advanced research center for nanolithography

NWO Institute ARCNL - Advanced Research Center for Nanolithography

The Advanced Research Center for Nanolithography (ARCNL) is a research centre that is part of NWO-I.   ARCNL is a public-private partnership between NWO (previously FOM), University of Amsterdam, VU Amsterdam and the company ASML. ARCNL is a new type of organisation unit within NWO that combines the best of both worlds: the scientific strength of NWO and its university partners with the application-oriented, demand-side management of a private party.

ARCNL's mission

ARCNL focuses on the fundamental physics and chemistry behind current and future technology for nanolithography, especially for application in the semiconductor industry. The scientific programme of ARCNL is closely related to the interest areas of ASML. A large part of this programme focusses on the physical and chemical processes that are vital for nanolithography with extreme ultraviolet (EUV) light. ARCNL's research is consolidated in three departments:

• Source: physics behind the generation of EUV light
• Metrology: new principles for producing accurate images and precise measurements
• Materials: materials physics for application in lithography instrumentation

ARCNL occupies an international position in the research in these three areas.

ARCNL formally started on 1 January 2014 and since 1 September 2015, it has been an independent organisation unit within NWO-I. ARCNL receives administrative and technical support from its neighbouring institute AMOLF.

Director ARCNL

Dr. Wim van der Zande 

Communication ARCNL

Petra Vastenhouw

Contact details

ARCNL Advanced Research Center for Nanolithography PO Box 93019 NL-1090 BA Amsterdam +31 20 851 7100 [email protected] Visiting address: Science Park 106 1098 XG Amsterdam YouTube

www.arcnl.nl

Follow ARCNL on

ARCNL: Advanced Research Center for Nanolithography

The Advanced Research Center for Nanolithography (ARCNL) is a research centre that is part of NWO-I. ARCNL is a public-private partnership between NWO, University of Amsterdam, VU Amsterdam, University of Groningen and the company ASML. ARCNL is a new type of organisation unit within NWO that combines the best of both worlds: the scientific strength of NWO and its university partners with the application-oriented, demand-side management of a private party.

ARCNL's mission

ARCNL focuses on the fundamental physics and chemistry behind current and future technology for nanolithography, especially for application in the semiconductor industry. The scientific programme of ARCNL is closely related to the interest areas of ASML. A large part of this programme focusses on the physical and chemical processes that are vital for nanolithography with extreme ultraviolet (EUV) light. ARCNL's research is consolidated in three departments:

• Source: physics behind the generation of EUV light
• Metrology: new principles for producing accurate images and precise measurements
• Materials: materials physics for application in lithography instrumentation

ARCNL occupies an international position in the research in these three areas.

ARCNL formally started on 1 January 2014 and since 1 September 2015, it has been an independent organisation unit within NWO-I. ARCNL receives administrative and technical support from its neighbouring institute AMOLF.

Contact details

ARCNL Advanced Research Center for Nanolithography PO Box 93019 NL-1090 BA Amsterdam +31 20 851 7100 [email protected] Visiting address: Science Park 106 1098 XG Amsterdam

Go the the ARCNL website

ARCNL is a public-private partnership between NWO (Dutch Research Council), University of Amsterdam, VU Amsterdam and the company ASML. ARCNL combines the best of both worlds: the scientific strength of NWO and its university partners with the application-oriented, demand-side management of the private party ASML. ARCNL focuses on the fundamental physics and chemistry behind current and future technology for nanolithography, especially for application in the semiconductor industry.

Subscribe to rss feed

Advanced Research Center for Nanolithography

• Date 6 hours 12 hours 1 day 3 days all
• Rank Last day 1 week 1 month all
• LiveRank Last day 1 week 1 month all
• Popular Last day 1 week 1 month all

Spreading focus for better imaging

Extreme Ultraviolet (EUV) light in microscopy offers the advantage of obtaining a high-resolution image combined with spectral information about the object under study. However, because EUV microscopy uses diffraction instead ...

Optics & Photonics

Jan 25, 2021

Detecting hidden nanostructures by converting light into sound

Researchers at ARCNL have found a way to detect nanostructures buried under many layers of opaque material using high-frequency sound waves induced by light. Their findings could have applications in the semiconductor manufacturing ...

Nanophysics

Jul 8, 2020

The exceptional origin of EUV light in hot tin plasma

Extreme ultraviolet light (EUV light) does not naturally occur on Earth, but it can be produced. In nanolithography machines, EUV light is generated using an immensely hot tin plasma. Researchers at ARCNL, in close collaboration ...

Plasma Physics

May 11, 2020

Fiber imaging beyond the limits of resolution and speed

Researchers at ARCNL and Vrije Universiteit Amsterdam have developed a compact setup for fast, super-resolution microscopy through an ultrathin fiber. Using smart signal processing, they beat the theoretical limits of resolution ...

May 8, 2020

Researchers report on helical soft-X-ray beams

Controlling the properties of light is of great importance for many areas of physics, including imaging and nanolithography. But for short wavelengths, such as soft X-ray radiation, such control over the phase of light has ...

Feb 18, 2020

E-mail newsletter

• Accommodation
• Communication
• Medical Care
• the Netherlands
• Your Family
• AcademicsNL for Ukraine
• Academic Careers

Research Centers

• Research Universities
• Top Institutes
• University Medical Centers
• Higher Education
• Universities of Applied Sciences
• Immigration
• Residence Permits
• What to do when
• Work Permits
• Excellence in Dutch Research
• Partners FactCards
• Research Funding
• Science Parks
• Top Sectors
• University Rankings

ARCNL, The Advanced Research Center for Nanolithography

While the academic setting and research style are geared towards establishing scientific excellence, the topics in ARCNL’s research program are intimately connected with the interests of the industrial partner ASML.

ARCNL is a public-private partnership established in 2014 by the Dutch Research Council (NWO), the University of Amsterdam (UvA), the Vrije Universiteit (VU) and semiconductor equipment manufacturer ASML. In 2022 the University of Groningen (RUG) joined as associate partner. The institute is located at Amsterdam Science Park.

ARCNL combines the best of two worlds. Being a research institute, ARCNL operates in the middle of the academic field. ARCNL shares its scientific output in peer-reviewed journals and at conferences. At the same time, the research questions are inspired by challenges from the semiconductor industry. Driven by Moore’s law, lithography technology developed and produced by ASML touches frequently on fundamental limits. Consequently, ARCNL researchers often find themselves working with extreme scientific challenges. ARCNL not only contributes to future technological developments but is also in the position to shorten the time between invention and possible application.

Source ARCNL

• arcnl.nl/our-mission

Tagged under

• Foundation for Fundamental Research on Matter
• research center
• Advanced Research Center for Nanolithography

Add this FactCard to your website

Is the information on this FactCard relevant to your audience?

Feel free to share this FactCard on your website. This is very easy and will enhance the service level to your visitors.

Simply check the preview, copy the embed code, paste it in your website and you are done!

Wageningen University & Research

Msm, maastricht school of management, embl, european molecular biology laboratory, knaw, royal netherlands academy of arts and sciences, nlr, netherlands aerospace centre, nizo, food research, knmi, royal netherlands meteorological institute, differ, dutch institute for fundamental energy research, nen, nederlands normalisatie-instituut, naturalis biodiversity center, mesa+ institute for nanotechnology, nwo institutes organisation, nwo, netherlands organization for scientific research, ihe delft, institute for water education, ipa, institute for programming research and algorithmics, dutch research institutes, energy transition tno, digital society institute, ib, international baccalaureate, techmed centre, kb, national library of the netherlands, estec, european space research and technology centre, kwr, watercycle research institute, nivel, netherlands institute for health services research, nikhef, national institute for subatomic physics, marin, maritime research institute netherlands, j.m. burgerscentrum (jmbc), sron, netherlands institute for space research, nki, the netherlands cancer institute, 4tu federation, the royal nioz, excellence in marine science, related decks, higher education in the netherlands, phd in the netherlands, water research, technical career, technical universities, health sector.

Stefan Witte Lab Website

• Publications

Latest news

In 2014, the Advanced Research Center for Nanolithography (ARCNL) has started. ARCNL is a new, public-private funded research center founded on the initiative of ASML, in collaboration with FOM, NWO, UvA, and VU University. The focus of ARCNL is to perform cutting-edge fundamental science relating to nanolithography. I will become group leader of the 'EUV Generation and Imaging' Group at ARCNL, in collaboration with Prof. Kjeld Eikema.

There are several exciting job opportunities for PhDs and postdocs. Check out the ARCNL website for details.

Lensless imaging with ultra-broadband high-harmonic sources

We developed a method that enables efficient lensless imaging with ultra-broadband light sources, using coherent pairs of pulses. This method works well with visible light sources, but especially with table-top extreme-ultraviolet radiation sources such as produced by high-harmonic generation. We recently published the paper on this invention in Light: Science and Applications !

Miniature phase contrast microscopes in Optics Letters

Our paper on the development of miniature lensless microscopes for live cell imaging has been published in Optics Letters ! Check out the paper for some high-res quantitative phase images of live neurons, recorded using only a few laser diodes and a camera.

FOM Valorisation grant

February 2014: together with Prof. Kjeld Eikema, I was awarded a Valorisation grant from FOM. This grant provides 50 kE in funding for our project entitled "Compact lensless microscopes for quantitative phase contrast imaging". Check out the press release here .

FOM Projectruimte Grant

January 2014: I received a 400 kE 'Projectruimte' grant from FOM (Foundation for Fundamental Research on Matter), for my proposal entitled "Beyond optical microscopy: Phase-contrast imaging of cells with a table-top soft-X-ray Microscope"!

Our research is centered around the theme of biophotonics and medical imaging. The aim is to improve microscopy and biomedical imaging by using advanced lasers and optical techniques. To this end, we work on a variety of projects that are all linked to this theme:

Soft-X-ray imaging with a table-top high-harmonic source

To this end, we are constructing a table-top soft-X-ray source based on high-harmonic generation (HHG), and are working on novel imaging methods that allow imaging with such sources. Since good-quality imaging optics for soft-X-rays are not (yet) available, we are focusing on so-called lensless imaging techniques: in lensless imaging, a diffraction pattern rather than an image of an object is recorded. In specific cases, an image of the object can then be reconstructed numerically, either through holographic detection or iterative phase retrieval algorithms.

A major limiting factor for coherent diffractive imaging is the ultra-broadband spectrum that is typically emitted by a HHG source, as a finite spectrum results in a blurry diffraction pattern. We have recently developed a method that enables spectrally resolved imaging with a HHG source (already down to 47 nm wavelength), while using the full source spectrum efficiently. We recently published our paper on these results in the new journal from the Nature Publishing Group, Light: Science and Applications . The online version can be found here ! This project is a collaboration with the group of Prof. Kjeld Eikema.

Ultrafast laser development and optical parametric chirped pulse amplification

Currently, we are setting up a new system, based on similar OPCPA technology, but now using a new pump laser approach based on quasi-CW diode pumped Nd:YAG amplifiers. This pulsed diode pumping approach enables much higher repetition rates than our previous, flashlamp-pumped amplifier system. The final OPCPA output will again produce terawatt peak power few-cycle pulses, but this time at a repetition rate of 300 Hz.

The Nd:YAG pump laser is now fully operational, and our paper describing it has recently been accepted by Optics Letters.

Lensless imaging and miniaturized microscopes

We are now introducing some of our lensless imaging concepts, which we originally developed for soft-X-ray microscopy, into an optical imaging device for live cell imaging. First results are highly encouraging: we find that by recording diffraction patterns at a few different wavelengths of the light, we can retrieve the phase in a very robust way, without the need for support constraints or moving objects in the microscope. We have applied this method to image live neurons in a culture dish, and we obtain quantitative phase contrast images of the cells at sub-2-micron spatial resolution, using a very small and robust lensless microscope. We have recently published these results in Optics Letters, the paper can be found here .

Nonlinear microscopy and biological imaging

I have also been involved in the development of nonlinear microscopy methods aimed at live brain imaging. We discovered that third-harmonic generation (THG) microscopy is an excellent method for live brain imaging with cellular resolution. A more detailed explanation of this project, including some cool pictures, can be found here , as well as in our paper in PNAS .

Coherent control and precision spectroscopy (PI Kjeld Eikema)

I have been involved in a series of experiments conducted by Kjeld Eikema's group, where we showed that coherent control techniques can be used to turn off this single-sided excitation, while the Doppler-free signal remains at full strength. In addition, complex spatially localized excitation patterns can be engineered, effectively allowing spatial coherent control.Furthermore, high-resolution Doppler-free spectroscopy can now be performed using direct frequency comb excitation, which has many advantages for e.g. XUV spectroscopy. These results were recently published in Nature Photonics (check out the January 2013 cover!) and PRL .

Full publication list (most recent first)

• The Scientist, Nov 2011, “Brainspotting”
• SciBX 4 (15), “Third-harmonic generation microscopy for guiding brain surgery”
• This week in PNAS, April 12 2011, “Live brain imaging without labels”

Current composition of the research team

Dr. Stefan Witte - Group leader

Daniel Noom, M.Sc. - PhD student

Dirk Boonzajer Flaes, M.Sc. - PhD student

Matthijs Jansen, B.Sc. - Masters student

Martijn Stoffels, B.Sc. - Masters student

Elias Labordus - Bachelors student

On the topic of X-ray microscopy, I collaborate with Prof. Dr. Kjeld Eikema, who heads the Ultrafast Laser Physics and Precision Metrology Group at VU University Amsterdam

Short CV of Dr. Stefan Witte

Former members and group alumni

Vasco Tenner - M.Sc. degree June 2013 (cum laude). Present: PhD student at Leiden University .

