SLS-2 Science Case - PSI

SLS-2

Science Case

Phil Willmott

Swiss Light Source

Paul Scherrer Institut

CH-5232 Villigen

Version 1.01

November 2016

ii

Contents

1 Introduction 1

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2 Basic concepts 13

3 Sources 17

3.1 Power density on optical components . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3.2 Small-period undulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3.3 Superconducting undulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.4 Twin bending-magnet sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.5 Superbends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3.6 Fast polarization switching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

4 Science case 31

4.1 Lensless imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

4.1.1 CXDI and ptychography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

4.1.2 Scanning SAXS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

4.1.3 X-ray photon correlation spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 39

4.2 Diffraction techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.2.1 Macromolecular crystallography . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.2.2 Microdiffraction, diffraction tomographies, and total scattering . . . . . . . . . 44

4.2.3 Diffraction studies under extreme conditions . . . . . . . . . . . . . . . . . . . 46

–iii–

iv CONTENTS

4.3 Full-field tomography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

4.4 Spectroscopies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

4.4.1 RIXS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

4.4.2 Photoelectron spectroscopies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

4.5 Time-resolved studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Chapter 1

Introduction

This document describes new scientific opportunities which would be made available to the users at the

Swiss Light Source, mainly as a consequence of an upgrade of the storage ring to a so-called “diffraction-

limited” low-emittance facility, but also because of timely innovations that could be implemented at

this time but which are not directly related to the improved emittance.

After this brief introduction and an outline of the operational principles of diffraction-limited

storage rings (DLSRs), the advantages with regards to the different manners in which the performance

improvement of DLSRs can be exploited will be detailed.

The three central ‘grand challenges’ addressed by research at the Paul Scherrer Institute include

Human health Fundamental questions in biology, from the atomic scale, addressing the structure,

function, and dynamical changes in proteins and other relevant biomolecules; to the hierarchical

nature and properties of constituent elements of the human body, such as neural tissue and

bones; to the noninvasive identification and treatment of tumours. These research drives also

encompass the neurosciences, personalized medicine, and the pharamceutical industry.

Energy and the environment A search for clean energy and its storage; sequestration of green-

house gases; and novel, cheap, and efficient catalytical systems.

Matter and materials Electron bonding in matter, particularly at the very small energy scales that

determine the properties of many emerging materials, particularly the strongly-correlated elec-

tron materials such as high-temperature superconductors, magnetic and multiferroic materials,

and Mott insulators.

–1–

2 INTRODUCTION

A critical parameter for studying this broad palette of systems using x-rays is brilliance. DLSRs

promise up to two orders of magnitude increase in brilliance at storage rings by reducing the horizontal

electron emittance by this factor, resulting in narrower x-ray beams with smaller source sizes, and a

consequent increase in the degree of coherence. This will have many positive impacts on x-ray science,

and also prompt parallel advances in associated technologies, not least in magnet-based x-ray sources.

Small source size This allows a smaller focus for a given optical magnification. This will be ben-

eficial for all scanning methods, including scanning x-ray fluorescence, absorption, micro-x-ray

diffraction, scanning x-ray transmission microscopy; and will also open opportunities to investi-

gate heterogeneous systems, or systems with domain structure, such as resonant inelastic x-ray

scattering.

Alternatively, a given desired spot size will translate to a longer working distance between

focussing optics and sample, allowing the introduction of more sophisticated sample environ-

ments, such as used in high-ressure, or ambient-pressure photoelectron spectroscopy, and in

high-pressure x-ray diffraction.

A small illuminated spot on a sample for which the emission spectrum is of interest, such as

in x-ray fluorescence using a von-Hamos arrangement, or in ‘hν2-RIXS’ directly translates to

higher spectral resolution.

Because the reduction in emittance is in the electrons’ orbital plane, the beam will be signifi-

cantly rounder, resulting in a less angularly anisotropic distribution of the coherent flux, thereby

reducing astigmatic artefacts in techniques such as phase-contrast tomography.

Parallel beam A parallel beam is essential for high-resolution studies using x-ray diffraction, in-

cluding macromolecular crystallography. In MX, there are are conflicting desires – one wants

a small spot size for the generally small crystals produced in MX, but also parallelity. The

approximately 40 times reduction in divergence can thus be invested in either or both of these

properties.

Scanning tomographies, that is, those which use a pencil beam and scanning to generate the

tomograpms, will also gain from the ability to produce small-cross-section and parallel x-ray

beams.

3

Coherence Lensless imaging techniques are becoming increasingly popular to investigate systems in

the ‘resolution gap’ between scanning x-ray methods (normally limited to a few tens of nm)

and electron microscopy (few Angstrom resolution). These methods depend intimately on the

coherent flux of the x-ray beam, which will increase by the same factor increase promised in

brilliance, to a few percent in the hard x-ray regime.

Novel source technologies As will be detailed in Section 3, the spectra produced by small-period

undulators at SLS-2 are much cleaner than those at the present facility, due primarily to the fact

that the beam width σex is an order of magnitude smaller. As a result, entire undulator harmonics

can be used for those experiments that do not demand the best spectral bandwidths, providing

thereby an additional factor of 100 or more photons. It is expected that somme methods,

particularly fast tomography and ptychography, will be able to record data at rates three orders

of magnitude faster or more, revolutionizing the breadth and complexity of experiments that

can be performed.

The above summary is outlined graphically in Figure 1.1.

The Swiss Light Source (SLS) first came online in 2001, providing photons from four beamlines to

expert users. By 2004, the SLS was running at near full capacity, covering photon energies from the

vacuum ultraviolet at around 10 to 15 eV at the Surface and Interface Spectroscopy (SIS) beamline,

through to hard x-rays in excess of 50 keV at the superbend TOMCAT beamline.

The SLS and APS were the first storage rings to be designed and optimized for top-up operation

[1]. This feature, in combination with sophisticated magnet-lattice alignment procedures and orbit

feedback loops meant that the SLS would become the benchmark synchrotron facility regarding how

close its performance was in terms of emittance to the theoretically best values – thanks to a coupling

ratio between the horizontal and vertical emittances of less than 5 × 10−4, it was possible by 2008

to obtain electron-beam emittances in the vertical direction as low as 3 pm·rad and an electron-

beam stability to below one micron [2]. By 2011, the coupling ratio had been further reduced to

2 × 10−4, resulting in a vertical electron emittance of only 1 pm·rad [3]. As is the case in all third-

generation synchrotron facilities, the horizontal electron-beam emittance is typically over two orders

of magnitude higher, due primarily to the lower limits set by the double- or triple-bend achromats

used to maintain the electron beam on a closed orbit. The horizontal electron-beam emittance of the

SLS is approximately 5 nm·rad.

4 INTRODUCTION

Figure 1.1: Case for DLSRs. As a consequence of a signifcant reduction in the horizontal emittance εx, many

experimental techniques will profit greatly, while in addition, new source technologies will emerge, driving the

science still more.

By 2010, a plot of the horizontal emittance would show the SLS sitting squarely amid a cluster of

third-generation facilities, including the Diamond Light Source (DLS, GB), the Shanghai Synchrotron

Radiation Facility (SSRF, China), Soleil (FR), and the Advanced Light Source (ALS, USA). Larger-

circumference facilities such as the ESRF, APS, and PETRA III had marginally superior emittances,

thanks to the smaller angles subtended by the bend achromats, although this was partly offset by

their generally higher storage-ring energies [Figure 1.2(a)].

The comparative performance of smaller, medium-energy storage rings and larger, higher-energy

facilities highlights the underlying machine physics that would be exploited in the development of

fourth-generation facilities, coined “diffraction-limited storage rings”, or DLSRs [4–7]. The lower

limit to the horizontal emittance scales with the cube of the subtended angle θBM of the bending

magnets, and with the square of the storage-ring energy E . As the circumference C is approximately

5

2 4 6 8Storage-ring energy [GeV]

0.1

1

10E

mitt

ance

[nm

.rad

]

0 1000 2000Circumference [m]

105

106

107

108

M [n

m.r

ad.m

3 /GeV

2 ]

CLSAusLS

SSRLElettra

Pohang IISLSALBA

ShanghaiBESSY II

SoleilALS

NSLS II

Diamond

PETRA III

ESRF

CHESS II

APS

SPring8

Soleil-II

Diamond-II

MAX-IVElettra2.0Sirius

ALS-U

ESRF-EBS

APS-U

SPring8-II

PETRA III

SPring8APS

ESRF

NSLS IICHESS II

Diamond

Shanghai

ANKA

Elettra

SoleilSLS

BeIIAus

SSRL Po IICLS ALBA

ALS

ALS-U

Elettra2.0SLS-2

Soleil-II

MAX-IVSirius

Diamond-II ESRF-EBSAPS-U

SPring8-II

(a) (b)

SLS-2

Figure 1.2: Performance of some synchrotron facilities worldwide. (a) Plot of horizontal emittance as a function

of storage-ring energy. Data points in blue circles represent third-generation facilities employing either double-

bend or triple-bend achromats, those given as yellow triangles are for operational and planned fourth-generation

facilities installed with multiple-bend achromats. (b) Plot ofM , the figure of merit for storage-ring performance,

in which the horizontal emittance is weighed by the cube of the ring circumference and the inverse square of

the storage-ring energy. The SLS and SLS-2 labels are highlighted in red.

proportional to the bending angle of each magnet dipole in the ring, a convenient figure of merit for

the performance of any given storage ring is therefore

M = εex C3/E2. (1.1)

A plot of this parameter is shown in Figure 1.2(b).

Until recently, a move to reduce θBM was hindered by the associated necessary increase in storage-

ring circumference, due on the one hand to the large increase in the number of magnet-lattice elements

and, on the other, to the introduction of accumulated misalignments between those elements. In

addition, a significant reduction in E would adversely impact the accessible range of photon energies.

Two relatively recent breakthroughs in technology, however, have allowed the construction of

magnet-lattice components with linear dimensions approximately a factor of three smaller than those

used in third-generation facilities (Figure 1.3). First, computer numerical control of machining of

6 INTRODUCTION

5 m

Figure 1.3: Miniaturization of magnet-lattice components (above, MAX-IV) has allowed approximately a factor

of three to four more elements per unit arc length compared to double- or triple-bend achromat systems typical

of third-generation facilities (below, SLS-1).

integrated magnet-lattice elements from single magnetic yokes to submicron accuracy have facilitated

miniaturization without compromising alignment. This means that seven or even nine bending mag-

nets can be inserted as single cells in a curved section of a storage ring where before only two or three

were possible. Such a sector of several bending magnets (BMs), quadrupoles, and sextupoles is called

a multibend achromat (MBA). Increasing the number of BMs by a factor X reduces the electron-beam

emittance by X3.

Secondly, pumping down to ultrahigh-vacuum conditions of the considerably narrower vacuum

tubing and vessels associated with this miniaturization has been made simpler by coating the vessel

walls with nonevaporable getter (NEG) material. NEGs are not, however, a fully necessary ingredient

(for example, the ESRF-EBS upgrade will not use them), and it remains an open question as to whether

the smaller apertures associated with NEGs will be offset or not by limitations of vacuum-chamber

impedances.

The first facility to exploit these technological advances is MAX-IV, a seven-bend achromat facility

(7-BA), which was officially inaugurated on the Summer Solstice of 2016. The 5-BA Sirius facility

in Brazil is expected to get “first light” in 2018 and take its first users in 2019. Importantly, several

other third-generation facilities, including the SLS, are planning or already undergoing upgrades to

DLSRs (the yellow triangles in Figure 1.2, see also Table 1).

