Edinburgh Research Explorer

Manipulating surface magnetic order in iron telluride

Citation for published version:Trainer, C, Yim, CM, Heil, C, Giustino, F, Croitori, D, Tsurkan, V, Loidl, A, Rodriguez, EE, Stock, C & Wahl,P 2019, 'Manipulating surface magnetic order in iron telluride', Science Advances, vol. 5, no. 3.https://doi.org/10.1126/sciadv.aav3478

Digital Object Identifier (DOI):10.1126/sciadv.aav3478

Link:Link to publication record in Edinburgh Research Explorer

Document Version:Peer reviewed version

Published In:Science Advances

General rightsCopyright for the publications made accessible via the Edinburgh Research Explorer is retained by the author(s)and / or other copyright owners and it is a condition of accessing these publications that users recognise andabide by the legal requirements associated with these rights.

Take down policyThe University of Edinburgh has made every reasonable effort to ensure that Edinburgh Research Explorercontent complies with UK legislation. If you believe that the public display of this file breaches copyright pleasecontact openaccess@ed.ac.uk providing details, and we will remove access to the work immediately andinvestigate your claim.

Download date: 13. Sep. 2020

Manipulating surface magnetic order in iron telluride1

C. Trainer,1 C. M. Yim,1 C. Heil,2 F. Giustino,2 D. Croitori,3, 4 V.2

Tsurkan,3, 4 A. Loidl,3 E. E. Rodriguez,5 C. Stock,6 and P. Wahl1, ∗3

1SUPA, School of Physics and Astronomy, University of St Andrews, North Haugh, St Andrews, Fife KY16 9SS, UK4

2Department of Materials, University of Oxford, Parks Road, Oxford OX1 3PH, UK5

3Center for Electronic Correlations and Magnetism, Experimental Physics V,6

University of Augsburg, D-86159 Augsburg, Germany7

4Institute of Applied Physics, Academy of Sciences of Moldova, MD 2028 Chisinau, Republic of Moldova8

5Department of Chemistry of Biochemistry, University of Maryland, College Park, Maryland 20742, USA9

6School of Physics and Astronomy, University of Edinburgh, Edinburgh EH9 3JZ, United Kingdom10

(Dated: June 14, 2018)11

Control and manipulation of emergent magnetic orders in strongly correlated electron materialspromises new opportunities for device concepts which exploit these for spintronics applications. Fortheir technological exploitation it is important to understand the role of surfaces and interfacesto other materials, and their impact on the emergent magnetic orders. Here, we demonstrate foriron telluride, the non-superconducting parent compound of the iron chalcogenide superconductors,determination and manipulation of the surface magnetic structure by low temperature spin-polarizedscanning tunneling microscopy. Iron telluride exhibits a complex magnetic phase diagram as afunction of interstitial iron concentration. Several theories have been put forward to explain thisphase diagram, which ascribe a dominant role either to interactions mediated by itinerant electronsor to local moment interactions. Through the controlled removal of surface excess iron, we canseparate the influence of the excess iron from that of the change in the lattice structure.

