Antiferromagnetic correlations in the metallic strongly correlated transition metal oxide LaNiO$_3$

TL;DR

This study reports the successful growth of large, high-purity LaNiO₃ single crystals and reveals an unexpected antiferromagnetic transition in their metallic state, contributing new insights into the magnetic properties of this material.

Contribution

It introduces a method for growing large, impurity-free LaNiO₃ single crystals and uncovers antiferromagnetic correlations in their metallic phase, previously unobserved.

Findings

Large, impurity-free LaNiO₃ single crystals were grown successfully.

An unexpected antiferromagnetic transition was observed in the metallic phase.

Crystals exhibit metallic properties with high oxygen pressure synthesis.

Abstract

The material class of rare earth nickelates with high Ni$^{3+}$ oxidation state is generating continued interest due to the occurrence of a metal-insulator transition with charge order and the appearance of non-collinear magnetic phases within this insulating regime. The recent theoretical prediction for superconductivity in LaNiO$_3$ thin films has also triggered intensive research efforts. LaNiO$_3$ seems to be the only rare earth nickelate that stays metallic and paramagnetic down to lowest temperatures. So far, centimetre-sized impurity-free single crystal growth has not been reported for the rare earth nickelates material class since elevated oxygen pressures are required for their synthesis. Here, we report on the successful growth of centimetre-sized LaNiO$_3$ single crystals by the floating zone technique at oxygen pressures of up to 150 bar. Our crystals are essentially free from Ni$^{2+}$ impurities and exhibit metallic properties together with an unexpected but clear antiferromagnetic transition.

Tables

Table 1. Supplementary Table 1: Crystal structure Refinement results of single crystal X-ray diffraction measurements of LaNiO 3  at room-temperature (space group R 3 ¯ ¯ 3 \mathit{\overline{3}} c ).

atom	occup.	$x$	$y$	$z$
La1	0.995(12)	0	0	0.25
Ni1	1.000(14)	0	0	0
O1	1	0.5457(4)	0	0.25
atom	U11 (Å2)	U22 (Å2)	U33 (Å2)
La1	0.00523(7)	0.00523(7)	0.00803(16)
Ni1	0.00304(13)	0.00304(13)	0.0055(3)
O1	0.0077(5)	0.0077(9)	0.0093(11)
atom	U12 (Å2)	U13 (Å2)	U23 (Å2)	Uiso (Å2)
La1	0.00261(4)	0	0	0.00616(7)
Ni1	0.00152(7)	0	0	0.00388(13)
O1	0.0038(5)	-0.0015(4)	-0.0029(7)	0.0082(6)

Table 2. Supplementary Table 2: Magnetic symmetry Basis vectors of the irreducible representations of the space group P ​ m ​ 3 ¯ ​ m 𝑃 𝑚 ¯ 3 𝑚 Pm\bar{3}m for sites 1 ​ b 1 𝑏 1b with propagation vector k = (0.25 0.25 0.25).

Ni ($1b$)			(1/2, 1/2, 1/2)
${\Gamma }_2$	${\psi }_1$	Re	(1, 1, 1)
${\Gamma }_3$	${\psi }_1$	Re	(1, -0.5, -0.5)
		Im	(0, -$\sqrt{3}$/2, $\sqrt{3}$/2)
	${\psi }_2$	Re	(0.5, -1, 0.5)
		Im	($\sqrt{3}$/2, 0, -$\sqrt{3}$/2)

Table 3. Supplementary Table 3: Magnetic structure Comparison of the measured and calculated intensities for our two models shown in Supplementary Figure 6. I obs : measured intensities, I cal , Γ 2 cal subscript Γ 2 {}_{\mathrm{cal},\Gamma_{2}} : calculated intensities for the longitudinal moment modulation, I cal , Γ 3 cal subscript Γ 3 {}_{\mathrm{cal},\Gamma_{3}} : calculated intensities for the helical moment arrangement.