Martijn Stoffels - B.Sc. project April 2013

Menno Pleijster - B.Sc. project Feb 2012

Further contact Details:

Office Room T0.61 Phone +31-(0)20-598 1508 Laser Lab Room KA 1.91a Phone +31-(0)20-598 7446

Links Faculty of Sciences Department of Physics LaserLaB

Section Biophotonics and Medical Imaging - VU University Amsterdam

Address : De Boelelaan 1081, 1081 HV Amsterdam, The Netherlands Telephone :+31(0)205981508 Fax :+31(0)205987992

View the latest institution tables

View the latest country/territory tables

Advanced Research Center for Nanolithography (ARCNL), Netherlands

Overall research output.

Overall Count and Share for 'Advanced Research Center for Nanolithography (ARCNL)' based on the 12-month time frame mentioned above.

View all articles

Research outputs by subject area

Note: Articles may be assigned to more than one subject area. Hover over the donut graph to view the Share for each subject.

Compare with other institutions

Nature Strategy Reports

Identify research insights to guide research strategy and grow your impact with our Nature Strategy reports.

Actionable insights into research performance.

Detailed analysis of strengths and weaknesses.

Covers all major disciplines and regions of the world.

Nature Navigator

Data-driven insights for informed research strategy decisions with AI summarisation technology.

Track trends, summarise articles effortlessly.

Spark serendipitous discoveries.

Uncover collaboration opportunities.

Collaboration

International vs domestic collaboration by share.

Note: Hover over the graph to view the percentage of collaboration.

Top 5 domestic collaborators with Advanced Research Center for Nanolithography (ARCNL) by Share

Collaborating institutions

Top 5 international collaborators with Advanced Research Center for Nanolithography (ARCNL) by Share

Related news.

Advertisement

Latest supplement

Nature index 2024 health sciences.

More money is being spent on medical R&D than ever before, but with few new drugs to show for it. The race is on to find technological solutions that will speed up drug development and deliver better clinical outcomes. Nature Index Health Sciences 2024 highlights the people and projects that are bringing fresh perspectives to some of the most complex challenges in medicine and provides an in-depth look at how one country – the United States – is dominating the field.

Relationships

Partner institutions.

Advanced Research Center for Nanolithography (ARCNL) is a research collaboration whose article contributions are accrued to its participating partner institutions below.

• NWO-I, the Institutes Organisation of NWO
• University of Amsterdam (UvA)
• VU Amsterdam

• Previous Article
• Next Article

I. INTRODUCTION

Ii. numerical simulations, a. code description, b. simulation parameters, iii. wavelength-dependent plasma dynamics, a. establishment of steady-state plasma flow, b. wavelength-dependent power partitioning, iv. intensity-dependent plasma dynamics, a. dependence of spatial plasma features on laser intensity, b. intensity-dependent power partitioning, v. summary and conclusion, acknowledgments, author declarations, conflict of interest, author contributions, data availability, simulations of plasmas driven by laser wavelengths in the 1.064—10.6 μ m range for their characterization as future extreme ultraviolet light sources.

• Split-Screen
• Article contents
• Figures & tables
• Supplementary Data
• Peer Review
• Open the PDF for in another window
• Reprints and Permissions
• Cite Icon Cite
• Search Site

D. J. Hemminga , O. O. Versolato , J. Sheil; Simulations of plasmas driven by laser wavelengths in the 1.064—10.6 μ m range for their characterization as future extreme ultraviolet light sources. Phys. Plasmas 1 March 2023; 30 (3): 033301. https://doi.org/10.1063/5.0125936

Download citation file:

• Ris (Zotero)
• Reference Manager

We characterize the properties of extreme ultraviolet (EUV) light source plasmas driven by laser wavelengths in the λ laser = 1.064 − 10.6   μ m range and laser intensities of I laser = 0.5 − 5 × 10 11 W cm−2 for λ laser = 1.064   μ m. Detailed numerical simulations of laser-irradiated spherical tin microdroplet targets reveal a strong laser-wavelength dependence on laser absorptivity and the conversion efficiency of generating in-band EUV radiation. For λ laser = 1.064   μ m irradiation, the increase in in-band radiation with increasing laser intensity is offset by only a minor reduction in conversion efficiency. Radiative losses are found to dominate the power balance for all laser wavelengths and intensities, and a clear shift from kinetic to in-band radiative losses with increasing laser wavelength is identified. Yet, with increasing laser intensity, such a shift is absent. We find that the existence of a maximum conversion efficiency, near λ laser = 4   μ m, originates from the interplay between the optical depths of the laser light and the in-band EUV photons for this specific droplet-target geometry.

Extreme ultraviolet lithography (EUVL) is driving mass production of today's most advanced integrated circuits (ICs). 1,2 Crucial to the success of this technology has been the development of a sufficiently powerful, stable, and “clean” source of EUV radiation 3 concentrated in a narrow 13.5 nm ± 1 % region where molybdenum/silicon multilayer mirrors exhibit high reflectance (the so-called “in-band” region). 4,5 This radiation is most efficiently generated in a laser-produced plasma (LPP) formed when high-intensity CO 2 laser light (laser wavelength λ laser = 10.6   μ m ⁠ ) is focused onto pre-shaped tin microdroplet targets. 6–12 This plasma contains large populations of Sn 11 + − Sn 15+ ions, which generate intense, narrowband EUV radiation through bound–bound atomic transitions. 13–19  

Nowadays, industrial EUV sources generate a remarkable 250 W of in-band EUV power. 20 Efforts to increase source powers beyond 600 W are now under way to facilitate increased wafer throughput. 21 While CO 2 laser-driven plasmas are the backbone of current EUV sources for high-volume manufacturing, 3,22 rapid developments 23 in solid-state lasers (which typically operate in the near- to mid-infrared wavelength range) make them a viable alternative 24–26 in the future due to their high efficiencies in converting electrical power to laser light and their potential for scaling to high average powers. 23,27 These developments have sparked significant interest in the study of short-wavelength laser-driven plasmas for nanolithography. 24–26  

Two crucial topics to be addressed in such studies are the impact of (i) laser wavelength λ laser and (ii) laser intensity I laser on the radiative and kinetic properties of the plasma. The kinetic properties are especially important in the context of debris generation, 28,29 which may limit the lifetime of optical components. The laser wavelength sets the critical electron density (the density at which the plasma becomes opaque to incident laser radiation) according to n e , cr ∝ λ laser − 2 ⁠ . This determines the densities for which EUV radiation is generated and must propagate through. High plasma densities or long path lengths generate large optical depths, which redistribute in-band energy into other channels, e.g., via spectral broadening. 30 While this effect can be negated by moving to long laser wavelengths, 31 the presence of steep electron density gradients in the plasma can lead to significant laser reflection from the critical surface and a loss of input laser energy. 32,33 Efficiency at different laser wavelengths is, therefore, determined by a trade-off in numerous underlying physical processes. The goal of the present work is to quantify this trade-off for laser wavelengths lying in the largely unexplored region between λ laser = 1.064   μ m (Nd:YAG laser) and 10.6   μ m ⁠ , although the existence of an optimum laser wavelength has been suggested previously in the EUV source community. 34,35 Furthermore, we aim to quantify the effect of a second trade-off between absolute power output and the efficiency of generating in-band radiation during illumination by a range of laser intensities.

In this article, we present a comprehensive characterization of the properties of laser-produced EUV source plasmas. The parameter space encompasses (i) laser wavelengths in the 1.064 ≤ λ laser ≤ 10.6   μ m range coupled with an optimum intensity scaling I laser ∝ λ laser − 1 and (ii) laser intensities in the 0.5 × 10 11 ≤ I laser ≤ 5 × 10 11 W cm −2 range for λ laser = 1.064   μ m ⁠ . We identify a strong laser-wavelength dependence on laser absorptivity and EUV generation efficiency, and, moreover, report on the early time establishment of steady-state plasma flows. The partitioning of laser power into plasma components (kinetic, internal, and radiated power) is quantified. In our case study, we delve into the factors underlying a maximum in the conversion efficiency (ratio of in-band power radiated into the laser-illuminated half-sphere to laser power) for λ laser ≈ 4   μ m laser irradiation, a promising wavelength region indicated previously. 34 It should be emphasized that this is not a global optimum; higher conversion efficiencies may be achieved at longer wavelengths by optimizing laser absorption by intricate target preparation, 20 which is not discussed in the presented case study. The partitioning of laser power into the plasma components is also quantified as a function of the laser intensity, where only weak dependencies are found for the kinetic and radiated power. Moreover, we report on the spatial profiles of various plasma parameters (temperature, electron density etc.) while the plasma exhibits steady-state flow. Finally, we observe an increase in in-band radiative power with increasing laser intensity at only a small cost to the overall efficiency.

We have performed numerical simulations of laser-produced tin plasmas using the two-dimensional radiation-hydrodynamics code RALEF-2D. 36–38 It has proven to be a powerful modeling capability for EUV source plasmas, providing useful insights on various topics such as tin droplet fragmentation, 39 droplet propulsion and deformation, 11,40 plasma properties, 19,33 and late-time plasma expansion. 29 In the following, we provide details of the code and the simulation parameters used in this study.

RALEF-2D solves the equations of single-fluid, single-temperature hydrodynamics including the processes of radiation transfer and thermal conduction. The hydrodynamic component of RALEF-2D is based on an upgraded version of the 2D CAVEAT code, 41 where the hydrodynamic equations are solved on a structured quadrilateral mesh using a second-order Godunov-type scheme. Radiation transfer, which is implemented using a symmetric semi-implicit method with respect to time discretization, 42,43 is modeled using the quasistatic, local thermodynamic equilibrium (LTE) radiation transport equation. 33 As in our previous works, 11,29 the angular dependence of the radiation intensity is modeled using the S n quadrature method with n  = 6. The spectral absorption coefficients required to solve this equation are derived from steady-state collisional-radiative modeling using the THERMOS code. 44,45

The equation-of-state of tin was constructed using the Frankfurt equation-of-state (FEOS) model, 46–48 which can model both low-temperature liquid–gas phase coexistence regions and high-temperature plasma states. 33 Laser light absorption and reflection are treated using a hybrid model combining a geometrical-optics ray-tracing approach in low-density plasma regions and a wave-optics approach in regions near and beyond the critical electron density. 49 Laser absorption coefficients are derived from the complex dielectric permittivity of the plasma as per the Drude model. 50  

The simulated cases consider laser irradiation of 30- μ m -diameter liquid tin droplets, close to the industry standard, with spatially constant laser fluences of 60 μ m in width. The laser pulses are temporally trapezoidal shaped with pulse lengths of 20 ns (rise and fall times of 0.2 ns). These experimental parameters are prototypical for recent simulation and experimental works alike (see, e.g., Refs. 9 , 25 , 30 , 32 , 33 , and 51–57 ). The laser wavelengths considered in the first part of this work are λ laser = 1.064 , 2 , 3 , 4 , 5 , 7 , and 10.6   μ m ⁠ . This encompasses two distinct regimes of laser absorption, where absorption occurs primarily in the (i) underdense corona (for small λ laser ⁠ ) or (ii) a narrow region near the critical surface (for long λ laser ⁠ ). 32 The laser intensity is scaled according to I laser = ( 1.4 × 10 11 ) / λ laser W cm −2 , an experimentally motivated scaling, which yields high conversion efficiencies for the laser wavelengths considered in this study. 25,51,58 We note the close similarity between this scaling and the optimum laser intensity I laser ∝ λ laser − 1.2 proposed by Nishihara et al. 59 In the second part of this study, we keep the laser wavelength fixed at λ laser = 1.064   μ m and vary the laser intensity in the range I laser = 0.5 − 5 × 10 11 W cm −2 to probe its effect on the radiative and kinetic properties of the plasmas.

Instantaneous (at t  = 18 ns) volume-specific laser deposition rate 49 normalized by the input laser power ζ abs (red) and in-band radiative power normalized by the input laser power ζ in − band (light blue–green) normalized by the input laser power for λ laser = (a) 1.064, (b) 2, (c) 4, and (d) 10.6 μ m ⁠ . The critical electron density (black contour) and fluid density ρ (dark blue) are indicated.

