needed for plant development and growth. The bacteria induced molecular changes in plants that led to a reduction in sodium ions and an increase in potassium ions transported to the plants’ shoots. In this way, the bacteria protected the plants from salt toxicity and helped them acquire the nutrition they needed. “Significantly, despite the fact that the five bacteria tested came from different plants were of different genera and possessed different plant-growth-promoting traits, they all had similar effects on the plants’ shoot and root biomass, ion distribution and transcriptional regulation of ion transporters,” says Eida.

The team is now testing the ability of some of their bacteria to promote wheat and barley growth in the field under saline and nonsaline irrigation conditions. They have a year’s worth of data with promising results, Eida says. 1. Eida, A.A., Ziegler, M., Lafi, F.F., Michell, C.T., Voolstra, C.R., Hirt, H. & Saad, M.M. Desert plant bacteria reveal host influence and beneficial plant growth properties. PLOS ONE 13, e0208223 (2018). 2. Eida, A.A., Alzubaidy, H.S., de Zélicourt, A., Synek, L., Alsharif, W., Lafi, F.F., Hirt, H., & Saad, M.M. Phylogenetically diverse endophytic bacteria from desert plants induce transcriptional changes of tissue-specific ion transporters and salinity stress in Arabidopsis thaliana. Plant Science 280, 228-240 (2019).

ABDUL AZIZ EIDA PH.D. STUDENT In Heribert Hirt’s lab, Abdul Aziz is earning his degree developing microbial solutions for sustainable agriculture and for alleviating climate change impacts on global food supplies.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

55


KAUST & Lockheed Martin

Collaborating on new flexible electronics to develop solar cells for fixed-wing unmanned aerial vehicles.

Solving complex challenges and advancing scientific discovery to accelerate tomorrow. 56

F a l l 2 0 19

innovation.kaust.edu.sa


TR A NSL ATING TEC H

MOFS CAN SENSE AND SORT TROUBLESOME GASES Fluorinated metal-organic frameworks make excellent

materials for selective sensing and removal of toxic gases. From astronauts and submariners to miners and rescue workers, people who operate in small enclosed spaces need good air quality to work safely and effectively. Electronic sensors, now developed by a KAUST team, can simultaneously detect at least three critical parameters that are important to monitor to ensure human comfort and safety. These new sensors use fluorinated metal-organic frameworks (MOFs) as the sensing layer. MOFs are porous materials comprising a regular array of metal atoms held together by small organic-molecule linkers to form a repeating cage-like structure. KAUST’s Mohamed Eddaoudi, who led the two studies of the sensor’s efficacy, explains that by altering the metal and organic components, MOFs can be tuned for applications ranging from gas separation and storage to catalysis and sensing. “Many people have attempted to develop simple, efficient, low-cost SO2, CO2 and H2O sensors without success,” say researchers Mohamed Rachid Tchalala, Youssef Belmabkhout and Prashant Bhatt, all from Eddoudi’s lab. The approach taken by Eddaoudi’s group was to develop a fluorinated MOF, which

Belmabkhout and Tchalala tested as sensor materials for these gases. Testing of these state-of-the-art materials was in collaboration with Khaled Nabil Salama and his team. The first study shows how the sensor can measure the concentration of carbon dioxide and the level of humidity in the air1. While the second study of the same fluorinated MOFs shows it can detect the harmful and corrosive gas sulfur dioxide, or even selectively remove it from powerplant flue gas2. Traces of SO2 are invariably present in the flue gas produced by factories and powerplants, and SO2 can poison materials developed to trap CO2 for carbon capture and storage,” say Belmabkhout and Bhatt. “AlFFIVE-1-Ni can soak up SO2 with an affinity 66 times higher than for CO2, while showing good stability to SO2 exposure.” The MOFs could also be used with two simple, low-cost, high-sensitivity sensor platforms. Quartz crystal microbalance (QCM) sensors that are coated with a thin film of either MOF detected the change in mass with the absorption of SO2, or water and CO2. Similarly, MOF-coated interdigitated electrode sensors detected a change in electronic properties with the

absorption of water and CO2. Both sensor platforms, the team showed, could monitor moisture and CO2 levels under real atmospheric conditions. “The signal is calibrated against CO2 concentration, humidity level and mixtures of both,” Tchalala explains. A QCM-based sensor could also detect SO2 in the air at levels of just 25 parts per million. The technology developed at the Advanced Membranes and Porous Materials Center is capable of detecting various gases with a high degree of selectivity and sensitivity. It was recently granted a US Patent3 . 1. Tchalala, M.R., Belmabkhout, Y., Adil, K., Chappanda, K.N., Cadiau, A., Bhatt, P.M., Salama, K.N., Eddaoudi, M. Concurrent sensing of CO2 and H2O from air using ultramicroporous fluorinated metal-organic frameworks: effect of transduction mechanism on the sensing performance. ACS Applied Material Interfaces 11, 1706−1712 (2019). 2. Tchalala, M.R., Bhatt, P.M., Chappanda, K.N., Tavares, S.R., Adil, K., Belmabkhout, Y., Shkurenko, A., Cadiau, A., Heymans, N., De Weireld, G., Maurin, G., Salama K.N. & Eddaoudi, M. Fluorinated MOF platform for selective removal and sensing of SO2 from flue gas and air. Nature Communications 10, 1328 (2019).