magnetic peak	Iobs	I${}_{\mathrm{cal},{\Gamma }_2}$	I${}_{\mathrm{cal},{\Gamma }_3}$
(0.25 0.25 0.75)	18.3	26.5	17.4
(0.75 0.75 0.25)	27.3	13.6	21.6
(0.75 0.75 0.75)	22.2	0	26.6

Equations

$$\Gamma ={\Gamma }_2\oplus {\Gamma }_3.$$

Full text

Antiferromagnetic correlations in the metallic strongly correlated transition metal oxide LaNiO3

H. Guo1,∗, Z. W. Li1,∗, L. Zhao1, Z. Hu1, C. F. Chang1, C.-Y. Kuo1,

W. Schmidt2, A. Piovano3, T. W. Pi4, O. Sobolev5, D. I. Khomskii6, L. H. Tjeng1

& A. C. Komarek1,†

• These authors contributed equally to this work

${\dagger}$: Alexander.Komarek@cpfs.mpg.de

• •

1 Max-Planck-Institute for Chemical Physics of Solids, Nöthnitzer Str. 40, Dresden D-01187, Germany

• •

2 Forschungszentrum Jülich GmbH, Jülich Centre for Neutron Science at ILL, CS 20156, 71 avenue de Martyrs, 38042 Grenoble, France

• •

3 Institut Laue-Langevin (ILL), 71 avenue des Martyrs, F-38042 Grenoble Cedex 9, France

• •

4 National Synchrotron Radiation Research Center (NSRRC), 101 Hsin-Ann Road, Hsinchu 30077, Taiwan

• •

5 Georg-August-Universität Göttingen, Institut für Physikalische Chemie, Tammannstrasse 6, D-37077 Göttingen, Germany

• •

6 Physics Institute II, University of Cologne, Zülpicher Str. 77, 50937 Cologne, Germany

Abstract

The material class of rare earth nickelates with high Ni3+ oxidation state is generating continued interest due to the occurrence of a metal-insulator transition with charge order and the appearance of non-collinear magnetic phases within this insulating regime. The recent theoretical prediction for superconductivity in LaNiO3  thin films has also triggered intensive research efforts. LaNiO3 seems to be the only rare earth nickelate that stays metallic and paramagnetic down to lowest temperatures. So far, centimetre-sized impurity-free single crystal growth has not been reported for the rare earth nickelates material class since elevated oxygen pressures are required for their synthesis. Here, we report on the successful growth of centimetre-sized LaNiO3 single crystals by the floating zone technique at oxygen pressures of up to 150 bar. Our crystals are essentially free from Ni2+ impurities and exhibit metallic properties together with an unexpected but clear antiferromagnetic transition.

Introduction

The rare earth nickelates RNiO3 (R $=$ rare earth, Y) with the high Ni3+ oxidation state have continued to attract enormous interest due to the famous bandwidth controlled metal-insulator (MI) transition and associated unusual charge and spin order phenomena occurring in this system [1, 2] with even the possibility for multiferroicity [3]. The description of the underlying physics of these phenomena turned out to be a true intellectual challenge and it is not a surprise that novel theoretical concepts have been (and still need to be) developed along the way [4, 5, 6, 7, 8, 9, 10, 11, 12]. More recently, the prediction of high-TC superconductivity in LaNiO3 based heterostructures [13] has triggered a flurry of new activities on LaNiO3-LaMO3 superlattices (M = other metal ion) [14, 15].