In the bottom halves of Fig. 1 , we show the net in-band radiated power per unit volume normalized by the input laser power, denoted as ζ in − band and labeled “+” for net emission and “−” for net absorption. This provides a local measure of the efficiency of converting laser light to in-band radiation. Of the four cases shown, the 4 μ m -driven plasma exhibits the highest ζ in − band , + ⁠ . We see that with increasing λ laser ⁠ , the region of net in-band emission ζ in − band , + (light blue regions) moves from regions with n e < n e , cr to regions with n e > n e , cr ⁠ . Furthermore, the regions of net absorption of in-band radiation (dark green regions) are located close to the droplet (a region with high density and low temperature). As discussed by Sunahara et al. , 60 the long mean free paths associated with such radiation can heat the high-density region and enhance the mass ablation rate. 61,62

The instantaneous partitioning of laser power during λ laser = 4   μ m illumination is shown in Fig. 2 . In Fig. 2(a) , we show the input P las (black), absorbed P abs (blue), reflected P ref (orange), and “escaped” P esc (brown) laser power components. The escaped component represents laser radiation that initially misses the target, a quantity which decreases rapidly as the plasma expands and starts absorbing incident laser radiation. After 5 ns, a “steady-state” plasma flow regime is established whereafter P abs and P ref attain near-constant values. This behavior is evident in the plasma-based components P kin (kinetic power, red), P rad (total radiated power, purple), P inb (in-band power, pink), and P int (internal power—derived from the specific Helmholtz free energy, 46 green) shown in Fig. 2(b) . It is well-known that plasmas containing high- Z ions exhibit large radiative losses, 63 and our simulations indicate that approximately 70% of the absorbed laser power is channeled into radiation. Moreover, we find that nearly 16 % of this radiation is concentrated in the in-band region, a surprisingly large fraction given the narrowness (0.27 nm) of this wavelength region.

Time-dependent partitioning of laser power into (a) laser- and (b) plasma-based components for λ laser = 4   μ m irradiation of a 30- μ m -diameter tin droplet.

Next, we quantify power partitioning as a function of laser wavelength. This enables a comprehensive characterization of the EUV plasma source conditions, where high laser absorptivities coupled with large in-band radiative losses and minimal kinetic losses are most desired. In Fig. 3(a) , we present the ratios P abs / P las (blue) and P ref / P las (orange) as a function of laser wavelength at steady-state conditions. We note the significant increase in laser reflectivity comparing λ laser = 1.064 and 10.6 μ m cases, which is due to the significant reduction of τ laser with increasing λ laser specific to the current target geometry.

Instantaneous (at t  = 18 ns) partitioning of (a) laser-based components (normalized by P las ⁠ ) and (b) plasma-based components (normalized by P abs ⁠ ), and (c) spectral purity SP P (yellow circles), conversion efficiency CE P (black squares), and CE P / ( P abs / P las ) (gray squares) as a function of laser wavelength. The dashed curves represent power-law fits to the data.

The internal, radiated, and kinetic components exhibit their own unique dependencies on laser wavelength. In Fig. 3(b) , we show the ratios P rad / P abs (purple squares), P kin / P abs (red circles), P inb / P abs (pink triangles), and P int / P abs (green inverted triangles). The dashed curves represent power-law fits to the data. The origin of these power laws is not exactly known, and they most likely originate from a complex interplay of radiation transport, laser absorption, and plasma expansion effects. With increasing λ laser (and therefore decreasing plasma density), the optical depth of EUV photons reduces from τ EUV ≈ 6 (Nd:YAG-driven plasma) through τ EUV ≈ 2 (4- μ m -driven plasma) to τ EUV ≈ 0.5 (CO 2 -driven plasma). 30,33 This limits the degree of spectral broadening and redistribution of in-band energy into other channels, which explains the observed increase in P inb / P abs with increasing λ laser and the behavior of the spectral purity SP P = P inb / P rad (defined in the full 4 π ⁠ ) presented in Fig. 3(c) . The influence of optical depth on spectral purity and conversion efficiency is further discussed in the work of Schupp et al. 25,30 As the relative fraction of radiative losses increases with increasing λ laser ⁠ , the balance dictates a corresponding decrease in kinetic losses.

The efficiency of producing in-band EUV radiation as a function of laser wavelength is shown in Fig. 3(c) . The conversion efficiency CE P (black squares) exhibits a concave dependence on λ laser with a maximum at λ laser = 4   μ m ⁠ . This maximum arises from the rather unique combination of the values of laser optical depth and optical depth of EUV photons. In essence, the plasma conditions are in a “sweet spot” intermediate to the extreme cases of high laser absorptivity/low spectral efficiency ( ⁠ λ laser = 1.064   μ m ⁠ ) and low absorptivity/high spectral efficiency ( ⁠ λ laser = 10.6   μ m ⁠ ). This explains the simulation results of Langer et al. , who identified an optimum for λ laser = 4.5 μ m irradiation of a one-dimensional tin vapor target, 34 with which our result is in agreement. It is worthwhile noting that the maximum is located on a rather flat part of the curve between 3 and 5 μ m ⁠ , and that the CE P increase from 1.064 to 2 μ m is rather substantial, in line with experimental observations. 58  

The strong dependence of conversion efficiency on laser absorptivity for λ laser > 4   μ m substantiates the opportunity to improve CE P for long laser wavelengths. In Fig. 3(c) , we plot the quantity CE P / ( P abs / P las ) (grey squares), which represents the conversion efficiency if the absorption fraction would be unity for all laser wavelengths. This quantity increases monotonically with increasing λ laser ⁠ , while reaching a plateau between λ laser = 7 and 10.6   μ m ⁠ .

In order to increase laser absorptivity for long λ laser ⁠ , one could pre-irradiate the target to convert it into a rarefied, spatially extended medium. This would decrease the plasma density gradient and subsequently increase the laser optical depth and, thus, its absorption in the plasma. Such target pre-shaping has been successfully applied in industrial applications, enabling high conversion efficiencies from CO 2 laser-irradiated tin targets. 3 That said, target shaping remains unexplored in the intermediate wavelength region considered in this work, and this may lead to substantial increases in CE P ⁠ .

As an extension to the characterization described above, we investigate the effects of laser intensity I laser on the radiative and kinetic properties of a plasma driven by λ laser = 1.064   μ m ⁠ . In a similar vein to Fig. 1 , we show in Fig. 4 the tin mass density ρ , critical electron density contour n e , cr ,   ζ abs (absorbed laser power per unit volume normalized by input laser power), and ζ in − band (in-band radiated power per unit volume normalized by input laser power) for four laser intensities I laser = { 0.5 , 1.4 , 2 , 5 } × 10 11 W cm −2 at t  = 18 ns. With increasing laser intensity, the critical electron density contour tends toward a more spherical shape, and its radius (as measured from the center of mass of the droplet, for instance) extends to larger distances. This leads to a reduction in the amount of laser radiation that “escapes” from the system, as is visible from the upper halves of the panels in Fig. 4 . In the following, we discuss and compare the properties of these plasmas in greater detail.

Instantaneous (at t  = 18 ns) volume-specific laser deposition rate 49 normalized by the input laser power ζ abs (red) and in-band radiative power normalized by the input laser power ζ in − band (light blue–green) normalized by the input laser power for I laser = (a) 0.5, (b) 1.4, (c) 2, and (d) 5 × 10 11 W cm −2 . The critical electron density (black contour) and fluid density ρ (dark blue) are indicated.

As steady-state plasma flows have been attained at t  = 18 ns, we can compare the spatial profiles of various plasma quantities for different laser intensities. Spatial profiles, taken along the z -axis, of all relevant variables are shown in Fig. 5 for I laser = { 0.5 , 2 , 5 } × 10 11 W cm −2 . The profiles generated by different laser intensities are shifted by the position of the droplet surface z droplet ⁠ , which depends on laser intensity as seen in Fig. 4 . Specifically, z droplet is defined as the point along the z -axis where ρ = 1 g cm −3 .

Spatial profiles along the laser axis (at t  = 18 ns) for three laser intensities I laser = { 0.5 , 2 , 5 } × 10 11 W cm −2 and fixed laser wavelength λ laser = 1.064   μ m ⁠ , shown as solid, dashed, and dotted lines, respectively. The profiles are shifted in space by the position of the droplet surface, z droplet ⁠ . Shown are (a) fluid density ρ (dark blue), fluid speed | v | (light green), and sound speed c s (brown); (b) electron density n e (cyan) and volume-specific laser deposition rate normalized by the input laser power ζ abs (red); and (c) temperature T (black), average charge state Z ion (orange), net in-band radiative emission power normalized by the input laser power ζ in − band , + (light blue), and net in-band radiative absorption power normalized by the input laser power ζ in − band , − (dark green).

Examining Fig. 5(a) , we see that the fluid density ρ looks similar for all laser intensities except in the region 0 < z − z droplet < 10   μ m ⁠ , where higher laser intensities yield higher fluid densities. In tandem, the fluid speed | v | shows the opposite behavior in this region, where higher intensities lead to lower speeds. However, for z − z droplet > 10   μ m ⁠ , the expected positive correlation between flow speed and intensity is established, and fluid density profiles overlap. In Fig. 5(a) , we also plot the sound speed c s = γ Z ion k B T / m ion ⁠ , where γ is the adiabatic index (taken to be γ = 1.167 ⁠ , following Ref. 32 ) and m ion = 118.71 amu is the isotope-averaged ion mass. The position of the sonic point (where | v | = c s ⁠ ) also moves to larger distance with increasing laser intensity. Shown in Fig. 5(b) are profiles of the electron density n e and the volume-specific laser deposition rate normalized by the input laser power ζ abs ⁠ , where the aforementioned shift of the position of the critical electron density n e , c r ≈ 10 21 cm −3 to larger distances with increasing I laser is clearly observed.

The fluid temperature profiles, shown in Fig. 5(c) , peak close to the positions of maximum laser absorption and are found to increase with increasing laser intensity (also see discussion in Sec. IV B ). The spatial profile of the average charge state Z ion shown in Fig. 5(c) follows that of the temperature, given the power law scaling Z ion ∝ T 0.6 ⁠ , see also Ref. 32 . Also presented in Fig. 5(c) is the volume-specific in-band radiative power normalized by the input laser power ζ in − band ⁠ , split into its positive (net emission) and negative (net absorption) branches. As before, the position of net absorption ζ in − band , − is located beside the droplet surface. The profile of the net emission of in-band radiation, ζ in − band , + ⁠ , exhibits a non-trivial dependence on laser intensity: with increasing I laser ⁠ , the spatial profile first broadens and then, for I laser = 5 × 10 11 W cm −2 , exhibits a local minimum near z − z droplet = 20   μ m ⁠ . This behavior can be understood from the spatial profiles of the average charge state Z ion and the fluid density ρ , respectively, the location and abundance of ion charge states responsible for in-band EUV emission ( ⁠ 11 < Z ion < 14 ⁠ ). 19 Comparing the region 10 < z − z droplet < 40   μ m for the three laser intensities, the case I laser = 0.5 × 10 11 W cm −2 is characterized by so-called “underheating” ( ⁠ Z ion < 11 ⁠ ) toward longer distances, while I laser = 5 × 10 11 W cm −2 is characterized by “overheating” ( ⁠ Z ion > 14 ⁠ ) toward shorter distances.

We now quantify the power partitioning as a function of laser intensity, the results of which are shown in Fig. 6 . From panel (a), we see that the absorbed power fraction remains near unity for the described intensity range. Moreover, the growth of the critical surface radius with increasing laser intensity leads to a reduction in the escaped component, while at the same time, we see an increase in the reflected component.

Instantaneous (at t  = 18 ns) partitioning of (a) laser-based components (normalized by P las ⁠ ) and (b) plasma-based components (normalized by P abs ⁠ ), and (c) spectral purity SP P (yellow circles), conversion efficiency CE P (black squares), and CE P / ( P abs / P las ) (gray squares) as a function of laser intensity.

The dependence of conversion efficiency CE P on laser intensity, shown in Fig. 6(c) , follows the same trend as the spectral purity ( ⁠ SP P ⁠ ) with a maximum at I laser = 2 × 10 11 W cm −2 . This is due to the near-independence of P rad / P abs and P abs / P las on laser intensity, the exact opposite of what was found in Sec. III B . As such, the effect of optimizing the absorption ratio by reducing the escaped laser component does not shift the found maximum as is clear from [ CE P / ( P abs / P las ) ] shown in Fig. 6(c) . Combining the dependences of P rad / P abs and CE P on I laser ⁠ , we see that increasing laser intensity can yield a significant increase in absolute in-band radiation without significantly influencing the conversion efficiency of the system.

In summary, we have investigated the power partitioning in a laser-produced tin plasma for laser wavelengths in the 1.064 ≤ λ laser ≤ 10.6   μ m range and laser intensities of 0.5 × 10 11 ≤ I laser ≤ 5 × 10 11 W cm −2 for λ laser = 1.064   μ m ⁠ . We have identified a strong laser-wavelength dependence of laser absorptivity and the location of EUV generation. With increasing laser wavelength, the power balance monotonically shifts from kinetic losses to in-band radiative losses. The decrease in laser absorption for long laser wavelengths, combined with a concurrent decrease in EUV optical depth, yields a non-monotonic behavior of the conversion efficiency, leading to an optimum at λ laser = 4   μ m ⁠ . EUV sources based on long laser wavelengths would, therefore, benefit from additional target preparation to ensure a higher absorption fraction. The influence of laser intensity on the power partitioning for λ laser = 1.064   μ m is found to be small: no significant shift between kinetic losses and radiative losses is seen. With increasing laser intensity, we find an increase in in-band radiative power at minor cost in terms of conversion efficiency.