3. Gas Sensor, United States Patent No. 10,001,448, June 19, 2018.

MOHAMED RACHID TCHALALA POSTDOC Mohamed Eddaoudi’s postdoc Rachid is interested in understanding the chemical structure, phytophysical behavior and transport properties of nanomaterials.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

57


TR A NSL ATING TEC H

DRIVING WATER DOWN NANOHIGHWAYS Removing water vapor from air and other gas mixtures, which is crucial for many industrial processes and air conditioning, could become cheaper and more effective through polymer membrane technology now developed at KAUST. “We have made a polymer film with extremely high permeability for water vapor while presenting an effective barrier for other gases,” explains former KAUST Ph.D. student now postdoc Faheem Akhtar. The researchers found a way to create tiny nanochannels in the membrane structure that they describe as “highways” for water molecules. The channels attract water and divert it away for extraction, leaving dry gases behind. “The water transport is extremely fast,” Akhtar adds. The membranes are composed of a commercial polymer called NexarTM. This is a block copolymer that assembles when short blocks of one repeating molecular unit become sequentially linked with short blocks of another type of unit. The chemical structure of the blocks controls the interaction with water vapor and other gases. The key innovation, however, was the discovery that the fine structure of bumps and ridges in the membranes can be controlled by varying the conditions in which the polymer self-assembles. Changing the solvents used

58

F a l l 2 0 19

during the polymer formation generates membranes with a variety of ordered or disordered channels. “Getting the right polymer morphology was very challenging and interesting,” says team leader Klaus-Viktor Peinemann. He explains that the polymer contains waterfriendly and water-repellent sections. When prepared using appropriate solvents, the water-friendly sections orient themselves like pearls on a string, forming the highways for water transport. “It took us a long time to find the right conditions,” Akhtar points out. To succeed, theoretical understanding of the chemical interaction between the chosen solvents and the polymer was combined with a fair bit of trial and error. Through science and perseverance, the researchers eventually identified a procedure to make ordered structures that yield a six-fold increase in water permeability compared to disordered membranes.

ILLUSTRATION BY XAVIER PITA.

Self-assembled channels in a polymer membrane could greatly enhance extraction of water from gases.


TR A NSL ATING TEC H

“T h e c h a n n e l s a t t ra c t wa t e r and divert i t a wa y f o r ex t ra c t i o n .”

Having demonstrated the basic potential of the membrane technology, the team now plans to scaleup the manufacturing process and to test it in realistic industrial applications. The commercial opportunities are considerable. More effective dehumidification methods could drastically reduce the energy consumption of an energy-intensive procedure. To complete this work, Akhtar collaborated with other experts in membrane science and experts in computational materials.

Schematic depictions of membranes with ordered (left) and disordered (right) channel structures.

Akhtar, F.H., Vovushua, H., Villalobos, L.F., Shevate, R., Kumar, M., Nunes, S.P., Schwingenschlögl, U. & Peinemann, K.V. Highways for water molecules: Interplay between nanostructure and water vapor transport in block copolymer membranes. Journal of Membrane Science 572, 641-649 (2018).

K AUST DISCOVERY

FAHEEM AKHTAR POSTDOC Faheem is now working on upscaling his membrane work for dehumidification coupled with adsorbents for energy-efficient cooling air conditioning systems.

d isc ove r y. k a u st. e d u .s a

59


TR A NSL ATING TEC H

PAPER SENSORS REMOVE THE STING OF DIABETIC TESTING Inkjet-printed device helps monitor a patient’s blood-sugar levels without painful needles.

60

F a l l 2 0 19

A technique that enables biologically active enzymes to survive the rigors of inkjet printing presents a promising alternative to routine blood screening exams faced by diabetic patients. The KAUST-led team used this approach to make disposable devices that can measure glucose concentrations in human saliva. Strips of pH-sensitive paper are commonly used to test whether a liquid is acidic or alkaline. Researchers are now working to apply similar principles to create paper sensors that quickly indicate disease biomarkers. Key to this approach is replacing traditional electronic circuitry in the sensors with low-cost plastics that can be manufactured quickly and in large quantities. Bioscientist Sahika Inal collaborated with electrical engineer Khaled Salama and materials scientist Derya Baran to use inkjet technology to produce sensors sensitive to small sugar concentrations in biofluids. Utilizing a commercial ink made from conducting polymers,


TR A NSL ATING TEC H Eloise Bihar looks on as the inkjet-printed glucose sensors are created.