I. INTRODUCTION12

Multiple competing interactions in strongly correlated13

electron materials lead to a plethora of emergent phases14

which are highly sensitive to external stimuli and offer15

tremendous potential for applications. Exploiting these16

requires interfacing them to the outside world, yet rela-17

tively little is known about the influence of reduced sym-18

metries and the interface itself. Iron telluride (Fe1+xTe)19

is such a strongly correlated electron material with a com-20

plex magnetic phase diagram as a function of excess iron21

concentration x.22

At low excess iron concentration x (x < 0.11), a bi-23

collinear antiferromagnetic (AFM) order with an order-24

ing wave vector qAFM = (0.5, 0, 0.5) (in units of the re-25

ciprocal lattice) is observed in a crystal structure with26

monoclinic distortion [1–3]. With increasing x, the crys-27

tal structure becomes orthorhombic for x > 0.11, ac-28

companied by a reduction in the difference in the lattice29

constants in the a and b directions [2–4]. This change30

is accompanied by the development of an incommensu-31

rate component of the magnetic order. For x > 0.14, the32

order becomes fully incommensurate and a helimagnetic33

spin spiral develops [1, 2].34

The bicollinear magnetic structure at low excess iron35

concentrations (x < 0.11) is well reproduced by density36

functional theory (DFT) calculations [5, 6], whereas ac-37

counting for the influence of interstitial iron has been38

more challenging. The incommensurate spiral structure39

can be reproduced by assuming that the interstitial Fe40

∗ wahl@st-andrews.ac.uk

atoms lead to electron-doping [7] or by considering ad-41

ditional nearest-neighbor coupling due to the randomly42

distributed interstitial Fe atoms [8]. Even for low inter-43

stitial iron concentrations, multi-q plaquette order has44

been predicted as a result of magnetic frustration and45

quantum fluctuations [9–11].46

It is only very recently that real space imaging of the47

surface magnetic order in iron telluride has been demon-48

strated by spin polarized scanning tunneling microscopy49

(SP-STM) [12–14]. In this work, we use atomic-scale50

imaging by low temperature SP-STM [15–17] to deter-51

mine and manipulate the surface magnetic order in iron52

telluride as a function of excess iron concentration x. Our53

results enable us to assess the impact of the structural54

distortion and excess iron concentration x on the mag-55

netic structure. By manipulating the excess iron con-56

centration of the surface layer, we discover a double-q57

magnetic order which is stabilized as the bulk crystal58

structure becomes orthorhombic.59

We use low temperature spin-polarized scanning tun-60

neling microscopy at temperatures T = 2K to study61

samples with excess Fe concentration x = 0.06 . . . 0.2.62

Ferromagnetic tips for SP-STM are prepared by picking63

up excess iron atoms from the sample surface [12, 13].64

This ability to manipulate the excess iron concentration65

in the surface layer offers the opportunity to manipulate66

the surface magnetic structure, and discriminate between67

the various scenarios for the influence of interstitial iron.68

Because the lattice constants of the surface layer remain69

locked to the crystal structure of the bulk material inde-70

pendent of the excess iron concentration in the surface71

layer, we can image the magnetic structure of strained72

FeTe in the surface layer.73

We will first demonstrate that SP-STM characteriza-74

2

2T

Δz

(pm

)

-10

10

(b)

(a)

a

b

ab

(c)

(d)

(e)

z (p

m)

-20

20

2T

qTe

qTe

qTeqa

FIG. 1. Spin-polarized STM of Fe1.06Te (a) TopographicSTM image taken with a non-magnetic tip ((14.5×14.5)nm2).Protrusions are excess Fe atoms. Inset: Fourier transform(FT) image of (a). Peaks which arise from the Te latticeare highlighted with solid circles. (b) Topographic SP-STMimage taken at the same position as (a) with a magnetictip. Stripes arise from the AFM order. Inset: FT im-age of (b), showing additional peaks due to the AFM or-der. (c-d) Topographic SP-STM images taken at the sameposition with the tip polarized along two opposite in-planedirections (2.5 × 3.5)nm2. Tunneling parameters for (a)-(d):V = 100mV, I = 50pA. (e) Difference image of (c),(d). Theheight difference is proportional to the spin polarization ofthe tunneling current. Inset at the bottom left: structuralmodel of the Fe1+xTe surface, showing the spin order in theFe lattice (red). Inset at top right: DFT calculation of themagnetic contrast due to the spin-polarization at the Fermienergy (see S1 [18]).