7

8 INTRODUCTION

Facility, Country E Circumference Current Lattice εex

λDL∗ hνDL

∗ M

[GeV] [m] [mA] [pm· rad] [nm] [eV] ∗∗

ALS-U, USA 2.0 197 500 9-BA 50# 0.62 2000 0.095

APS-U, USA 6.0 1104 200 7-BA 67 0.84 1476 2.5

DLS-II, England 3.0 562 300 4, 5, or 6-BA? 270− 100 3.39− 1.26 366− 984 5.33− 1.97

Elettra2.0, Italy 2.0 260 400 6-BA? 230− 280? 2.89− 3.52? 429− 352 1.01− 1.23?

ESRF-EBS, France 6.0 844 200 7-HBA‡ 140 1.76 704 2.34

MAX-IV, Sweden 3.0 528 500 7-BA 200− 330†† 2.51− 4.15†† 494− 299 3.27− 5.40††

SLS-2, Switzerland 2.4 290 400 7-BA 137 1.88 659 0.64

Soleil-II, France 2.75 354 N.A. 6/7BA? ∼ 200? ∼ 2.5? ∼ 496? ∼ 1.2?

Sirius, Brazil 3.0 518 350 5-BA 250 3.14 395 3.86

SPring8-II, Japan 6.0 1436 100 5-BA 190 2.39 519 15.6

Table 1.1: Selected parameters and performance metrics of DLSRs in the design phase (D), under construction (C), or in operation

(O).

∗ The diffraction-limited photon wavelength λDL = 4πεexmatches the horizontal electron emittance. The corresponding diffraction-

limited photon energy EDL[eV] = 1239.8/λDL[nm].

∗∗ Units: ×106 [nm·rad·m3/GeV2].

# For fully coupled beam. Natural lattice emittance ≈ 100 pm·rad.

† Early design phase.

‡ HBA: hybrid bend achromat.

†† With/without insertion devices.

9

A cursory view of Figure 1.2 immediately yields some interesting information. First, implementation

of multibend achromats in DLSRs generally improves the emittance (and thereby the brilliance) by

one to two orders of magnitude compared to third-generation facilities. Interestingly, despite their

large circumferences and associated very small bending-magnet angles, the magnet-lattices of the high-

energy DLSRs (APS-U, ESRF-EBS, and SPring8-II) have been designed more conservatively, allowing

larger dynamic apertures and less aggressive focusing (sextupole optics), thus minimizing nonlinear

effects. As a consequence, their figures of merit M in Figure 1.2(b) are more modest. Their advantage,

of course, lies in their ability to access much higher photon energies1.

In Figure 1.2(b), the ALS-U and SLS-2 perform particularly well, the former primarily because of

the introduction of on-axis swap-out injection [8] which allows the use of especially small-cross-section

vacuum vessels, the latter due to a combination of innovations. The relatively small circumference of

the SLS’ storage ring means more aggressive solutions to the magnet lattice are required in order to

obtain low horizontal emittances that are competitive with other planned upgrades. An entirely novel

solution to achieve this involves the inclusion of six “antibend” magnets interlaced between the dipoles

in each 7-BA [9,10], plus bending magnets with strong longitudinal variations in their magnetic field

strengths (referred to as longitudinal-gradient bends, LGBs). The antibends are so designed that

the zero-dispersion nodes of the electron beam coincide with the peak magnetic-field positions of the

LGBs, where most photon emission occurs. An added benefit is that the antibends collectively act as

a distributed damping wiggler, reducing the emittance still further.

The reduction of the horizontal emittance in the SLS upgrade by a factor of approximately 40 lends

many potential opportunities and will drive exciting innovations along the full technology chain, from

the sources (in particular, undulators), through improved x-ray optics (regarding imperfections and

aberrations), detector technology, to handling of large data sets. These are discussed in context in the

following sections.

1The critical energy scales with the square of E .

10 INTRODUCTION

Bibliography

[1] A. Streun, M. Boge, M. Dehler, C. Gough, W. Joho, T. Korhonen, A. Ludeke, P. Marchand,

M. Munoz, M. Pedrozzi, L. Rivkin, T. Schilcher, V. Schlott, L. Schulz, and A. Wrulich, Commis-

sioning of the Swiss Light Source, Proc. PAC 1, 224 (2001).

[2] A. Andersson, M. Boge, A. Ludeke, V. Schlott, and A. Streun, Determination of a small vertical

electron beam profile and emittance at the Swiss Light Source, Nucl. Instrum. Methods A 591,

437 (2008).

[3] M. Aiba, M. Boge, N. Milas, and A. Streun, Ultra low vertical emittance at SLS through systematic

and random optimization, Nucl. Instrum. Methods A 694, 133 (2012).

[4] M. Eriksson, A. Andersson, S. Biedron, M. Demirkan, G. Leblanc, L. Lindgren, L. Malmgren,

H. Tarawneh, E. Wallen, and S. Werin, MAX 4, a 3 GeV light source with a flexible injector, in

EPAC 2002, page 686, 2002.

[5] M. Eriksson, L. J. Lindgren, M. Sjostrom, E. Wallen, L. Rivkin, and A. Streun, Some small-

emittance light-source lattices with multi-bend achromats, Nucl. Instrum. Methods A 587, 221

(2008).

[6] M. Borland, Progress toward an ultimate storage ring light source, J. Phys. Conf. Ser. 425,

042016 (2013).

[7] E. Weckert, The potential of future light sources to explore the structure and function of matter,

IUCrJ 2, 230 (2015).

[8] L. Emery and B. M, Possible long-term improvements to the Advanced Photon Source, Proc.

PAC 1-5, 256 (2003).

–11–

12 BIBLIOGRAPHY

[9] A. Streun, The anti-bend cell for ultralow emittance storage ring lattices, Nucl. Instrum. Methods

A 737, 148 (2014).

[10] A. Streun and A. Wrulich, Compact low emittance light sources based on longitudinal gradient

bending magnets, Nucl. Instrum. Methods A 770, 98 (2015).

Chapter 2

Basic concepts

The horizontal electron (zero-current) emittance of SLS-2 is predicted to be approximately εxe =

137 pm·rad, and is expected to increase marginally to 150 pm·rad with intrabeam scattering, where

εxe = σx

e′ · σxe, and σx

e′ and σxe are the standard deviations of the electron-beam divergence and

size, respectively1.

The SLS-2 magnet lattice will contain twelve 7-BAs and twelve straights. The length of the

straights depends on the lattice periodicity – in the original period-3 (P-3) concept, there would

be three short, three medium, and three long straights, which can accommodate two canted undula-

tors each. In the newer P-12 concept, all straights would be identically long at 5.3 m. The horizontal

electron-beam emittance would be expressed at the centres of the LGBs as a narrower, more diver-

gent beam (6.5 µm × 21 µrad), and as a broader, but less divergent beam at the undulator centres

(12 µm × 11 µrad).

The vertical and horizontal photon-beam divergences from undulators are both equal to

σx,y′p =

[

γ(mN)1/2]

−1

, (2.1)

where γ is the Lorentz factor (approximately equal to 4700 for the SLS), m is the harmonic number,

and N is the number of undulator periods. A small-period undulator serving a hard x-ray beamline

may contain approximately N = 150 periods (300 poles), in which case σx,yp′ will vary between 17 µrad

for the fundamental (n = 1) to 4.3 µrad at the 16th harmonic, that is, from approximately 50 % larger

1Note the relationship between the full-width at half-maximum, FWHM, and the standard deviation is FWHM =√8 ln 2σ = 2.355 σ

–13–

14 BASIC CONCEPTS

0.01 0.1 1 10 100Photon energy [keV]

10-4

10-3

10-2

10-1

100

f coh

3rd gen.4th gen.

Figure 2.1: The coherent fraction fcoh as a function of photon energy for a 2-m undulator at a third-generation

facility, for which ǫex= 5 nm rad, ǫe

y= 1 pm rad, σe

x= 100 µm, and σe

y= 2 µm; and for a fourth-generation

facility (ǫex= 100 pm rad, ǫe

y= 1 pm rad, σe

x= 10 µm, and σe

y= 2 µm).

than σxe′ to about 40 %, respectively.

In each lateral direction, the total beam sizes and divergences are convolutions of the electron-beam

and diffraction-limit (photon) components, that is

σx,y =[

(σex,y)

2 + (σp)2]1/2

, (2.2)

σ′

x,y =[

(σ′ex,y)

2 + (σ′p)2]1/2

. (2.3)

Consequently, in the horizontal direction, both electron- and photon contributions are significant,

while in the vertical direction, the divergence is entirely dominated by the photons. It is therefore

important that at any one position the phase-space ellipses are approximately matched. Interestingly,

adjustments of the beta function (σ/σ′, also a measure of the aspect ratio of the phase-space ellipses)

in the P-12 period lattice appears to improve the total emittance by up to 65 % compared to P-3,

depending on the position along the ring.

The intrinsic ‘single-electron’ radiation from an axially extended source (undulator) of length L is

defined by the source size σp = (λL)1/2/4π and divergence σ′p = (λ/L)1/2. The diffraction-limited

emittance σp · σ′p = λ/4π, of the order of 10 pm·rad for hard x-rays. When the electron emittance

is equal to the photon emittance, the storage ring is close to being fully diffraction limited. From

Equation 2.5 below, the coherent fraction under these conditions is 25 %. Reducing the horizontal

15

electron emittance much more than this brings only marginal benefits.

The intrinsic photon-beam size σpx,y generated by an undulator can be calculated from the lower,

or diffraction, limits of the divergence and beta function, and is equal to β · σp′x,y. This turns out to be

of the order of a micron.

The coherent fraction of a synchrotron source has contributions from the electron beam and the

natural emittance of the emitted radiation, and is given by

fcoh =

[

1 +

(

σex

σp

)2] [

1 +

(

σ′ex

σ′p

)2] [

1 +

(

σey

σp

)2]

1 +

(

σ′ey

σ′p

)2

−1/2

. (2.4)

=ε2p

εx · εy(2.5)

For SLS-2 in the hard x-ray regime, this is dominated by the first term describing the ratio of the

electron- to photon-beam size in the orbital plane. From the specifications given for SLS-2, a coherent

fraction of the order of 5 % at 1 A is expected, some two orders of magnitude greater than presently

achievable (Figure 2.1).

16 BASIC CONCEPTS

Chapter 3

Sources

3.1 Power density on optical components

A comparison between simulations of the power densities of radiation produced by identical undulators

in SLS and SLS-2 leads to the, at first seemingly surprising, result that the absorption profiles are

essentially identical, despite the fact that the emittance of SLS-2 is some forty times smaller. This is

because the condition for constructive interference for the angular distribution of the off-axis radiation,

given by

mλm(θ) =λu

2γ2

(

1 +K2

2+ γ2θ2

)

(3.1)

does not depend on the emittance of the central cone. Here, λm is the photon wavelength of the

mth harmonic, λu is the undulator period, γ is the Lorentz factor of the storage ring, K is the ratio

of the maximum angular deviation of the electron beam induced by the undulator and the natural

opening angle 1/γ, and θ is the angle of observation of the central axis. An example for undulator

radiation is shown in Figure 3.1. Importantly, this means that, for a given undulator type, thermal

management of beamline optics, in particular the first element after the front end (typically a mirror

or monochromator) at SLS-2 should be very similar to the present conditions.

3.2 Small-period undulators

The reduced horizontal emittance associated with DLSRs provides opportunities beyond tighter fo-

cussing, higher coherent fraction, and/or a larger working distance for a given focal spot size. In

–17–

18 SOURCES

[kW

/mra

d ]

[kW

/mra

d ]

Pow

er d

ensi

ty

Pow

er d

ensi

ty

X’ [mrad] X’ [mrad]

Y’ [mrad]Y’ [mrad]

(a) (b)

Figure 3.1: Comparison of the heat load as a function of divergent angles X ′ and Y ′ for SLS (left) and SLS-2

(right). Note that these are almost identical. K = 1.7; 140 undulator poles.

particular, the reduced electron-beam cross-section allows for smaller magnetic gaps of the sources.