Apart from LaNiO3 all RNiO3 (R $=$ Pr-Lu, Y) compounds exhibit a MI transition at lower temperatures [1, 2] with an insulating antiferromagnetic ground state. With decreasing R-ionic radius the octahedral tilts become larger. Thus, the Ni-O-Ni bond angles become smaller which alters the electronic bandwidth and the magnetic exchange interactions. Whereas an enhancement of the insulating properties and an increase of the MI transition temperature TMI can be observed for decreasing R-ionic radius, the antiferromagnetic transition temperature and the strength of the exchange interactions decrease with decreasing R-ionic radius and Ni-O-Ni bond angles. With La having the largest ionic radius of the series one might expect the strongest antiferromagnetic properties for LaNiO3. However, so far it was reported that LaNiO3 does not show any magnetic order and, thus, violates the trend. It remains a paramagnetic metal down to lowest temperatures with an enhanced effective mass [1, 2, 16, 17].

Here, we report on the successful growth of large LaNiO3 single crystals by the floating zone technique. Electrical resistivity and Hall effect measurements on our single crystals show that LaNiO3 is intrinsically not a bad metal as recently discussed for the RNiO3 (R $=$ Pr-Lu, Y) compounds in their paramagnetic phase [10]. On the contrary, we found that LaNiO3 has a high conductivity. Moreover, we were able to observe bulk antiferromagnetism in magnetization, specific heat and neutron scattering experiments. Thus, LaNiO3 appears to be a highly metallic and antiferromagnetic transition metal oxide - a rather rare combination in oxides. Special about LaNiO3 is that it is also very close to an insulating state, making LaNiO3 an intriguing quantum material, probably close to a quantum critical point, where strong local electronic correlations at the Ni sites are likely to interfere in an intricate manner with Fermi surface effects.

Results

Crystal growth and characterization

Using high oxygen pressures of 130-150 bar we were able to grow LaNiO3 single crystals with the floating zone technique, see inset of Fig. 1(a). Further details can be found in the Methods Section and in the Supplementary Notes 1 & 2, in the Supplementary Figures 1 & 2 and in the Supplementary Movie 1. Powdered crystals exhibit no impurity phases within the accuracy of powder X-ray diffraction measurements, see Fig. 1(a). The single crystallinity of our LaNiO3 crystal is confirmed by a series of Laue diffraction patterns from different directions and across the length of this crystal as well as by single crystal X-ray diffraction, see inset of Fig. 1(a) and Supplementary Figure 1 as well as Supplementary Note 3, Supplementary Figure 3 and Supplementary Table 1. Due to the occurrence of a cubic to rhombohedral phase transition (Pm$\mathit{\overline{3}}$m $\rightarrow$ R$\mathit{\overline{3}}$c) somewhat below 1100 K [2] our floating zone grown single crystals are twinned.

Fig. 1(b) shows the Ni-L2,3 X-ray absorption spectroscopy (XAS) data of the LaNiO3 single crystal taken at 300 K and 90 K together with that of a NiO single crystal serving as a Ni2+ reference compound. We have removed the La-M4 white lines located at 850.6 eV from the LaNiO3 spectra using the La-M4 spectrum of LaCoO3. It is well known that XAS spectra at the L2,3 edge of transition metal oxides are highly sensitive to the valence state, in particular, an increase of the valence state of the transition metal ion by one causes a shift of the XAS L2,3 spectra by one or more eV towards higher energies [18]. The more than one eV higher energy shift between the spectra of NiO and LaNiO3 indicates the formal Ni3+ valence state in LaNiO3 [18, 19], with the note that the spectrum can not be interpreted in terms of a Ni 3d7 configuration, but, rather by a coherent mixture of 3d8 and 3d8\underline{L}$$\underline{L} configurations [4, 5, 9, 12], where each $\underline{L}$ denotes a hole in the oxygen ligand. Here we can exclude Ni2+ impurities in our LaNiO3 single crystal - otherwise the sharp main peak of the Ni2+ impurity spectrum would have been visible as a sharp shoulder at the leading edge [18]. At the O-K edge, the pre-edge peak is shifted by about one eV to higher energies when going from Ni3+ to Ni2+ [18, 19]. Fig. 1(c) shows the O-K XAS spectra of our LaNiO3 single crystal (blue) and the NiO reference compound (red). The O-K XAS spectra demonstrate even more clearly that there is no spectral feature from Ni2+ impurities and thus that our as-grown LaNiO3 single crystals are highly stoichiometric. This is further confirmed by thermogravimetric and inductively coupled plasma optical emission spectroscopy measurements, see Methods section and the Supplementary Note 2.