tion of Fe1+xTe as a function of excess iron concentra-75

tion x reproduces the magnetic phase diagram found by76

bulk characterization techniques, and then show how the77

magnetic order changes by manipulating the excess iron78

concentration of the surface layer.79

II. RESULTS80

A. Surface magnetic order of Fe1+xTe81

Figure 1(a) shows a topographic STM image of the82

surface of Fe1.06Te obtained with a non-magnetic tip. In83

the image, the square lattice corresponds to the Te atoms84

in the top surface layer, and bright protrusions to the ex-85

cess Fe atoms. Imaging the surface with a magnetic tip86

leads to the appearance of an additional stripe-like mod-87

ulation [Fig. 1(b)] running along the crystal b axis, with88

a periodicity along the crystal a axis twice that of the89

Te lattice. This stripe ordering results in the develop-90

ment of an additional peak in the Fourier transform at a91

wavevector of qAFM = (0.5, 0) (relative to the reciprocal92

unit cell, qTe = (1, 0), (0, 1) are due to the atomic lat-93

tice, compare Fourier image insets of Figs. 1(a) and (b)).94

This additional modulation arises from the bicollinear95

AFM order in Fe1+xTe with low levels of excess Fe con-96

centration (x < 0.1) [12–14]. The additional modulation97

appears due to tunneling magnetoresistance: for a fixed98

tip-sample distance, the current depends on the relative99

magnetization of the tip and the sample. The images100

shown here are acquired using a feedback loop which reg-101

ulates the tip-sample distance such that the current stays102

constant, resulting in a magnetic contrast in the topo-103

graphic SP-STM image. The strength of this contrast104

depends on the relative magnetizations of tip and sam-105

ple and on the spin polarization of the charge carriers in106

the tip and the sample at the Fermi energy.107

Imaging the magnetic structure of Fe1.06Te with108

the same ferromagnetic tip but with its magnetization109

aligned along two opposite in-plane directions results110

in phase reversal in the appearance of the stripe-like111

modulation in SP-STM images[Figs. 1(c) and 1(d)] [12–112

14]. Subtraction of these SP-STM images produces a113

real-space image of the magnetic structure, Fig. 1(e).114

The height difference ∆z is proportional to the spin-115

polarization of the tunneling current, and provides in-116

formation on the local magnetic order projected onto the117

magnetization direction of the tip. The inset in Fig. 1(e)118

shows for comparison a spin-polarized image simulated119

based on a DFT calculation, showing excellent agreement120

with the experimental data (see S1 [18]).121

Using the above approach and aligning the magnetiza-122

tion of the tip along three orthogonal directions, it be-123

comes possible to reconstruct the surface magnetic struc-124

ture of a sample [19] – provided that the sample magnetic125

structure remains unaltered while changing the magneti-126

zation of the tip. For Fe1.06Te, we find that the magnetic127

moments possess an out-of-plane component, pointing128

into a direction 28± 3◦ away from the surface plane (see129

S2 and Fig. S1 [18]). The magnetic ordering wave vector130

we find at low excess iron concentrations is in excellent131

agreement with previous SP-STM [12–14] and neutron132

scattering [1, 2] studies, as well as with calculations [5, 6].133

The out-of-plane angle of the magnetic order found here134

differs from that observed in neutron scattering, where135

the magnetization of the iron atoms is parallel to the b136

direction, but is fully consistent with previous SP-STM137

studies [14].138

At high excess Fe concentrations x > 0.11, the139

structure of Fe1+xTe transforms from monoclinic to or-140

thorhombic. Figure 2(a) shows a topographic image of141

the surface of Fe1.16Te, with three images of the magnetic142

order obtained with the tip magnetized in three different143

angles in the b-c plane [Figs. 2(b)-2(d)]. As the magneti-144

3

(d)

Displacement (nm)

Δz

(pm

)

-90 -40 10 60� (°)

0

1

Pha

se (

rad)

2

3

0 1 2 3 4

0

20

40

6060°

30°

0°

-30°

-60°

-90°

a

b

-90°

-30°

30°

(c)

(b)

Δz

(pm

)

25

-25

(a)100z

(pm

)

-100

(e)

(f)

b

cB�

FIG. 2. Spin spiral in Fe1.16Te. (a) Topographic SP-STMimage ((8.2 × 4.6)nm2, V = 50mV, I = 200pA). (b)-(d),magnetic images taken at out-of-plane angles θ = −90◦, −30◦

and 30◦. Close inspection reveals that the positions of themaxima of the magnetic order shift as a function of out-of-plane angle θ. (e) line cuts through magnetic images as shownin (b)-(d) along a for different out-of-plane field angles. Theline cuts show the shift of the maxima of the stripes. (f) Plotof the phase of the stripes shown in (e) as a function of fieldangle θ. The phase has been extracted using the maximummarked by an arrow in (e). Measurements were taken at anin-plane angle φ = 120◦ from the crystal a axis.

