Arenberg 2013

Foto cover: Leen Cuypers

Table of contents

Introduction 1

AALBERS ManuelThe real estate/financial complex 3

CEULEMANS GrietChallenges of secondary school science curriculum design, course materials development, and the training of high quality chemistry teachers 9

DELABIE AnneliesNanoengineered thin films 15

JACQUEMYN HansPlant survival in a changing world 21

JANSSENS EwaldElectron, spin, and atom dynamics in atomic clusters 27

POLLIN Sofie Connecting everything 33

VAN DER BORG Jan Toerisme en economie 39

VAN DORPE PolNanobiophotonics 45

VANSTEENWEGEN Pieter From personalised city trip planning to robust railway timetables 51

VERHELST Marian Making chips smarter 57

1

Introduction

Young and new talent is a key for a learning, searching and innovating organization like the KU Leuven. Annually many colleagues start their career at our institute. This 24th brochure continues a long tradition: it offers new professors and young faculty members of the Science, Engineering and Technology (SET) Group the opportunity to present themselves and their research areas.

In the past these short descriptions have proven to be a very interesting and quick way to get to learn each other. Read them, contact and meet each other. Often this leads to opportunities for mutual collaboration and support in the different research domains and education programmes of the Group. Especially in the view of the recent integration this brochure can contribute to strengthening the cross-fertilisation between the different departments and units, research centres and the application-oriented and curiosity-driven research.

We hope that this brochure may be helpful to you and your co-workers and we take this opportunity to wish you all the success with your academic endeavours and assignments.

Georges GielenVicerector Science, Engineering and Technology

Inleiding

Jong en nieuw talent is de toekomst voor een lerende, zoekende en innoverende organisatie zoals de KU Leuven. Jaarlijks starten vele collega’s aan onze instelling. Deze 24e uitgave van de Arenbergbrochure zet een jarenlange traditie verder: het geeft nieuwe professoren en jonge academici van de Groep Wetenschap & Technologie (W&T) de kans om zichzelf en hun onderzoeksgebied aan u voor te stellen.

In het verleden is al gebleken dat deze korte teksten een interessante en snelle weg zijn om met elkaar kennis te maken. Lees ze door, neem contact op, maak kennis. Vaak ontstaan hierdoor kansen tot wederzijdse samenwerking en ondersteuning in de verschillende onderzoeksdomeinen en onderwijsprogramma’s binnen de Groep. In het licht van de recente integratie kan deze brochure nog meer bijdragen tot kruisbestuiving tussen de verschillende departementen en afdelingen, onderzoekscentra en het toepassingsgericht en nieuwsgierigheidsgedreven onderzoek.

Wij hopen dat deze brochure nuttig kan zijn voor u en uw medewerkers en wensen u via deze weg heel veel succes in uw academische activiteiten en opdrachten.

Georges GielenVicerector Wetenschap & Technologie

2

3

Manuel B. Aalbers was born in the Netherlands on October 25, 1977.

He studied urban planning and sociology at the University of Amsterdam (UvA) and received a PhD in Human Geography from UvA in 2006. After spending 6 months at the University of Milan-Bicocca and 18 months at Columbia University (New York), he returned to UvA. In 2012 he received a Starter Grant from the European Research Council (ERC) to build up a research team on ‘The Real Estate/Financial Complex’ and in February 2013 he was appointed associate professor at the Geography Division at KU Leuven.

His main research interest is in the intersection of housing and finance. He has published on redlining, social and financial exclusion, urban policy, gentrification, the privatisation of social housing, financialisation, and the Anglophone hegemony in academic research and writing. He is the author of Place, Exclusion, and Mortgage Markets (Wiley-Blackwell, 2011) and the editor of Subprime Cities: The Political Economy of Mortgage Markets (Wiley-Blackwell, 2012). He is also the associate editor of the Encyclopedia of Urban Studies (2010) and of the geography journal TESG.

4

The real estate/financial complex

Real estate and finance were at the roots of the global economic crisis that started in 2007 and continues to drag on. States and their many institutions, including central banks and financial regulators, have also been seen as complicit to the crisis, either because they did too little (e.g. lack of regulation) or too much (e.g. promoting homeownership). The connections between real estate (both residential and non-residential), finance and states still remain underresearched and undertheorized.

A subset of economics has focused on real estate finance, but generally their models are too abstracted in the eyes of other social scientists and their analysis is often more akin that of cheerleader of the market in general and the financial sector’s expansion into different real estate markets in particular. Furthermore, state institutions are either ignored or seen as frustrating the market mechanism. To further our understanding of the connections between real estate, finance and states, we therefore need more attention to scale and politics.

Alternatively, we look for inspiration in the interdisciplinary fields of political economy and urban studies. Work in various political economy traditions has done a great deal of research into the connection between finance and states, but they have often ignored a crucial sector: housing and real estate more generally speaking. There is also a tradition of work focusing on the interaction between real estate and states, usually concentrating on the involvement of municipalities in real estate projects.

Just like real estate gets little attention in political economy, finance is often ignored in this latter tradition that we could label ‘critical real estate studies’ or ‘urban political economy’. Moreover, this tradition has its roots in urban studies and is very microfocused, while the various political economy traditions are very macrofocused. In other words, we not only need a stronger connection between finance and real estate, we also need a stronger connection between different scales: local/urban, national/global.

I here propose a new metaphor that can help us to centre attention on the connections between real estate, finance and states: the real estate/

5

financial complex, akin the military/industrial complex. In his farewell address as USA President, Dwight Eisenhower (1961) warned his country about the military/industrial complex and thereby coined the metaphor commonly used to refer to policy and monetary relations between armed forces, the industrial sector and legislators, including political contributions, extensive lobbying, approval for defence spending, and beneficial legislation and oversight of the industry.

Despite discourses of withdrawing states, absent states and failed states, the hand of the state in its many guises is visible everywhere in real estate, finance and its connections. In introducing the metaphor of the real estate/financial complex, I will present a number of propositions that on the one hand show the many facets of the metaphor and on the other indicate what research should look into in order to come to a fuller understanding of the real estate/finance/states triangle.

These propositions often resemble hypotheses and could easily be rephrased as research questions, e.g. ‘to what extent have real estate and finance in the previous decades grown as part of the economy at the relative expense of other sectors of the economy?’ or ‘how and why do countries follow different pathways in the development of a real estate/financial complex?’

Proposition 1 Both real estate and finance have grown as parts of the economy at the relative expense of other sectors of the economy.

Proposition 2 Real estate and finance are increasingly connected. Of course, real estate has always been a capital-intensive sector, but the difference is that today the organisations in real estate and finance are also more integrated. According to the financialisation thesis, capitalism has become increasingly dependent on the growth of finance to enlarge money capital as a result of the stagnation of the real economy.

Proposition 3 The trend of propositions 1 and 2 will go in the same direction in most developed countries and rapidly growing economies, but they follow different pathways, reflecting different historical socioeconomic structures and policy choices. Indeed, politics, geography and history matter. There is a lot of research into path dependencies, political geographies, varieties of capitalism and welfare state typologies, but most of this literature has ignored non-residential real estate and real estate/financial/state connections.

6

Proposition 4 It follows that states at different scales have often enabled the growth of real estate, finance and their interdependence. The real estate/financial complex is organised like its inspiration, the military/industrial complex, including powerful lobby groups and think tanks, with many shared interests such as a strong focus on economic growth. Large companies within these sectors and selective state institutions play a strong role in the organisation of the complex. It is here that concepts such as regulatory capture, cognitive locking and cognitive closure come into play.

Proposition 5 The real estate/financial complex is not only organised at the national level, but at different scales, including the urban. It is here that the metaphor touches upon a rich urban studies literature on gentrification, megaprojects, starchitecture, city competition, and in particular growth coalitions/machines. What is needed is a connection between local case studies on real estate/state ties on the one hand and the literature on the financialisation on the other, building on existing studies on ‘form follows finance’, ‘towers of capital’ and ‘subprime cities’.

Proposition 6 There is a global dimension to the real estate/financial complex. Although there is a lot of literature on the globalisation of finance, little is known about the globalisation of real estate. How has the binary between Immobilien/immobilier (the German and French words for real estate) and liquidity been overcome, and what does this mean for the nature of both real estate and finance? Furthermore, how are the global and the urban connected through the real estate/financial complex?

Proposition 7 The restructuring of all three elements of the real/estate financial complex has resulted in a ‘great risk shift’ in which households are increasingly dependent on financial and housing markets for their long-term security. There is not only a shift towards the financial sector, but non-financial companies, governmental institutions and households are also increasingly expected to think in financial terms.