There has been a move in the last decade to smaller-period undulators [1, 2]. From Equation 3.1,

the condition for constructive interference for on-axis undulator radiation is

mλm(θ) =λu

2γ2

(

1 +K2

2

)

. (3.2)

Hence, in order to access higher photon energies, particularly at medium-energy storage rings such as

the SLS, one should minimize λu. Because a given magnetic material will generate a weaker magnetic

field along the undulator’s central axis if the undulator period is reduced, short-period undulators

require particularly high-remanence materials. This can produce technological obstacles that must be

overcome, discussed in more detail in Section 3.3. The lowest practical undulator period size for SLS

has been realized with the U14 undulator at the Materials Science beamline [1, 2].

Shorter periods do, however, mean that more pole pairs can be packed into a given straight section.

The intrinsic on-axis bandwidth of the mth harmonic of an undulator is given by ∆ν/ν = 1/mN ,

where N is the number of undulator periods. In addition, the coherent fraction increases as N2.

The reduced lateral extent of the electron beam in the orbital plane (From σex ∼ 200 µm for SLS to

σex ∼ 10 µm for SLS-2) brings it closer to the size of the lateral excursions induced by the undulator’s

magnet array (∼ 1 µm). This means that, for a given acceptance slit size to the beamline, the off-

axis undulator radiation (visible as lobes in the low-energy flanks of the undulator spectral peaks)

is substantially suppressed, thus resulting in much cleaner spectra (see Figure 3.2), with relative

bandwidths of the order of 5 × 10−3, only a factor of 30 higher than radiation monochromatized by

3.2. SMALL-PERIOD UNDULATORS 19

0 10 20 30 40Photon energy [keV]

1016

1017

1018

1019

1020

1021

1022

Bril

lianc

e [p

h/s/

0.1%

BW

/mm

2 /mra

d2 ]

SLS-2SLS

1

2

3

4

5

1719

Figure 3.2: Comparison of a U12 undulator brilliance spectrum inserted at the SLS and SLS-2, all other

conditions being equal (K = 1.6, 240 poles). Note first that the spectrum maxima of the U12 at SLS-2 are

approximately a factor 40 greater than at SLS; while the overlap with off-axis lobes at the higher harmonics is

largely suppressed at SLS-2, permitting even at 40 keV the use of the entire harmonic bandwidth. The relative

bandwidth of the SLS-2 spectrum ∆ν/ν decreases from approximately 0.007 at the fundamental to 0.004 at

the m = 19th harmonic, less steeply than the theoretical 1/m dependence, due to there still being a nonzero

contribution from off-axis radiation. Note that ‘leakage’ of the off-axis features into the on-axis spectrum is far

more pronounced for the SLS, in the form of distinct lobes on the low-energy flanks of the on-axis maxima, that

follow Equation 3.1.

Si(111). The opportunity thus presents itself, especially for techniques that do not require ultra-high

monochromacity (e.g., x-ray absorption and phase-contrast imaging, including ptychography), to use

the full flux of the undulator harmonic without further monochromatization.

The reduced electron-beam size has a further, related, benefit in that the poles for SLS-2 can be

shaped and made to contain a soft-magnet central block which ‘funnels’ the magnetic field lines to

induce higher fields on axis, but maintain the same overall forces (that is, not exerting more strain on

the support structures, see Figure 3.3).

The use of an entire and unfiltered undulator harmonic still requires the removal of the other

20 SOURCES

Figure 3.3: By encasing a soft magnetic material (light blue) in three high-remanent magnets, their fields can

be funneled and higher flux densities achieved along the central axis. The magnetic forces in total should remain

unchanged, meaning the present support structures could be kept. This adaptation is made possible by the

reduced lateral extent of the electron beam in SLS-2 by approximately a factor of 6 compared to SLS.

harmonics. This can be achieved using a double multilayer monochromator (DMM). The condition

for diffraction nominally means that higher harmonics 2ω, 3ω · · · are also selected in addition to the

fundamental frequency ω. So selection, for example, of the fourth undulator harmonic would also

select for the eighth, twelfth, . . . harmonics. However, these can be essentially entirely suppressed in

DMMs exploiting refraction effects and a careful choice of the multilayer parameters. The transmission

of x-rays across the interface between two materials is associated with small but significant changes

in angles due to refraction, particularly for shallow angles and multilayer periodicities Λ substantially

larger than the x-ray wavelengths. Precisely, the selected wavelengths at different harmonics m are

given for a multilayer as being

λm = 2(Λ/m) sin θ(

1− κΛ2/m2)

, (3.3)

where κ is a material-dependent constant, but always assumes values close to 2 × 10−3 nm−2. Be-

cause Λ is an order of magnitude larger than typical lattice constants of most inorganic crystals,

refraction effects thus become more pronounced; and because the correction term κΛ2/m2 is inversely

proportional to m2, it is the first maximum (m = 1) which is most affected.

Harmonic suppression is thus achieved by using different material combinations for the two mul-

tilayers in the DMM. This results in the most pronounced energy difference for a given incident

angle being at the fundamental (m = 1). By a small adjustment of one multilayer periodicity Λ,

the fundamental can be made to overlap, while subsequent harmonics have differences in their angles

3.3. SUPERCONDUCTING UNDULATORS 21

substantially larger than their bandwidths, and those values deviate even more from integer multiples

of the fundamental. In addition, if the thickness ratio of the ‘reflection’ (high-density) sublayer dr and

the ‘spacer’ (low-density) sublayer ds is chosen so that it can be expressed as a simple integer ratio

h/l, reflections which are integer multiples of (h+ l) will be systematic absences, that is, they will be

suppressed. So, by using one multilayer with h/l = 1 and another for which h/l = 1/2, the second,

third, and fourth harmonics can be effectively suppressed (see Figure 3.4).

3.3 Superconducting undulators

A technological problem arises with the reduction of the undulator periodicity λu to values below

approximately 14 mm. As a rule of thumb, magnetic materials with a remanent field strength Br less

than approximately 1.2 T remain stable. For example, the 1.6 T remanent-field magnets employed at

the U14 undulator of the Materials-Science beamline require cooling to approximately 130 K in order

to remain stable. At a minimum gap of 3.8 mm, the on-axis magnetic-field amplitude B0 = 1.18 T,

for which K = 1.55 [1, 2] (see Figure 3.5).

The modest storage-ring energy of the SLS (and SLS-2) means that large-K undulators are required

if photon energies much above 20 keV are to be accessed. But K = 0.934λu[cm]B0[T] and thus scales

linearly with λu, while B0 drops with λu for a given remanent field. A reduction of λu to values

lower than 14 mm would thus appear to either require remanent fields associated with significant

thermal-stability issues, or gap sizes of the order of 2 mm or less.

This problem may be resolved by using a novel configuration in which superconducting ring elements

are employed [see Figure 3.6(a)]. By completely closing the gap and then opening up again, a current

is induced in the superconducting rings measured in kiloAmperes, leading to an enhancement in the

K-value by approximately a factor of two compared to a ‘normal’ undulator, depending on the gap size

[see Figure 3.6(b)]. Development of such a source is presently being investigated by a team composed

of members of the undulator group at the PSI and collaborators at SPring8.

3.4 Twin bending-magnet sources

In the triple-bend achromats used at SLS-1, the angular separation betwen dipoles is 10o, while that

between the LGBs in SLS-2 is 4.29o.

22 SOURCES

10-4

10-2

100

10-4

10-2

100

Ref

elct

ivity

5000 10000 15000 20000 25000 30000Photon energy [eV]

10-10

10-8

10-6

10-4

10-2

100

ML1: B4C/Ru

dRu/dΛ = 1/3

Λ = 4.83 nm

50 periods

ML2: C/W Λ = 5 nmdW/dΛ = 1/2

50 periods

ML1 x ML2

Figure 3.4: An example of a multilayer monochromator in which one multilayer is fabricated from B4C and

Ru with dr/Λ = 1/3 (dr/ds = 1/2) and the second from carbon and tungsten (dr/Λ = 1/2). Integer multiples

of the fundamental are shown as coloured vertical lines up to the fifth harmonic. At these energies, the highest

reflectivity of the higher harmonics in the bottom curve is approximately 10−7.

Soft x-rays are characterized by having a high absorption coefficient (even for air), while their mirror

reflectivity varies strongly with deflection angle and photon energy. Although soft x-ray beamlines

tend to cover a photon energy range of only one or two keV, the low-energy limit is typically about

an order of magnitude lower than the upper limit. Therefore, the design of soft x-ray beamlines tend

to be constrained strongly by the geometry of the optical path. Strong deflection by the mirrors and

3.4. TWIN BENDING-MAGNET SOURCES 23

Undulator period [mm]K

−va

lue

3 mm gap

4 mm gap

Figure 3.5: Plot of K-values as a function of undulator period for 3- and 4-mm gaps and magnets with a

remanent field of 1.6 Tesla.

K

/K

sup

norm

Gap size [mm]

(a) (b)

Figure 3.6: Superconducting undulators. (a) Schematic figure of the magnet configuration of a proposed

undulator with enhanced K-values. The permanent magnets (blue and green) are poled against one another

and the field lines pass through the centres of the superconducting rings (in yellow). (b) Enhancement in K

compared to the same setup without superconducting rings, as a function of gap size.

monochromator gratings will only efficiently transmit the lower photon energies and hence restrict the

upper energy limit. The use of more glancing reflections results in a significant transmission of higher

orders of the monochromator, spoiling the spectral purity in the low photon-energy range. An obvious

24 SOURCES

0o+4

M: Ni, 175175 (MgF )

178 (Rh) 178 (Al)

o

o

o

o

o

0o+4

M: Ni, 175175 (MgF )

178 (Rh) 178 (Al)

o

o

o

o

o

0o+4

M: Ni, 175175 (MgF )

178 (Rh) 178 (Al)

o

o

o

o

o

Grating

2

500 − 1500 eV

200 − 600 eV

1.5 − 3 keV

2

2

M: Rh, 178

M: Rh, 178

M: Rh, 178o

o

o

Figure 3.7: Configuration of a twin bending-magnet soft x-ray beamline for three photon-energy ranges. Cour-

tesy B. Watts.

solution would be to allow variations in the geometry of the beamline; however many implementations

of this idea have issues with mirror alignment and stability and/or with physically moving the required

vacuum enclosures. Therefore, soft x-ray beamlines normally accept heavy compromises, and are

optimized for specific uses but consequently have to accept poor performance in other areas.

A novel soft x-ray beamline design concept that allows multiples geometries, without introducing

complexity or stability issues is to utilise multiple sources that feed into a single monochromator.

This allows the optical path requirements to be separated and individually optimized, supplying the

endstation with a wider photon energy range, a higher spectral purity and also higher photon flux than

is possible with conventional beamline designs. A concept design is modelled on the PolLux beamline

that uses a bending-magnet source and a horizontally deflecting spherical grating monochromator.

The ‘Twin Source’ beamline uses two bending magnets, each with its own toroidal first mirror and

entrance slit, arranged such that the two beams converge at the surface of the spherical grating and

the diffracted beams then go to the same exit slits and endstation (Figure 3.7). While one upstream

branch uses a first mirror with a small deflection that efficiently transmits photons with energies up

to 3 keV, the other uses a stronger deflection angle and a coating material with a cut-off energy

3.5. SUPERBENDS 25

chosen to increase the spectral purity in the lower part of the photon-energy range. Just as PolLux

provides a choice of two diffraction gratings, optimized for different energy ranges, the Twin Source

beamline would provide three, optimized for the low (200 - 600 eV), middle (500 - 1500 eV), and high

(1.5 - 3 keV) photon-energy ranges. Switching energy ranges therefore would require opening and

closing shutters between sources (both first mirrors would be constantly illuminated) and exchanging

diffraction gratings (by vertical translation). These operations are fast and simple, require minimal

realignment of mirrors, and do not compromise the stability of the optics. Since bending magnets tend

to be under-utilized at the SLS, the Twin Source beamline would provide a significant performance

increase at minimal extra cost.