zation of the tip rotates, the magnetic contrast in the dif-145

ference image translates along the a direction similarly to146

what was observed in previous SP-STM studies of other147

systems with spin spiral magnetic orders[20]. This evi-148

dences the presence of a unidirectional spin spiral in the149

surface propagating along the a axis with spins rotating150

in the b-c plane. Analysis of the line-cuts taken from the151

difference images [Fig. 2(e)] reveals the spin spiral with152

a wave-vector of q = (0.43, 0), slightly incommensurate153

with the crystal lattice (see S3 and Fig. S2 [18]). The154

spin spiral found here is in full agreement with that de-155

tected in bulk samples at high excess iron concentrations156

x > 0.12 by neutron scattering.157

The key result from this section is that SP-STM at the158

surface of Fe1+xTe yields consistent result with neutron159

scattering. Differences appear merely in the details, such160

as the out-of-plane component of the ordered moment.161

B. Manipulation of the excess iron concentration162

In addition to imaging the surface magnetic structure163

at the atomic scale, STM also allows us to manipulate164

the surface composition, which for the case of Fe1+xTe165

can be achieved by fast, high tunneling current scanning166

(∼ 2nA) at a slow feedback loop response time. The re-167

sulting change in composition is illustrated in Fig. 3: the168

Fe interstitial iron atoms originally present on the surface169

layer of Fe1+xTe are removed, leaving the remaining Fe170

interstitials at the lower part of the surface layer intact171

(for details see S4 [18]). As a consequence, this provides172

an opportunity to study the magnetic order in the sur-173

face layer with only half of the excess Fe concentration174

present, but with the same lattice structure as that of the175

bulk. As the coupling between the FeTe layers is rather176

weak, understanding the magnetic structure of a manip-177

ulated surface layer provides us with information about178

how the bulk of the material would respond to similar179

conditions.180

Figure 3(c) shows an SP-STM image of the surface181

of Fe1.12Te, on one half of which all the surface excess182

Fe atoms have been removed by fast, high-current scan-183

ning, whereas they are still present in the other half. The184

Fourier transform of the area shown in Fig. 3(c) plotted185

in Fig. 3(d) reveals a pair of magnetic ordering peaks,186

one at the bicollinear ordering vector q = (0.5, 0) and187

one at an incommensurate vector q = (0.39, 0). Fourier188

filtering the topographic image [Fig. 3(c)] at q = (0.5, 0)189

[Fig. 3(f)] and q = (0.39, 0) [Fig. 3(g)] reveals that the190

bicollinear order is concentrated to the areas where the191

excess Fe atoms at the surface have been mostly re-192

moved whilst the incommensurate order is confined to193

areas where the initial surface excess Fe concentration has194

been left untouched. This incommensurability has also195

been observed in neutron scattering conducted at similar196

excess Fe concentrations. The STM image in Fig. 3(c)197

demonstrates that while the interstitial iron concentra-198

tion of the surface layer can be manipulated using STM,199

the lattice structure remains commensurate with the bulk200

as no additional superstructure is seen which would arise201

from a structural incommensurability of the surface layer202

with the bulk.203

C. Impact of excess iron concentration on surface204

magnetic order205

Detailed investigation of the magnetic structure of the206

area cleaned of excess iron in Fe1.12Te (compare Fig. 3)207

reveals a complex picture: Figures 4(a) and 4(b) show208

images of the magnetic order in the surface layer pro-209

jected onto two different magnetization directions of the210

tip, in-plane and out-of-plane. While the image obtained211

with an in-plane magnetization of the tip, Fig. 4(a), re-212

veals only the unidirectional bicollinear order, measure-213

ments with an out-of-plane magnetized tip [Fig. 4(b)]214

reveal domains of checkerboard-like double-q order. In215

the Fourier transformation [Fig 4(c)], the component of216

the magnetic order along the a axis is characterized by a217

sharp peak at q = (0.5, 0), whereas the new component218

along b manifests itself as a broad peak at q ∼ (0, 0.5),219

reflecting its localized nature. Unlike the single-q mag-220

netic order found at low excess Fe concentration, which221

has the magnetization of the iron atoms oriented in op-222

posite directions in the a-c plane, in the double-q order223

the magnetization is also modulated in the b direction,224

4

FIG. 3. Manipulation of surface excess iron concentration. Model of Fe1+xTe (a) before and (b) after removal of excessiron. Red (yellow) spheres are Fe (Te) atoms. Dashed open circles mark the interstitial/excess Fe atoms at the surface layerthat are removed during surface manipulation, leading to a 50% reduction in the concentration of Fe interstitials in the surfacelayer (to x/2) compared to that of the bulk (x). (c) SP-STM image of Fe1.12Te ((65.3 × 28.2)nm2, V = 150mV , I = 50pA),showing an area cleaned of surface excess iron (blue) next to one where the excess iron has been left untouched (green). (d)Fourier transform of (c) showing magnetic peaks due to the bicollinear order (blue arrow) and at an incommensurate position(green arrow). (see Supplementary fig. S5 for Fourier transforms of regions with high and low excess iron concentrationsat the surface). (e) Line-cut from the Fourier transform in (d) taken along the aTe direction. Peaks corresponding to thebicollinear order at q = (0.5, 0) and the incommensurate order q = (0.39, 0) are highlighted by a blue arrow and green arrowrespectively. (f,g) Maps of the intensity of the magnetic order at the wave vector of the bicollinear order (q = (0.5, 0)) and ofthe incommensurate order (q = (0.39, 0)). Both have been obtained through Fourier filtering at the respective wave vector,and then low pass filtering of the modulus. The maps show that bicollinear order predominantly exists in regions that havebeen cleaned of Fe while the incommensurate order is dominant in regions where the Fe is still present.