7

Fig. 1: The real estate/financial complex (thanks to colleague Chris Kesteloot for visualising the concepts/literatures)

8

9

Griet Ceulemans was born on January 7, 1970. She is mother of two boys, Finn (°2008) and Rameau (°2009).

She graduated as pharmacist (KU Leuven 1992). Her doctoral research focused on medicinal chemistry (synthesis of artificial DNA). Throughout 1997 she was postdoctoral fellow at the Scripps Research Institute in San Diego, California, and honorary fellow of the Belgian American Educational Foundation. In 1998 she joined imec as project manager of the biosensors group.

In 2000 Griet Ceulemans redirected her career towards chemistry education. At the University of Antwerp she became part-time professor in 2003. For about ten years, she taught general chemistry courses at the Faculty of Pharmaceutical, Biomedical and Veterinary Sciences (FFBD). She negotiated and implemented several projects to improve bachelor education in FFBD. She co-authored ‘Oefeningenboek algemene chemie’ (LannooCampus 2011) with KU Leuven Professor of chemistry, Lucien Viaene.

In October 2012 Griet Ceulemans was appointed part-time professor at the KU Leuven (Department of Chemistry). She is responsible for teacher education in natural sciences, in the subject of chemistry. In relation to this academic teaching responsibility, she tries to start up her own research group.

10

Challenges of secondary school science curriculum design, course materials development, and the training of high quality chemistry teachers

Chemical education research (CER) is by nature interdisciplinary, just as chemistry research needs to apply tools and hence substantial knowledge of physics and biology. A truly scientific approach of CER requires a combination of academic knowledge generated in different disciplines such as psychology, sociology, pedagogy, science education, chemistry as well as chemical education, with its two aspects: chemistry teaching and chemistry learning. CER can also appeal to very diverse interested groups: chemical education researchers, chemistry teachers, students of chemistry, funding agencies, government, at the national as well as European and international level.

The Programme for International Student Assessment, better known as PISA, administrated by the Organisation for Economic Cooperation and Development (OECD); and the Trends in International Mathematics and Science Study (TIMSS) organized by the International Association for the Evaluation of Educational Achievement (IEA) indicate this interest. CER might address issues of primary school, secondary school, higher education or informal learning. For all these reasons it is clear that any in-depth research effort will be based on clear choices. CER at KU Leuven wishes to address problems teachers face in secondary school chemistry education in Flanders. This entails research to support secondary school science curriculum design, course materials development, and the training of high quality science teachers, within the Flemish context, for the subject of chemistry.

A source of inspiration in this area is the EU report by the Nuffield foundation ‘Science Education in Europe: Critical Reflections’ (Osborne and Dillon, 2007). The project investigated why in recent times fewer young people seem to be interested in science and technical subjects. Its message is clear: the challenge is to reimagine science education: to consider how it can be made fit for the modern world and how it can meet the needs of all students; those who will go on to work in scientific and technical subjects, and those who will not.

More recently the ROSE project (Relevance Of Science Education, 2010) reported mainly regarding the affective dimensions of school science. It is a worrying observation of this project that in many countries where the students are on top of the international TIMSS and PISA score

11

tables, they tend to score very low on interest for science and attitudes to science. These negative attitudes may be long-lasting and in effect rather harmful to how people later in life relate to science and technology as citizens. In accordance with these reports and societal evolution, Flemish secondary school educational policy, has been working since several years on the optimisation of the standards of the natural sciences, be it the integrated science education or the separate science subjects chemistry, physics and biology.

This has resulted in the adaptation of the standards grade 1 of secondary education in 2010, with progressive consequences in 2012 for grade 2 and forthcoming reform in 2014 for grade 3. Along the same line, attention has been given to the development of standards for the natural sciences for technical and art secondary education in the second and third grade of secondary education (currently in development).

In Flanders, as in many countries, currently ‘science for the citizen’ is fiercely advocated as the road to be taken for mainstream curriculum design and course material development, in all science subjects in secondary school education. But what specific subjects, if any, from the academic science curriculum would then remain in such a science for the citizen curriculum? And in what way should they best be presented, to be motivating, understandable, relevant and attitude shaping in relation to the secondary school students?

It has been shown repeatedly that regarding chemistry, small scale experiments to be provided as demonstration or student practical, are always, whatever the subject, important to arouse interest and generate more thorough understanding. CER at KU Leuven has a considerable tradition, and continues to invest highly, in the development and optimisation of experimental protocols suitable for use in secondary school education. As much as possible, ever new course materials, and the opportunity to try them hands-on, are made available to in-service teachers at a yearly schooling event known as ‘building blocks for up-to-date chemistry teaching’.

At the same time, an attempt is made to go and explore new roads. When designing science curricula, and developing secondary school science course materials, the aspect of providing students with ‘context’ is gaining a lot of interest. ROSE delivered an ‘interest in science inventory’ that indicated, that topics that are close to what is often found in science curricula and textbooks have low scores on the rating of interest among young learners from Europe and other well developed countries. ‘No context’ or ‘school context’ is of low interest, also ‘everyday context’ is of low interest; biographies of (often male, old and

12

dead!) scientists is of low interest. On the other hand boys’ interests is the technical, mechanical, electrical, spectacular, violent, explosive. Girls’ interests is health and medicine, beauty and the human body, ethics, aesthetics, speculation (and the paranormal). ‘Space, life’ is among the most popular for girls and boys. It might be interesting to investigate in what specific conditions these findings can be most fruitfully exploited to make science education more relevant and meaningful.

As for Flemish conditions, it will be necessary to work within the boundaries of the limited number of science hours in secondary schools. This makes teacher cooperation over the subject boarders (natural sciences, mathematics, technology, geography) an important issue. Also the fact should be considered that Flemish teachers are possibly not specifically educated or intensely trained in the subjects they teach in practice. Finally, in Flanders’ science education development and practice, educational chemistry expertise is underrepresented. Only 2 out of 18 schools of higher education, that offer science teacher education, provide chemistry curricula.

Only at the Flemish universities, chemistry is equally present with biology and physics teacher education. The latter teacher educations prepare specifically for second and third grade science. In practice it is therefore unlikely that 12- or 13 year-old pupils come in contact with a science teacher with solid chemistry background.

13

14

15

Annelies Delabie was born in Kortrijk in 1975. She is married and mother of two boys.

She graduated in chemistry at KU Leuven in 1997. In 2001, she obtained a PhD degree in science from KU Leuven. Her PhD thesis was entitled ‘Computational Study of Transition Metal Ions in Zeolites: Structure, Spectroscopy and Reactivity’.

In 2001, she joined imec, the research institute for nanoelectronics and nanotechnology in Leuven, to explore novel materials and their deposition for applications in nanoelectronics. Within this research, she is focusing on the fundamentals and applications of thin films and their deposition techniques, in particular Atomic Layer Deposition (ALD), using experimental as well as theoretical methods.

Since 2012, she is also part-time associate professor at the Chemistry Department at KU Leuven in the research group ‘Nanoengineered Thin Films’. She is external internship supervisor for students in the first master chemistry.

16

Nanoengineered thin films

Nanoengineered thin films are nanometer scale thin films with material or surface properties that are designed towards specific applications and realized in a controllable manner, i.e. by controlling the composition, atomic structure, morphology and/or crystallinity of the material. They are of growing interest for applications in nanotechnology and nano-electronics. For example, the performance of nanoelectronic devices, such as transistors in computer chips and memory devices, is being improved by the introduction of novel materials with enhanced properties, by miniaturisation of the device dimensions, as well as by new device concepts, often implying an evolution towards three dimensional (3D) structures.

Creating nanoengineered thin films with well controlled structure and properties implies that one wants to assemble films with atomic growth control of thickness and structure. The availability of deposition techniques that enable such atomic growth control is therefore the key element for the successful exploration of nanoengineered thin films. The deposition techniques should ideally also be compatible with the increased importance of 3D-structures. As such, Atomic Layer Deposition (ALD) is of particular interest because its deposition principle inherently provides growth control at the atomic level. It originates from the use of self-limiting chemisorption reactions of gas phase precursors with a substrate. By using sequential exposures, gas phase reactions between the precursors are avoided and the thin film grows only by surface reactions. This provides the basis for the growth control, as well as the conformality of ALD: the deposited film can follow the shape of the underlying substrate even on extremely complex nanostructures.