3.5 Superbends

It is planned to include three longitudinally-graded superbends, which would most probably continue

to serve an MX beamline, a high-energy XAS beamline, and a tomography beamline. The P-12 sym-

metry is only insignificantly affected by their inclusion, and indeed, by the fact that each superbend’s

maximum field strength will be individually tailored to the beamline’s needs – the XAS station needs

only to access approximately 35 to 40 keV, for which a 3-T magnet would suffice; the MX beamline

rarely uses radiation above 19 keV; while the tomography beamline (presently TOMCAT) would like

to access as high energies as possible. For this last beamline, a magnetic-field strength of 5.4 T (up

from 2.9 T at present) is currently being considered. The improvements in flux and accessible energies

are summarized in Figure 3.8.

Lastly, it is mentioned that even higher superbend magnetic fields are possible and could extend

the accessible photon energies to beyond 100 keV. Additionally, any of the ‘normal’ LGBs can be

exchanged for super-LGBs without perturbing significantly the lattice.

3.6 Fast polarization switching

Many spectroscopic experiments that investigate element-specific magnetism or structure, such as,

for example, XMCD photoelectron emission microscopy (PEEM), require switching of polarizations.

This is presently performed by APPLE undulators, where subsets of the magnet array are shifted

longitudinally with respect to the each other, and/or by opening and closing in series two undulators

26 SOURCES

0 20 40 60 80 100Photon energy [keV]

1011

1012

1013

Pho

tons

/s/m

rad2 /0

.1%

BW

2.9 T5.4 T

0 20 40 60 80 100Photon energy [keV]

0

10

20

30

40

Flu

x @

5.4

T/F

lux

@ 2

.9 T

Figure 3.8: Comparison of fluxes for the present 2.9-T superbend at TOMCAT and a proposed 5.4-

longitudinally-graded superbend at SLS-2.

of opposite polarization sense. This typically requires switching times of the order of minutes and

seconds, respectively.

Several new approaches are being considered in order to reduce these switching times, preferably

down to a either a fraction of a second, or to switching rates in excess of a hundred Hz. The former

improvement woould reduce dead time between polarization-dependent experiemnts, while the latter

would implement lock-in methods to improve temporal resolution and data quality. The first could

be implemented by a simple acceleration of the relative shift of the magnet subarray positions in

the APPLE undulator by the use of hydraulics, reducing switching to a few seconds or less. The

significantly faster second approach might be solved by introducing a set of kicker magnets, as shown

in Figure 3.9.

Note that none of these novel switching scenarios depend intimately on a low-emittance ring, but

would nonetheless be a timely technological upgrade that would mesh conveniently with the storage-

ring upgrade.

3.6. FAST POLARIZATION SWITCHING 27

BL

k k k kk1 2 3 4 5

Figure 3.9: In a kicker-switching setup, the polarization of the radiation entering the beamline (BL) could be

rapidly switched by kicking the electron beam with k1, k2, and k3 (radiation from the downstream undulator

selected, shown here), with k3, k4, and k5 (upstream undulator selected).

28 SOURCES

Bibliography

[1] P. R. Willmott et al., The Materials Science beamline upgrade at the Swiss Light Source, J. Synch.

Rad. 20, 667 (2013).

[2] M. Calvi, T. Schmidt, A. Anghel, A. Cervellino, S. J. Leake, P. R. Willmott, and T. Tanaka,

Commissioning results of the U14 cryogenic undulator at the SLS, J. Phys. Conf. Ser. 425, 032017

(2013).

–29–

30 BIBLIOGRAPHY

Chapter 4

Science case

An improved source brilliance through a reduction in horizontal emittance will benefit almost all

synchrotron-based experiments. The improved emittance can be exploited in different manners for

any given method, that is, as a more parallel beam, or as a tighter focus, (low- or high beta-function,

respectively). This is a convenient metric to use to discuss scientific gains likely to emerge as a

consequence of the upgrade to the SLS.

Section 4.1 is concerned with experimental techniques that take advantage of the increased co-

herent illumination lent by a parallel beam, in particular coherent x-ray diffractive imaging (CXDI),

ptychography, scanning SAXS, and x-ray photon-correlation spectroscopy (XPCS).

Section 4.2 covers diffraction techniques, which utilize the improved emittance in a variety of ways

that are less easy to generalize; while full-field tomography is discussed in Section 4.3. Spectroscopies

and spectroscopic imaging are covered in Section 4.4; although scanning microspectroscopies clearly

invest the improved emittance by producing micron- and nanometer sized focal spots, other methods

too will pursue this approach in operando, extreme-conditions, domain-mapping, and other investiga-

tions.

4.1 Lensless imaging

With the exception of scanning SAXS, all of the techniques described in this section depend on the

coherence to encode the necessary information to image objects – the action of an x-ray lens with

sufficient magnifying power is performed by phasing algorithms that depend on the illumination of

–31–

32 SCIENCE CASE

the entire object being coherent1.

4.1.1 CXDI and ptychography

Lensless-imaging techniques, in particular CXDI and ptychography [1,2], bridge the spatial-resolution

gap between crystal diffraction and electron microscopies on the Angstrom and sub-Angstrom scale,

and nanoprobe methods such as scanning x-ray transmission microscopy (STXM), photoelectron emis-

sion microscopy (PEEM), and tomography on the tens of nanometers to micron scale. STXM relies

on tight focussing down to as low as approximately 10 nm (which thus determines the spatial res-

olution), though it is more typically a few tens of nanometres; while CXDI and ptychography have

linear illumination spot sizes often two orders of magnitude larger. Their present ultimate resolutions

are, however, of the order of a few nanometers – this is provided by the information encoded in the

interference between the scattered radiation (‘speckle’) [3]. The use of DLSR-radiation will most likely

provide the technological breakthrough to facilitate this and other developments.

The speed of data acquisition and/or the maximum resolution which can be achieved in lensless

imaging depends intimately on the coherent flux Fcoh provided by the beamline. In addition to the

increased coherence associated with the reduced horizontal electron emittance, the following relatively

straightforward improvements could be implemented to enhance Fcoh.

• Smaller period undulators with relatively large K-values. An increase in K increases the flux

of higher harmonics at high photon energies. In addition, for a given straight-section length,

more undulator periods N can be installed. The coherent fraction increases with the square of

N . In this manner, it is estimated that one could increase Fcoh by an order of magnitude. A

caveat to this is that in order to resolve speckle, the ratio of the desired spatial resolution to

the field of view (i.e., illumination spot) should be greater than the relative bandwidth. Using

a bandwidth of, say, 5 × 10−3 for an experiment demanding a resolution of 10 nm would thus

limit the illumination spot to 2 µm or smaller.

• Use of the entire undulator harmonic bandwidth (see Section 3.2) would increase the flux by

approximately a factor of 30. The transverse coherence length is unaffected by an increase in

the photon beam’s spectral bandwidth.

1Note that coherence does not imply that the radiation be a perfect plane wave. The exact description of the wavefront

can be determined in a nested iterative loop within the main algorithm used to regain the phases of the scattered signal.

4.1. LENSLESS IMAGING 33

• Presently, focussing down to approximately one or a few microns requires the use of Fresnel zone

plates (FZPs). The efficiency of FZPs in the hard x-ray regime is at best approximately 25 %,

and more typically 10 %. The option to use novel, in-house-developed kinoform lenses, KB-,

or other achromatic reflection optics presents itself, again potentially increasing the flux on the

sample by a further order of magnitude.

Together, these technological improvements associated with the storage ring (×40), the undulator

(×10), the monochromator (×30), and the focussing optics (×10) leads to a total potential increase

in coherent flux on the sample by as much as a factor of 100′000 or more!

Moving down the technological chain, such a huge gain in available flux demands that this is

matched by the detector performance and sample manipulation. The intensity of scattered signal in

the so-called Porod regime drops off with the fourth power of the scattering vector Q, hence, for a

given lower limit to the detector sensitivity, an increase in flux by 105 corresponds to an increase in

the Q-range by over a factor of 15, bringing the accessible resolution in theory down to the Angstrom

regime. This, however, would require sample scanning and rotation with accuracies and circles of

confusion of the same magnitude, and may also not be dose efficient compared to electron microscopy.

The increased flux can be more profitably invested elsewhere. Presently, the OMNY stage developed

at the PSI represents the cutting-edge technology, at a little over 10 nm [4], over one order of magnitude

larger than would be required for sub-nm experiments. In addition, the unfiltered speckle patterns

would require an enhanced dynamic range and photon count rate presently unobtainable using state-

of-the-art detectors. Although this problem can be resolved by attenuating the central region of the

detector with an appropriately designed absorbing disc [5] (a somewhat ‘subtractive’ solution), the

use of a more divergent beam would use a larger fraction of the available photon flux. In any case, the

availability of so much additional flux should trigger the development and commissioning of suitable

integrating detectors, such as the Jungfrau.

With this in mind, a more relevant metric than ultimate resolution2 is instead how many voxel

elements there are within the reconstruction of the extended object under investigation (see Figure 4.1

[6]) and how rapidly these can be recorded3, particularly for the case of the scanning technique of

2For example, most biological samples require a resolution of 10 nm or greater. Biology is known for its hierarchical

ordering, from macromolecular structures on the tens of nanometer scale, to intracellular structures on the micron scale,

to macroscopic biological features that may be of the order of millimeters or larger.3An estimate provided by the cSAXS beamline of a target acquisition rate using a DLSR is of the order of one

34 SCIENCE CASE

20 mµ

Figure 4.1: The ‘Eiger selfie’. Detail of a ptychographic reconstruction of a region of the electronics board of

an Eiger detector, recorded with an Eiger detector with 6.2-keV radiation. The resolution of the reconstruction

was 41 nm. Adapted from [6].

ptychography.

Samples investigated using ptychography may extend over several hundred microns or more in linear

extent, hence even a resolution of 100 nm would in such cases equate to some 109 voxels. In many

cases, therefore, the enhanced flux would be far better invested in rapid data-acquisition rates (∼ kHz

or higher) and scanning velocities (∼ mm/s), than in an attempt to improve the spatial resolution.

This would result in an accelerated rate of resolved spatial elements per second, which will thereby

set their own stringent specifications on the sample stage and detector. Perhaps the most attractive

feature of ptychography is its ability to cover a large range of length scales, which makes it ideal for

investigating hierarchical structures with nested length scales, not least in complex heterogeneous and

biological samples (Figure 4.2).

A serious obstacle, especially for biological samples, to obtaining reproducible and meaningful data

teravoxel per day. This would equate to a three-dimensional object with linear dimensions of 100 µm being resolved

down to 10 nm.

4.1. LENSLESS IMAGING 35

Figure 4.2: Top left: overview of different components making up hydrated cement. The field of view is

110 µm. Top right: Image of calcium-silicate hydrate (C-S-H) particulates. Note that although they have

extents measured in a few microns, their boundaries consist of dendritic structures with widths some two to

three orders of magnitude smaller. Bottom: the hierarchical nature of structures within cortical bone. From [7]

and [8].

36 SCIENCE CASE

5 10 15 20 25 30Energy [keV]

10-10

10-9

10-8

10-7

10-6

10-5

δ, β

0

0.2

0.4

0.6

0.8

1

Tra

nsm

issi

on

δ

β

Tr

flesh chitin bone cement basalt Ti-alloy10 µm

100 µm

1 mm

1 cm

10 cm

1 m

Atte

nuat

ion

leng

th

100 keV502510

Figure 4.3: Left: the attenuation lengths of some scientifically and technologically interesting materials at

different photon energies. Right: the transmission curve and values for the refractive-index decrement δ and

absorption index β for flesh. The chemical formula for flesh was taken to be C12H60O25N5, with a density of

1.07 g cm−3.

is radiation damage; this is no less true for lensless imaging. An interesting avenue to explore would

thus be to move to higher photon energies – for example, the transmission of 1 mm of flesh rises from

12 % at 6.2 keV (2 A) to 80 % at 12.4 keV (1 A), see Figure 4.3. The increased coherent flux of

DLSRs would more than compensate for the reduced coherence volume at higher photon energies (it

scales as the inverse third power).