We would like to thank Wim van der Zande for useful discussions. This project has received funding from the European Research Council (ERC) Starting Grant No. 802648. This work has been carried out at the Advanced Research Center for Nanolithography (ARCNL). ARCNL is a public–private partnership with founding partners UvA, VU, NWO-I, and ASML and associate partner RUG. This work made use of the Dutch national e-infrastructure with the support of the SURF Cooperative using Grant No. EINF-1043 and EINF-2947.

The authors have no conflicts to disclose.

Diko J. Hemminga: Formal analysis (lead); Investigation (lead); Visualization (lead); Writing – original draft (lead). Oscar Oreste Versolato: Conceptualization (equal); Funding acquisition (lead); Project administration (equal); Supervision (supporting); Writing – review & editing (supporting). John Sheil: Conceptualization (equal); Project administration (equal); Supervision (lead);Writing – review & editing (lead).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Citing articles via

Submit your article.

Sign up for alerts

• Online ISSN 1089-7674
• Print ISSN 1070-664X
• For Researchers
• For Librarians
• For Advertisers
• Our Publishing Partners  
• Physics Today
• Conference Proceedings
• Special Topics

pubs.aip.org

• Privacy Policy
• Terms of Use

Connect with AIP Publishing

This feature is available to subscribers only.

Sign In or Create an Account

Accessibility Links

• Skip to content
• Skip to search IOPscience
• Skip to Journals list
• Accessibility help
• Accessibility Help

Click here to close this panel.

Purpose-led Publishing is a coalition of three not-for-profit publishers in the field of physical sciences: AIP Publishing, the American Physical Society and IOP Publishing.

Together, as publishers that will always put purpose above profit, we have defined a set of industry standards that underpin high-quality, ethical scholarly communications.

We are proudly declaring that science is our only shareholder.

High-energy ions from Nd:YAG laser ablation of tin microdroplets: comparison between experiment and a single-fluid hydrodynamic model

D J Hemminga 1,2 , L Poirier 1,2 , M M Basko 3 , R Hoekstra 1,4 , W Ubachs 1,2 , O O Versolato 1,2 and J Sheil 5,1

Published 12 October 2021 • © 2021 The Author(s). Published by IOP Publishing Ltd Plasma Sources Science and Technology , Volume 30 , Number 10 Citation D J Hemminga et al 2021 Plasma Sources Sci. Technol. 30 105006 DOI 10.1088/1361-6595/ac2224

Article metrics

2377 Total downloads

Share this article

Author e-mails.

[email protected]

Author affiliations

1 Advanced Research Center for Nanolithography, Science Park 106, 1098 XG Amsterdam, The Netherlands

2 Department of Physics and Astronomy, and LaserLaB, Vrije Universiteit, De Boelelaan 1081, 1081 HV Amsterdam, The Netherlands

3 Keldysh Institute of Applied Mathematics, Miusskaya Square 4, 125047 Moscow, Russia

4 Zernike Institute for Advanced Materials, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands

Author notes

5 Author to whom any correspondence should be addressed.

D J Hemminga https://orcid.org/0000-0002-4038-3646

L Poirier https://orcid.org/0000-0001-7177-8926

M M Basko https://orcid.org/0000-0001-8809-8601

R Hoekstra https://orcid.org/0000-0001-8632-3334

W Ubachs https://orcid.org/0000-0001-7840-3756

O O Versolato https://orcid.org/0000-0003-3852-5227

J Sheil https://orcid.org/0000-0003-3393-9658

• Received 12 July 2021
• Revised 23 August 2021
• Accepted 27 August 2021
• Published 12 October 2021

Peer review information

Method : Single-anonymous Revisions: 1 Screened for originality? Yes

Buy this article in print

We present the results of a joint experimental and theoretical study of plasma expansion arising from Nd:YAG laser ablation (laser wavelength λ = 1.064  μ m) of tin microdroplets in the context of extreme ultraviolet lithography. Measurements of the ion energy distribution reveal a near-plateau in the distribution for kinetic energies in the range 0.03–1 keV and a peak near 2 keV followed by a sharp fall-off in the distribution for energies above 2 keV. Charge-state resolved measurements attribute this peak to the existence of peaks centered near 2 keV in the Sn 3+ –Sn 8+ ion energy distributions. To better understand the physical processes governing the shape of the ion energy distribution, we have modelled the laser-droplet interaction and subsequent plasma expansion using two-dimensional radiation hydrodynamic simulations. We find excellent agreement between the simulated ion energy distribution and the measurements both in terms of the shape of the distribution and the absolute number of detected ions. We attribute a peak in the distribution near 2 keV to a quasi-spherical expanding shell formed at early times in the expansion.

Export citation and abstract BibTeX RIS

Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence . Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

1. Introduction

Laser-produced plasmas (LPPs) formed on tin microdroplets are now established as the light source of choice in new-generation lithography machines for high-volume manufacturing of integrated circuits below the 10 nm node [ 1 – 3 ]. Their incorporation in modern-day lithography machines relies on their ability to provide sufficiently high fluxes of short-wavelength radiation to enable the patterning of nanometre-scale features on integrated circuits.

Under optimum conditions, spectra recorded from tin LPPs exhibit an intense, narrowband emission feature centered near an extreme ultraviolet (EUV) wavelength of 13.5 nm [ 4 – 8 ]. The superposition of millions of lines arising from transitions between complex configurations in n = 4 shell Sn 11+ –Sn 15+ ions are the atomic origins of this light [ 9 – 13 ]. Importantly, this feature overlaps with the 2% reflective bandwidth (13.5 ± 0.135 nm—the so-called 'in-band' region) of molybdenum/silicon multilayer mirrors (MLMs) [ 14 ]. Such mirrors are an integral component of EUV lithography tools, transporting EUV photons from the light source to their final destination at the wafer stage.

One key aspect of industrial EUV light source development has focussed on optimising the photon output of LPP EUV light sources. To-date, efforts have concentrated on increasing (i) EUV power and (ii) the so-called conversion efficiency (CE—the ratio of in-band EUV energy emitted into the half sphere back towards the laser to input laser energy) of the light source [ 1 , 7 ]. To meet the high power levels required for industrial applications, a dual-pulse irradiation scheme is employed [ 15 ]. First, a low-intensity prepulse is used to deform the droplet into an elongated disk-like target. This target is then irradiated by a second, high-energy CO 2 laser pulse ( λ = 10.6  μ m) which generates a hot, EUV-emitting plasma. This EUV light is then focussed by an MLM (known as the collector mirror ) to an exit port of the light source vessel whereupon it enters the scanner tool for use in the lithographic process.

A second and no-less crucial aspect of EUV light source development has focussed on the design and implementation of so-called debris mitigation schemes . In the context of the current application, plasma expansion will lead to the bombardment by tin ions on the plasma-facing collector mirror. The combined effects of sputtering and ion implantation will, over time, degrade the performance of the collector mirror and reduce EUV throughput. In an industrial setting, the light source vessel is typically filled with a background hydrogen gas to stop energetic ions from reaching the collector mirror [ 1 , 16 , 17 ]. One can also introduce a strong magnetic field in the region surrounding the droplet to deflect plasma ions away from the collector mirror [ 18 – 20 ]. A comprehensive understanding of the characteristics of the plasma expansion (distribution of ions over kinetic energy, angular distribution of ions, etc.) can greatly assist the design of effective debris mitigation schemes.

A number of studies examining tin plasma expansion have been performed over the past two decades. In the mid-2000's, Murakami et al [ 21 ] and Fujioka et al [ 22 ] demonstrated that ion energy distributions recorded from minimum-mass plasmas driven by 10 ns-long Nd:YAG ( λ = 1.064  μ m) pulses can be described using a model of isothermal plasma expansion. In this model, the distribution of the number of ions N as a function of kinetic energy E is written

Plasmas driven by shorter, ps-duration pulses [ 23 , 24 ] exhibit ion energy distributions whose shapes are better described by the planar isothermal expansion model of Mora [ 25 ]. In this case the ion energy distribution reads

Other work on the topic of tin plasma expansion has explored, for example, the role of laser pulse duration and laser wavelength on the ion energy distribution [ 26 – 28 ], the angular distribution of ion kinetic energies [ 29 – 31 ], the suppression of fast ions using a low-energy prepulse [ 32 , 33 ], and the role of electron–ion recombination during the expanding phase of the plasma [ 34 ]. It is important to note that the vast majority of these studies have investigated plasma expansion from laser-irradiated planar tin targets rather than from industrially-relevant droplet targets. Much work still remains to be done on this latter topic.

The goal of the present study is to investigate plasma expansion in the form of emission of energetic charged particles from Nd:YAG-irradiated tin microdroplet targets. This study serves to complement recent work on photon emission from solid-state laser-driven EUV light source plasmas [ 7 , 35 , 36 ]. In contrast to the current industrial implementation, solid-state laser-driven plasmas may not require the use of a pre-pulse for efficient EUV production [ 7 ]. As such, they are a promising candidate for future laser-driven EUV light source plasmas. First we present measurements of the ion energy distributions using an electrostatic analyser (ESA). These measurements reveal the existence of peaks near 2 keV in the high-energy tails of the Sn 3+ –Sn 8+ ion energy distributions. These features combine to yield a peak near 2 keV in the charge-state summed ion energy distribution. To elucidate the origin of this peak, we have performed two-dimensional radiation-hydrodynamic simulations of the plasma formation and its subsequent expansion using the radiative arbitrary Lagrange–Eulerian fluid dynamics in two dimensions (RALEF-2D) code. The ion energy distribution obtained from the simulations compares favourably to the measurements both in terms of the shape of the distribution and the absolute number of detected ions. We attribute the peak in the ion energy distribution to a high-velocity, quasi-spherical expanding shell formed at early times in the plasma expansion. The current work advances on the work presented in references [ 8 , 23 ] to provide a quantitative understanding of absolutely-calibrated measurements via radiation-hydrodynamic modelling of the expanding plasma, beyond the aforementioned idealized plasma expansion models.

The layout of this paper is as follows: in section  2 we discuss the experimental setup and provide details of the ion energy distribution measurements. This is followed by a description of the single-fluid, single-temperature model implemented in the RALEF-2D code and a brief discussion of the simulation parameters. In section  4 we discuss the results of the simulations, focussing on the temporal and spatial evolution of the speed and ion number density profiles in the expansion. In section  5 we compare the ion energy distribution obtained from the simulations with our experimental measurements. Comparisons are drawn with the predictions of well-known analytical models of plasma expansion into vacuum. Finally, we summarise this work in section  6 .

2. Experimental setup, method and results

In the experiments, tin droplets were dispensed from a droplet generator mounted at the top of a vacuum chamber (backing pressure ∼10 −7 mbar). The droplet generator consists of a heated (260 °C) molten tin reservoir connected to a nozzle. The diameter of the tin droplets used in the experiment was 28 μ m. Upon crossing the centre of the chamber, the droplets pass through a light sheet created by a He:Ne laser. Light scattered by the droplets was detected by a photomultiplier tube which triggered the plasma-generating laser pulse and the acquisition apparatus. Plasmas were generated by focussing the output of a commercial Nd:YAG laser system onto the tin droplets. The laser pulses exhibited Gaussian-like temporal and (focussed) spatial laser profiles. The temporal full-width at half-maximum (FWHM) was 10 ns and the FWHM of the focussed pulses was approximately 60 μ m. Employing a laser pulse energy of 60 mJ resulted in a laser power density on the targets of I L ≈ 2 × 10 11 W cm −2 . This choice of power density is known to yield optimum CE's for Nd:YAG-driven tin plasmas [ 7 ]. We note that this particular choice of laser parameters, combined with the given droplet diameter, will not lead to the full ablation of the tin droplet.

Charge-state resolved ion energy distributions were measured using an ESA. The opening aperture of this device was located 1.12 m away from the droplet targets and was positioned at 60° with respect to the incident laser axis. The ESA consists of a radial electric field deflection region followed by a calibrated channeltron detector. The radial electric field between the two electrodes of the ESA selects charge states based on the ratio of their kinetic energy E to charge state Z according to E / Z = 5 × U ESA where U ESA is the voltage across the ESA electrodes (measured in volts). A time-of-flight (ToF) analysis is used to obtain charge-state resolved ion counts for a given E / Z . By scanning the ESA voltage U ESA over a desired range, one can obtain charge-state resolved ion energy distributions. The ESA-ToF measurements have been benchmarked against charge-state integrated measurements made using a Faraday cup (FC) which was positioned at an angle of −60° with respect to the incident laser axis.