ALD nowadays is a widely used thin film deposition technique for inorganic materials, with hundreds of ALD chemistries developed including oxides, nitrides, chalcogenides, and metals. Still, little is known about the reaction mechanisms of the surface reactions that provide the basis of this growth control. Fundamental understanding of these reactions is required if one wants to realize specific structural and compositional variations in nanoengineered thin films. We therefore combine in situ and ex situ experimental investigations with theoretical investigations of ALD processes. Many of our previous and ongoing studies focus on high dielectric constant materials (HfO2, Al2O3 and other oxides) on

17

semiconductor surfaces (silicon, germanium, IIIV materials) due the technological importance of these systems in nanoelectronic devices. Continuous, pinhole-free insulating dielectric layers with a thickness close to a single nanometer are required, with a low electrical defect density at and near the semiconductor interface.

However, some ALD precursors are not sufficiently reactive with the initial semiconductor surface. The film nucleates only at defect sites, resulting in an island morphology instead of a smooth and continuous thin film. Proper surface functionalisation is required to ensure a fast growth evolution from sub-monolayer coverage to fully continuous film. Recent investigations also focus on metals such as ruthenium (Ru), a candidate electrode material in memory devices, where we investigate the reactivity of different precursors for Ru ALD (Fig. 1).

The ‘ALD principle’ can also be used for organic thin films, and recently also for organic-inorganic hybrid thin films in the so-called Molecular Layer Deposition (MLD). The wealth of building blocks from organic and inorganic origin, combined with the atomic level control provided by MLD,

Fig. 1: Chemisorption reaction mechanism for RuCpPy (Cp = cyclopentadienyl = C5H5, Py = pyrrolyl = C4H4N) on a TiN(100) surface, calculated by Density Functional Theory. The RuCpPy precursor first interacting through the Py ligand, next it dissociates in RuCp and Py fragments. In contrast, the RuCp2 precursor only physisorbs on the TiN surface, explaining the experimentally observed different nucleation behaviour for the RuCp2 and RuCpPy precursors.

18

offers a unique and versatile route for atomically engineered organic-inorganic hybrid thin films. Such organic-inorganic hybrid thin films show an immense potential for tunable as well as novel material properties as a result of the interface and/or interplay between the organic and inorganic components. We are currently investigating MLD of hybrid organic-inorganic thin films with tunable optical properties for application in advanced lithography (Fig. 2a). Other possible applications include energy storage, organic electronics, advanced memories, encapsulating barrier layers and many more.

On the other hand, the Chemical Vapour Deposition (CVD) technique deposits a thin film on a substrate by chemical reactions of gas phase precursors, not being restricted to self-limiting surface reactions, as opposed to the ALD or MLD techniques. Although neither the growth control at the atomic level nor the conformality are inherently ensured in CVD, they can be realized in specific cases through careful control of the reaction kinetics, if sufficiently understood.

A recent illustration is the CVD of crystalline superlattices, consisting of alternating periods of silicon (Si) layers and atomic layers of elements such as oxygen (O). Such superlattices are investigated to potentially replace the conventional Si semiconductor channel material in future transistors. An anisotropy in the Si band structure is introduced with an enhanced carrier mobility in the direction parallel to the O-planes. Designing deposition processes for crystalline Si/O superlattices is a

Fig. 2a: Cross-sectional Trans-mission Electron Microscopy (TEM) images of (a) a hybrid organic-inorganic alucone thin film conformally deposited by MLD on a three-dimensional Si/SiO2 structure

Fig. 2b: a fully epitaxial five-period Si/O superlat-tice deposited by CVD on a Si(100) substrate.

2a

2b

19

major challenge, as epitaxial stacking of silicon on top of atomic layers of O is required. In epitaxy, the deposited crystalline thin film follows the crystalline order of the underlying substrate. However, O is generally viewed as a disturbing contaminant that hinders epitaxial deposition of Si on Si, either introducing defects in the growing crystal or by converting to a polycrystalline deposition.

We recently demonstrated that epitaxial deposition of Si can be enabled on O layers by carefully controlling the O content and its chemical bonding. Our insights lead to a deposition method for fully epitaxial Si/O superlattices with well controlled Si thickness and O content (Fig. 2b). Further research focusses on fundamental understanding of which O bonding configurations allow epitaxial Si deposition and the electrical and optical characterisation of Si/O superlattices.

To conclude, the fabrication of novel nanoengineered thin films by either ALD or CVD requires a detailed fundamental understanding of the chemical reactions occurring during thin film deposition, such that the desired structural and compositional variations can be exploited. Such insights can be obtained by combining in situ and ex situ experimental investigations with theoretical investigations based on computational chemistry.

20

21

Hans Jacquemyn was born in Leuven (Belgium) on October 22, 1974. He lives together with Elke Verbruggen and their three children in Lovenjoel.

In 1999, he graduated as ‘bio-ingenieur in het land- en bosbeheer’ at KU Leuven. He started his PhD studies at KU Leuven in 2000 and graduated in 2004 with a thesis on the impact of forest fragmentation on the spatial distribution and landscape genetics of forest plant species. From 2004 until 2010, he continued his research at KU Leuven as a postdoctoral fellow of the Fund for Scientific Research (FWO) – Flanders. During this period, he focused mainly on the population biology of terrestrial orchids, including the processes impacting on spatial genetic structure, population dynamics and sexual reproduction. In 2010, he was awarded an ERC starting grant focusing on the role of mycorrhizal associations in orchid diversification.

Since October 2012, he has a full-time position as BOF-ZAP professor at the Ecology, Evolution and Biodiversity Conservation Section of the Biology Department at KU Leuven.

22

Plant survival in a changing world

Plants are ubiquitous, occurring from the poles to the tropics, from low altitudes to mountain tops. The number of plant species has recently been estimated between 400.000 and 420.000 species, of which over 287.000 plant species have been described (about 23% of all known species on Earth). They can be found in all kinds of growth forms, ranging from small herbs to huge trees. Some plant species are very short-lived and flower only once, whereas others can survive for centuries and reproduce many times. Perhaps the most striking feature of plants is their enormous diversity in floral forms, which has fascinated many researchers for decades, leading to classic books including Christian Konrad Sprengel’s ‘Das entdeckte Geheimnis der Natur im Bau und in der Befruchtung der Blumen’ (1793) and Charles Darwin’s ‘The Different Forms of Flowers on Plants of the Same Species’. Both books still serve as a rich source of inspiration for evolutionary biologists and ecologists worldwide.

Although plants are an indispensable part of the Earth’s ecosystem and fulfill important ecosystem services, many of them have become rare and endangered, and some have gone extinct. Habitat loss and fragmentation, deterioration of local growth conditions, climate change, pollution and the introduction of exotic species are regarded as the most important threats to plant survival. Currently, over 8.500 plants are listed as threatened, which is about 2% of the world’s described plants. However, it is very likely that this figure is a serious underestimation of the total number of threatened plant species. Based on published threatened species lists and extrapolation techniques based on hotspot area and plant species endemism of threatened species in tropical countries, the total number of species that is threatened with extinction has been estimated to vary between 34 and 45% of all plant species. My first research line aims at elucidating the impact of various threats to plant survival on plant population viability. To assess the long-term persistence of plant populations and to elucidate the factors that potentially impact on plant population viability, we develop mathematical models that allow predicting the impact of changing environmental conditions on the long-term survival and the spatial spread of plant species. These models provide powerful tools to understand the factors that may drive populations to extinction or that facilitate rapid expansion in space and time. We employ these models to a variety of model systems, including rare orchids, grass species, and species with particular breeding systems such as dioecious

23

and distylous herbs (Fig. 1). With this knowledge, we also try to provide reserve managers with practical guidelines to aid conservation of rare plant species. Currently, we are also developing continuously size-structured integral projection models (IPMs) that allow assessing the impact of climate change on population viability and predicting optimal (evolutionarily stable) reproductive strategies, including size at reproduction and size-dependent reproductive effort, under various climate change scenarios.

Fig. 1: Overview of plant species for which demographic models were developed in order to assess their long-term viability (a-c) or to calculate rates of invasive spread using both demographic data and information on dispersal distances (d). a) Lady orchid (Orchis purpurea) is a long-lived, iteroparous orchid species that reaches the northern edge of its distribution area in Belgium. In Flanders, the species is very rare, with only 15 known po-pulations. b) The Cowslip (Primula veris) is a distylous plant species that typically occurs in mesic calcareous grasslands and that has shown a strong in decline in distribution area in Flanders during the last fifty years. c) Lady’s slipper orchid (Cypripedium calceolus) is an extremely rare orchid species that can be typically found in shaded woodland and mountainous areas. d) Purple moor grass (Molinia caerulea) is a grass species that rapidly expands in space and succeeds in occupying large areas of heathland after fire.