leading to the formation of a spin-spiral-like order. A225

model which reproduces this behaviour is described in226

S6, equation S1. In the orthorhombic phase at higher ex-227228

cess iron concentration we find an even stronger change229

of the magnetic structure after removing the surface in-230

terstitial iron atoms: Fig. 5(a), obtained from a surface231

layer of Fe1.1Te on a Fe1.2Te sample, shows a topographic232

image together with three images of the magnetic order233

[Figs. 5(b) to 5(d)] for the tip magnetized along two in-234

plane [Figs. 5(b) and 5(c)] and an out-of-plane direction235

[Fig. 5(d)]. All three magnetic images reveal a double-236

q magnetic order, of which both components, along the237

a and b directions, are characterized by commensurate238

wave-vectors of q = (0.5, 0) and (0, 0.5). As revealed by239

their Fourier images [insets of Figs. 5(b) to 5(d)], the240

strength of both components varies with the magnetiza-241

tion of the tip. To resolve the magnetic structure, we242

have rotated the magnetization of the tip in a plane par-243

allel to the surface in steps of 18◦ through 180◦, and244

recorded a magnetic image for each angle (for details245

see S5 [18]). The results are summarized in Fig. 5(e).246

The intensities of both components [I(qa), I(qb)] vary as247

| cosφ|, and reach their maxima at ∼ 30◦ from the b axis.248

Both exhibit an almost identical angular dependence un-249

der the in-plane rotation. However, when the magneti-250

zation of the tip is rotated out of the surface plane the251

strength of the component of the magnetic order along252

a varies strongly, while the component along b remains253

unchanged [Fig. 5(f)]. This can be accounted for in a254

model for the magnetic structure consisting of two spin255

spirals along the [110] directions (see S7, eq. S2 and Fig.256

S7 [18]). The spin spirals alternate between clockwise257

and counter-clockwise winding on alternate rows of Fe258

atoms.259

III. DISCUSSION260

By combining spin-polarized imaging with the ability261

to manipulate the surface Fe atoms, our results enable262

5

us to extract a comprehensive picture of the magnetic263

phase diagram of iron telluride and assess the impact of264

interstitial iron on the magnetism in this material.265

In Fig. 6(a), we show models of the magnetic struc-266

ture deduced from our study for surface layers after re-267

moval of the interstitial iron, effectively in a surface layer268

of Fe1+x/2Te on Fe1+xTe. In the monoclinic phase, we269

observe the same magnetic order after removal of inter-270

stitial iron as in the bulk. With reduced asymmetry of271

the lattice constants in the a and b directions, we see272

an increasing deviation of the spins from the b direction.273

At the transition of the bulk from the monoclinic to the274

orthorhombic structure at x = 0.12, the surface layers275

exhibit patches of double-q order, with apparent compe-276

tition between the bicollinear order in the a direction and277

a developing SDW order in the b direction. At high ex-278

cess iron concentration in the bulk, the cleaned surface279

layer adopts a double-q magnetic order.280

This can be best seen in the phase diagram in Fig. 6(b),281

where we compare the incommensurability seen in neu-282

tron scattering with the appearance of magnetic order at283

a second q-vector along b in our SP-STM measurements284

(a)

qTe

qa

qb

qTe

(c)

0 0.2 0.4 0.6 0.8 1

q/qTe

0

1

2

3

I (a.

u.)

-10 0 10�z (pm)

-5 0 5�z (pm)

LO HI

(b)

(d)

2 T2 T2 T2 T

qTe

qb

qa

FIG. 4. Magnetic order in a surface layer of Fe1.06Te onFe1.12Te (a) Image of the magnetic order ((20.5 × 20.5)nm2)projected onto an in-plane direction of the magnetizationas indicated by the arrows. (b) As (a), taken with thesame tip with an out-of-plane direction of the magnetization(V = 100mV, I = 50pA). (c) FT image of (b), with intensityat the center suppressed for clarity. Peaks due to the AFMorder along a and b are marked with pink and cyan circles,respectively. (d) Normalized line cuts taken from the originalong a (red) and b (blue) in (c). Dashed lines indicate thepositions qTe and qAFM along both a and b directions.

θ (°)

Δz

(pm

)

16

-16

-27°-50

50

z (p

m)

a

b

qTe (b)

(d)(c)

(a)

5 T27°

5 T

5 T

0 20 40 60 80 100 120

I (a.

u.)