The total ion energy distribution d 2 N /d E  dΩ was derived from the measurements via

In figure  1 we present the results of our charge-state resolved ion energy distribution measurements. Examining this figure, we first note that the distributions associated with Sn 1+ and Sn 2+ ions are rather broad, spanning energies in the range 0.03–2 keV. The kinetic energy for which the ion energy distribution peaks, E peak , clearly increases with increasing charge state. Both distributions also exhibit a near-exponential fall-off for E > E peak . While the aforementioned trend in E peak continues for Sn 3+ and Sn 4+ , we note the emergence of a second, high-energy peak located just below 2 keV (this peak is also present in the Sn 2+ distribution although it is less pronounced than in the Sn 3+ and Sn 4+ distributions). With increasing charge state this peak grows in intensity until E peak ≈ 2 keV in the Sn 5+ and Sn 6+ ion energy distributions (we also make a tentative observation of two peaked features in the Sn 7+ ion energy distribution near 1.6 and 2.5 keV, respectively). While we do detect Sn 8+ ions in the experiments, the amplitude of the ion energy distribution is an order-of-magnitude lower than the Sn 7+ distribution. No traces of higher charge states could be reliably detected. Importantly, the kinetic energy associated with this high-energy feature is independent of charge state. Shown in red in figure  1 is the total ion energy distribution obtained by summing the individual Sn 1+ –Sn 8+ ion energy distributions. Three distinct regions emerge: (i) a near-plateau in the ion energy distribution between 0.03–1 keV (ii) a peak near 2 keV followed by (iii) a sharp fall-off for E > 2 keV. Finally, we note that the EUV-generating tin charge states Sn 11+ –Sn 15+ , whilst generated in the hot, dense region of the plasma, are not detected in the measurements. The absence of Sn 11+ –Sn 15+ charge states may in part be attributed to the process of recombination, whereby free electrons in the expanding plasma recombine with these ions through processes such as three-body or radiative recombination [ 34 ].

Figure 1.  Experimental measurements of the distribution of the number of ions over ion kinetic energy is shown. Charge-state resolved ion kinetic energy distributions for Sn 1+ –Sn 8+ are shown as dashed colored lines. The total ion energy distribution, shown in red, is obtained by summing the distributions of the individual charge states.

Download figure:

3. Radiation hydrodynamic simulations and the RALEF-2D code

In order to elucidate the dynamics of the plasma expansion and its influence on the ion energy distribution, we have undertaken numerical modelling of the plasma formation, growth and subsequent expansion using radiation-hydrodynamic simulations. In the following, we discuss the underlying assumptions of the single-fluid, single-temperature approach adopted in the present work and we provide details of the simulations we have performed with the RALEF-2D code.

3.1. Single-fluid single-temperature radiation hydrodynamics

We have chosen to model the plasma formation and its subsequent expansion using a single-fluid, single-temperature hydrodynamic model including the effects of radiation transport and thermal conduction. In this approach, the free electrons and ions are treated as a single fluid having a single temperature T e = T ion = T . Although more complex approaches such as the two-fluid, two-temperature [ 38 ] or single-fluid, two-temperature models [ 39 – 42 ] have been pursued, the single-fluid, single-temperature description should be adequate for the current purposes. For one, simulations of Nd:YAG-irradiated lithium, plastic and gold targets performed by Sunahara and Tanaka [ 39 ] indicate a rather small difference (less than 20%) between T e and T ion in the plasma. This behaviour has also been observed in simulations of laser-driven tin plasmas [ 43 ]. Second, the moderate ionisation degrees ( Z ≈ 11–15) found in EUV source plasmas implies that the free-electron contribution to the pressure p e = Zn ion kT e (in the ideal gas approximation) far outweighs the ion contribution to the pressure p i = n ion kT e ( n ion is the ion number density). As such, the ion temperature will play a near-negligible role in the context of the current study.

The equations of single-fluid, single-temperature hydrodynamics take the form

In the above equations, ρ is the fluid mass density, v is the fluid velocity, p = p e + p i is the pressure, E = e int + | v | 2 /2 is the mass-specific total energy (sum of the internal and kinetic energy contributions), S T represents thermal conduction, S R is the volume-specific heating rate provided by thermal radiation and S ext represents any external energy sources, e.g. energy deposition from a laser beam, ion beam, etc.

3.2. RALEF-2D

We have performed radiation-hydrodynamic simulations using the RALEF-2D code. This code was originally developed to provide theoretical support for laser-plasma experiments at GSI Darmstadt under moderate laser intensities ≲10 13 –10 14 W cm −2 [ 47 , 48 ]. More recently, the code has found application in modelling laser-driven plasma sources of EUV light [ 13 , 49 – 52 ]. The hydrodynamic component of the code is based on an upgraded version of the fully-explicit CAVEAT code for ideal hydrodynamics [ 53 , 54 ]. RALEF-2D solves the single-fluid, single-temperature hydrodynamic equations (equations ( 4 )–( 6 ) in two spatial dimensions on a structured quadrilateral mesh in either Cartesian ( x , y ) or axisymmetric ( z , r ) coordinates using a second-order Godunov-type method [ 54 ]. The axisymmetric coordinate system has been used in the present simulations.

In the RALEF-2D code, radiation transfer and heat conduction are coupled into the fluid energy equation using a symmetric semi-implicit method with respect to time discretisation [ 55 ]. The code solves the LTE radiation transfer equation in the quasi-static approximation using pre-tabulated absorption coefficients generated with the THERMOS code [ 56 , 57 ]. The equation of state (EOS) of tin was built using the Frankfurt equation of state (FEOS) model which can treat both low-temperature liquid-gas phase coexistence regions as well as high-temperature plasma states [ 58 ]. The FEOS model supplies the RALEF-2D code with key thermodynamic quantities such as the pressure, mass-specific internal energy as well as the average charge state of the plasma.

The simulations were performed on a computational mesh shaped in a half-disk consisting of multiple blocks with initially distinct properties. A simplified representation of the mesh is shown in figure  2 . Centered in the origin of the ( z , r ) coordinate system we define a tin 'droplet' having a mass density of 6.9 g cm −3 and a temperature of 592 K (0.051 eV). As in the experiments, the droplet diameter is set to 28  μ m. The bulk of the droplet is defined in a mesh section constructed as a rectangle stretched to a half-disk with dimensions of 45 × 90 mesh cells. As shown in figure  2 , the outer region of the droplet exhibits a more refined mesh structure. Here, the length of each successive mesh cell along the radial direction decreases with increasing r . The mesh cell length on the outer droplet boundary is approximately 10 nm. Outside the droplet the mesh is filled with a tin vapour having a mass density of 10 −12 g cm −3 . This section is divided into quadrilateral cells by approximate concentric and radial edges and extends to a radial distance of 10 mm.

Figure 2.  A simplified representation of the mesh structure employed in the RALEF-2D simulations. The liquid droplet, shown in purple, is assigned an initial mass density of 6.9 g cm −3 and a temperature of 592 K (0.051 eV). The region outside the droplet is filled with a tin vapour having a mass density of 10 −12 g cm −3 .

In essence, the experimental laser parameters have been replicated in the simulations. The laser beam is circular and coaxial to the positive z -axis. The simulations have employed unpolarized laser light. The laser absorption coefficient was calculated from the complex dielectric permittivity of the plasma [ 59 ].

The ion energy distribution is extracted from the fluid simulation by considering mass flow through the computational mesh. The main fluid variables density and velocity are assigned to each mesh cell throughout the simulation and are converted to the quantities mass and speed (velocity magnitude). These variables are cell-centered and form the basis of keeping track of the fluid throughout the simulation. As the curved boundary of the computational mesh is defined as a free-outflow boundary, matter flowing out of the mesh leaves the computational domain; it leaves the simulated area in space. This is closely related to the treatment of the ion energy distribution by RALEF-2D. Mass flowing out of the mesh is recorded ('binned') in the (i) energy bin corresponding to its speed and (ii) the angular bin corresponding to the angle between the laser axis and the escape velocity vector. This module is called at every hydrodynamic time step, summing the number of particles equivalent to the outflowing mass. This procedure explicitly constructs the distribution of the number of particles into 360 predefined discrete energy bins in the range [1, 20 × 10 3 ] eV. The bin width increases exponentially with increasing energy. The angular domain is divided into 36 bins (over 180°). In the current simulation we consider mass flow into two angular bins extending over the range [55°, 65°]. The duration of the simulation is 1 μ s which allows accounting for ions with energies down to ∼70 eV leaving the computational domain in this time window.

4. Plasma formation and expansion

In figure  3 we present the evolution of the plasma expansion through the variables speed and ion number density, where the pseudocolour indicates the magnitude of these variables. At distances larger than 0.5 mm the velocity vector effectively points radially outwards. The ion number density n ion is obtained from n ion = ρN A / A , where ρ is the fluid mass density, N A is Avogadro's constant and A is the atomic weight of tin. For visibility, we reflect the ion number density information into the lower plane (this is possible as the simulations were performed using the axisymmetric ( z , r ) coordinate system). We define t = 0 ns as the time that the laser pulse is switched on in the simulations. The left column shows times t = {11, 15} ns, the middle column t = {25, 35} ns and the right column the late times t = {60, 120} ns. The laser propagates along the positive z axis (laser axis) and its (local) intensity is represented by the black shading seen in the t = {11, 15} ns frames. Frames grouped in the same column, e.g. t = {11, 15} ns share the same axial and radial domains. In order to follow the plasma expansion in space, we increase the axial and radial coordinate domains in the t = {25, 35} and {60, 120} ns frames.

Figure 3.  Two-dimensional profiles of the speed | v | and ion number density n ion during laser ablation and subsequent plasma expansion are shown. The color scale represents the size of the variable. The laser illuminates from the left along the z axis, illustrated in the t = {11, 15} ns frames by the black shaded band. The black dashed line in the t = 11 ns frame corresponds to the 'lineout' (see main text for description) along which the speed and ion number density profiles shown in figures  5 and 6 are taken.

The overall dynamics of plasma formation and expansion, as displayed in figure  3 , can be qualitatively described as a succession of two distinct bursts of laser-induced ablation from the droplet surface. These two bursts are clearly identified in the t = 15 ns and t = 25 ns frames as two concentric red regions exhibiting high speed. In the following two subsections we describe the formation and evolution of these two ablation bursts.

4.1. Initial burst of laser-induced ablation

The initial burst of laser-induced ablation forms in the first 2–3 ns after the laser pulse is turned on. Initially, the laser pulse has an intensity I ≈ 3 × 10 8 W cm −2 which lies only moderately above the ablation threshold of liquid tin. In figure  4 we plot one-dimensional (1D) profiles of the mass density ρ (black), temperature T (red), fluid speed | v | (orange dashed curve) and mass-specific heating rate of the laser q (green) along the negative laser axis (starting from the droplet center) at times t = (a) 0.5, (b) 1 and (c) 1.4 ns, respectively. These data are extracted along a so-called 'lineout' taken at θ = 0° with respect to the axial coordinate (laser) axis.

Figure 4.  One-dimensional profiles of the mass density ρ (black), temperature T (red), speed | v | (orange dashed curve) and mass-specific laser heating rate q (green) along the negative axial coordinate axis for times t = (a) 0.5, (b) 1 and (c) 1.4 ns.

As is evident from figure  4 (b), the temperature of the ablated plasma T ≈ 1–2 eV exceeds the critical temperature of tin ( T critical ≈ 0.5 eV) by only a moderate factor. In addition, we notice the emergence of a 'hump' in the speed profile at d 0° ≈ 15  μ m which is coincident with the location of the peak value of q , the mass-specific laser heating rate. Material associated with this hump accelerates and eventually 'catches up' with the initially-ablated material. By t ≈ 2.5 ns, the speed profile exhibits a near-linear dependence on distance d , i.e., | v | ∝ d and the speed of the front edge of the plasma cloud stabilizes to | v front | ≈ 33 km s −1 . This expanding plasma cloud drives a shock into the low-density ( ρ = 10 −12 g cm −3 ) ambient gas which, however, has a negligible effect on the overall expansion dynamics.

4.2. Second burst of laser-induced ablation

The increase in laser intensity (and subsequent increase in the mass-specific laser-heating rate q ) as t → 15 ns generates a second burst of ablation visible in the upper halves of the t = 11, 15 and 25 ns frames in figure  3 . This second burst is characterized by a significantly higher ablation rate, density and temperature ( T > 30 eV) of the ejected plasma whose leading edge quickly reaches a speed | v front | ≈ 60 km s −1 .

To better understand the dynamics of this ablation burst along the line-of-sight of the ESA and FC devices, we show in figure  5 the (a) speed and (b) ion number density profiles along a lineout taken at θ = 60° with respect to the laser axis at times t = 5 (black), 10 (red), 15 (blue), and 20 ns (green), respectively. This lineout is shown as a black dashed line in the t = 11 ns frame in figure  3 . The shaded vertical bars indicate the location of local maxima in the number density lineouts and serve to guide the eye between both variables. From figure  5 we see that as the second and more powerful ablation burst rams into the background plasma (left behind by the initial ablation burst), it rakes up material into a quasi-spherical expanding shell. This shell is evident as a hump in the t = 15 and 20 ns n ion profiles in figure  5 (b). The deposition of energy (via laser absorption) and the expansion that follows can be likened to the effects of a fast piston pushing a gas; the shockwave launched by the piston effectively sweeps up material in front of it, driving a compression wave.

Figure 5.  One-dimensional profiles of the (a) speed and (b) ion number density along the lineout shown in figure  3 at t = 5 (black), 10 (red), 15 (blue) and 20 ns (green).