1a

1c

1b

1d

24

To survive in human-dominated landscapes, plants rely on several aboveground and belowground interactions such as pollination, herbivory, and interactions with mycorrhizal fungi. Since human interference is likely to have a profound impact on these biotic interactions, my second research line aims at getting insights into the impact of human-induced environmental changes on plant-fungus interactions and plant-pollinator interactions. We particularly focus on the symbiotic interactions between orchids and their mycorrhizal fungi. Using next generation sequencing we aim at elucidating the architecture of plant-fungus networks in both terrestrial and epiphytic orchids.

We also explore the role of mycorrhizal fungi in affecting orchid rarity and investigate whether and how mycorrhizal fungi contribute to population viability. In cases where several orchids co-occur, we investigate whether mycorrhizal fungi drive niche partitioning and therefore contribute to orchid coexistence. Besides, we investigate how human-induced impoverishment of pollinator faunas affects plant–pollinator interactions and limits pollen availability. More specifically, we investigate whether plants are capable of adapting to newly created pollinator environments, for example by changing important floral traits that increase fitness and thus population viability (Fig. 2). Finally, we aim at elucidating the ecological role of nectar-borne microorganisms (yeasts and bacteria) in affecting plant-pollinator interactions.

Fig. 2: Differences in flower morphology in plants of Centaurium erythraea originating from natural populations growing in two contrasting pollinator environments, with a) larger and more herkogamous flowers in pollinator-rich environments, and b) smaller flowers that lack herkogamy in pollinator-poor environments.

2a 2b

25

26

27

Ewald Janssens was born in Leuven on March 22, 1978. He is married to Marjoly Sintobin and is father of Quinten (2004) and Margot (2007).

In 2000, he obtained a Master degree in Physics at KU Leuven. He started his PhD studies at the Laboratory of Solid State Physics and Magnetism as a research assistant of the Fund for Scientific Research (FWO)-Flanders under supervision of Peter Lievens and Roger Silverans, and graduated in 2004 with a thesis on the electronic and geometric structure of transition metal doped silver and gold clusters.

From 2004 to 2007 he continued the study of metal and metal oxide clusters in the gas phase as a postdoctoral researcher of the FWO. In 2006 he obtained a mobility grant for a research stay of one year at the Freie Universität Berlin. From 2007 to 2010 he worked in the financial service industry at Deloitte Enterprise Risk Services, specialized in financial risk management and valuation of financial instruments.

In 2010 he returned to the academic sector as a full-time visiting lecturer at the Department of Physics and Astronomy at KU Leuven. Between 2011 and 2013 he also had a partial appointment as guest lecturer at the Hogeschool-Universiteit Brussel.

As of September 2012 Ewald is appointed as associate professor at the Department of Physics and Astronomy. His research focuses on electron, spin, and atom dynamics in photo-excited metal clusters.

28

Electron, spin, and atom dynamics in atomic clusters

The ongoing miniaturisation in microelectronics and the desire for high device speeds in technological applications direct condensed matter physics to nanometre length scales and ultrafast dynamics, respectively. Where top-down methods use lithographic processes to write nanoscale features, bottom-up methods assemble atomic scale building blocks, such as organic molecules, carbon fullerenes, and atomic clusters. Clusters, defined as small pieces of matter consisting of a few up to a few thousands of atoms, have size and composition specific properties. Although governed by the same fundamental laws, clusters have structural, optical, magnetic, and catalytic properties that are different from and in several cases superior to those of their macroscale analogues.

The conduction electrons of a metallic cluster are delocalised over the entire volume. Quantisation effects due to the small cluster size lead to discrete energy levels and an electronic shell structure, which can be described by electronic shell models of different degrees of sophistication. Using well controllable production techniques, one can tailor the composition and morphology of clusters at the atomic level. Flexible adjustment of properties is achieved by studying bimetallic clusters, since mixing two types of atoms allows uncoupling the number of atoms and the number of valence electrons. Besides metallic clusters, also transition metal oxide clusters have attracted at lot of attention because of their novel structural and magnetic properties. For instance, iron oxide clusters have very large magnetic exchange energies, rendering them interesting as sub-nanometre magnets. The coupling between localized magnetic moments on the transition metal atoms is controlled via superexchange interaction through the bridging oxygen atoms. Therefore the magnetic configuration, i.e. ferromagnetic, antiferromagnetic, or non-magnetic, of transition metal oxide clusters strongly depends on the oxygen concentration.

Earlier studies have shown that alloying two metals in a cluster is much more than averaging them. The properties of mixed clusters cannot be obtained by interpolating the behaviour of the corresponding pure metals. The stability of the bimetallic particles is governed by the interplay between electronic and structural effects. For instance, combined experimental and theoretical evidence demonstrated the enhanced stability of the cobalt-doped silver cluster Ag10Co+. This

29

cluster has a symmetric endohedral geometry with a closed 18-electron electronic shell structure (Fig. 1a). It was also shown that the electronic state of metal oxide clusters can be systematically modified by doping. Upon exchange of a V by Ti atom in doublet V6O15

−, the polyhedral caged structure remains unchanged (Fig. 1b), but the isomorphous substitution results in a singlet V5TiO15

− cluster that is isoelectronic with the neutral V6O15 cluster.

At present, there is solid scientific knowledge to treat nanomaterials on the length scale of single atoms. On the other hand, the manipulation of nanomaterials on their relevant time scale is still in its infancy. Even more, the characteristic times of several physical processes and their dependence on quantum size effects are still unknown. To study fast physical processes, the timing resolution of the experimental probes has to be at least of the same order of magnitude as their characteristic times. Analyzing motions faster than the blink of an eye (0.1 s) became only possible with the technological development of snapshot photography and stroboscopy.

Famous motion pictures of a galloping horse (Muybridge, 1878) and a bullet passing through an apple (Edgerton, 1964) demonstrated millisecond to microsecond time resolution. With the advent of ultrafast lasers in the 1980s, this resolution became 10 orders of magnitude better, reaching the femtosecond scale, the time scale of atoms and molecules in motion. Besides promising forecasts for controlling reaction dynamics with ultrafast optical lasers, recent developments of sub-picosecond pulses in the X-ray regime using synchrotron sources (e.g., ALS in Berkely and Bessy in Berlin) and using free electron lasers in the X-ray (e.g., FLASH and XFEL in Hamburg and LCLS in Stanford) and the far-infrared regime (e.g., FELIX in Nijmegen and CLIO in Paris), bring exciting prospects for studying the dynamics of nanosystems.

These facilities may bring the dream of ‘making a molecular movie’, i.e., imaging nanoscale structures in real time, to realisation. Following photon absorption multiple competing relaxation pathways (fluorescence, fragmentation, electron transfer, intersystem crossing, vibrational relaxation, electron-spin and electron-phonon relaxation…), each with their own characteristic rates (see Fig. 2), can take place.

In the regime where every atom counts, the study of clusters as function of their size is a fantastic playground for the fundamental study of dynamic processes. Clusters can be studied in molecular beams, i.e., in absence of any support or medium. In addition, state-of-the-art experimental cluster sources allow carefully controlling the initial state of gas phase clusters (size, charge, temperature) and the clusters are

30

small enough to be simulated with ab initio computational approaches. Understanding their size and composition dependent dynamics is key to be able to predict (and eventually steer) the nanosystem’s evolution.

My future research will focus on the strong coupling between electron, spin, and atomic dynamics in metal and metal oxide clusters. Ideally I would like to drive systems out of equilibrium and record at any point in time thereafter the position of the atoms and the electronic and magnetic states. This way multidimensional energy surfaces will be mapped out and one can learn how complex systems evolve in time on these surfaces. Interesting research questions are related to the transitions between ground and excited states with a different magnetic coupling or electron distributions.

Can magnetic transitions be triggered by electronic excitation following laser irradiation? If yes, how do these transitions affect the geometry of the clusters? How is energy released when spin relaxation takes place (fragmentation, radiation, intramolecular vibrational redistribution…)? What are the involved time scales for rearrangement of atoms, spins, and electrons?

Fundamental understanding of the magnetic, electronic, and structural properties and their interplay in metal and metal oxide clusters is not only a scientific goal, but is also required information to downscale component dimensions when designing next generation devices, e.g., in fast magnetic recording devices and spintronics.

Fig. 1a: Photofragmentation yield of cobalt doped silver clusters, AgnCo+, showing a pro-nounced cluster size dependence. The intensity drop after Ag10Co+ indicates that this size is particularly stable. The calculated endohedral cage structure and the spin configuration of Ag10Co+ are shown on the right.