1

1.8

1.6

1.4

1.2

(f)I(qa)I(qb)

0-50 50 100� (°)

150

1.8

0.6

1.0

1.4

I (a.

u.)

(e) �max �maxa b

qa

qb

FIG. 5. Plaquette order in Fe1.1Te on Fe1.2Te. (a) Non-magnetic image (for details see S5 and Fig. S6 in 18). Inset:FT image showing peaks due to the Te lattice. (b) Image ofthe magnetic structure for φ = −27◦. AFM order can be seenalong both lattice directions. Inset: FT image of (b), withpeaks due to AFM order in a (b) direction marked with pink(blue) circles. (c), (d) As (b), with (c) φ = 27◦ and (d) out-of-plane magnetic field (θ = 90◦). Insets show the correspondingFT images. Images (a)-(d) were recorded in the same area((24.1 × 24.1)nm2, V = −40mV, I = 100pA). (e) Integratedintensities of the magnetic peaks in the FT image as a functionof in-plane angle φ. The horizontal dashed line indicates theintegrated intensity of a point away from the magnetic peaks.Blue (red) markers are the intensities of the qb (qa) peak.Solid lines are numerical fits of I = I0| cos (φ− φ0)|+ c to thedata. A red (blue) vertical line indicates the in-plane fielddirection parallel to the a (b) axis, a cyan dashed vertical linethe in-plane direction of maximum intensity (φ = 117◦). (f)As (e), plotted as a function of out-of-plane angle θ at in-planeangle of maximum intensity of the magnetic order, φ = 117◦.Data shown here was recorded in the same location and withthe same tip.

and the anisotropy of the crystal structure. To this end,285

we plot the ratio of the intensities I(q) of the magnetic286

order in the b direction Ib = I(q = (0, 0.5)) and in the287

a direction Ia = I(q = (0.5, 0)), as a function of excess288

6

FIG. 6. Magnetic structures and phase diagram of a surface layer of Fe1+x/2Te. (a) Model structures of surface layersof Fe1.03Te/Fe1.06Te, Fe1.06Te/Fe1.12Te and Fe1.1Te/Fe1.2Te that are consistent with the SP-STM results. The magnetic unitcell in each case is highlighted (for details of the model see S7 [18]). Arrows indicate Fe spins and are colored blue if theyhave a positive component along b and red if they have a negative component. Blue and gray spheres represent the upper andlower Te atoms. (b) Phase diagram of the magnetic order in the surface layer after removal of excess iron (with concentrationFe1+x/2Te) as a function of bulk excess Fe concentration x. Red dots represent the ratio of the intensities I(qb)/I(qa) of themagnetic order along the lattice directions a and b taken from the Fourier transforms of the STM data. The blue diamondsshow the wave vector of the magnetic order in terms of the lattice spacing from neutron scattering from Ref. 2, the blackdotted lines depict the ”mixed spin density wave (SDW)” phase defined there. The grey background highlights how the Telattice parameters change with excess Fe concentration and is defined by Ψ = aTe

bTe− 1, evaluated from values given in Ref. 2.

(c)-(e) (7 × 7nm2) SP-STM images of out-of-plane polarization of a surface layer of (c) Fe1.03Te in the monoclinic phase and(d) Fe1.06Te and (e) Fe1.1Te in the orthorhombic phase.