In figures  6 (a) and (b) we plot the speed and ion number density profiles along the θ = 60° lineout at times t = 30 (orange), 50 (light blue), 80 (purple) and 120 ns (dark green). As the plasma expands, we note that the two speed profiles merge to form a single profile exhibiting a near-linear dependence on distance. A linear dependence of speed on distance, i.e., | v | = d / t is exactly the asymptotic t → ∞ behaviour of any 'explosion-like' expansion [ 60 ]. Our simulations therefore recover the late-time behaviour expected from such an expansion. The evolution of the ion number density profiles over time and space is less dramatic. First, we note that the quasi-spherical expanding shell observed in figure  5 (b) persists throughout the expansion. As time evolves the shell is observed to broaden, a direct consequence of the non-constant speed profile across the shell. It is interesting to note that the spatial variation of the ion number density up to the quasi-spherical shell appears to follow a power law of the form n ion ∼ d − n where 2 < n < 3. Unlike the speed profile, a universal analytic form for n ion as t → ∞ does not exist for an explosion-like expansion. As discussed by Zel'dovich and Raizer [ 60 ], the asymptotic solution | v | = d / t is satisfied for n ion = ϕ ( d / t )/ t 3 where ϕ ( d / t ) is an arbitrary function of d / t . This function can only be evaluated through numerical simulations of the system and is case-specific.

Figure 6.  One-dimensional profiles of the (a) speed and (b) ion number density along the lineout shown in figure  3 at t = 30 (orange), 50 (light blue), 80 (purple) and 120 ns (dark green). The gray dashed line illustrates an analytic form for the speed | v | ∝ d .

Before proceeding with the comparison of experimental and simulated ion kinetic energy distributions, we wish to make two remarks. First, recall that the validity of the single-fluid description of a plasma relies on the condition λ D ≪ L where λ D is the Debye length and L is the characteristic flow length. We have calculated λ D and L for the late time case t = 120 ns and have found that λ D ≈ 1–10  μ m and L ≈ 500  μ m in the vicinity of the density hump. These results validate the use of the quasi-neutral hydrodynamic approach in the current context. It is important to mention that the mechanism of ion acceleration in the other extreme case, i.e., plasmas in which λ D ≫ L is often referred to as 'Coulomb explosion' [ 38 ]. Second, recall that in the simulations the region outside the droplet is filled with a low-density tin vapour having ρ = 10 −12 g cm −3 . The experiments, on the other hand, have been performed in a near-vacuum environment. This choice of density is sufficiently low as not to distort the vacuum-like expansion we wish to emulate in our simulations. This is evident from figure  6 where we do not observe a decrease of the peak velocity as the fluid propagates through the low-density background gas.

5. Ion kinetic energy distributions: experiment and simulation

We now wish to compare our measured ion energy distribution with that obtained from the simulations. These two quantities are compared in figure  7 . The experimental data, shown in red, corresponds to the total ion energy distribution (also shown in red in figure  1 ) and the solid black curve is the ion energy distribution obtained from the RALEF-2D simulations. As described in section  3.2 , the RALEF-2D ion energy distribution is obtained by recording mass flow into two angular bins subtending an angle 55° < θ < 65° with respect to the laser axis. From the two-dimensional computational mesh we have calculated the three-dimensional solid angle Ω S ≈ 0.95 sr by revolving the arc length in the mesh around the laser axis. The RALEF-2D ion energy distribution, once corrected for this solid angle, is then convolved with bin-specific Gaussian functions having full width at half-maxima equal to 5% of the lower boundary of the energy bin. The purpose of this convolution is to account for processes which may broaden the distribution, e.g. mass distribution of tin isotopes.

Figure 7.  The distribution of the number of ions over ion kinetic energy. The experimental ion energy distribution is shown in red (solid curve) and the RALEF-2D ion energy distribution is shown in black (solid curve). Also illustrated are the predictions of the analytic models of Murakami (green dash-dot) [ 21 ], Mora (green dash) [ 25 ] and the Riemann wave (green solid) [ 60 ].

It is clear from figure  7 that the RALEF-2D ion energy distribution closely resembles the experimental measurements. First, the simulation reproduces the high-energy peak observed in the experimental data near 2 keV. This high-energy feature originates from fast-moving material associated with the quasi-spherical expanding shell (a tin ion with | v | = 57 km s −1 has a kinetic energy E ≈ 2 keV). Second, the simulations reproduce the near-constant behaviour of the experimental ion energy distribution in the 0.07–1 keV range. Fluctuations observed in the RALEF-2D ion energy distribution most likely arise from spatial fluctuations in the density during the expansion (visible in figure  6 for t = 120 ns). Above 2 keV, both the simulations and experimental data exhibit a sharp fall-off with increasing kinetic energy. This fall-off is sharper in the case of the simulations, which do not predict any ions having kinetic energies above 3 keV. It is also interesting to note that, within the limits of the experimental uncertainties, the simulations provide a reliable prediction for the absolute number of ions detected in the experiments.

We show in figure  7 the predictions of the models of Murakami (equation ( 1 )) and Mora (equation ( 2 )). Guided by the work of Torretti et al [ 13 ], we have taken Z = 12, k B T = 35 eV and N 0 = 2 × 10 12 . For the model of Murakami we have chosen a spherical expansion ( α = 3) and a value ln[ R ( t )/ R 0 ] = 4. We also provide in figure  7 the ion energy distribution arising from a planar isentropic expansion, better known as the Riemann wave [ 60 ]:

Here γ is the adiabatic index (we have taken γ = 4/3 [ 61 ]) and E max = 2 Zk B T e γ /( γ − 1) 2 is the maximum ion kinetic energy. Both the planar isothermal expansion model of Mora and the Riemann wave solution predict a similar monotonic decrease in d N /d E with increasing E which is not observed in the experimental measurements. The shape of the experimental data is qualitatively better described by the spherical isothermal expansion model of Murakami, which exhibits a slow rise in d N /d E up to a peak at E peak = E 0 /2 ≈ 1.7 keV. The fall-off in d N /d E at energies above E peak is far less steep compared to the experimental data.

The reason why the models of Murakami and Mora and the Riemann wave solution cannot be expected to reproduce the current experimental distribution ultimately lies in the plasma density profiles adopted in these analytic models. The function ϕ ( d / t ) obtained with the RALEF-2D simulations is significantly more complex than the Gaussian and exponential density profiles assumed in the models of Murakami and Mora, respectively. It is the interaction of many complex processes (2D expansion of a non-uniformly heated, non-planar (radiating) plasma) that ultimately determines the function ϕ ( d / t ).

Four possible causes have been identified which may contribute to the observed differences between the experimental and simulated ion energy distributions. The first cause is the effect of numerical diffusion in the large mesh cells at larger mesh radii. This has been partly tackled by increasing the radial detail in the mesh at larger distances. The three other causes are inherent to the ansatz of the simulation code. As mentioned in section  3.2 , RALEF-2D uses Godunov's method for the Lagrangian phase of each hydrodynamic cycle. As the internal energy component of the total energy determines the pressure, rounding errors can propagate especially if the internal energy is small. A third possible cause is related to the EOS model employed in RALEF-2D. The EOS model adopted in this work assumes LTE ionization throughout the entire simulation. This assumption breaks down at late times in the expansion when ionization and recombination processes cease to exist, leading to the well-known 'freezing' of charge states [ 34 ].

We note that the simulated domain (10 mm) is much smaller than the experimental flight path (∼1 m) and the assumption is made that neither the experimental nor simulated ion energy distributions change significantly between the two distances. This assumption is supported by two arguments: (1) within the simulated spatial scale the velocity profile attains its asymptotic 'triangle-like' shape | v | = d / t before leaving the mesh, having converged to the late-time behaviour; (2) on the experimental side the aforementioned freezing of charge states will occur on a length scale similar to that of the simulation spatial scale [ 34 ] and, thus, no significant changes over the remaining flight path will occur in our high-vacuum environment. Keeping these remarks in mind, the results presented in this paper demonstrate that the single-fluid single-temperature approach implemented in RALEF-2D can (i) reproduce the general shape of the experimental ion energy distribution and (ii) provide a reliable prediction for the absolute number of ions detected in the experiments.

6. Conclusion

We have undertaken a joint experimental and theoretical study of plasma expansion arising from Nd:YAG laser irradiation of tin microdroplets. The experimentally-recorded ion energy distribution is found to exhibit a complex, non-monotonic dependence on ion kinetic energy. Charge-state resolved measurements of the ion energy spectra reveal the existence of peaks centered near 2 keV in the Sn 3+ –Sn 8+ distributions. Two-dimensional radiation-hydrodynamic simulations performed using a single-fluid single-temperature approach are shown to reproduce the overall shape of the experimentally-recorded ion energy distribution and provide a reliable prediction for the absolute number of ions detected in the experiments. The existence of a peak in the experimental ion energy distribution near 2 keV is attributed to the formation of a quasi-spherical expanding shell at early times in the plasma expansion. Our interpretation of the plasma dynamics in terms of two distinct bursts of laser-induced ablation indicates that the observed ion energy distribution would in general be sensitive to the temporal profile of the laser pulse. The results of the present work are therefore specific for a Gaussian temporal profile with a laser intensity varying on the timescale of ∼10 ns.

Acknowledgments

J Sheil would like to thank A Sunahara for sharing his simulation results and for useful discussions. We would also like to thank W van der Zande for useful discussions. This project has received funding from European Research Council (ERC) Starting Grant No. 802648. Part of this work has been carried out at the Advanced Research Center for Nanolithography (ARCNL), a public-private partnership of the University of Amsterdam (UvA), the Vrije Universiteit Amsterdam (VU), NWO and the semiconductor equipment manufacturer ASML. Part of this work was carried out on the Dutch national e-infrastructure with the support of SURF Cooperative.

Data availability statement

The data that support the findings of this study are available upon reasonable request from the authors.

April 16, 2024

10 min read

New Prostate Cancer Treatments Offer Hope for Advanced Cases

Major discoveries during the past 10 years have transformed prostate cancer treatment, enabling it to proceed even for the most advanced form of the disease

By Marc B. Garnick

David Cheney

D eciding how to diagnose and treat prostate cancer has long been the subject of controversy and uncertainty. A prime example involves prostate-specific antigen (PSA) testing, a blood test for a telltale protein that can reveal cancer even when the patient has no symptoms. After its introduction in the early 1990s, PSA testing was widely adopted—millions of tests are done in the U.S. every year. In 2012, however, a government task force indicated that this test can lead to overtreatment of cancers that might have posed little danger to patients and so might have been best left alone.

While arguments for and against PSA testing continue to seesaw back and forth, the field has achieved a better grasp on what makes certain prostate cancers grow quickly, and those insights have paved the way for better patient prognoses at every stage of the disease, even for the most advanced cases. A prostate cancer specialist today has access to an enhanced tool set for treatment and can judge when measures can be safely deferred.

The importance of these advances cannot be overstated. Prostate cancer is still one of the most prevalent malignancies. Aside from some skin cancers, prostate cancers are the most common cancers among men in the U.S. Nearly 270,000 people in America will be diagnosed with prostate cancer this year, and it is the fourth most common cancer worldwide. Fortunately, the vast majority of patients will live for years after being diagnosed and are more likely to die of causes unrelated to a prostate tumor.

On supporting science journalism

If you're enjoying this article, consider supporting our award-winning journalism by subscribing . By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.

At its most basic level, prostate cancer is a malignancy that occurs in the prostate gland, which produces fluid that mixes with sperm from the testicles to make semen. The prostate is located in front of the rectum, below the bladder and above the penis, and cancer in the gland has four major stages.

Early on, localized tumors show no evidence of extension beyond the prostate gland. A second, “regionally advanced” form of the disease remains close to the prostate. Then there are metastatic prostate cancers, which spread outside the gland to other parts of the body. Treatment of tumors in this category has benefited from improved diagnostic imaging tests. In fact, with these tests, cancer specialists have characterized the fourth category, oligometastatic prostate cancer, a disease stage on a continuum between localized prostate cancer and more broadly dispersed metastatic disease. Major discoveries in the past 10 years have transformed the way we approach each type of prostate cancer, and these advances are likely to continue for decades to come.

The first treatment steps for people with localized cancer involve risk stratification. Through this process, a physician gauges the likelihood of a cancer’s being eliminated or cured by local treatment (usually surgery or radiation) and, if it does abate, of its returning. A physician determines the risk based on PSA results, physical examination of the prostate gland and inspection of cells from the biopsied tumor.

The right course of action for a patient with elevated PSA levels continues to undergo constant revision. Until five to seven years ago, a physician evaluated a person with high PSA by feeling their prostate gland for potentially cancerous abnormalities. Invariably, the next step would be a needle biopsy—an uncomfortable procedure in which the physician obtains snippets of prostate tissue through the rectum.

But we now have a way to biopsy through the perineum—the area between the back of the scrotum and the anal-rectal area. Thanks to technical improvements, it can be done in an outpatient setting without general anesthesia or sedation. The technique reduces the patient’s risk of infection and need for antibiotics because it doesn’t disrupt the bacterial flora in the rectum. In a recent study, researchers compared outcomes in patients who underwent a trans­rectal biopsy and received antibiotics with those for people who had a transperineal biopsy with minimal to no antibiotics. They found the two approaches comparable in terms of complications from infections.

Even more exciting is the prospect of eliminating biopsies altogether. When a patient has an abnormal PSA value but their rectal examination shows no obvious evidence of cancerous deposits, physicians can now use magnetic resonance imaging (MRI) to look at the prostate and surrounding tissue. MRI scans are best for identifying clinically significant cancers—those that, if left untreated or undiagnosed, could eventually spread. MRI can also uncover more extensive cancer spread or tumors in unusual locations such as the front of the prostate.