Lastly, data transfer rates will need to be of the order of 10 GByte/s or higher to accommodate

kHz frame rates and a sufficient dynamic range.

The setup for ptychography is very similar to that for scanning transmission x-ray microscopy

(STXM), the primary difference being that in STXM, the focal spot is normally made to be as tight as

possible, and that no attempt is made to spatially resolve the transmitted signal with an area detector.

The PolLux beamline (bending magnet, 250 – 1600 eV) can routinely produce spatial resolutions of

the order of 40 nm (using FZPs) and has even achieved values of the order of 10 nm under exceptional

circumstances. A natural development associated with the upgrade of the SLS would be to provide

soft x-ray ptychography (SXP) in addition to STXM [9]. This would require an SXP endstation to

operate at an undulator source.

SXP can provide spatially resolved chemical component distributions with nanometer accuracy,

4.1. LENSLESS IMAGING 37

LCP RCP

(b)

Difference

(a)

Figure 4.4: (a) Chemical-contrast soft x-ray ptychographic mapping of LiFePO4 nanoparticles, highlighting

left, the lithiated, and right, the delithiated regions. The long axis of the particle is approximately 900 nm.

From [9]. (b) Magnetic contrast of nanoparticles found in bacteria through XMCD-SXP. From [10].

by exploiting the full complex refractive index (n = 1 − δ + iβ) and operating in the pre-edge and

XANES regions of the element(s) of interest. Because the refractive-index decrement δ (or alternatively

f1 = f0 + f ′) changes significantly even before the absorption edge, this can be exploited to provide

phase contrast, while additionally being dose efficient. Absorption contrast is provided exactly at the

absorption maximum. This has been demonstrated and compared favourably to ‘standard’ STXM

results for LiFePO4 nanoparticles, around the L3-edge of iron [9]. By exploiting differences in the

XANES spectra between the original compound and delithiated FePO4, detailed chemical mapping

highlighting delithiated regions can be produced [Figure 4.4(a)]. In a further study, nanosized magnetic

grains in magnetotactic bacteria of approximately 30 nm diameter could be highlighted through x-ray

magnetic circular dichroic (XMCD) contrast at the same L3 iron edge at 707 eV [10] [Figure 4.4(b)].

It seems very probable that the resolution of SXP may be pushed with relative ease using DLSRs

to the few-nm scale, making such techniques highly competitive with conventional full-field XMCD,

38 SCIENCE CASE

Figure 4.5: Scanning SAXS of a human trabecular bone. (a) Computer tomogram of the bone using standard

filtered backprojection. (b) 3D tensor image of the orientation of the collagen fibrils within the bone ultra-

structure. (c) One tomographic slice taken from (b). The colour scale indicates the ratio of anisotropic to total

scattering, while the orientation of the rods indicates the main direction of alignment. (d) and (e) are examples

of high and low degrees of orientation, respectively. The scale bars in (a) and (b) equal 0.5 mm. From [11].

especially for static or only slowly changing systems.

Lastly, the diffraction-limited photon wavelength λDL of SLS-2 is predicted to be approximately

2 nm (see Table 1), corresponding to 600 eV. The coherent fraction of an SXP station would thus be

able to exploit a close to 100 % coherent beam.

4.1.2 Scanning SAXS

Scanning SAXS is the partially coherent sister technique of ptychography, in which general features,

shapes, anisotropies, and length scales are investigated, rather than recording a detailed and contiguous

extended image (see Figure 4.5).

Although the total flux generated at a beamline is not expected to increase for SLS-2, scanning

SAXS will benefit from the improved emittance, whereby both small spot sizes (spatial resolution) and

a low-divergence beam are used, thereby offering high resolution in both real- and reciprocal space.

4.1. LENSLESS IMAGING 39

t + N t∆

t + t∆t + 2 t∆

t + 3 t∆

tdynamicsample

Figure 4.6: X-ray photon correlation spectroscopy probes dynamics of systems by observing the temporal

changes in the speckle pattern they produce.

In addition, the implementation of the complementary improvements in source- and optics technology

described in the list in the previous section means that faster acquisition rates will be possible, by

as much as three orders of magnitude. This will significantly strengthen the emerging technique of

scanning SAXS for the study of hierarchically structured materials in both the life- and materials

sciences.

4.1.3 X-ray photon correlation spectroscopy

XPCS is presently a promising probe of nanoscale fluctuations up to frequencies in the kHz range [12]

based on intensity fluctuations in the far-field speckle pattern of the system under investigation (see

Figure 4.6). Depending on the sample system under investigation, XPCS is sensitive to structural

fluctuations ranging between several hundred nanometers and a few Angstroms, while the largest

system that can be probed is given by the transverse coherence length, of the order of a few tens of

micron in the vertical direction, and an order of magnitude smaller in the horizontal direction.

A serious current limitation of XPCS is its modest temporal resolution. XPCS measures the

correlation between two photons rather than interference of a photon with itself. This means that, in

contrast to ‘normal’ diffraction phenomena, the signal-to-noise ratio scales with intensity rather than

the scattering amplitude. The temporal resolution can be pushed towards the microsecond regime if

point detectors or avalanche photodiode arrays such as the AGIPD detector [13] are employed rather

40 SCIENCE CASE

than standard 2D detectors, which have readout times presently limited to the kHz range. Extension

of this to tens or even hundreds of MHz would allow experiments to be performed on many systems

with characteristic time scales that are presently inaccessible, such as magnetic spin reorientation

(109 Hz), ferroelectric-domain switching (∼ 108 Hz), or structural phase transitions (105 − 109 Hz).

The signal-to-noise ratio S/N for the autocorrelation function between two speckle patterns sepa-

rated by a time ∆t is given by

S/N = Fcoh (T ∆t nxny)1/2 , (4.1)

where T is the total accumulated recording time, and nx×ny is the number of pixels in the detector that

record ‘usable’ signal. So, for a given acceptable lower limit to S/N , anX times increase in the coherent

flux would permit an X2 reduction in the minimum time separation ∆t. For a predicted 40 times

increase in Fcoh for SLS-2, this therefore corresponds to an improvement in temporal resolution by

over three orders of magnitude. A probable bottleneck to fully exploiting this substantial improvement

will be detector technology, especially for area detectors. Another important aspect to consider is the

fact that synchrotron radiation has a time structure, with pulses arriving at 500 MHz, and pulse

lengths expected to be approximately 200 ps for SLS-2. As soon as one begins to probe processes on

time scales of the order of a train of a few pulses (of the order of 10 ns), this time structure must

be considered. It has been suggested that one could perform XPCS within the duration of a single

x-ray pulse [12], though the detector technologies required to achieve this (i.e., with readout times

significantly shorter than the pulse width, of the order of 10 ps) means that such experiments lie much

further into the future.

PRovided that these technological challenges are successfully addressed, radiation-damage issues

are likely to remain. For example, in the case of XPCS performed in the forward-scattering (SAXS)

geometry, the scattered intensity drops as the fourth power of the scattering vector Q. Studies of fast

fluctuations and at short length scales (large Q) are inherently impaired by the low intensity, requiring

often prohibitively intense incident beam intensities to compensate for this. The applicability of XPCS

is likely therefore to be limited to radiation-tolerant systems and either slow processes or phenomena

that include intensity enhancements (above the Q−4-dependence of ‘standard’ XPCS), such as the

use of Bragg peaks or crystal-truncation rods, further limiting the scope of sample types that lend

themselves to this method.

4.2. DIFFRACTION TECHNIQUES 41

4.2 Diffraction techniques

4.2.1 Macromolecular crystallography

Third generation synchrotron radiation facilities have revolutionized modern macromolecular crystal-

lography (MX). Atomic resolution structures of biological molecules and their complexes have been

determined at an unprecedented speed with more than 100′000 structures now deposited in the pro-

tein data bank (www.pdb.org). However, one important class of protein – membrane proteins –

which accounts for one third of all proteins and two thirds of medicinal drug targets, is extremely

under-represented (1 - 2%) [14]. This is in a large part due to the hydrophobic nature of membrane

proteins, which makes crystallization difficult and often limits the crystal size to the micron-scale.

Small crystals require micron or submicron beam focussing, which, until the advent of DLSRs, meant

divergences could be 0.1o in the horizontal plane, or even larger, which weakens the diffraction signal

at high resolution. MX at DLSRs will therefore be a technique that individually tailors case-for-case

the division of the improved emittance between divergence and source size. From a given diffraction

volume, only a limited amount the diffraction signal can be obtained before radiation damage sets in.

For microcrystals, which would individually be insufficient for a complete data set, diffraction data

from multiple crystals must be merged for structure determination [15]. Prompted by successes in

serial femtosecond crystallography (SFX) at XFELs [16], there has been a thrust in the last five years

towards similar approaches using synchrotron radiation in a technique coined serial millisecond crys-

tallography (SMX), or synchrotron serial crystallography (SSX) [17–21]. SSX can be carried out at

both room-temperature and cryogenic temperature, requiring novel techniques in sample preparation,

delivery, and data collection and processing.

Recently, room-temperature crystallography is experiencing a renaissance through SSX. Despite

the one to two orders of magnitude reduction in the highest tolerable dose compared to cryocooled

MX, RT-SSX offers several advantages, including sampling conformational landscape [22], bypassing

the need for cryoprotectants, and the possibility of investigating dynamic processes down to the

millisecond or even smaller time scales (Figure 4.7). In addition, and very importantly with regards to

the tight but parallel-beam focusing attainable with DLSR-radiation, it has been demonstrated that,

in contrast to cryo-MX [23], there is a positive correlation between dose rate and maximum tolerable

integrated dose in room-temperature MX, which enables measuring more useful diffraction within a

limited tolerable dose (0.1 - 0.5 MGy), the latter increasing by approximately a factor of six when

42 SCIENCE CASE

Figure 4.7: Comparison of bacteriorhodopsin solved by SSX and conventional cryocrystallography. The protein

backbone of the T-SMX structure (purple) superimposes well with the cryostructure (blue). Bottom left and

right: Subtle but clear differences can however be seen in the residues of the retinal omit maps. From [18].

increasing the dose rates from 0.5 to 5 MGy/s [24,25].

The well-established technique of cryo-crystallography at third-generation synchrotron beamlines

has been extended to SSX. In addition to two orders of magnitude extended x-ray dose (∼ 20 MGy),

cryogenic methods have unsurpassed advantages in sample preparation, storage, and transportation,

which allows one to preserve crystals in their optimal state prior to beamtime. The advantages of

SSX at cryo-temperatures has been demonstrated with the structure determination of both soluble

and membrane proteins with crystals as small as a few microns in size (Figure 4.8) [20, 26]. The

use of small micron-sized crystals furnishes one further benefit: it appears that radiation damage

per unit volume is further reduced compared to larger crystals, as the photoelectrons produced by

the initial absorption process and the subsequent secondary electrons can escape with a significant

probability [27,28]. The threshold size below which this begins to bestow an experimentally significant

advantage appears to be a few microns [29].

4.2. DIFFRACTION TECHNIQUES 43

Figure 4.8: In meso and in situ serial crystallography at cryogenic temperature (IMISXcryo). Top left: IMISX-

cryo chip mounted on a goniometer with the crystal-laden mesophase bolus positioned in the cryostream. Top

right: Images of crystals in IMISXcryo wells. Bottom: Structures of four membrane proteins solved using the

IMISXcryo method. From [26].