Fig. 1b: Infrared multiphoton fragmentation spectrum of V5TiO15- (left) and the corresponding

harmonic vibrational spectrum calculated with density functional theory (right), which assigns the polyhedral caged structure (inset) of this cluster.

1a 1b

31

Fig. 2: Schematic illustration of some physical processes that can take place after photoex-citation of a cluster and their corresponding time scales.

32

33

Sofie Pollin was born in Brugge in 1979. After her engineering studies at the Department of Electrical Engineering (ESAT) - KU Leuven, she started a PhD at imec. During the PhD, she excelled in crossdisciplinary cooperation, did an internship at National Semiconductor in Santa Clara and was a visiting scholar at UC Berkeley. She obtained a PhD degree in 2006, with honors and continued her research on wireless communication, energy-efficient networks, cross-layer design, coexistence and cognitive radio at UC Berkeley. She obtained a BAEF and Marie Curie Fellowship for that work. In November 2008 she returned to imec to become a principal scientist in the green radio and cognitive radio teams.

Since October 2012 she is tenure track assistant professor at the Department of Electrical Engineering (ESAT) at KU Leuven within the TELEMIC division. Her research centers around networked systems that require networks that are ever more dense, heterogeneous, battery powered and spectrum constrained. She is married to Bart Adams and has two daughters Liselore (2009) and Jozefien (2012).

34

Connecting everything

We live in a connected world. It is expected that by 2025, integrated components will be approaching molecular limits and may cover complete walls. Every device will have a wireless connection, hence leading to trillions of connected devices and sensors, the sensory swarm. This sensory swarm will enable a tsunami of novel applications to improve environment monitoring, health supervision, traffic monitoring, emergency support or elderly assistance to name a few.

Today, we notice that the goal is far from being realized, and the biggest reason for this is the lack of high performance networking solutions with sufficient lifetime, caused by spectrum and battery constraints. I have been passionate about improving spectrum and energy efficiency of practical smart adaptive networks during my PhD and postdoc research, as summarized below, and will explain how this will be translated towards a vision for dense networked systems at KU Leuven.

Smart adaptive networks for spectrum and energy efficiencyInterest in wireless networking technology has been increasing over the last decades. To cope with the demand, network design has been focusing on increasing the throughput offered to some very demanding applications, at the cost of energy expenditure. While peak performance keeps improving with each new generation of wireless technology, typical performance is improving at a much slower rate due to interference and coexistence problems. We see a trend from optimizing peak spectral efficiency (bits/Hz) to considering area spectral efficiency (bits/Hz/m2). In addition, energy consumption is becoming more and more a constraint, and energy efficiency (energy/bit) is considered a major design criterion. To improve these metrics, is it commonly agreed that networks should become more adaptive, allowing to improve energy consumption by delivering the just required performance at minimal energy, or similarly delivering the just required throughput at minimal interference.

One of the major challenges towards smart adaptive networks is the complexity of understanding the problem as a whole. Energy consumption depends to a large extent on the analog and digital circuits that take care of the wireless signals and their processing. The circuit and implementation limitations should be understood when designing

35

meaningful solutions for energy efficient networks. Solving this requires a close collaboration with experts on RF IC modeling and implementation, which is already an expertise of the ESAT-TELEMIC research group. This knowledge should be complemented with expertise on cross-layer modeling and optimisation of wireless networks, with key towards understanding the behaviour and energy efficiency of important wireless standards (see Fig. 1a for an energy model of IEEE 802.15.4 obtained during my PhD work). Together with colleagues from ESAT-TELEMIC, we can hence fully understand wireless communication systems and solve system level optimisation problems effectively.

To solve the spectrum scarcity problem specifically, cognitive radios were proposed which are ‘radios that are able to sense the spectral environment over a wide frequency band and exploit this information to opportunistically provide wireless links that best meet the user communication requirements.’ Such radios ensure that spectrum is used optimally at each location and time, boosting spectral efficiency.

Any view on cognitive radio however has to cope with a series of very challenging problems. They require techniques to sufficiently monitor or learn the interference coming from other traffic sources, and effectively adapt while maintaining the required performance. The run-time overhead of those complex algorithms should be acceptable. Next, a pure physical layer approach to cognitive radio is typically not possible, and decisions and behaviour of other users, which are dictated by the protocols and applications, policy and regulation, should be taken into account.

During my research at UC Berkeley, I studied game theory, policy and regulation, to enable a complete view on cognitive, smart adaptive networks.

Fig. 1a: performance and energy modeling is key to optimize and improve wireless networks. Spectrum sensing hardware is key towards improve spectrum use. Fig. 1b: a picture of the sensing engine

1a 1b

36

Experimental validation is very important also in the context of cognitive radio, to convince policy makers and drive regulation. During my research at imec, we worked towards a sensing engine implementation that could be used to verify and experiment with solutions for spectrum sensing. This sensing engine is now also part of the EU FIRE (Future Internet Research and Experimentation) test bed CREW (Cognitive Radio Experimentation World), which is a test bed designed to support EU research on cognitive radio. A picture of the sensing engine can be found in Fig. 1b.

Dense networked systemsThe use of wireless communication is expected to continue increasing dramatically. More and more data will be stored in the cloud, and users will want to access that content continuously using wireless communication technology combined with wearable electronics such as smart watches or glasses. How to deal with this capacity explosion, especially indoor, is one of the most important challenges related to wireless communication and requires a radical rethinking of the wireless communication infrastructure.

As most wireless systems now operate close to Shannon capacity, large gains at physical layer are not possible any more, and the only viable option is to increase the spatial reuse, i.e., allowing multiple spatially separated users to make use of the spectrum simultaneously. Similar to the switch in wired networks from a shared medium to a fully switched medium, we expect a radical context change in wireless communication technology that will allow the establishment of fully dedicated links from the infrastructure to each mobile terminal. The switched wireless network will learn the parameters of its destination, and autonomously adapt.

To achieve this, we have to rely on physical layer techniques such as beamforming or many antenna solutions, and complement them with a novel network view where multiple access points (AP) serve a single user (Fig. 2b), instead of the current practice where multiple users are served by a single AP (Fig. 2a). Controlling and coordinating many APs takes time, and can only happen at relatively long timescales, while sudden changes are possible in a wireless context (e.g., interference) that require immediate adaptation. In addition, obtaining context information, i.e., spectrum sensing, is expensive, and should be minimized. To achieve sufficient lifetime, energy should be optimized, or power transmitted wirelessly. In summary, the aim is always to achieve very high throughput communication, combined with very long lifetimes. Even more ambitious, we aim to build proof-of-principles for our ideas, and have impact by proving our ideas in hardware. For this, we are building a novel, unique, state-of-the-art test bed for dense networked systems. Clearly, sufficient challenges for our highly skilled and motivated PhD students in the coming years!

37

Fig. 2: Traditional wireless architecture where each user is served in a different time or frequency slot; system throughput remains constant and per user throughput decreases with 1/N. (a) Switched, contention-free wireless communication solution.

2a 2b

38

39

Jan van der Borg is aangesteld op de Toerisme Vlaanderen Leerstoel Tourism Management en werkzaam bij het Departement Aard- en Omgevingswetenschappen van de KU Leuven. Daarnaast is hij professor in de toegepaste economie, in het bijzonder in de Toeristische Economie aan de Universiteit Ca’Foscari van Venetië.

Hij behaalde zowel zijn doctoraal examen als zijn PhD in de regionale economie aan de Erasmus Universiteit Rotterdam. Zijn onderzoek richt zich op de studie van de economische dimensie van het verschijnsel toerisme en in het bijzonder op de duurzaamheid van toeristische ontwikkeling voor stedelijke en regionale economieën.

Speciale aandacht, in het bijzonder toen hij adviseur van de EG, UNESCO en de Raad van Europa was, werd er in zijn werk besteed aan het verschijnsel toerisme in Europese kunststeden, zoals Brugge en Venetië. De opgedane bevindingen kwamen goed van pas bij het verdere advieswerk voor lokale, regionale en nationale overheden op het gebied van destination management, duurzame ontwikkeling en toeristisch beleid.

Hij heeft meer dan 100 artikelen, hoofdstukken van boeken en bijdragen aan conferentiebundels op zijn naam staan. Daarnaast is hij columnist voor een aantal Italiaanse kranten.

40

Toerisme en economie

De toeristische industrie is nu al de grootste (8% van de jobs is in toerisme) en, ondanks de aanhoudende economische malaise, de snelst groeiende industrie van de wereldeconomie. Wereldwijd zijn er in 2013 volgens de UNWTO zo’n 1.078 miljoen mensen op vakantie gegaan.