A KAUST-led group has developed a technique that enables biologically active enzymes to survive the rigors of inkjet printing.

membrane that repels the negative charges present in most interfering species enabled measurement of only the relevant glucose levels in saliva samples from volunteers. Experiments showed that the top coating gave the sensor an unprecedented shelf life—the enzyme could be kept alive and active for a month if stored in a sealed bag. These results are encouraging the team to expand the

“P r i n t i n g t h e e n z y m e i s t r i c k y — i t ’s s e n s i t i v e to variations of t e m p e r a t u r e .”

the team printed microscale electrode patterns onto glossy paper sheets. Next, they printed a sensing layer containing an enzyme, glucose oxidase, on top of the tiny electrodes. The biochemical reaction between available glucose and the enzyme creates electrical signals easily correlated to blood-sugar levels. “Paper is porous, which makes it challenging to print conducting and biological inks that are dissolved in water,” says Eloise Bihar, a postdoctoral researcher at KAUST and the first author of the study. “Printing the enzyme is tricky as well—it’s sensitive to variations of temperature, the voltage applied at the cartridge, and the pH of the ink.” After optimizing the enzyme-printing conditions, the researchers had another obstacle to tackle. While fluids, such as sweat or saliva, contain enough sugar for monitoring purposes, they also contain molecules, such as ascorbic acid, that interfere electrically with conducting polymers. Coating the sensor with a nafion polymer

capabilities of this approach by incorporating different enzymes into the sensing layer. “Optimization never ends in engineering, so we are trying to make this system more robust to detect other metabolites in biofluids,” says Inal. “We are also looking to integrate printed and self-powered energy devices into the sensors, giving us a more user-friendly platform that eliminates external batteries or wires.” Bihar, E., Wustoni, S., Pappa, A.M., Salama, K.N., Baran, D. & Inal, S. A fully inkjet-printed disposable glucose sensor on paper. npj Flexible Electronics 2, 30 (2018).

ELOISE BIHAR POSTDOC In Sahika Inal’s working group, Eloise focuses on the fabrication of printed autonomous biosensors for monitoring human metabolites. Her goal is to develop cutaneous, minimally invasive and comfortable sensors that can be mounted on the skin.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

61


TR A NSL ATING TEC H

BETTER RACING CAR DESIGN THROUGH AN INDUSTRY PARTNERSHIP The Formula 1 race track is the ultimate testbed for a KAUST researchers’ latest work.

Improving elite racing car design is an exciting application of computational fluid dynamics for KAUST’s Matteo Parsani, as he brings next-generation algorithms into the real world through a partnership with McLaren Racing. Parsani is applying his computational and numerical expertise to help the Formula 1 team improve the aerodynamic design of their car. The project is one part of a long-term multicomponent partnership between KAUST and McLaren Racing that started in 2018. The partnership will develop new technologies to drive forward the Formula 1 team in the short term, but everyday vehicles, such as road cars and commercial airlines, are set to benefit in the slipstream. Parsani’s work, at KAUST’s Extreme Computing Research Center, focuses on developing novel, robust and highly accurate high-order algorithms for modeling fluid flow around objects to run in current- and next-generation supercomputer architectures. “We are pushing the boundaries of our research by applying our algorithms to an advanced object like a Formula 1 car,” Parsani says. With Formula 1 cars separated from their rivals by fractions of a second per lap, every little detail counts when seeking a performance advantage.

62

F a l l 2 0 19

Parsani’s group creates robust, highorder algorithms that solve nonlinear PDEs important to industry.

“To win in Formula 1, you have to be good at everything,” explains Jonathan Neale, McLaren Group Chief Operating Officer. Formula 1 is at the cutting edge of areas as diverse as fuel design, composite materials science, hydraulics and sensor design, to name just a few. “It is not possible for us to match the depth of knowledge you have here in some of those domains,” Neale said. “Great teams and championships are made on great partnerships.” “A Formula 1 car driving around a corner is effectively running in curved air, and that’s impossible to do in a


TR A NSL ATING TEC H

“D e v e l o p i n g a mesh for a highly complex shape is a real c h a l l e n g e .”