iron concentration. The doping dependence of this ra-289

tio Ib/Ia shows that the magnetic order at q = (0, 0.5)290

appears once the material has undergone the transition291

from a monoclinic crystal structure to an orthorhombic292

one.293

The magnetic structure of the surface iron telluride294

layer adopts a staggered magnetic order in the a-b plane295

with the spins on the iron atoms alternating between296

fixed angles pointing away from and parallel to the b axis297

direction, while the component along the c axis alter-298

nately points out of (or into) the a-b plane. This model299

of the magnetic order consists effectively of a pair of co-300

existing spin spirals along the Fe-Fe ([110] and [110]) di-301

rection where the spirals alternate between right and left302

handedness for every other row of Fe atoms along [110].303

The order remains commensurate and is similar to the304

plaquette order predicted theoretically in scenarios where305

interstitial iron is neglected and the lattice structure ap-306

proaches tetragonal symmetry [9, 10] or when biquadratic307

exchange interactions are included [21]. We note that308

there may also be an influence of Djaloshinskii-Moriya309

interactions on the surface magnetic structure.310

Our results suggest that a description of the interstitial311

iron atoms as modifying the local couplings, and thus the312

magnetic order, is more appropriate than there being an313

overall charge doping which changes the nesting of the314

band structure. The latter would imply that removal of315

surface excess iron recovers the magnetic order found at316

lower excess iron concentrations, continuously changing317

the ordering wave vector, which however is not what we318

observe.319

Our findings show some parallels to the C4 magnetic320

order found in the iron pnictide superconductors when321

the magnetostructural phase transition is suppressed [22–322

24]. In both cases, the reduction of the structural asym-323

metry results in formation of double-q magnetic order,324

though based on a different order in the undoped com-325

pound.326

IV. CONCLUSION327

Our measurements demonstrate how atomic manipu-328

lation combined with spin-polarized scanning tunneling329

7

microscopy can be used to understand the influence of330

defects on emergent orders [25] in strongly correlated331

electron materials, and provide a new path to control332

emergent phases by atomic manipulation.333

EXPERIMENTAL METHODS334

STM Experiments335

The STM experiments were performed at 2K using a336

low temperature STM equipped with a vector magnet337

that enables application of magnetic fields of up to 5 T338

in any direction with respect to the tip-sample geome-339

try [26]. To obtain a pristine, impurity-free surface for340

imaging, Fe1+xTe samples were prepared by in-situ cleav-341

ing at a temperature of ∼ 20K. Pt-Ir tips were condi-342

tioned by field-emission on a Au(111) sample. Ferromag-343

netic tips used for SP-STM measurements were prepared344

by picking up the interstitial Fe atoms from the Fe1+xTe345

sample in the STM [12]. The magnetization of the tip346

was controlled through the applied magnetic field. The347

influence of the magnetic field on the magnetic structure348

of Fe1+xTe is assumed to be negligible due to substantial349

magneto-crystalline anisotropy [12]. The magnetization350

direction of the magnetic tip is denoted by (θ, φ), where θ351

represents the out-of-plane angle, and φ the in-plane an-352

gle measured from the crystal a axis of the FeTe sample.353

All STM images were obtained at 2K.354

Sample growth355

Single crystals of Fe1+xTe were grown by the self-flux356

method [27, 28]. The excess iron concentrations reported357

here have been determined using energy-dispersive x-ray358

(EDX) analysis. Throughout the main text, the excess359

iron concentration of bulk samples (i.e. before removal of360

surface excess iron) refers to the off-stoichiometric part x361

of the composition of the material as extracted in EDX,362

which in principle can originate either from interstitial363

iron or a tellurium deficiency. Interstitial or excess iron364

refers to iron between the FeTe layers or at the surface.365

Characterization of the crystals indicates that excess iron366

concentration and interstitial iron concentration are iden-367