Temperature dependence of physical properties

In Fig. 2(a) we show the temperature dependence of the lattice parameters of our LaNiO3 single crystal that has been powdered and measured by means of powder X-ray diffraction. There is no clear indication for the presence of a structural anomaly which otherwise occurs readily in the other nickelates RNiO3 with smaller rare earths (R $=$ Pr-Lu, Y) when cooling through the metal-insulator transition [1, 2].

The electrical resistivity $\rho$ of our LaNiO3 single crystal that is shown in Fig. 2(b) is in the $\muup\Omega$ cm range, reaching $\sim$6 $\muup\Omega$ cm at low temperatures, i.e. distinctly lower than that of LaNiO3 powder samples [1, 2, 16, 17] or LaNiO3 thin films [14] or of crystals grown under 50 bar oxygen pressure [20]. Thus, our LaNiO3 single crystals are more metallic, see also Supplementary Note 4 and Supplementary Figure 4(a). Also the measurement of the Hall effect on the single crystal LaNiO3 (for temperatures between 2 K and 300 K) yields high carrier densities of about 4.8$\times$1028 m*-3*, corresponding to $\sim$2.7 (close to 3) electrons per formula unit, consistent with the 3+ valence of the Ni ions in LaNiO3. Up to $\sim$35 K the low-temperature behavior of the resistivity is Fermi-liquid-like, $\rho$($T$) = $\rho_{0}+AT^{n}$, with $\rho_{0}$ $\sim$ 6.45 $\muup\Omega$ cm, $A$ = 1.62$\cdot$10*-3* $\muup\Omega$ cm K*-2* and $n$ $\sim$ 2.0, see Fig. 2(b). The value of $A$ is of the same order as reported in literature [21]. A fit over the entire temperature range gives an exponent $n$ $\sim$ 1.50(1) with the residual resistivity $\rho_{0}$ $\sim$ 4.83$\muup\Omega$ cm, similar to the value of the fit at low temperatures. Most probably this behavior is a signature that LaNiO3 is close to a quantum critical point (cf. the results for PrNiO3 under pressure [21]).

Fig. 2(c) displays the temperature dependence of the magnetic susceptibility $\chi$ of our LaNiO3 single crystal, see also Supplementary Note 4 and Supplementary Figure 4(b). First of all, we notice that it shows a significantly smaller low-temperature upturn than reported previously for powder and ceramic samples [16, 17], which confirms unprecedented high quality of our single crystals. Surprisingly, $\chi$ exhibits an anomalous kink at $\sim$157 K which we take as an indication for a hitherto unknown antiferromagnetic transition in LaNiO3. That this anomaly is not simply caused by a signal from a tiny fraction of a magnetic impurity phase (which is so small that it is not visible in our powder X-ray diffraction measurements) can be excluded by our specific heat ($C_{p}$) measurements. As can be seen in Fig. 2(d) there is a small but clearly visible anomalous peak at $\sim$157 K in $C_{p}/T$. Also the resistivity data excludes that the anomaly in the susceptibility is caused by the presence of oxygen deficient LaNiO3 minority phase that becomes antiferromagnetic and insulating at low temperatures [22, 23]: we do not observe an upturn or a slowing down of the decrease in the resistivity on cooling. Moreover, our samples have resistivities in the $\muup\Omega$ cm range and have conductivities higher than reported so far. All these support the notion that the transition is an intrinsic bulk property and not due to an impurity phase. The low-temperature behaviour of the specific heat supports a Fermi-liquid type of the ground state of LaNiO3, $C_{p}$($T$) = $\gamma T+\beta T^{3}$, with $\gamma$ $\sim$ 17 mJ mol*-1* K*-2*, consistent with the value 18 mJ mol*-1* K*-2* reported before [17], and with $\beta$ = 1.87(5)$\cdot 10^{-4}$ J mol*-1K-4*, thus yielding a Debye temperature $\theta_{D}=(12\pi^{4}NR/5\beta)^{1/3}$ = 373 K, where $N$ is the number of atoms in the chemical formula and $R$ is the ideal gas constant.