wind tunnel and very challenging to do in computational fluid dynamics,” says Neale. Enter Parsani’s expertise in numerical analysis and computational fluid dynamics. The aim is to provide more accurate aerodynamics simulations that better represent physical phenomena by fully exploiting the capabilities of supercomputers. The nuts and bolts of Parsani’s project are to conceptually create a mesh and then to use robust and very efficient highorder algorithms to help describe airflow around the subject. “Developing a mesh for a highly complex shape is a real challenge,” says Parsani. And then there’s the nonlinearly stable algorithm to design and implement in a high-performance computing fashion. Until recently, this was very difficult: researchers would adapt and modify high-order linearly stable algorithms. But calculations that run with these repurposed algorithms are fragile; they can “blow up” and fail to give an answer. As a result, the industry has never adopted them, Parsani says. Parsani leads one of a handful of groups around the world creating robust, highorder algorithms specifically to solve nonlinear PDEs. “Industry wants them to be robust and fast,” Parsani says. “We are rapidly taking important steps toward robustness. We are implementing these novel and adaptive algorithms in a high-performance computing framework, which performs very well on some of the largest supercomputers in the world, including KAUST’s Shaheen XC40 supercomputer,” he adds. To earn his racing stripes, Parsani and his group will run several test-case aerodynamic problems to ensure the code is accurate before it will be used to help in the future with any redesign of the car. Such is the level of competition in Formula 1—the drive for performance gains is relentless. “By the time we have built a car it is obsolete because the R&D team has already moved on,” says Neale. But it’s a relentless drive for improvement that will provide broader benefits. The Formula 1 trickle-down effect is a key reason to engage with the sport, Parsani says. Seatbelts and antilock brakes are just some of the technologies that started off there. The next applications for Parsani’s models could be designing more efficient aerodynamic components for commercial aircraft.

K AUST DISCOVERY

d isc ove r y. k a u st. e d u .s a

63


TR A NSL ATING TEC H

The team incorporated the switch into a tunable antenna printed on a flexible substrate.

ONE SWITCH TO RULE THEM ALL

Inkjet-printed switches make multiple frequency bands easier and cheaper to manage in wireless devices. Frequency-tunable communication modules, such as antennas and filters, are expected to help miniaturize wireless devices. Researchers at KAUST have created switches that enable control over these modules in response to stimuli. Mobile devices to support multiple standards, such as a global positioning system and a global system for mobile communications, require antennas that are capable of covering several frequency bands. “Radio-frequency switches are the key to realizing cost- and space-saving frequency-tunable antennas and filters,” says Ph.D. student Shuai Yang, who worked on the project with his supervisor Atif Shamim. Commercially available radio-frequency switches have performance limitations and involve convoluted fabrication approaches that require expensive

64

F a l l 2 0 19

materials and tools. Now, Shamim’s team has developed a cost-effective inkjet-printing method to generate switches. “Just as for newsprint, the cost of printed electronics is extremely low,” says postdoctoral fellow Mohammad Vaseem, who is also an author on the paper. The switches consisted of thermally and electrically responsive single layers of vanadium dioxide. The researchers synthesized vanadiumdioxide nanoparticles with a specific crystal arrangement to create the desired ink. They printed two different switch configurations that could be triggered thermally and electrically. The performance of these switches was comparable to their nonprinted analogues, but at much lower cost. As a proof-of-concept experiment, the team incorporated the switch in a tunable antenna printed on a flexible substrate.

At room temperature, the antenna could work in the range corresponding to future 5G communications. When heated, the switch allowed the antenna to operate in the WiFi and Bluetooth range. “When fully printed electronics become mature for industrialization, our switch will be useful for mass-producing smartphones and other wireless devices,” says Yang. In preparation for these developments, the team is building an accurate switch model for computer simulations. “We are also working on improving the performance of the switch,” adds Vaseem. Yang, S., Vaseem, M. & Shamim, A. Fully inkjet-printed VO2-based radio-frequency switches for flexible reconfigurable components. Advanced Materials Technologies 4, 1800276 (2018).


Join #SciCafeKAUST FACEBOOK.COM/KAUSTOFFICIAL


“Fascinated by aerodynamics from the age of five, I’m developing new algorithms, solvers and physical models with powerful mathematical properties and integrating them into next-generation supercomputing systems to resolve complex questions in the field.”

Destination KAUST:

Partnerships that bring big ideas to life

Through an accelerated partnership with McLaren Racing, computational scientist Matteo Parsani and his group are modeling air flow around the complex shape of McLaren’s Formula 1 car. Their simulations are geared to launch McLaren’s cars forward to achieve the fraction of a second per lap that separates them from their competitors.

Go to discovery.kaust.edu.sa to learn more about Parsani’s work on computational fluid dynamics.

discovery.kaust.edu.sa

Profile for KAUST

KAUST Discovery - Issue 8  

KAUST Discovery - Issue 8  

Advertisement