Another benefit of MRI procedures is that they identify fewer clinically insignificant cancers—those that are unlikely to cause problems and might best be left alone. In this case, failure to detect certain cancers is a good thing because it spares people unnecessary treatment. In some medical centers in the U.S. and many in Europe, a physician will perform a biopsy only if the MRI scan does reveal evidence of clinical significance. Studies that have compared the two diagnostic approaches—routine biopsy for all patients with elevated PSA levels versus biopsies based on abnormal MRI findings—found they are similarly effective at detecting clinically significant cancers.

Once a patient is diagnosed with prostate cancer, what happens next? For decades the debate over treatment has been just as contentious as the debate over diagnosis. Fortunately, new research from the U.K. has provided some clarity. Investigators there studied several thousand people with elevated PSA levels whose prostate biopsies showed cancer. These patients were randomized to receive surgical removal of the cancerous gland, radiation treatments or no active treatment at all. At the end of 15 years of comprehensive follow-up, about 3 percent of patients in each group had died of prostate cancer, and nearly 20 percent in each group had died of unrelated causes.

Based on the results of this study and others, more people are now being offered “active surveillance” after a prostate cancer diagnosis, in which treatment is either delayed or avoided altogether. Careful monitoring of patients who have not undergone surgery or radiation is becoming more common; it is now being extended even to those with more worrisome tumors. The monitoring involves a range of measures: PSA testing every three to six months, physical examination of the prostate gland and assessment of the patient’s urinary symptoms. Those tests are followed by repeat biopsies at increasing intervals, as long as there are no significant pathological changes.

If a cancer is identified as having either intermediate- or high-risk features, doctors need to track its progression, usually with bone scans using radio­­pharma­ceut­i­cals and with abdominal-pelvic computed tomography (CT) scans, which may show any spread in the areas to which prostate cancer most often metastasizes. Unfortunately, these techniques are not sensitive enough to reliably detect cancer in structures less than a centimeter in diameter, such as lymph nodes. Consequently, small areas of metastatic disease may go undetected. These cases are said to be “understaged.”

Understaging can now be studied through more precise diagnostic testing. Typically patients whose disease is understaged are not treated until the cancer becomes detectable through symptoms such as urination problems or pain. The disease then may require intensive therapies, and there is less of a chance of long-term remission. One technology that can help address understaging is advanced scanning that combines radiodiagnostic positron-emission tomography (PET) with CT.

These scans can detect molecules commonly found in prostate cancer cells, such as prostate-specific membrane antigen (PSMA). If PSMA is present outside the prostate gland, such as in pelvic lymph nodes, the affected areas can be identified, and a plan can be made for targeted radiation treatments or surgical removal.

Let’s consider how PET-CT scanning can be used in clinical practice. One of my patients, a 68-year-old man, was diagnosed with prostate cancer that was localized but had high-risk features. The traditional diagnostic bone and CT scans did not show any evidence of cancer spread outside the prostate. A PET-CT scan for PSMA, however, did reveal the presence of several small deposits of cancer cells in well-defined areas of the pelvis, indicating the cancer had spread to the lymph nodes. This finding prompted treatment that included radiation therapy in the prostate gland and the cancerous lymph nodes, as well as androgen-deprivation therapy (ADT), a treatment that reduces levels of testosterone, the hormone that enables prostate cancer to grow and progress.

The more precise identification of small tumor deposits in a limited number of pelvic lymph nodes—diagnosed as oligometastatic prostate cancer—enabled a new use for an old technology in oncology called metastasis-directed therapy (MDT), which targets cancer-containing lymph nodes or bony areas with radiation. At times, surgical removal of the abnormal lymph nodes may also be incorporated into MDT. Recently published studies on the use of MDT in conjunction with conventional treatments show, in some cases, long-term remission lasting through years of follow-up. Until recently, such a scenario was unthinkable for people whose prostate cancer had spread to their lymph nodes. My patient had the PSMA scan and MDT, as well as a relatively short course of ADT. He is cancer-free for now.

Precise identification of small metastatic deposits has other positive benefits. ADT has for decades been the mainstay for treating many forms of prostate cancer. Patients must continue the therapy for years, sometimes for the rest of their lives. Side effects of ADT are similar to those experienced during menopause. In fact, “andropause” is the term that captures the effects of ADT. Lower levels of testosterone are accompanied by a multitude of symptoms, including but not limited to loss of libido, erectile dysfunction, weight gain, hot flashes, bone loss, cognitive impairment, mood changes, diminished energy, and worsening of preexisting heart and vascular problems.

Studies of MDT for oligometastatic prostate cancer have raised the question of whether ADT could be delayed, administered for a shorter duration or even omitted in patients who otherwise would have required it. By strategically deploying traditional forms of localized treatment—usually surgery to remove the prostate gland or radiation—with added MDT for oligometastatic disease, doctors can significantly shorten the duration of ADT or potentially eliminate it. Such an approach would have been difficult to imagine five years ago. Longer-term follow-up studies will help scientists determine whether some people diagnosed in this fashion can go into an extended remission.

F or advanced forms of prostate cancer that have spread to other parts of the body, ADT has been the main treatment. Physicians historically have generally recommended surgical removal of the testicles—the primary source of testosterone—or the administration of other hormones that block the production and action of testosterone. In the mid-1980s I was involved with research on drugs called luteinizing hormone–releasing hormone analogues that lowered testosterone by shutting off the signal in the brain that instructs the testicles to make testosterone. Today newer agents have been added that further lower and block testosterone’s action.

The goal of prostate cancer treatment at later stages is to eliminate multiple sources of testosterone. As noted earlier, testosterone in the body comes predominantly from the testicles; the adrenal glands also produce a small amount. But prostate cancer cells can evolve to produce their own androgens. Testosterone and its active form, dihydrotestosterone (DHT), traverse the membranes of prostate cancer cells and interact with androgen receptors in the cytoplasm, a cell’s liquid interior. The receptors then transport DHT to the nucleus, where it instructs the cancer cell to grow, replicate and spread.

Traditional ADT does little to affect either the production of testosterone by the adrenal glands or androgen-producing prostate cancer cells, and it doesn’t block the activity of androgen receptors. But new approaches to ADT may address these shortcomings. Drug combinations that affect all these processes have substantially improved survival in people with metastatic prostate cancer—and, more important, patients are able to tolerate these more intensive treatment programs.

Instead of just one drug to decrease testosterone, new standards for treatment prescribe combinations of two or even three drugs. In addition to traditional ADT, there are medications such as do­cetaxel, a chemotherapy, and other new drugs that can block the production of testosterone by the adrenal glands or cancer cells or stop it by interfering with the activity of androgen receptors. All these drug combinations have resulted in meaningful improvements in survival.

Yet another therapy for advanced disease involves the identification of PSMA-expressing cancer cells that can be targeted with pharmaceuticals designed to deliver radioactive bombs. An injectable radiopharmaceutical can be delivered selectively to these cells, leaving healthy cells mostly unaffected. This therapy, lutetium-177-­PSMA-617 (marketed as Pluvicto), has been approved by the U.S. Food and Drug Administration for the treatment of prostate cancer that has become resistant to other forms of ADT and chemotherapy. It is likely to become an important therapy for even earlier stages of prostate cancer.

Genetics and genomic testing of patients and cancers have also helped in the quest for improvement of symptoms and longer survival. Some genetic mutations that are known to increase the risk of breast and ovarian cancer have also been associated with a heightened risk of prostate cancer. Testing for such mutations is becoming much more common, and patients who have them can be treated with specific therapies that block their deleterious effects, leading to better outcomes.

An understanding of the type of mutation is also critical—for both patients and their family members. Germline mutations are inherited from a patient’s biological parents by every cell in the body. These mutations can be passed along to the patient’s children. A somatic mutation, in contrast, is not inherited but develops in the cancer itself. Targeted therapies designed specifically to correct the effects of either germline or somatic mutations have produced significant improvements in patient longevity. Some of the most commonly recognized cancer mutations—either somatic or germline—are those in BRCA genes, which have been associated with early-onset breast and ovarian cancer.

When researchers studied cancer in families with BRCA mutations, they uncovered many cases of prostate cancer. This finding led to the discovery that BRCA mutations appeared in both men and women in these families. The mutations change the way DNA is repaired, introducing defects that can result in cancer formation. Drugs have now been developed that treat cancers linked to the BRCA mutations. Several such drugs—those in a class called poly­(ADP-ribose) polymerase (PARP) inhibitors—have recently received FDA approval for use as a treatment in people with these mutations. This research has led to more widespread genetic testing of patients with prostate cancer and, when germline mutations are found, family genetic counseling.

All these advances have occurred over the past decade—an incredibly short interval in the context of cancer oncology. Current options for early-stage prostate cancer enable physicians and patients to feel more at ease with conservative choices rather than immediate interventions with negative side effects. For patients whose cancers are advanced at initial diagnosis or progress and become metastatic, the treatment of oligometastases now often leads to long-term remission and requires fewer treatments with harmful systemic side effects. For those with more widespread metastatic disease, their cancer can now be managed with improved therapeutics based on a better understanding of disease biology. These new strategies have begun to transform this once rapidly fatal disease into a chronic condition that people can live with for years or even for their full life expectancy.

Marc B. Garnick is Gorman Brothers Professor of Medicine at Harvard Medical School and Beth Israel Deaconess Medical Center in Boston. He is editor in chief of Harvard Medical School’s 2024–2025 Report on Prostate Diseases.

Find Info For

• Current Students
• Prospective Students
• Research and Partnerships
• Entrepreneurship and Commercialization

Quick Links

• Health and Life Sciences
• Info Security and AI
• Transformative Education
• Purdue Today
• Purdue Global
• Purdue in the News

April 3, 2024

SK hynix announces semiconductor advanced packaging investment in Purdue Research Park

SK hynix announced Wednesday (April 3) semiconductor advanced packaging investment in Purdue Research Park. From left to right: Indiana Gov. Eric Holcomb; Kwak Noh-jung, SK hynix president and CEO; Woojin Choi, SK hynix executive vice president; Arati Prabhakar, director, White House Office of Science and Technology Policy, and assistant to the president for science and technology; Mung Chiang, Purdue University president (speaking); Arun Venkataraman, U.S. Department of Commerce assistant secretary; U.S. Sen. Todd Young; Hyundong Cho, ambassador of the Republic of Korea to the United States; David Rosenberg, Indiana secretary of commerce; Mitch Daniels, Purdue Research Foundation chairman. (Purdue University/Kelsey Lefever)

The company's facility for AI memory chips marks the largest single economic development in the history of the state

WEST LAFAYETTE, Ind. — SK hynix Inc. announced Wednesday (April 3) that it plans to invest close to $4 billion to build an advanced packaging fabrication and R&D facility for AI products in the Purdue Research Park. The development of a critical link in the U.S. semiconductor supply chain in West Lafayette marks a giant leap forward in the industry and the state. 

“We are excited to build a state-of-the-art advanced packaging facility in Indiana,” said SK hynix CEO Kwak Noh-Jung. “We believe this project will lay the foundation for a new Silicon Heartland, a semiconductor ecosystem centered in the Midwest Triangle. This facility will create local, high-paying jobs and produce AI memory chips with unmatched capabilities, so that America can onshore more of its critical chip supply chain. We are grateful for the support of Gov. Holcomb and the state of Indiana, of President Chiang at Purdue University, and of the broader community involved, and we look forward to expanding our partnership in the long run.”

SK hynix joins Bayer, imec, MediaTek, Rolls-Royce, Saab and many more national and international companies bringing innovation to America's heartland. The new facility — home to an advanced semiconductor packaging production line that will mass-produce next-generation high-bandwidth memory, or HBM, chips, the critical component of graphic processing units that train AI systems such as ChatGPT — is expected to provide more than a thousand new employment opportunities in the Greater Lafayette community. The company plans to begin mass production in the second half of 2028.

The project marks SK hynix’s intention for long-term investment and partnership in Greater Lafayette. The company’s decision-making framework prioritizes both profit and social responsibility while promoting ethical actions and accountability. From infrastructure developments that make accessing amenities easier to community empowerment projects such as skill development and mentorship, the SK hynix advanced packaging fabrication marks a new era of collaborative growth.

“Indiana is a global leader in innovating and producing the products that will power our future economy, and today’s news is proof positive of that fact,” said Indiana Gov. Eric Holcomb. “I’m so proud to officially welcome SK hynix to Indiana, and we’re confident this new partnership will enhance the Lafayette-West Lafayette region, Purdue University and the state of Indiana for the long term. This new semiconductor innovation and packaging plant not only reaffirms the state’s role in the hard-tech sector, but is also another tremendous step forward in advancing U.S. innovation and national security, putting Hoosiers at the forefront of national and global advancements.” 

Investment in the Midwest and Indiana was spurred by Purdue’s excellence in discovery and innovation and its track record of exceptional R&D and talent development through collaboration. Partnerships among Purdue, the corporate sector, and the state and federal government are essential to advancing the U.S. semiconductor industry and establishing the region as the Silicon Heartland.

“SK hynix is the global pioneer and dominant market leader in memory chips for AI,” Purdue President Mung Chiang said. “This transformational investment reflects our state and university’s tremendous strength in semiconductors, hardware AI and development of the Hard-Tech Corridor. It is also a monumental moment for completing the supply chain of the digital economy in our country through the advanced packaging of chips. Located at Purdue Research Park, the largest facility of its kind at a U.S. university will grow and succeed through innovation.”