Neutron scattering experiments

The availability of sizeable LaNiO3 single crystals also enabled us to study this intriguing system by means of neutron diffraction and inelastic neutron scattering. These experiments were performed at the Thales, IN12 and IN8 spectrometers at the ILL in Grenoble, France. Within elastic scans we were able to observe quarter-integer peaks at low temperatures, see Fig. 3(a). These quarter-integer peaks (in pseudocubic notation) resemble those found in powder neutron diffraction experiments within the insulating regime of RNiO3 [24]. The propagation vector observed for the insulating antiferromagnetic regime of RNiO3 amounts to (1/4 1/4 1/4) in pseudocubic notation (or (1/2 0 1/2) in orthorhombic notation) [2]. Moreover, the study of the temperature dependence of these quarter-integer peaks indicates an onset temperature which coincides with the magnetic ordering temperature T${}_{\text{N}}$ that we observed in susceptibility and specific heat measurements of LaNiO3, see Fig. 3(b).

The magnetic origin of these quarter-integer peaks in LaNiO3 could be also confirmed by polarized neutrons at the IN12 spectrometer. In Fig. 4(a,b) $L$-scans across two quarter-integer peaks are shown for the three spin flip (SF) channels and for a non-spin-flip channel. Only in the spin flip channels neutron scattering intensities $\sigma_{x_{i}x_{i}}$ can be detected. This unambiguously shows the magnetic nature of these quarter-integer reflections in LaNiO3. Although we can not detect a symmetry lowering from our high resolution powder X-ray diffraction measurements

• see Fig. 2(a) - the pseudocubic propagation vector for rhombohedral LaNiO3 is the same as the pseudocubic propagation vector for the orthorhombic insulating nickelates RNiO3 ($R$ = Pr-Lu, Y). Based on the magnetic symmetry analysis for the high symmetry cubic structure with space group Pm$\mathit{\overline{3}}$m and for the propagation vector (1/4 1/4 1/4) a helical magnetic structure with moments spiraling perpendicular to the propagation vector is consistent with our single crystal neutron data, see Supplementary Note 5, the Supplementary Tables 2 & 3 and the Supplementary Figure 6. The size of the ordered moment amounts to $\sim$0.3 $\muup_{\mathrm{B}}$ which indicates that magnetism in LaNiO3 is a bulk property and does not originate from a tiny (insulating) impurity phase. This small magnetic moment might explain why paramagnetic properties have been reported for LaNiO3 in the past [1, 2, 16, 17].

We also have been able to study and observe the magnetic excitations in LaNiO3 by means of inelastic neutron scattering thereby also providing further support that the antiferromagnetism is a bulk property. As can be seen in Fig. 5, magnetic excitations are clearly visible up to at least 12 meV. With increasing energy transfer the magnetic peaks become somewhat broader and damped which is indicative for the appearance of fluctuations in this itinerant antiferromagnetic system. These fluctuations could be also responsible for a reduced ordered moment of $\sim$0.3 $\muup_{\mathrm{B}}$ in LaNiO3.