tical within the errors of our measurements.368

ACKNOWLEDGMENTS369

CT, CMY and PW acknowledge funding from EP-370

SRC through EP/L505079/1 and EP/I031014/1, and371

CS through EP/M01052X/1. VT, AL and JD acknowl-372

edge funding from the Deutsche Forschungsgemeinschaft373

(DFG) via the Transregional Collaborative Research374

Center TRR 80 (Augsburg, Munich, Stuttgart). CH375

acknowledges support from the Austrian Science Fund376

(FWF) project No. J3806-N36 and the Vienna Science377

Cluster. FG acknowledges support from the Leverhulme378

Trust (Grant RL-2012-001) and the UK EPSRC Research379

Council (Grant No. EP/M020517/1). Underpinning380

data will be made available at DOI:10.17630/.381

[1] W. Bao, Y. Qiu, Q. Huang, M. A. Green, P. Zajdel,382

M. R. Fitzsimmons, M. Zhernenkov, S. Chang, M. Fang,383

B. Qian, E. K. Vehstedt, J. Yang, H. M. Pham, L. Spinu,384

and Z. Q. Mao, Phys. Rev. Lett. 102, 247001 (2009).385

[2] E. E. Rodriguez, C. Stock, P. Zajdel, K. L. Krycka, C. F.386

Majkrzak, P. Zavalij, and M. A. Green, Phys. Rev.B 84,387

064403 (2011).388

[3] C. Koz, S. Roßler, A. A. Tsirlin, S. Wirth, and389

U. Schwarz, Phys. Rev. B 88, 094509 (2013).390

[4] J. Warmuth, M. Bremholm, P. Hofmann, J. Wiebe, and391

R. Wiesendanger, npj Quantum Materials 3, 21 (2018).392

[5] F. Ma, W. Ji, J. Hu, Z. Y. Lu, and T. Xiang, Phys. Rev.393

Lett. 102, 177003 (2009).394

[6] M. D. Johannes and I. I. Mazin, Phys. Rev. B 79,395

220510(R) (2009).396

[7] S. Ducatman, R. M. Fernandes, and N. B. Perkins, Phys.397

Rev. B 90, 165123 (2014).398

[8] C. Fang, B. A. Bernevig, and J. Hu, EPL 86, 67005399

(2008).400

[9] S. Ducatman, N. B. Perkins, and A. Chubukov, Phys.401

Rev. Lett. 109, 157206 (2012).402

[10] J. K. Glasbrenner, I. I. Mazin, H. O. Jeschke, P. J.403

Hirschfeld, and R. Valentı, Nat. Phys 11, 953 (2015).404

[11] S.-C. Tang, M.-C. Ding, and Y.-Z. Zhang, Sci. Rep. 6,405

19031 (2016).406

[12] M. Enayat, Z. Sun, U. R. Singh, R. Aluru, S. Schmaus,407

A. Yaresko, Y. Liu, V. Tsurkan, A. Loidl, J. Deisenhofer,408

and P. Wahl, Science 515, 653 (2014).409

[13] U. R. Singh, R. Aluru, Y. Liu, C. Lin, and P. Wahl,410

Phys. Rev. B 91, 161111(R) (2015).411

[14] T. Hanke, U. R. Singh, L. Cornils, S. Manna, A. Kamla-412

pure, M. Bremholm, E. M. J. Hedegaard, B. B. Iversen,413

P. Hofmann, J. Hu, Z. Mao, J. Wiebe, and R. Wiesen-414

danger, Nat. Commun. 8, 13939 (2017).415

[15] R. Wiesendanger, H.-J. Guntherodt, G. Guntherodt,416

R. J. Gambino, and R. Ruf, Phys. Rev. Lett. 65, 247417

(1990).418

[16] M. Bode, Rep. Prog. Phys 66, 523 (2003).419

[17] R. Wiesendanger, Rev. Mod. Phys. 81, 1495 (2009).420

[18] See Supplemental Information at http://...421

[19] K. F. Zhang, X. Zhang, F. Yang, Y. R. Song, X. Chen,422

C. Liu, D. Qian, W. Luo, C. L. Gao, and J. F. Jia, Appl.423

8

Phys. Lett. 108, 061601 (2016).424

[20] M. Bode, M. Heide, K. von Bergmann, P. Ferriani,425

S. Heinze, G. Bihlmayer, A. Kubetzka, O. Pietzsch,426

S. Blugel, and R. Wiesendanger, Nature 447, 190 (2007).427

[21] H.-H. Lai, S.-S. Gong, W.-J. Hu, and Q. Si,428

arXiv:1608.08206 [cond-mat] (2016), arXiv: 1608.08206.429

[22] S. Avci, J. M. Allred, O. Chmaissem, D. Y. Chung,430

S. Rosenkranz, J. A. Schlueter, H. Claus, A. Daoud-431

Aladine, D. D. Khalyavin, P. Manuel, A. Llobet, M. R.432

Suchomel, M. G. Kanatzidis, and R. Osborn, Phys. Rev.433

B 88, 094510 (2013).434

[23] A. E. Bohmer, F. Hardy, L. Wang, T. Wolf, P. Schweiss,435

and C. Meingast, Nat. Commun. 6, 7911 (2015).436

[24] J. M. Allred, K. M. Taddei, D. E. Bugaris, M. J.437

Krogstad, S. H. Lapidus, D. Y. Chung, H. Claus, M. G.438

Kanatzidis, D. E. Brown, J. Kang, R. M. Fernandes,439

I. Eremin, S. Rosenkranz, O. Chmaissem, and R. Os-440

born, Nat. Phys. 12, 493 (2016).441

[25] H. Alloul, J. Bobroff, M. Gabay, and P. J. Hirschfeld,442

Rev. Mod. Phys. 81, 45 (2009).443

[26] C. Trainer, C. M. Yim, M. McLaren, and P. Wahl, Rev.444

Sci. Instrum. 88, 093705 (2017).445

[27] V. Gnezdilov, Y. Pashkevich, P. Lemmens, A. Gusev,446

K. Lamonova, T. Shevtsova, I. Vitebskiy, O. Afanasiev,447

S. Gnatchenko, V. Tsurkan, J. Deisenhofer, and A. Loidl,448

Phys. Rev. B 83, 245127 (2011).449

[28] V. Tsurkan, J. Deisenhofer, A. Gunther, C. Kant,450

M. Klemm, H.-A. Krug von Nidda, F. Schrettle, and451

A. Loidl, Eur. Phys. J. B 79, 289 (2011).452