Ondanks het enorme belang van de sector voor de samenleving wordt er in de academische wereld nog maar weinig belangstelling besteed aan het verschijnsel toerisme. De KU Leuven maakt hierop een positieve uitzondering. Al jaren is er binnen de Divisie Geografie van het Departement Aard- en Omgevingswetenschappen een klein groepje onderzoekers dat zich bezighoudt met toeristisch onderzoek en dat een, voor Vlaanderen unieke, master in het toerisme verzorgt. Toerisme Vlaanderen draagt hier al jaren een steentje aan bij en financiert in deze context een leerstoel Tourism Management waar ondergetekende tijdelijk op aangesteld is.

Een aantal belangrijke en actuele thema’s komen in mijn onderzoek en het onderwijs terug:

1. Tourism and Global Change: de wereld is snel aan het veranderen en het toerisme ondergaat het merendeel van de consequenties van deze veranderingen nogal passief. De veranderingen zijn niet alleen sociaal-economisch van karakter (economische crisis in Europa, de verschuiving van het economische zwaartepunt van de wereld naar Azië en Zuid-Amerika, enz.), maar hebben ook betrekking op het veranderende klimaat en milieu. Als we deze globale veranderingen beter begrijpen en vervolgens zelfs kunnen voorzien, dan kunnen wetenschappers en ondernemers vervolgens ook de veranderingen in de toeristische markt beter begrijpen en eventueel voorspellen. Dit komt het beleid van toeristische organisaties, zowel in de publieke als private sector, uiteraard ten goede;

2. Tourism and Innovation: the toeristische industrie is allesbehalve ‘cutting-edge’. Omdat het overgrote deel van het toeristische bedrijfsleven erg klein en familiegedreven is, is de intensiteit van adoptie en van innovatie diffusie erg laag. Een gedegen analyse van de mechanismen die product- en procesinnovatie in de toeristische industrie bepalen en van de factoren die dit proces beïnvloeden, kan

41

leiden tot een gerichte innovation policy die de toeristische industrie competitiever maakt;

3. Tourism and Infrastructure: de relatie tussen infrastructurele investeringen en toeristische ontwikkeling is een voor de hand liggende. Veel onderzoek naar de precieze relatie tussen beide zaken is nog niet gedaan. Vooral het onderscheid tussen fysieke en immateriële infrastructuur is hier van belang. Vlaanderen is een uitstekende case om deze relatie te onderzoeken;

4. Visitor Management: the-state-of-the-art: een belangrijke ingrediënt van het toeristisch beleid is het zogeheten visitor management. Doel van de visitor management strategie is het toerisme zo intelligent mogelijk te gebruiken om zo sociaaleconomische doelstellingen te verwezenlijken. Aan de hand van een internationaal vergelijkbaar onderzoek kan op een efficiënte manier informatie worden verzameld ten aanzien van de state-of-the-art met betrekking tot visitor management;

5. Impact of Mega Events: steden en regio’s blijven investeren in zogenaamde mega-events. De resultaten die bereikt worden vallen vaak tegen, vooral op middellange termijn, en zeker gezien de enorme investeringen. Met de kandidatuur van Mons/Bergen als culturele hoofdstad van Europa in 2015 is een studie naar de effecten van zulke evenementen voor de lokale samenleving extra interessant, vooral ook voor de instanties en de bedrijven die direct en indirect bij de organisatie betrokken zullen zijn. CBA en IOA maken een belangrijk, maar niet exclusief deel uit van de toolbox waarmee het probleem zal worden aangepakt;

6. Governance of Tourism: toerisme is een uiterst complex en transversaal fenomeen. Het heeft vele raakvlakken en dimensies met andere sociaaleconomische activiteiten en kan hiermee zowel een synergetische als conflicterende effecten genereren. In het algemeen komt deze complexiteit zelden terug in het design van de governance van het toerisme. Dit geld voor alle bestuurlijke niveaus. Onderzoek naar concrete oplossingen voor bestaande governance problemen maakt ook DMOs efficiënter en hun toeristische beleid effectiever;

7. Polarisatie van de toeristische vraag: het lijkt erop dat de toeristische vraag de laatste jaren aan het polariseren is. Na een periode van diversificatie en fragmentatie van de massatoeristische vraag, heeft het er alle schijn van dat de uiteinden van de vraag steeds langer worden (vandaar de term principe van de ‘long tail’) en dat deze ontwikkeling ten koste gaat van het grijze middensegment. De vraag naar B&Bs en naar luxury resorts groeit gestaag terwijl het twee- en driesterrenhotel

42

het steeds moeilijker heeft. De budget airlines groeien samen met de vijfsterrenairlines en de nationale luchtvaartmaatschappijen hebben het steeds moeilijker.

Dezelfde ontwikkelingen zien we terug in de catering en de car rental sector. Wat nog interessanter is, is het feit dat de veranderende consumptiepatronen de klassieke relaties tussen vraag, prijs en inkomen, zo kenmerkend voor de traditionele micromodellen, steeds zwakker lijken te worden. Onderzoek naar deze polarisatie van consumptiepatronen en de factoren die deze trend bepalen, vooral wat betreft de toeristische vraag, is nog erg schaars. Een beter begrip van deze tendens is van fundamenteel belang voor het werkveld;

8. Toerisme en WEB2.0: over de impact van nieuwe technologieën, met name die in de ICT-sfeer, is al heel wat bekend. Wat de impact van het WEB2.0, een verzamelnaam voor alle sociale internetapplicaties, op de toeristische vraag en aanbod precies zal zijn is echter nog onduidelijk. Dat er door het toeristische bedrijfsleven en door DMOs in deze applicaties en hun toepassingen voor smartphones en tablets moet worden geïnvesteerd, is inmiddels wel duidelijk, maar wat exact het business model is dat de investeerders in staat stelt deze investeringen ook daadwerkelijk terug te verdienen, is nog onduidelijk.

Wat voor Google, Twitter en Facebook lijkt te gelden, lijkt minder waar vaar hun toeristische declinaties. Onderzoek naar dit fenomeen helpt de toeristische industrie competitiever te worden en de DMOs slagvaardiger;

9. Stakeholdergerichte branding: verschillende toeristische gebruikersgroepen hebben zo hun eigen kijk op een bestemming. Dit maakt het hebben van een ongedifferentieerde brand niet altijd wenselijk. In sommige gevallen is er zelfs sprake van een conflictsituatie. Een goed voorbeeld zijn de koffieshops in Nederland: een selling point voor jongeren; een doorn in het oog van de behoudende bezoeker.

Het differentiëren van een brand of zelfs specifiek brandingbeleid per groep van stakeholders en/of marktsegmenten ligt voor de hand. Onderzoek naar de praktische mogelijkheden van gedifferentieerde branding helpt DMOs en toeristische ondernemingen hun bestemmingen en producten beter te positioneren en dus makkelijker aan de man te brengen;

10. Optimal Use of Natural and Cultural Heritage: natuurlijk en historisch erfgoed is niet reproduceerbaar en vaak in publieke handen. Om deze redenen kunnen we geen toevlucht zoeken tot de marktwerking om het

43

optimale gebruiksniveau door toeristen te bepalen en te garanderen. Niet optimaal gebruik van erfgoed is wijdverbreid. Erfgoed is vaak of ondergebruikt of overgereikt. Beide verschijnselen zijn vormen van unsustainable tourism development. Onderzoek naar de grenzen aan het gebruik heeft zich vooralsnog gericht tot situaties van overgebruik en tot de definiëring van de zogenaamde carrying capacity. Verder onderzoek is echter nog nodig om bestemmingen tegen hun eigen toeristische enthousiasme te beschermen. Onderzoek naar de ondergrens van de sustainability is nog erg schaars maar van strategisch belang voor al die bestemmingen die erover denken hun toeristische vocatie aan te scherpen.

44

45

Pol Van Dorpe was born in Leuven on November 2, 1978 and is married to Julita Jarmuz and is father of Lisa and Mattes.

He graduated as Master of Science in Electrical Engineering, specialized in microelectronics, in 2001. In 2001 he started to work on his PhD in the field of spintronics at imec and KU Leuven being funded by an IWT fellowship. In 2006 he received his PhD in electrical engineering at KU Leuven and after a short stay in Stanford University, he pioneered the activities on plasmonics for biosensing applications as an FWO post-doctoral fellow in imec (2006-2012). Since October 2012 he is a principal scientist of imec and has also been appointed as part-time (10%) associate professor (hoofddocent) at the Physics Department at KU Leuven. His work has led to over 60 peer reviewed papers and multiple patents. His current research interests are focused on the application of integrated photonics to miniaturise biophotonic devices.