In SSX, multiple micron- or submicron-sized crystals are exposed to synchrotron radiation before

being destroyed through radiation damage. The crystals are delivered to the beam via one of several

methods: by a fast-moving liquid jet; a slow-moving lipidic cubic-phase (LCP) ‘toothpaste’ extrusion;

mounted on a membrane; or as a collection of crystals in a crystallization well specifically designed to

produce a low x-ray background for in situ measurements. Fresh material can then be delivered at a

controlled manner adjusted to administer the maximum tolerable dose for each crystal. For example,

even in the first (far from optimized) proof-of-principle measurements using LCP streams, just 300 µg

of lysozyme were required, far less than needed in SFX [17].

44 SCIENCE CASE

Lysozyme, of course, is available as millimeter-sized crystals; one does not need to resort to SSX

for this ‘drosophila fly’ of the MX-world. There are, however, many unexplored protein types, in

particular membrane proteins and their subclass, G-protein-coupled receptors (GPCRs), that in gen-

eral crystallize only up to the micron-scale, due to their hydrophobic nature (although they do grow

more readily in the near-native LCP medium [30]). SSX therefore lends itself ideally to investigations

of these biomolecules (Figure 4.8) [26], and will function far more efficiently with the micron- and

submicron-sized beams promised by SLS-2.

Since many membrane proteins are expected to be novel, experimental phasing is required to reveal

their structures. The recent advances in native-SAD phasing have led to great advances in de novo

phase determination [31]. The development of DLSRs provides a timely opportunity to optimize

existing MX beamlines for native-SAD experiments, which require tender X-rays down to 3 keV and

sample environments with minimum background scattering and absorption [32].

Even higher photon intensities are possible by increasing the relative bandwidth of the incident

radiation. Presently, this is typically ∆ν/ν = 1.4× 10−4 for Si(111) monochromatized radiation. The

relationship between this and the angular spread ∆θ (assuming perfectly parallel radiation and zero

mosaicity) is

∆ν

ν=

∆θ

tan θ. (4.2)

So, for example, at a value of 2θ = 60o, bandwidth-smearing of the Bragg peak from the Darwin

width of Si(111) would amount to approximately 2∆θ = 0.01o. Except for the very largest-unit-cell

samples, an increase in the energy bandwidth and consequent increase in flux by a factor of 10 could

therefore be easily tolerated. Combined with detectors capable of recording at hundreds of Hz, SSX

with MX beamlines at DLSRs will not only enable biologists to study structure and function of largely

unexplored protein families, but also pave the way for high-throughput, structure-based drug discovery

of membrane proteins.

4.2.2 Microdiffraction, diffraction tomographies, and total scattering

Spatial mapping of crystalline phases is a powerful method in industrial applications, such as in studies

of cracking, corrosion, and other failure mechanisms in metals and alloys. In order to understand and

follow the evolution of cracks and other failure mechanisms, one needs to record the 3-dimensional

4.2. DIFFRACTION TECHNIQUES 45

300 mµ

Figure 4.9: (Left) Diffraction-contrast tomography and (right) PDF-contrast tomography will provide a novel

tool for investigations of industrial and catalytic systems. From [36] and [37].

grain-boundary properties throughout the sample. Diffraction-contrast tomography [DCT, see Fig-

ure 4.9(a)] is capable of providing this information, and will benefit directly from the small focal-spot

sizes offered by DLSRs [33]. DCT functions by a narrow micron-sized x-ray beam illuminating a line

of polycrystalline material. Each volume unit (voxel) along that line produces its own powder pattern,

and a detector records the incoherent sum of these patterns. The sample is then rotated to produce

a ‘diffraction-contrast’ sinogram, and the procedure repeated for different sample heights. Note also

that complementary information can be recorded in parallel in the conceptually similar technique of

fluorescence tomography [34,35], with which trace-element analysis and oxidation states can be probed

in three dimensions (see Section 4.4).

In addition, the enhanced fluxes at higher energies provided by short-period undulators (see Sec-

tion 3.2) will mean thicker and/or higher-density samples that are more relevant to industrial samples

can be investigated.

In many catalytic processes, catalyst nanoparticles such as Pt, CeO2, or Ni, are deposited on

macroscopically sized porous pellet supports. The efficiency of a catalytic reactor depends intimately

on the efficiency such structures, which in turn depend on the nanoparticle size distribution, shape,

and catalyst–support interactions, to name just three parameters. Because of the poor crystallinity

46 SCIENCE CASE

and intrinsic small size of the catalytic nanoparticles, their structural study requires techniques which

access large scattering vectors, such as the pair-distribution function (PDF), which is complementary

to the spectroscopic method of EXAFS. As in the case of DCT, the photon-hungry technique of PDF-

contrast tomography [PDF-CT, see Figure 4.9(b)] will become a viable technique once the hundredfold

or so increase in brilliance associated with DLSRs and short-period undulator radiation becomes

available. In addition, PDF patterns generally have very broad features, which do not require the high

monochromacity of Si(111) monochromators, so they could profit from the larger relative bandwidth

of a full undulator harmonic (see Section 3.2).

4.2.3 Diffraction studies under extreme conditions

The powder-diffraction station of the SLS already carries out high-pressure experiments up to ap-

proximately 10 GPa (10′000 atmospheres). Dedicated beamlines at, for example, the ESRF and APS

are now approaching tera-Pascal pressures, some hundred times larger [38]. Associated with this are

technological obstacles, such as the difficulty in maintaining hydrostatic conditions (through the use of

helium as the pressure-transfer medium) and the usually very small volumes that can be probed. It is

therefore a moot point as to whether the Paul Scherrer Institute should attempt to compete in terms

of the highest pressures that can be offered. Instead, a combination of the use of an area detector

(such as the Pilatus 6M) and microstrip 1-D detector (such as the 120-degree Mythen detector) for

studies at more modest pressures up to a maximum of a few tens of GPa may be a more profitable

route to follow. The Mythen detector should allow very precise equations of state to be determined, as

its high resolution guarantees very accurate unit-cell determinations. The area detector would allow

one to follow phase transitions quantitatively, and obtain full structural determination through precise

intensity measurements. More accurate studies could also reveal subtle changes associated with the

bonding and other valence electrons (see Figure 4.10) [39].

For crystallites with dimensions of the same order of magnitude as the beam size, of the order of a

few microns, single-crystal studies also become possible, especially for DAC cells with large exit-angle

apertures and for the increased flux of high photon-energies above approximately 25 keV produced by

small-period undulators.

4.3. FULL-FIELD TOMOGRAPHY 47

Figure 4.10: The electron-density maps of doubly bridged annulene at different hydrostatic pressures. PDFT =

periodic DFT calculations; XCWFN = x-ray constrained wave functions computed at Hartree-Fock level, con-

strained with the experimentally measured diffraction intensities; MM = multipolar expansion, with coefficients

refined against the experimentally determined intensities. Note the induced asymmetry of the electron-density

distribution of the perimeter bonds, and consequent enhanced reactivity at 7.7 GPa. From [39].

4.3 Full-field tomography

X-ray computed tomography (XCT) on the micron scale has burgeoned in the last decade and a half,

not least in the fields of biological, palaeontological, and industrial imaging. Phase-contrast XCT has

proved to be especially powerful for low-dose experiments of low-density biological samples in the hard

x-ray regime.

Approximately 50 % of the samples studied to date at the TOMCAT (TOmographic Microscopy

and Coherent rAdiology experimenTs) beamline have had linear dimensions of 1 mm or smaller. This

means that for such investigations, over 99 % of the available flux from the broad fan generated by the

superbend presently installed at TOMCAT (∼ 40× 5 mm2 at 20 m) goes missing for a field of view,

for example, of 1 mm2. Migration to a short-period undulator (in particular the superconducting

undulator detailed in Section 3.3, see also Figure 4.12) would therefore be highly beneficial, including

48 SCIENCE CASE

Figure 4.11: Cutaway visualization of the thorax of a living blowfly, showing the five steering muscles (green to

blue) and power muscles (yellow to red). The data were recorded in a stroboscopic mode to achieve an effective

full tomographic frame rate of approximately 1500 Hz, or 10 times the wing beat frequency. From [40].

0 10 20 30 40 50Peak harmonic energy [keV]

1012

1013

1014

1015

Flu

x in

1 m

m2 @

30

m [p

h/s/

0.15

BW

]

U14, SLS-1U14, SLS-2U12, SLS-2

Figure 4.12: Comparison of the flux incident on a central 1×1 mm2 area at a distance of 30 m from the source,

for the present U14 undulator at the Materials Science beamline; for the same undulator operating at SLS-2;

and for a U12 undulator at SLS-2. Courtesy Marco Calvi, Paul Scherrer Institut.

4.3. FULL-FIELD TOMOGRAPHY 49

exposure time = 150 ms

U14, Si(111) SB 2.9 T, Ru/C ML SB 2.9T, Si(111)

exposure time = 100 msmean: 27631, SD: 2663mean: 27044, SD: 2468

exposure time = 2000 msmean: 4171, SD: 444

Figure 4.13: Comparison of radiographs of a carbon-fibre composite material recorded using Si(111)-

monochromatized undulator radiation from the U14 at the Materials Science beamline, SLS (left); Ru/C

multilayer monochromatized (2 % BW) superbend radiation from TOMCAT (middle); and the same super-

bend radiation but again monochromatized using Si(111). The photon energy was 17.1 keV. The sample was

35 m from the undulator source, and 25 m from the superbend source. From these figures it is apparent that

a combination of undulator radiation and a multilayer monochromator would provide an approximately 200

times increase in areal flux density compared to a 2.9-T-superbend and the same multilayer monochromator.

Courtesy C. Schleputz.

access to higher photon energies; an enhanced coherent flux through its square dependence on the

number of undulator periods; and a multilayer monochromator bandwidth of the order of ∆ν/ν =

1/100 (see Figure 4.13). Such a beamline (for a given monochromator design) would provide several

orders of magnitude more flux on the sample and further drive the world-leading efforts pioneered by

the TOMCAT staff in time-resolved (fast) tomography ( [40], see Figure 4.11).

A two-lens beam-expander could be employed to increase the usable field of view from an undulator

source, with the caveat that the flux density will decrease accordingly. This might allow samples with

linear dimensions up to a centimeter to be studied at an undulator beamline.

Nonetheless, experiments on substantially larger samples would require a superbend beamline. It

is planned to increase the on-axis field from 2.9 T to 5.4 T, thereby shifting the critical energy from

11.1 keV to 20.6 keV, and providing a fivefold increase in flux at 50 keV compared to the present

setup (see Section 3.5).

The transverse coherence function of SLS-2 will be significantly more isotropic, which will facilitate

50 SCIENCE CASE

homogeneous edge enhancement in phase-contrast tomography, and should thereby enable more so-

phisticated reconstruction algorithms which go beyond a linear approximation, which presently only

makes use of the first edge interference fringe. In general, the larger coherent fraction will increase

the contrast in phase-contrast and thus increase the information content that can be extracted per

incident photon. SLS-2 is therefore expected to strongly benefit phase-contrast imaging in all the

variants offered at TOMCAT, including Talbot interferometry and Zernike techniques.

Lastly, the impact of a higher coherent fraction can be exploited to reduce sample doses. The photon

energy associated with a given desired coherence is shifted to higher values for DLSRs compared to

third-generation machines by a factor close to the improvement in emittance (see Figure 2.1 and

Equation 2.5). This will be offset to a degree by the reduced sensitivity of the scintillator and/or

detector at these higher energies, but nonetheless, it is anticipated that there will be a substantial net

gain.

4.4 Spectroscopies

Scanning microspectroscopies, such as x-ray absorption (including XANES and EXAFS), x-ray fluores-

cence (XRF), and other chemical-imaging techniques; and resonant inelastic x-ray scattering (RIXS),

will all profit from the enhanced spatial resolution provided by the tighter focal spot sizes of DLSRs.