In 1990 the U.S. was producing nearly 40% of the world’s semiconductors. However, as manufacturing moved to Southeast Asia and China, the U.S. global output of semiconductor manufacturing has fallen to closer to 12%.

“SK hynix will soon be a household name in Indiana,” said U.S. Sen. Todd Young. “This incredible investment demonstrates their confidence in Hoosier workers, and I’m excited to welcome them to our state. The CHIPS and Science Act opened a door that Indiana has been able to sprint through, and companies like SK hynix are helping to build our high-tech future.” 

To aid in bringing semiconductor manufacturing closer to home and shoring up global supply chains, the U.S. Congress introduced the Creating Helpful Incentives to Produce Semiconductors for America Act, or CHIPS and Science Act, on June 11, 2020. Signed by President Joe Biden on Aug. 9, 2022, it funds holistic development of the semiconductor industry to the tune of $280 billion. It supports the nation's research and development, manufacturing, and supply chain security of semiconductors.

“When President Biden signed the bipartisan CHIPS and Science Act, he put a stake in the ground and sent a signal to the world that the United States cares about semiconductor manufacturing,” said Arati Prabhakar, President Biden’s chief science and technology advisor and director of the White House Office of Science and Technology Policy. “Today’s announcement will strengthen the economy and national security, and it will create good jobs that support families. This is how we do big things in America.”

Purdue Research Park, one of the largest university-affiliated incubation complexes in the country, unites discovery and delivery with easy access to Purdue faculty experts in the semiconductor field, highly sought-after graduates prepared to work in the industry, and vast Purdue research resources. The park also offers convenient accessibility for workforce and semitruck traffic, with access to I-65 just minutes away.

This historic announcement is the next step in Purdue University’s persistent pursuit of semiconductor excellence as part of the Purdue Computes initiative. Recent announcements include these

• Purdue University Comprehensive Semiconductors and Microelectronics Program
• A strategic partnership with Dassault Systèmes to improve, accelerate and transform semiconductor workforce development
• European technology leader imec opens innovation hub at Purdue
• The nation’s first comprehensive Semiconductor Degrees Program
• Purdue continues to create unique lab-to-fab ecosystem for the state and country
• Green2Gold, a collaboration between Ivy Tech Community College and Purdue University to grow Indiana’s engineering workforce

What they’re saying

• “This decision by a world-renowned, best-in-class company represents a dramatic fulfillment of Purdue’s duty to serve the state as not only its premier academic institution but also its No. 1 economic asset. It’s also a gratifying validation of our Discovery Park District initiative to bring new opportunities to our students, faculty and Greater Lafayette neighbors. Today marks the Purdue ecosystem’s latest and greatest, but assuredly not its last, contribution to a more prosperous Indiana and a stronger America.” — Mitch Daniels, chairman of the board, Purdue Research Foundation
• “On behalf of my fellow trustees, we are pleased to welcome SK hynix Inc. to the Purdue Research Park. Their arrival will significantly strengthen Purdue University’s dual commitments to educating the next generation of workforce leaders in semiconductors and supporting the national security of our nation.” — Michael Berghoff, chair, Purdue Board of Trustees
• “The impact of SK hynix is more than the creation of high-paying careers for Hoosiers. Undergraduates will have opportunities for internships, co-op and full-time employment when they graduate. Graduate students and faculty will work closely with SK hynix researchers, not only on basic research, but also to accelerate the transition of research into pilot production and manufacturing. This is just the beginning. As other companies see what’s happening here in the heart of the heartland, they’ll come too, and a significant new cluster of semiconductor manufacturing and research will emerge.” — Mark Lundstrom, chief semiconductor officer, Purdue University
• “West Lafayette is thrilled to join our national efforts to bring the semiconductor industry to the United States through President Biden’s CHIPS and Science Act. This partnership will leverage Purdue University’s science and research expertise with SK hynix’s innovation in semiconductor technology. The impact on West Lafayette will enable us to continue to provide the high level of service our community expects and to increase our quality-of-life amenities for the region so we can attract and retain the excellent graduates of Purdue University. In addition, SK hynix’s global dedication to net zero carbon emissions by 2050, water process reduction and recycling, and zero-waste-to-landfill programs aligns with our community’s commitment to environmental stewardship. We are grateful for SK hynix’s investment and commitment to West Lafayette and for our partners Purdue University, Purdue Research Foundation, the city of Lafayette, Tippecanoe County and the Greater Lafayette region.” — Erin Easter, mayor of West Lafayette
• “The pandemic disruption has shown the reliance on semiconductors, with production concentrated in limited regions around the world. Greater Lafayette has worked continuously and cooperatively for years to position ourselves for an opportunity of this magnitude, and we look forward to the long-term economic impact this will have on our communities. The collaborative efforts between cities and county governments, Purdue University, the state of Indiana and Sen. Todd Young’s office is a testament to these efforts. Our joint investments in infrastructure, innovation, along with quality-of-life initiatives, have contributed to this venture becoming a reality. We look forward to working with and welcoming SK hynix to Greater Lafayette!” — Tony Roswarski, mayor of Lafayette
• “Ivy Tech, as Indiana’s largest postsecondary institution, is focused on building Indiana talent pipelines aligned to employers and emerging industries which strengthen Indiana’s economy. The microelectronics industry will play a key role in Indiana’s success, which is why we are pleased to work with SK hynix and Purdue to provide training, credentials and degrees designed for the semiconductor industry. SK hynix’s commitment to Indiana reinforces that we all win when we address complex issues through strong partnerships." — Sue Ellspermann, president, Ivy Tech Community College
• “Semiconductors and microelectronics are at the forefront of focus for Purdue Research Foundation. I am pleased to welcome SK hynix to Indiana and start the hard work of ensuring this is the best business decision that SK hynix has ever made.” — Brian Edelman, president, Purdue Research Foundation
• “The Alliances team is thrilled to welcome SK hynix to the Purdue ecosystem, and we look forward to empowering them to thrive here in Indiana with all the immense assets Purdue and Greater Lafayette offer. We look forward to forging a strong relationship with mutual value for SK hynix, Purdue Research Foundation and the broader Greater Lafayette community for many years to come.” — Gregory Deason, senior vice president of alliances and placemaking, Purdue Research Foundation
• “During my time at Purdue Research Foundation, we have consistently been successful in assisting our partners like Saab in developing complex builds well ahead of schedule and within budget. I look forward to building on our excellent track record with SK hynix to help them in creating a state-of-the-art facility which best meets their unique needs.” — Richard Michal, senior vice president of capital projects and facilities, Purdue Research Foundation

About SK hynix Inc.

SK hynix Inc., headquartered in Korea, is the world’s top-tier semiconductor supplier offering Dynamic Random Access Memory chips (“DRAM”), flash memory chips (“NAND   flash”)   and CMOS Image Sensors (“CIS”) for a wide range of distinguished customers globally. The Company’s shares are traded on the Korea Exchange, and the Global Depository shares are listed on the Luxembourg Stock Exchange. Further information about SK hynix is available at   www.skhynix.com ,   news.skhynix.com .  

About Purdue Research Foundation

Purdue Research Foundation is a private, nonprofit foundation created to advance the mission of Purdue University. Established in 1930, the foundation accepts gifts, administers trusts, funds scholarships and grants, acquires and sells property, protects and licenses Purdue's intellectual property, and supports creating Purdue-connected startups on behalf of Purdue. The foundation operates Purdue Innovates, which includes the Office of Technology Commercialization, Incubator and Ventures. The foundation manages the Purdue Research Park, Discovery Park District, Purdue Technology Centers and Purdue for Life Foundation.

For more information on licensing a Purdue innovation, contact the Office of Technology Commercialization at [email protected] . For more information about involvement and investment opportunities in startups based on a Purdue innovation, contact Purdue Innovates at [email protected] .

About Purdue University

Purdue University is a public research institution demonstrating excellence at scale. Ranked among top 10 public universities and with two colleges in the top four in the United States, Purdue discovers and disseminates knowledge with a quality and at a scale second to none. More than 105,000 students study at Purdue across modalities and locations, including nearly 50,000 in person on the West Lafayette campus. Committed to affordability and accessibility, Purdue’s main campus has frozen tuition 13 years in a row. See how Purdue never stops in the persistent pursuit of the next giant leap — including its first comprehensive urban campus in Indianapolis, the new Mitchell E. Daniels, Jr. School of Business, and Purdue Computes — at https://www.purdue.edu/president/strategic-initiatives . 

Media contact:

Tim Doty, [email protected]

Note to journalists:   Photo, b-roll and sound bites from this announcement will be available for media use on   Google Drive .

Research News

Communication.

• OneCampus Portal
• Brightspace
• BoilerConnect
• Faculty and Staff
• Human Resources
• Colleges and Schools

Info for Staff

• Purdue Moves
• Board of Trustees
• University Senate
• Center for Healthy Living
• Information Technology
• Ethics & Compliance
• Campus Disruptions

Purdue University, 610 Purdue Mall, West Lafayette, IN 47907, (765) 494-4600

© 2015-24 Purdue University | An equal access/equal opportunity university | Copyright Complaints | Maintained by Office of Strategic Communications

Trouble with this page? Disability-related accessibility issue? Please contact News Service at [email protected] .

Your browser is too old to optimally experience this website. Upgrade your browser to improve your experience.

dr. Stefan Witte

Back to all people

Group leader: EUV Generation & Imaging

• Privacy statement
• Disclaimer & Cookies
• Publications
• Accessibilty statement

Visiting address

• Science Park 106
• 1098 XG Amsterdam
• The Netherlands

Delivery address

• Science Park 102

Mail address

• PO Box 93019
• 1090 BA Amsterdam
• T. +31 20 8517100
• E-mail: [email protected]

© ARCNL 2024

• Vanderbilt welcomes new executive director of Advanced Computing Center for Research and Education

Media Inquiries

• 615-322-6397 Email

Latest Stories

• WATCH: The Wond’ry builds a climate-positive future with Climate Innovation Accelerator
• WATCH: Class of 2024 students build belonging in personal ways

Apr 19, 2024, 3:37 PM

The former director of scientific information technology at the National Institute of Allergy and Infectious Diseases, Christian Presley, has been appointed executive director of Vanderbilt University’s Advanced Computing Center for Research and Education. His appointment began April 15.

The search for a new director began last autumn as part of the university’s dedication to providing the best computing resources for cutting-edge research at Vanderbilt. ACCRE recently expanded its data storage , and the center will play a pivotal role in the new College of Connected Computing , a transformative college dedicated to computer science, AI, data science and related fields.

“Christian Presley’s appointment as ACCRE’s new executive director marks a pivotal moment for Vanderbilt’s computing landscape,” Provost C. Cybele Raver said. “With his exceptional record in IT management and forward-thinking approach, he’s ideally suited to guide ACCRE forward.”

In his previous roles, Presley managed comprehensive IT and computing services, overseeing significant scientific data resources and supporting advanced research infrastructures. He was instrumental in leading the rollout of NIAID’s newest flagship high-performance computer cluster, a 15-petabyte expansion of their high-speed storage and the creation of high-performance graphical group workstations in the data center.

He was also responsible for leading teams for direct support of scientific instrumentation, where the balance of easy and remote access to facilitate research must be balanced with ever changing cybersecurity, as well as outreach to engage researchers directly and address needs. His work with the scientific community there spanned cryo-electron microscopy to AI-assisted protein folding.

Presley is a Nashville native and alumnus of Hume-Fogg High School. “I am very excited to be back in Nashville at such a pivotal time for the growth of research computing at Vanderbilt,” he said. “With so many new initiatives in AI and growing needs for computing across all academic areas, there is so much ACCRE can offer to campus. I see ACCRE as an important partner for both research and education for the entire Vanderbilt community.”

Before his time at NIAID, he led the IT and research computing initiatives at the Institute for Bioscience and Biotechnology Research, a partnership between the University of Maryland and the National Institute of Standards and Technology, where he was pivotal in developing and managing teams across web development, desktop support, systems administration, research computing and data analytics/machine learning. His extensive experience and strategic vision are integral to advancing Vanderbilt’s computing capabilities and supporting the university’s mission of fostering breakthrough discoveries and innovations.

“Our faculty-led search committee and I were all struck by the unique combination of leadership skills, technical expertise and strategic thinking that Christian brings to this role,” said Vice Provost for Research and Innovation Padma Raghavan. “His deep expertise and experience in high performance computing, data analytics, machine learning and more have prepared him exceptionally well to provide the critical leadership needed to take research computing at Vanderbilt to the next level.”

Visit the ACCRE website for more information on ACCRE and its services.

Keep Reading

Vanderbilt announces search for new ACCRE executive director

Vanderbilt to establish a college dedicated to computing, AI and data science

Vanderbilt hosts SmartComp 2023, prestigious IEEE international conference on smart computing

Explore story topics.

• Advanced Computing Center for Research and Education
• C. Cybele Raver
• Christian Presley
• College of Connected Computing
• computer science
• data science
• Discovery Vanderbilt
• MyVU latest headlines
• Padma Raghavan