46

Nanobiophotonics

Nanophotonics is a field of research, exploiting the interaction of light with nanostructures for a number of new applications.

One of the most successful branches of this field is based on high index contrast waveguides, which has led to the now maturing field of silicon photonics. Light waveguides are well known and widely used nowadays in fiber optics. Fibers are light guides that function based on total internal reflection. Due to the limited index contrast that is feasible in this material system, the dimensions of the core of the fibers are always several micrometers thick and the bend radius of such waveguides is in the order of centimeters. High density integration of photonic functionality on chip is, hence, not feasible with this material set. In semiconductor processing, however, we have a set of materials that provides high index contrast and shows transparency and limited dispersion in relevant wavelength regions, such as the telecom wavelengths (silicon) or the visible and near-infrared (silicon nitride). Due to the high index contrast, light can be guided and manipulated on chips, leading to numerous new applications, e.g. in high density processing of optical signals in telecom networks, in chip-to-chip and board to board communication (replacing copper wiring), and, increasingly in miniaturizing bulky biophotonic systems.

Indeed, photonics is widely used in the life sciences, in microscopy, imaging, sensing and spectroscopy. Molecular fluorescence is employed standardly as a tool for high resolution imaging and high sensitivity detection schemes (such as in genome sequencing or mapping). Often these optical (microscope) systems are bulky, slow and low throughput. Simultaneously, semiconductor processing has reached the era of nano-electronics. Complex electronic circuits can be printed with nanometer accuracy. Similarly, also photonic circuitry can be fabricated with the same accuracy and with the same stable set of materials. Combining high index contrast waveguide photonics at visible frequencies with high density electronic chips can generate a wealth of new, highly integrated optical applications.

Miniaturized spectrometers, microscopes, fluorescence detectors and single molecule fluorescence tracking devices are just a few examples of the possibilities that face us. There is a strong opportunity to combine

47

innovative photonic devices with biological applications. Ultimately, this can be a game changer for health care, where optical systems are expensive and therefore less prone for consumer applications. New, highly integrated and miniaturised optical systems can enable to broaden the use of these applications to the consumer space.

A second increasingly popular branch of nanophotonics is based on nanopatterned metals. Traditionally, metals are not assessed as particularly interesting for photonic applications, mainly due to their strong absorption and/or reflection of electromagnetic radiation. However, the application of an electromagnetic field at the interface between a dielectric and a metal can give rise to a collective oscillation of electronic charges along that interface: a surface plasmon polariton. This phenomenon has applications in various areas.

As surface plasmon polaritons (or surface plasmons) exhibit much shorter wavelengths than the light that excites them, surface plasmons offer the opportunity to guide light in much smaller structures than common dielectric waveguides. Because of this, plasmon polaritons are interesting candidates to function as information carriers in nanoelectronic-photonic circuits, where high density, high speed communication at the nanoscale is necessary. Moreover, the properties of the surface waves strongly depend on the dielectric environment, enabling several sensing applications. Nanostructuring metals allow for the excitation of ‘local plasmons’, which, depending on the geometry, strongly enhance the local electromagnetic field. These kind of structures are called ‘nanoantennas’.

These antennas can be designed to exhibit very special optical properties. We have spent quite some time on understanding the microscopic behaviour of these antennas and by carefully simulating the behaviour of custom shaped nanostructures, we have succeeded in unraveling the different loss mechanisms that contribute to the total loss of the plasmonic resonance. This has allowed us to come up with new designs leading to improved functionality. There are several applications for these antennas, such as local refractive index sensing, highly directive scattering, directive guiding of fluorescent emission and nanosized optical modulators.

In the case of local refractive index sensing, the antenna resonance strongly depends on the refractive index of the surrounding medium in its near field. Hence, its resonance position is susceptible to changes upon changes in the local dielectric environment. We can take advantage of this effect to build nanoplasmonic biosensors, which respond to binding of specific biomolecules on the metal surface. In order to reach

48

low detection limits, such binding events need to result in a measurable shift. This can be achieved when the shift is appreciable compared to the linewidth of the resonance. Therefore, we have focused on optimizing both the shift and the linewidth of plasmonic resonators, an at that time unexplored field. Another physical phenomenon taking place near plasmonic nanoantennas, excited at their resonant wavelength is the strong increase of the local electric field. It is well known that this can lead to strong enhancement of Raman scattering – a vibrational technique allowing direct molecular identification, but usually suffering from low signals.

This enhancement (usually called surface enhanced Raman scattering (SERS)) can even lead to single molecule sensitivity. Nevertheless, real applications were hindered by the irreproducibility of the metal aggregates typically used for SERS. We have designed and fabricated novel types of nanostructures, that can either be fabricated in a cost-effective way on large substrates, or specific nanostructures integrated with nanofluidics to controllably investigate single molecules.

49

50

51

Pieter Vansteenwegen was born in Leuven on November, 9, 1979. He is married to Sofie Beersmans and father of Sien (2007), Adriaan (2011) and Lena (2013).

He graduated as Master of Science in Mechanical Engineering and obtained a PhD in operations research at KU Leuven Faculty of Engineering in 2008 with a dissertation entitled ‘Planning in Tourism and Public Transportation’. After two years of postdoctoral research funded by the Fund for Scientific Research (FWO) at KU Leuven, he became an assistant professor in October 2010 at the Department of Industrial Engineering at UGent.

After a research stay at UPC in Barcelona he is, since October 2012, an assistant professor at the Centre of Industrial Management, Traffic and Infrastructure at the Mechanical Engineering Department. His research aims at developing more robust public transportation services and metaheuristics for solving complex logistic planning problems. He teaches about public transportation, operations research and distribution logistics in the new Master of Science in Engineering: Traffic, Logistics and Intelligent Transportation Systems (VLITS).

52

From personalised city trip planning to robust railway timetables

Operational researchIn the engineering field of operational research, mathematical modelling and quantitative techniques are developed in order to support decision making or to optimise the performance of a system. Operational research is often concerned with determining a maximum (such as profit or robustness) or a minimum (such as cost or risk). This research domain is situated between the fields of economics, psychology, mathematics and statistics and uses elements from all these domains. Originally it was applied in the domain of production planning, but nowadays it is also frequently used in other domains such as, for example, logistics, financing and public transportation.

An optimisation problem that needs to be solved, will typically be modelled by an objective function (what exactly needs to be optimised?) and a number of constraints (which limitations should be considered?). Then, an appropriate solution approach should be developed. Exact solution approaches always obtain the optimal solution but are mostly too time and memory consuming to be applied on real life instances. Therefore, (meta)heuristics are frequently developed since they are much faster and also lead to high quality solutions. A metaheuristic will quickly evaluate a high number of small changes to a current solution and select the best change to be actually applied. When no more improvements are feasible, the current, ‘locally optimal’ solution is disturbed in some way. Then again, small improvements are applied, in the hope to arrive at a new and better local optimum. This process of trial and error is repeated until a stopping criterion is reached. For complex real life problem instances, this process can obtain, when designed carefully, solutions which are only a few percent away from the optimal solution, in less than a second.

In my research and teaching, I mainly apply operational research techniques to complex logistical problems and (the operational planning of) public transportation. Both will be explained in more detail in the next sections.

From personalised city trip planning…Many interesting and complex logistical problems can be modelled with (variants of) the orienteering problem. In this problem, a set of potential customers is given, each with a profit. The goal is to determine a route, limited in length (constraint), which visits a selection of customers and maximises the sum of the collected scores (objective function).

53

Two decisions need to be made in an integrated way: finding the best possible selection of customers and determining a route between these selected customers, that respects the available time. The complexity of this problem can be illustrated by the fact that for a small problem with only 20 customers, a computer, which can evaluate one billion solutions per second, needs 210 years to evaluate all possible solutions.

My interest in the orienteering problem was triggered by the wish to develop an application that proposes personalised tourist trips. Most tourists staying in a city or region are limited in time and cannot visit every interesting point-of-interest. Therefore, a selection will have to be made and a sequence will have to be decided in order to maximise the tourist’s pleasure (objective function). We define this problem as the ‘tourist trip design problem’ and this can be modelled by the orienteering problem. Obviously, a number of extra constraints or objectives are taken into account: opening hours, limited budget for entrance fees, avoid or impose certain combinations of attractions, lunch breaks, automated hotel selection, integrating pubic transportation, etc. All of these are tackled and a metaheuristic solution approach is designed to solve problems with around 200 possible points-of-interest near to optimality in around one second. This research is put into practice and the personalised city trip planner is available for the public at www.citytripplanner.com.