This will in turn drive improvements in the performance of optical elements, in particular the slope er-

rors in K-B systems, and the fabrication technologies associated with diffractive (FZP) optics [42,43].

An important ‘flip-side’ to this coin is the potential substantial increase in working distances — for

the same focal spot size, the distance from the focussing optics to the sample can be increased in

DLSRs by the same factor as the improvement in brilliance, that is, up to two orders of magnitude.

This allows for much more sophisticated sample environments (Figure 4.14) in techniques such as

high-pressure photoelectron spectroscopy (HiPPES), also called ambient-pressure photoelectron spec-

troscopy (APPES) [41] and other in operando methods.

4.4.1 RIXS

Resonant inelastic x-ray scattering (RIXS) is a rapidly developing experimental method in which

photons resonant with electronic transitions are inelastically scattered from matter [44]. RIXS has

many analogies with Raman spectroscopy in the visible and infrared regime.

4.4. SPECTROSCOPIES 51

Binding energy [eV]

Figure 4.14: Left: APPES instrument at MAX IV. Right: O 1s spectra on Pt(111) investigating CO2 reduction

at 0.15 mbar O2 pressure as a function of Pt(111) temperature. From [41].

The energy difference between the inelastically scattered photon and the incident photon can range

between meV and a few eV. This energy loss, plus the momentum transfer and change in polariza-

tion between the incoming and outgoing photon can be directly associated with the production of an

electron-hole pair, induced by one or more important processes related to electronic and supercon-

ducting properties such as phonon, magnon, or low-energy electronic excitations.

The energy resolution of a RIXS experiment is defined by the spot size on the sample in the

dispersive direction of an analysis grating spectrometer. The much improved focal spot size of DLSRs

can improve the resolution in principle by this amount [45], although this is likely to be limited through

the slope errors and aberrations of the mirror and grating optics. It is planned to ameliorate this at

52 SCIENCE CASE

Spherical grating

Woltermirror

V−RMH−RMSample

Monochromatorfocal plane

Figure 4.15: Left: Parallel acquisition of the incident and scattered radiation in ‘hν2’-RIXS. The height of the

imaged dispersed beam on the sample is of the order of 100 µm, while the width may be less than a micron.

From [46]. Right: Overview of XAS and RIXS measurements of a 30-nm NdNiO3 film grown on NdGaO3. (a)

XAS of the insulating (blue) and metallic (red) phases. (b) and (c) The corresponding RIXS spectra. Note the

gap in the RIXS signal of the insulating state. From [47].

SLS-2 by extending the beamline by over a factor of two to approximately 100 m, and increasing the

demagnification. In this manner, it is planned to obtain 10-meV resolution with the option of domain

scanning with submicron-sized beams on the sample. This would open up the possibility of following,

among other things, in operando magnetic or ferroelectric domain switching (see Figure 4.15).

Another interesting development in RIXS is to disperse the incident radiation hνin from the grating

monochromator and image this onto the sample (see Figure 4.15). In the direction perpendicular to

4.4. SPECTROSCOPIES 53

the dispersed beam, the focus remains tight, thereby preserving the resolution. By then dispersing the

scattered radiation hνout with a second grating, the image formed on an area detector allows the par-

allel detection of hνin and hνout [46,48], enabling efficient high-resolution, time-resolved experiments.

4.4.2 Photoelectron spectroscopies

Angle-resolved photoelectron spectroscopy (ARPES) is the pre-eminent tool for studying the electronic

structure of crystalline solids. In the last two decades or more, ARPES has been employed to study

correlation effects, such as between orbital order, charge, and spin, in systems as diverse as the

colossal magnetoresistive manganites and the high-temperature cuprate superconductors. Potential

benefits in the long-term include new multifunctional devices and the physics of surfaces and interfaces,

spintronics, unconventional/high-Tc superconductors, and even quantum computing.

A nanofocus ARPES station at the Swiss Light Source would complement the present and soon-to-

be ugraded setup by this push to higher spatial resolution. Presently, experiments at the SIS beamline

of the SLS are limited to objects which are single-crystal and single domain on the ten to 100-micron

scale, and preferably have a well-defined cleaving plane.

A 100-nm focus would allow access to many novel systems and experimental types that are im-

possible now, including the study of phase separation in doped Mott insulators or in magnetism;

polycrystalline samples (containing crystallites with typical dimensions larger than the focus, i.e., of

the order of a micron or larger);the impact of strain and its mapping on high-temperature supercon-

ductors, topological insulators [49] or even tribological materials [50]; in operando studies of device

structures [51]; or the effect of vortex motion on the spectral lineshapes in cuprate superconductors

such as BISCCO [52]. In addition, the perennial problem of preparing a clean surface by crystal cleav-

ing could be circumvented simply by selecting a ‘good region’ of a randomly broken sample surface.

Lastly, a common feature of, for example, ferromagnetic or ferroelectric materials, is the spontaneous

emergence of domains on the micron and submicron scale. It is difficult to predict in such com-

plex systems how these domains form from a bottom-up approach, and hence microspectroscopic or

nanospectroscopic ARPES (µ-ARPES and nano-ARPES, respectively) would be invaluable tools in

the study of such systems and their theoretical understanding.

For the photon energies typically used in ARPES, the emittance is entirely diffraction limited (see

Table 1), at approximately 1.9 nm·rad. As the source size at the undulators is expected to be of the

54 SCIENCE CASE

CHA

bellows

Figure 4.16: The MAESTRO nano-ARPES chamber at the ALS. The CHA is rotatable around its own axis

(dot-dashed line) and also around the vertical axis containing the sample surface.

order of 10 µm, a spot size on the sample of, for example, 100 nm, would require a demagnification

of 100, which for the diffraction-limited emittance of 1.9 nm·rad, equates to divergence of the order

of 1 o. Clearly, secondary optics (Fresnel zone plate, Schwarzschild optics, or ellipsoidal mirrors) and

a stable sample (vibration-free and highly thermally stabilized) stage are required. Because of the

need to demagnify by such a large amount, a nano-ARPES station would benefit from being as far as

possible from the source.

A chamber under the name of ‘MAESTRO’ has been designed at the ALS for nano-ARPES (Fig-

ure 4.16), which might serve as a prototype for a similar machine at SLS-2. The sample is fixed

and kept steady by being fixed to the floor, while the CHA electron analyzer rotates both around the

sample and also around the axis of its entrance aperture. The addition of a retractable scanning probe

4.5. TIME-RESOLVED STUDIES 55

microscope could provide additional spatial resolution.

Time-resolved- and/or three-dimensional (kx, ky, and electron energy) studies using either a camshaft

mode or a pseudo single-bunch mode (see Section 4.5) are also envisaged by using a time-of-flight de-

tector in conjunction with a 2-dimensional delay-line detector.

4.5 Time-resolved studies

Time-resolved studies at storage rings can be realized in different manners (see Figure 4.17). The

hybrid camshaft mode has the advantage that ‘normal’ users can operate at reasonable average photon

fluxes, while those beamlines wanting to perform time-resolved experiments can receive a trigger from

the storage-ring control centre, in order to synchronize their experiments with the isolated bunch.

This does require, however, that those beamlines performing time-resolved experiments using this

camshaft bunch must gate their detectors and/or install expensive and fabricationally challenging

chopper wheels, which must absorb the photon flux originating from the main bunch train. The time-

window of the camshaft mode at the SLS is ±90 ns. Because of the technical challenges for users

in the camshaft mode, certain facilities, such as the ALS, reserve a fraction of the yearly beamtime

exclusively for single-bunch-mode experiments, and all users requiring this are scheduled for this time.

A novel approach to this technological problem is to displace the camshaft bunch in the direction

perpendicular to the plane of the storage ring, by a short-pulse ‘kicker’ magnet, that induces this

bunch to oscillate up and down [Figure 4.17(c)]. This ‘pseudo single-bunch’ (PSB) mode thus spatially

decouples the camshaft pulse from the main pulse train, so that the former can be selected by a simple,

stationary aperture. The temporal separation of the arrival of successive camshaft bunches can be

adjusted from the minimum of one for every full cycle (i.e., the storage-ring circumference divided

by the speed of the electrons ≈ c) to several milliseconds, by adjusting the ring tune [53] in a so-

called kick-and-cancel scheme. Many experiments such as PEPICO coincidence spectroscopy require

time-of-flight electron-energy spectrometers, and would profit enormously from the existence of a PSB

mode.

The PSB mode is particularly attractive for complementary experiments to those carried out at

SwissFEL, in particular for processes characterized by several different time-scales, such as in photo-

synthesis. At time-scales much larger than 100 ps, time-resolved studies using synchrotron radiation

come into their own, such as will become possible using RT-SSX [54](see also Section 4.2.1).

56 SCIENCE CASE

(b)

(d)

(c)

(a)

Figure 4.17: Different filling modes in a synchrotron storage ring. (a) Normally, the ring is filled with bunches

of electrons equally spaced from one another by a few nanoseconds. (b) Some time-resolved experiments may

require that only a single bunch of electrons is in the storage ring. (c) In the hybrid ‘camshaft’ mode, a bunch

(often containing more charge) is isolated by approximately ±100 ns by dropping bunches on either side. (d)

Problems associated with the limited time in which detectors can be gated and/or the high precision required

by mechanical choppers to block the main bunch train in the camshaft mode are overcome in the pseudo single-

bunch mode, in which the camshaft bunch is spatially separated from the main bunch train using a short-pulse

kicker magnet, which induces it to execute an oscillatory motion (shown in green).

Bibliography

[1] H. M. L. Faulkner and J. M. Rodenburg, Movable aperture lensless transmission microscopy: A

novel phase retrieval method, Phys. Rev. Lett. 93, 023903 (2004).

[2] P. Thibault, M. Guizar-Sicairos, and A. Menzel, Coherent imaging at the diffraction limit, J.

Synch. Rad. 21, 1011 (2014).

[3] X. H. Zhu, T. Tyliszczak, H. W. Shiu, D. Shapiro, D. A. Bazylinski, U. Lins, and A. P. Hitchcock,

Magnetic studies of magnetotactic bacteria by soft x-ray STXM and ptychography, AIP Conf. Proc.

1696, 020002 (2016).

[4] M. Holler, A. Diaz, M. Guizar-Sicairos, P. Karvinen, E. Farm, E. Harkonen, M. Ritala, A. Men-

zel, J. Raabe, and O. Bunk, X-ray ptychographic computed tomography at 16 nm isotropic 3D

resolution, Sci. Reports 4, 3857 (2014).

[5] R. N. Wilke, M. Vassholz, and T. Salditt, Semi-transparent central stop in high-resolution X-ray

ptychography using Kirkpatrick-Baez focusing, Bone 69, 490 (2013).

[6] M. Guizar-Sicairos, I. Johnson, A. Diaz, M. Holler, P. Karvinen, H. C. Stadler, R. Dinapoli,

O. Bunk, and A. Menzel, High-throughput ptychography using Eiger: scanning X-ray nano-

imaging of extended regions, Opt. Express 22, 14859 (2014).

[7] J. C. da Silva, P. Trtik, A. Diaz, M. Holler, M. Guizar-Sicairos, J. Raabe, O. Bunk, and A. Menzel,

Mass density and water content of saturated never-dried calcium silicate hydrides, Langmuir 31,

3779 (2015).

–57–

58 BIBLIOGRAPHY

[8] H. D. Barth, M. E. Launey, A. A. MacDowell, J. W. Ager, and R. O. Ritchie, On the effect

of X-ray irradiation on the deformation and fracture behavior of human cortical bone, Bone 46,

1475 (2010).