Discussion

LaNiO3 appears to be a rare case of an antiferromagnetic and metallic transition metal oxide with a fully 3D crystal and electronic structure. Other systems like (La,Sr)3Mn2O7 and Ca3Ru2O7 have a lower electronic and structural dimensionality where the antiferromagnetic order is resulting from a stacking of ferromagnetic layers [25, 26]. So far, the chromate system $A$CrO3 ($A$ = Ca, Sr) and very recently, RuO2 are known to be the only other intrinsically antiferromagnetic and metallic transition metal oxides with such a fully 3D crystal and electronic structure [27, 28]. The systems CaCrO3 and LaNiO3 have in common that the oxidation state of the transition metal ion is very high. Thus, the oxygen $2p$ to transition metal $3d$ charge transfer energy here is apparently negative [5, 12, 29] resulting in extreme $2p$-$3d$ covalency where the presence of holes in the oxygen band can effectively prevent the opening of the conductivity gap and at the same time mediate strongly the magnetic exchange interactions. However, in contrast to CaCrO3 [27] the rare earth nickelate LaNiO3 in single crystalline form is much more metallic and shows conductivities in the $\muup\Omega$ cm range. This is probably also true for RuO2 [28]. Unique for LaNiO3 is that it is close to the insulating phase of RNiO3 (R $=$ Pr-Lu, Y) indicating the importance of strong correlation effects, making it rather exceptional among all transition metal oxides.

Although charge order - which plays a very important role in nickelates RNiO3 with small rare earth ions R - cannot be observed in our present XRD and neutron diffraction measurements - see Fig. 4(c) - the presumably associated symmetry lowering of the structure that goes along with the antiferromagnetic order may produce only very weak new peaks in diffraction experiments. Some hints for this may in fact be found in a recent pair distribution function study using neutron diffraction [30]. Nevertheless, with the charge order effects being so weak, one could infer that LaNiO3 is perhaps better described using Fermi surface arguments [7, 28] while the other insulating nickelates RNiO3 with the smaller $R$-ionic sizes and with higher charge ordering temperatures can be more intuitively understood in terms of local charge or bond disproportionations [5, 6, 8, 9, 11, 12]. Note, that recent ab-initio calculations [32] reproduced metallic and antiferromagnetic state of LaNiO3, and also predicted charge (or rather bond) disproportionation, whose magnitude is however below our experimental detection limit. One needs special dedicated experiments to probe for the eventual symmetry lowering with the appearance of inequivalent Ni ions. However, our main conclusion - the existence of antiferromagnetic ordering in highly metallic single crystals of LaNiO3, is confirmed by these calculations. According to our findings we now also present a tentative RNiO3 phase diagram in Fig. 6.

Methods

Chemical synthesis

The LaNiO3 single crystal was grown under high oxygen pressures of 130-150 bar with a growth speed of 6-7 mm h*-1* in a one-mirror furnace from Scidre that was equipped with a 5000 W Xe-lamp and with counter-rotation of feeding and seeding rods - see Supplementary Movie 1. The seeding and feeding rods were synthesized by pressing appropriate mixtures of La2O3 and NiO into rods with roughly 10 cm length and 6 mm diameter. These rods were sintered at 800∘C-1000∘C for several days. The temperature of the melting zone was measured in-situ by means of a pyrometer and is close to 1800∘C at $\sim$140 bar p${}_{\mathrm{O_{2}}}$.

X-ray diffraction

Laue diffraction measurements were performed on a multiwire real-time back-reflection Laue camera from Multiwire Laboratories; see also the Supplementary Note 1 and Supplementary Figure 1.

Powder X-ray diffraction measurements were performed on a Bruker D8 Discover A25 diffractometer which is equipped with a Johansson monochromator for Cu K${}_{\alpha_{1}}$ radiation. A closed cycle helium cryostat (Phenix of Oxford Cryosystems) was used for temperature-dependent measurements.