54

… to robust railway timetables… All over the world, public transportation operators are under pressure to improve their services and save costs at the same time. Therefore, operations research techniques are frequently used in the domain of public transportation. My research focuses on improving the robustness (objective function) of the Belgian railway system, in cooperation with Infrabel. Robustness is defined as minimising the total passenger travel time in practice, taking into account passenger numbers, expected delays and missed transfers.

Obviously, many constraints should be taken into account as well: safety regulations, capacity of the infrastructure, minimal driving times, available crews and vehicles, etc. Currently, a mathematical model is under construction that can optimise the timetable for the complete Belgian railway network, taking into account all considerations of both Infrabel and NMBS (objectives and constraints). Based on some preliminary results, the percentage of missed transfers can be reduced from around 11% to 2% and the total passenger travel time (of which around 70% corresponds to fixed driving times) can be reduced by 7%, only by improving the timetable.

Furthermore, the capacity of the railway system in the North-South-corridor in Brussels and the station area of Antwerp is evaluated. A combination of exact techniques, metaheuristics and simulation is developed in order to optimise the usage of the capacity and to increase the robustness of the system in station areas.

55

…and other applications.Similar optimisation techniques from the field of operational research can be applied to improve the robustness of public bus transport timetabling and network design. Obviously, this is more complicated than for railways since buses share the network with cars and are much more affected by congestion. Another application area is the dynamic repositioning of bikes in the bike sharing systems. This repositioning is required to provide enough empty spaces (or bikes) in each station so that customers can leave (or collect) their bike at the station they prefer.

Master of Science in Engineering: Traffic, Logistics and Intelligent Transportation Systems (VLITS)In 2011, we started a new master on traffic, logistics and ITS. Next to traffic engineering and traffic management, the students in this master learn how operational research can be applied to optimise logistic and public transportation networks.

56

57

Marian Verhelst was born in Leuven (Belgium) on May 15, 1980. She is married to Thomas Plomteux and mother of Marie, Emma and Laura.

In 2003, she graduated as Master of Science in Electrical Engineering at KU Leuven, specializing microelectronics and started PhD studies at KU Leuven on ultra-low power reconfigurable wireless transceivers, funded by the Fund for Scientific Research (FWO)-Flanders. In the summer of 2005, she was a visiting scholar at the Berkeley Wireless Research Center (BWRC) of UC Berkeley, USA. After graduation in 2008, she joined Intel Labs, Hillsboro OR, USA, doing research on digital assistance of wireless radio frontends.

As of October 2012, she is an assistant professor with a research focus on self-adaptive and self-learning circuits and systems. Outside of her research interests, she is passionate about science communication. She is the lead of project InnovationLab, creating STEM projects for high schools, and is a member of the Young Academy of Belgium.

58

Making chips smarter

In the coming decade, the continuous miniaturisation and cost reduction of sensing, processing and communication hardware will allow to equip virtually any object or surface with advanced sensing capabilities. This will fundamentally change the way users interact with objects. Firstly, users will no longer have to adapt to a particular interface, but multiparameter sensing interfaces can submit to natural interaction mechanisms of users e.g., through voice or gestures. Secondly, objects can in turn benefit from context awareness, and user perception to deliver context-based services. The real breakthrough of this ubiquitous sensing vision is however only feasible though the availability of miniature, invisible sensing devices, autonomously operating on scarce energy budgets, and efficiently dealing with the massive data overload from all these embedded sensors. Yet they have to be always-on, ever-listening, and ready to be triggered by our actions. The energy budget currently required for such continuous sensing and embedded sensor data processing is fundamentally holding back the birth of such sensory swarm.

The major roadblock towards more energy efficient embedded sensors is the inherent lack of information awareness of current sensing devices. Nowadays’ static sensing systems consume energy regardless of the effective information content present in, or the fraction of information actually used from, the observed signal (Fig. 1). Typically sensing interfaces amplify and sample the incoming signal at fixed worst-case settings and process the result on a DSP processor. For example, a camera-equipped sensing node will sample a fixed amount of frames per second, at a fixed pixel resolution per frame, and with a fixed post-processing object recognition algorithm running on a processor. The static system is designed for its worst-case conditions, and its resulting high energy consumption simply impedes always-on sensing. This is apparent from present-day devices, where for instance already a button has to be pushed before activating the acoustic processing for iPhone’s Siri speech app.

The aim of my research is to overcome this roadblock and enable the widespread deployment of always-on sensing devices for smart environments. I do this through combining runtime reconfigurability and embedded machine learning to enable context-aware operation of the sensor node.

59

Run-time reconfigurability for context-adaptivityTo be able to deal with the sensory data overload in a energy-efficient way, it is important to discard irrelevant data as soon as possible, and extract only information-carrying features. It has, for example, been well established that processing the raw data within the sensing device is far more cost-effective compared to transmitting raw data to a central data-collecting node. Taking this one step further, also within a sensing node relevant information should be extracted as close to the raw sensor as possible to avoid power wastage.

Traditionally, such data reduction would be achieved through application-specific optimisations that perfectly tailor local information extraction towards the current application requirements and environment constraints. However, the emerging applications of targeted ubiquitous sensing systems are very dynamic, characterized by rapidly varying operational contexts. As an example, both the operating conditions and the performance requirements of sensors integrated in clothes will vary hugely when their owner is sleeping, running or not wearing the clothes at all. Tailoring efficiency of such ubiquitous sensing systems to particular operating conditions is hence unfeasible at design time and there is a strong need towards run-time reconfigurable systems.

Fig. 1: Information-aware sensor interfaces autonomously adapt to the operating context to always extract relevant information at minimal energy cost.

60

Run-time reconfigurable systems are electronic systems whose behaviour, and resulting energy consumption, can be dynamically adapted at run-time. This allows to at any time configure the sensing system according to the current context it operates in, and the amount of information present in the sensed signal. These context- and information-aware sensors as such reduce sensing and processing resources when the incoming signal information content is low, to save energy and compress the incoming data.

Embedded machine learning to control self-adaptivityA second challenge for these run-time reconfigurable systems lies in the control of their configurability knobs. One cannot expect the user to control the mass of embedded sensing systems operating in his environment. Widespread applicability requires the embedded sensors to autonomously derive context information, and link sensed context to their own optimal configuration settings. It is again not possible to predict this optimal context-configuration relationship off-line due to the limited visibility into the dynamic operational context. As a result, the targeted sensor nodes should be equipped with embedded intelligence to at run-time optimize their own behaviour, and learn from experience. Moreover, the overhead of this embedded intelligence should be minimal, to ensure net energy gains. We pursue this through the embedding of machine learning on-chip, at circuit level. This is done in collaboration with the DTAI Machine Learning Group of Computer Science, KU Leuven.

Application targets and CubiqLabIt is clear that to achieve this vision of context-aware environments, through such self-scalable distributed sensor nodes, expertise from a range of research areas will have to be combined. I therefore founded a collaboration framework with young researchers in complementary research labs within KU Leuven.

Fig. 2: www.esat.kuleuven.be/CubiqLab/

61

Under the new CubiqLab umbrella (Context-aware UBIQuitous electronics), we join forces with the TELEMIC team (Sofie Pollin, networking), STADIUS team (Alexander Bertrand, distributed and adaptive processing), DTAI team (Wannes Meert, machine learning), and MICAS team (myself, reconfigurable hardware).

Within CubiqLab, we study the above concepts in a generic way, and apply them to several key sensor domains. Currently, CubiqLab focuses on acoustic sensing, and image sensing. Self-scalable audio and image sensors enable to bring these technologies to new, energy scarce applications, such as remote surveillance, ambient assisted living, and wearable electronics, to name a few. Additionally, in collaboration with imec I apply these concepts to biomedical sensor interfaces.

I look forward to continue to build out my research lab around self-learning and self-adaptive sensors and processors, and especially towards the many opportunities it offers to collaborate with various application partners. I am excited about the challenges in this newly developing field, and eager to develop solutions to make such technology a reality.

Foto’s binnenwerk: Rob Stevens

FACULTEIT INGENIEURSWETENSCHAPPEN

Kasteelpark Arenberg 1 bus 22003001 HEVERLEE, België

tel. + 32 16 32 13 50fax + 32 16 32 19 82

info@eng.kuleuven.bewww.eng.kuleuven.be

v.u.

: Jel

le D

e B

orge

r, K

aste

elpa

rk A

renb

erg

1, 3

001

Hev

erle

e