[9] D. A. Shapiro, Y. S. Yu, T. Tyliszczak, J. Cabana, R. Celestre, W. L. Chao, K. Kaznatcheev,

A. L. D. Kilcoyne, F. Maia, S. Marchesini, Y. S. Meng, T. Warwick, L. L. Yang, and H. A.

Padmore, Chemical composition mapping with nanometre resolution by soft X-raymicroscopy,

Nature Phot. 8, 765 (2014).

[10] A. P. Hitchcock and M. F. Toney, Spectromicroscopy and coherent diffraction imaging: focus on

energy materials applications, J. Synch. Rad. 21, 1019 (2014).

[11] M. Liebi, M. Georgiadis, A. Menzel, P. Schneider, J. Kohlbrecher, O. Bunk, and M. Guizar-

Sicairos, Nanostructure surveys of macroscopic specimens by small-angle scattering tensor to-

mography, Nature 527, 349 (2015).

[12] O. G. Shpyrko, X-ray photon correlation spectroscopy, J. Synch. Rad. 21, 1057 (2014).

[13] B. Henrich, J. Becker, R. Dinapoli, P. Gottlicher, H. Graafsma, H. Hirsemann, R. Klanner,

H. Krueger, R. Mazzocco, A. Mozzanica, H. Perrey, G. Potdevin, B. Schmitt, X. Shi, A. K.

Strivasta, U. Trunk, and C. Youngman, The adaptive gain integrating pixel detector AGIPD: a

detector for the European XFEL, Nucl. Instrum. Meth. A 633, S11 (2011).

[14] W. A. Hendrickson, Atomic-level analysis of membrane-protein structure, Nat. Struct. Mol. Biol.

23, 464 (2016).

[15] J. L. Smith, R. F. Fischetti, and M. Yamamoto, Micro-crystallography comes of age, Curr. Op.

Struct. Biol. 22, 602 (2012).

[16] H. N. Chapman et al., Femtosecond x-ray protein crystallography, Nature 470, 73 (2011).

[17] S. Botha, K. Nass, T. R. M. Barends, W. Kabsch, B. Latz, F. Dworkowski, L. Foucar,

E. Panepucci, M. Wang, R. L. Shoeman, I. Schlichting, and R. B. Doak, Room-temperature se-

rial crystallography at synchrotron X-ray sources using slowly flowing free-standing high-viscosity

microstreams, Acta Cryst. D 71, 387 (2015).

BIBLIOGRAPHY 59

[18] P. Nogly et al., Lipidic cubic phase serial millisecond crystallography using synchrotron radiation,

IUCrJ 2, 168 (2015).

[19] F. Stellato, D. Oberthur, M. Liang, R. Bean, C. Gati, O. Yefanov, A. Barty, A. Burkhardt, P. Fis-

cher, L. Galli, R. A. Kirian, J. Meyer, S. Panneerselvam, C. H. Yoon, F. Chervinskii, E. Speller,

T. A. White, C. Betzel, A. Meents, and H. N. Chapman, Room-temperature macromolecular

serial crystallography using synchrotron radiation, IUCrJ 1, 204 (2014).

[20] C. Gati, G. Bourenkov, M. Klinge, D. Rehders, F. Stellato, D. Oberthur, O. Yefanov, B. P.

Sommer, S. Mogk, M. Duszenko, C. Betzel, T. R. Schneider, H. N. Chapman, and L. Redecke,

Serial crystallography on in vivo grown microcrystals using synchrotron radiation, IUCrJ 1, 87

(2014).

[21] C.-Y. Huang, V. Olieric, P. Ma, E. Panepucci, K. Diederichs, M. Wang, and M. Caffrey, In meso

in situ serial X-ray crystallography of soluble and membrane proteins, Acta Cryst. D 71, 1238

(2015).

[22] J. A. Fraser, H. van dem Bedem, A. J. Samelson, P. T. Lang, J. M. Holton, N. Echols, and T. Al-

ber, Accessing protein conformational ensembles using room-temperature X-ray crystallography,

Proc. Natl. Acad. Sci. U.S.A. 108, 16247 (2011).

[23] E. F. Garman, Radiation damage in macromolecular crystallography: what is it and why should

we care, Acta Cryst. D 66, 339 (2010).

[24] M. Warkentin, J. B. Hopkins, R. Badeau, A. M. Mulichak, L. J. Keefe, and R. E. Thorne, Global

radiation damage: temperature dependence, time dependence and how to outrun it, J. Synch.

Rad. 20, 7 (2013).

[25] R. L. Owen, N. Paterson, D. Axford, J. Aishima, C. Schulze-Briese, J. Ren, E. E. Fry, D. I.

Stuart, and G. Evans, Exploiting fast detectors to enter a new dimension in room-temperature

crystallography, Acta Cryst. D 70, 1248 (2014).

[26] C.-Y. Huang, V. Olieric, P. Ma, N. Howe, L. Vogeley, X. Liu, R. Warshamanage, T. Weinert,

E. Panepucci, B. Kobilka, K. Diederichs, M. Wang, and M. Caffrey, In meso in situ serial X-ray

crystallography of soluble and membrane proteins at cryogenic temperatures, Acta Cryst. D 72,

93 (2016).

60 BIBLIOGRAPHY

[27] C. Nave and M. A. Hill, Will reduced radiation damage occur with very small crystals?, J. Synch.

Rad. 12, 299 (2005).

[28] R. Sanishvili, D. W. Yoder, S. B. Pothineni, G. Rosenbaum, S. Xu, S. Vogt, S. Stepanov,

O. Makarov, S. Corcoran, R. Benn, V. Nagarajan, J. L. Smith, and R. F. Fischetti, Radia-

tion damage in protein crystals is reduced with a micron-sized X-ray beam, Proc. Nat. Am. Soc.

108, 6127 (2011).

[29] J. M. Holton and K. A. Frankel, The minimum crystal size needed for a complete diffraction data

set, Acta Cryst. D 66, 393 (2010).

[30] M. Caffrey, A comprehensive review of the lipid cubic phase or in meso method for crystallizing

membrane and soluble proteins and complexes, Acta Cryst. F 71, 3 (2015).

[31] Q. Liu and W. A. Hendrickson, Crystallographic phasing from weak anomalous signal, Curr. Op.

Struct. Biol. 34, 99 (2015).

[32] A. Wagner, D. Ramona, K. Henderson, and V. Mykhaylyk, In-vacuum long-wavelength macro-

molecular crystallography, Acta Cryst. D 72, 430 (2016).

[33] P. Bleuet, E. Welcomme, E. Dooryhee, J. Susini, J.-L. Hodeu, and P. Walter, Probing the structure

of heterogeneous diluted materials by diffraction tomography, Nat. Mater. 7, 468 (2008).

[34] M. D. de Jonge and S. Vogt, Hard X-ray fluorescence tomography – an emerging tool for structural

visualization, Curr. Op. Struct. Bio. 20, 606 (2010).

[35] M. D. de Jonge, C. G. Ryan, and C. J. Jacobsen, X-ray nanoprobes and diffraction-limited storage

rings: opportunities and challenges of fluorescence tomography of biological specimens, J. Synch.

Rad. 21, 1031 (2014).

[36] A. King, G. Johnson, D. Engelberg, W. Ludwig, and J. Marrow, Observations of intergranular

stress corrosion cracking in a grain-mapped polycrystal, Science 321, 382 (2008).

[37] S. D. M. Jacques, M. Di Michiel, X. H. Kimber, S A J andYang, R. J. Cernik, A. M. Beale, and

S. J. L. Billinge, Pair distribution function computed tomography, Nat. Comms. 4, 2536 (2013).

BIBLIOGRAPHY 61

[38] M. I. McMahon, High-pressure X-ray science on the ultimate storage ring, J. Synch. Rad. 21,

1077 (2014).

[39] N. Casati, A. Kleppe, A. P. Jephcoat, and P. Macchi, Putting pressure on aromaticity along with

in situ experimental electron density of a molecular crystal, Nat. Comms. 7, 10901 (2016).

[40] S. M. Walker, D. A. Schwyn, R. Mokso, M. Wicklein, T. Muller, M. Doube, M. Stampanoni,

H. G. Krapp, and G. K. Taylor, In vivo time-resolved tomography reveals the mechanics of the

blowfly flight motor, PLOS Bio. 12, e1001823 (2014).

[41] J. Schnadt, J. Knudsen, J. N. Andersen, H. Siegbahn, A. Pietzsch, F. Hennies, N. Johansson,

N. Martensson, G. Ohrwall, S. Bahr, S. Mahl, and O. Schaff, The new ambient-pressure X-ray

photoelectron spectroscopy instrument at MAX-lab, J. Synch. Rad. 19, 701 (2012).

[42] F. Siewert, J. Buchheim, T. Zeschke, M. Stormer, G. Falkenberg, and R. Sankari, On the charac-

terization of ultra-precise X-ray optical components: advances and challenges in ex situ metrology,

J. Synch. Rad. 21, 968 (2014).

[43] M. Yabashi, K. Tono, H. Mimura, S. Matsuyama, K. Yamauchi, T. Tanaka, H. Tanaka,

K. Tamasaku, H. Ohashi, S. Goto, and T. Ishikawa, Optics for coherent X-ray applications,

J. Synch. Rad. 21, 976 (2014).

[44] L. J. P. Ament, M. van Veenendaal, T. P. Devereaux, J. P. Hill, and J. van den Brink, Resonant

inelastic x-ray scattering studies of elementary excitations, Rev. Mod. Phys. 83, 705 (2011).

[45] T. Schmitt, F. M. F. de Groot, and J.-E. Rubensson, Prospects of high-resolution resonant X-

ray inelastic scattering studies on solid materials, liquids and gases at diffraction-limited storage

rings, J. Synch. Rad. 21, 1065 (2014).

[46] V. N. Strocov, Concept of a spectrometer for resonant inelastic X-ray scattering with parallel

detection in incoming and outgoing photon energies, J. Synch. Rad. 17, 631 (2010).

[47] V. Bisogni, S. Catalano, R. J. Green, M. Gibert, R. Scherwitzl, Y. Huang, V. N. Strocov, P. Zubko,

S. Balandeh, J.-M. Triscone, G. Sawatzky, and T. Schmitt, Ground state oxygen holes and the

metal-insulator transition in the negative charge transfer rare-earth nickelates, Nat. Comms. 7,

14000 (2016).

62 BIBLIOGRAPHY

[48] T. Warwick, Y.-D. Chuang, D. L. Voronov, and H. A. Padmore, A multiplexed high-resolution

imaging spectrometer for resonant inelastic soft X-ray scattering spectroscopy, J. Synch. Rad. 21,

736 (2014).

[49] Y. Liu, Y. Y. Li, S. Rajput, D. Gilks, L. Lari, P. L. Galindo, M. Weinert, V. K. Lazarov, and

L. Li, Tuning Dirac states by strain in the topological insulator Bi2Se3, Nat. Phys. 10, 294 (2014).

[50] M. Morstein, P. R. Willmott, H. Spillmann, and M. Dobeli, From nitride to carbide: control of

zirconium-based hard materials film growth and their characterization, Appl. Phys. A 75, 647

(2002).

[51] E. Rotenberg and A. Bostwick, microARPES and nanoARPES at diffraction-limited light sources:

opportunities and performance gains, J. Synch. Rad. 21, 1048 (2014).

[52] M. Naamneh, J. C. Campuzano, and A. Kanigel, The electronic structure of BSCCO in

the presence of a super-current: Flux flow, doppler shift and quasiparticle pockets, 2016,

arXiv:1607.02901v1.

[53] C. Sun, G. Portmann, M. Hertlein, J. Kirz, and D. S. Robin, Pseudo-single-bunch with adjustable

frequency: a new operation mode for synchrotron light sources, Phys. Rev. Lett. 109, 264801

(2012).

[54] M. Schmidt, Time-resolved crystallography at x-ray free electron lasers and synchrotron light

sources, Synch. Rad. News 28(6), 25 (2015).

BIBLIOGRAPHY 63