Single crystal X-ray diffraction measurements have been performed on a twined single crystal of LaNiO3 using a Bruker D8 VENTURE single crystal X-ray diffractometer equipped with a bent graphite monochromator for Mo $K_{\alpha}$ radiation (about 3$\times$ intensity enhancement) and a Photon CMOS large area detector. A crystal with roughly 20 $\muup$m diameter has been measured and a multi-scan absorption correction has been applied to the data (minimum and maximum transmission: 0.6184 and 0.7519 respectively). 12328 (observed) reflections (H: -12 $\rightarrow$ 13, K: -13 $\rightarrow$ 9 and L: -32 $\rightarrow$ 31) have been collected with an internal R-value of 6.70%, a redundancy of 36.8 and with 98.85% coverage up to 2$\Theta_{max}$ = 123.6∘. For the refinement the Jana2006 program package was used. The Goodness of fit of our crystal structure refinement amounts to 1.93 and the R- and weighted R-values amount to 2.29%  and 5.79% respectively. The refinement of the Ni and La occupancies yields an almost stoichiometric composition: La*0.995(12)Ni1.000(14)*O3. The structural parameters are listed in Supplementary Table 1 and the crystal structure is visualized in the Supplementary Figure 3.

Composition determined by ICP and TG measurements

Inductively coupled plasma optical emission spectroscopy (ICP-OES) measurements yields the following composition of LaNiO3: La1.001(11)Ni0.999(4)O3+δ. Moreover, thermogravimetric measurements confirm an almost perfectly stoichiometric oxygen content with $\delta=-0.002$.

X-ray absorption spectroscopy

XAS measurements have been performed at the 08B beamline of the National Synchrotron Radiation Research Center (NSRRC), Taiwan.

Magnetization measurements

The magnetic properties were studied using a Quantum Design Inc. MPMS-5XL SQUID magnetometer.

Electrical conductivity

The measurements of electrical resistivity and specific heat were carried out using a four-probe and a standard thermal relaxation calorimetric method in a Quantum Design Inc. Physical Property Measurement System (PPMS).

Neutron measurements

Unpolarized neutron measurements have been performed at the IN8 and Thales Spectrometers at the ILL in Grenoble, France. For the elastic and inelastic measurements on the Thales spectrometer a pyrolytic graphite (PG) (002) monochromator and analyzer as well as a velocity selector were used ($k_{\mathrm{f}}$ = 1.8 Å*-1*). On the IN8 spectrometer very first quick inelastic scans have been made using a Si monochromator and a Flatcone analyzer array. Polarized neutron measurements have been performed at the IN12 Spectrometer at the ILL in Grenoble, France. A velocity selector was used for choosing the incident neutron wavevector $k_{\mathrm{i}}$ of 2.25 Å*-1*. A horizontally and vertically focussing PG (002) monochromator and Heusler (111) analyser have been used. The incident neutron beam was polarized by a transmission polarizer (cavity) in the neutron guide and the measured flipping ratio amounts to 22.2.

Data availability

The authors declare that the data supporting the findings of this study are available within the paper and its supplementary information files.

Acknowledgement

We thank O. Stockert, G. A. Sawatzky and A. Fujimori for valuable discussions. We thank M. Schmidt and his team for thermogravimeteric measurements. We thank G. Auffermann and her team for ICP measurements. We acknowledge A. Todorova and G. Ryu for support on preparatory work. We thank the team of H. Borrmann for initial room-temperature powder X-ray diffraction measurements used for phase analysis. The research in Dresden and Cologne is partially supported by the the Deutsche Forschungs-Gemeinschaft through FOR 1346.

Competing Interests

The authors declare that they have no competing financial interests.

Author Contributions

Conceiving experiments and project management: A. C. K.; physical experiments: H. G., Z. W. L., L. Z., Z. H., C. F. C., C. Y. K., A. T., W. S., A. P. T. W. P., O. S. and A. C. K.; chemical synthesis: A. C. K., H. G.; interpretation and manuscript writing: A. C. K., D. I. K. and L. H. T.

H. G. and Z. W. L. are equally contributing authors.

Correspondence

Correspondence and requests for materials should be addressed to A.C.K. (email: Alexander.Komarek@cpfs.mpg.de).

Supplementary Note 1

Laue X-ray diffraction

We have performed Laue diffraction measurements for different spots on one of our LaNiO3 single crystals. As can be seen in Supplementary Figure 1, for each studied side (left and right side) the diffraction patterns are the same for all positions (spots) over the entire length of the crystal. This indicates, that our crystal is a single crystal. Also our single crystal has no NiO impurities visible by eyes in powder X-ray diffraction measurements (see Fig. 1 in the main text).

Supplementary Note 2

Thermogravimetric measurements

Supplementary Figure 2 shows a typical thermogravimetric (TG) measurement of one of our nickelate samples that was heated to 900*∘C in a flow (100ml/min.) of 20% H2 and 80% Ar gas. The oxygen concentration amounts to $\sim$3.005 according to the graphical analysis and $\sim$2.998 as derived from weighing initial and final masses before and after the reduction of the sample. Both values basically indicate the absence of Ni2+* impurities in our LaNiO3 crystals. The small difference between these values is comfortably within the typical error of TG measurements and can be attributed to the impact of the gas flow which is affecting the buoyant force that has non-negligible influence on the weighing of the sample mass during the reaction.

Supplementary Note 3

Crystal structure

Moreover, we determined the atomic positions of LaNiO3 by means of single crystal X-ray diffraction. The structural parameters are listed in Supplementary Table 1 and the (obtained) crystal structure is visualized in Supplementary Figure 3.

Supplementary Note 4

Physical properties

We also re-measured the magnetic susceptibility and electrical resistivity for six different single crystals from three batches that we have grown, see Supplementary Figure 4.

If the crystal is a single crystal and not polycrystalline it shows the kink in the magnetic susceptibility and a low value of the electrical resistivity, see Supplementary Figure 4. However, if the crystal is not single crystalline, we also observe a magnetization without any kink, similar as in Ref. [20]. Therefore, we could imagine that magnetic ordering only appears in LaNiO3 single crystals where domain sizes are very large and samples are very pure.

Finally, there seems to be no field dependence (up to 9 T) of the electrical resistivity of LaNiO3 - see Supplementary Figure 5. Since the single crystals are twined (and pseudocubic) also no direction dependence on the magnetic field direction can be expected.

Supplementary Note 5

Magnetic structure

Since our single crystal is fully twinned we decided to make a magnetic symmetry analysis for the high symmetry cubic (undistorted perovskite) structure with space group Pm$\mathit{\overline{3}}$m. The integrated intensities of the (0 0 1) and (1 1 0) nuclear peaks and (0.25 0.25 0.75), (0.75 0.75 0.25) and (0.75 0.75 0.75) magnetic peaks have been obtained from Gaussian fits to the corresponding Q scans (note, that an analyzer was used in our triple-axis measurements,) and normalized by the monitor and by the attenuation factor (plexi: 7.5) for the nuclear reflections. The scale factor was determined from the nuclear intensities and fixed for the subsequent magnetic structure refinement. The magnetic structure was determined by irreducible representation (IR) analysis for space group $Pm\bar{3}m$ and for a propagation vector (0.25 0.25 0.25). The reducible magnetic representation for the Ni ions at the $1b$ site is decomposed into two IRs as:

[TABLE]

The basis vectors for the IRs are listed in Supplementary Table 2. and a comparison of observed and calculated intensities is listed in Supplementary Table 3. For the cubic symmetry the appearance of the (3/4 3/4 3/4) reflection excludes the model with longitudinal modulation of the magnetic moments (see Supplementary Figure 6(a)) which is also reflected by the large R- and weighted R values that amount to 65.0% and 69.4%.

The model with helical arrangement of the magnetic moments perpendicular to the propagation vector (see Supplementary Figure 6(b)) is consistent with our data (R- and weighted R values amount to 16.3% and 17.7% respectively) and yields an ordered moment of about 0.3 $\muup_{\mathrm{B}}$.