betaFeedback

WorksPhysics and Astronomy

75^{75}As NMR study of the antiferromagnetic Kondo lattice compound CeNiAsO

Abstract

We revisit the magnetic properties of the antiferromagnetic Kondo lattice CeNiAsO by 75^{75}As nuclear magnetic resonance measurements. Our results confirm two successive antiferromagnetic transitions of Ce moments at TN1=9.0(3)T_{N1}=9.0(3) K and TN2=7.0(3)T_{N2}=7.0(3) K. Incommensurate and commensurate antiferromagnetic orders are suggested for TN2<T<TN1T_{N2}<T<T_{N1} and T<TN2T<T_{N2} respectively, consistent with previous neutron and muon experiments. A Knight shift anomaly, characterized by the failure of K(T)−χ(T)K(T)-χ(T) scaling, is observed below T∗∼15T^*\sim15 K, which gives a measure of the onset of coherent c−fc-f correlations. This energy scale is further confirmed by the spin-lattice relaxation rate (1/T11/T_1). The analysis of spin dynamics also reveals a quasi-two-dimensional character of spin fluctuations in CeNiAsO. This work paves the way for further 75^{75}As nuclear magnetic resonance studies under pressure.

Introduction

The ZrCuSiAs-structured rare-earth oxypnictides not only boosted a new wave of research boom on high-TcT_{\mathrm{c}} superconductors[1, 2, 3, 4], but also provided a novel playground for investigating strongly correlated electronic effects and quantum critical phenomena. Depending on the specific chemical element in the TmT_{\mathrm{m}} and PnP_{\mathrm{n}} sites, the family of CeTmPnT_{\mathrm{m}}P_{\mathrm{n}}O (TmT_{\mathrm{m}}=transition metal, PnP_{\mathrm{n}}=P, As) displays a wide spectrum of physical properties. The member CeFeAsO shows a spin-density-wave (SDW) type antiferromagnetic (AFM) transition of Fe-3d3d electrons near 140 K, and superconductivity can be induced by appropriate F−^{-} doping when the SDW order is suppressed[2]. At a much lower temperature ∼4\sim4 K, the Ce local moments also order antiferromagnetically[2, 5, 6, 7]. In contrast, the analogue compound CeFePO is a nonsuperconducting heavy-fermion metal with pronounced ferromagnetic correlations[7, 8, 9, 10, 11]. For TmT_{\mathrm{m}}=Co, strong interplay between Co-3d3d and Ce-4f4f magnetism was observed in both CeCoPO[12] and CeCoAsO[13], where the Co-3d3d electrons show ferromagnetic orders near 75 K, while Ce-4f4f electrons are on the border to magnetism with enhanced Sommerfeld coefficient. The appearance of 3d3d magnetism in these compounds complicates the study of the 4f4f physics.

Unlike the cases for TmT_{\mathrm{m}}=Fe or Co, the 3d3d electrons in the Ni counterpart were shown to be nonmagnetic[14, 15]. This makes CeNiAsO an ideal candidate to explore the heavy-fermion properties of the 4f4f electrons. Our previous studies based on magnetic susceptibility and specific heat measurements revealed that the Ce moments in CeNiAsO undergo two successive AFM transitions at TN1≈9.3T_{\mathrm{N1}} \approx9.3 K and TN2≈7.3T_{\mathrm{N2}} \approx7.3 K [see also in Fig. 1(c)]. The Sommerfeld coefficient was found to be ∼203\sim203 mJ/mol·K, indicative of substantial 3d3d-4f4f hybridization. Applying physical pressure to CeNiAsO or substituting As with smaller P ions suppresses the AFM orders and leads to a quantum critical point (QCP) at pc≈0.65p_{\mathrm{c}} \approx0.65 GPa or xc≈0.4x_{\mathrm{c}} \approx0.4, respectively[16]. An abrupt change in the Hall coefficient was revealed at pcp_{\mathrm{c}}, characteristic of local (or Kondo-breakdown) quantum criticality[17, 18, 19, 20] around which the suppression of order parameter is also accompanied with breakdown of Kondo effect and hence a Fermi-surface reconstruction. Recently, magnetic structures of the two AFM ordered states were reported by Wu et al by a combination of neutron scattering and μ\muSR experiments[21]. They found that for T<TN1T < T_{\mathrm{N1}}, a second order phase transition yields an incommensurate SDW order with a wave vector q=(0.44,0,0)\mathbf{q} = (0.44, 0, 0); while for T<TN2T < T_{\mathrm{N2}}, the magnetic structure evolves into a coplanar commensurate AFM order with q=(0.5,0,0)\mathbf{q} = (0.5, 0, 0). The schematic of these magnetic structures is shown in Figure 1(a). Furthermore, their μ\muSR measurements on CeNiAs1−x_{1-x}Px_xO manifested that the commensurate order exists only for x≤0.1x \leq0.1, therefore the QCP is probably between the incommensurate SDW and a paramagnetic phase[21], which is not expected in the general framework of local quantum criticality. Additional microscopic local experiments are invited to confirm the magnetic properties of CeNiAsO.

Image not extracted

Figure 1. (a) Local environment of As in CeNiAsO. The red arrows denote Ce moments, which order in an incommensurate SDW with q=(0.44,0,0)q=(0.44,0,0) for TN2<T<TN1T_{N2}<T<T_{N1}, and a commensurate AFM with q=(0.5,0,0)q=(0.5,0,0) for T<TN2T<T_{N2}[21]. In SDW order, Ce moments might have a small zz component that is not shown here. The blue arrows indicate the dipolar field at As sites calculated for specific magnetic structures. (b) Simulated magnetization based on CEF theory shows c as the hard axis, and ab as the easy plane. The open symbols are experimental results on aligned sample, red squares - B0⊥cB_0 \perp c, blue circles - B0∥cB_0 \parallel c. The inset depicts CEF splitting. (c) Temperature dependence of magnetic susceptibility measured at B0=10B_0=10 T, with field applied parallel and perpendicular to c, respectively. The upper inset shows d(Tχ)/dTd(T\chi)/dT for non-aligned powder measured at 0.1 T (left), and C/TC/T (right); the lower inset shows d(Tχc)/dTd(T\chi_c)/dT for aligned powder measured at 10 T. The two magnetic transitions at TN1T_{N1} and TN2T_{N2} are clearly seen. Both TN1T_{N1} and TN2T_{N2} are relatively lower when under a high field of 10 T.

Herein, we employed 75As^{75}\mathrm{As} nuclear magnetic resonance (NMR) measurements on field-aligned CeNiAsO powders, and both static and dynamic spin susceptibilities are investigated. NMR spectra confirm two kinds of AFM orderings below TN1=9.0(3)T_{\mathrm{N1}} = 9.0(3) K and TN2=7.0(3)T_{\mathrm{N2}} = 7.0(3) K, respectively. Both NMR shift (KK) and spin-lattice relaxation rate (1/T11/T_1) manifest anisotropic magnetic properties. The hyperfine interaction and the influences of the Kondo effect on KK and 1/T11/T_1 are also discussed.

Experimental details

Polycrystalline CeNiAsO sample of high purity was synthesized by solid-state reaction as described previously[15]. The sample quality was verified by powder X-ray diffraction (XRD, PANalytical) with Cu Kα\alpha radiation, and no impurity trace could be detected. Specific heat and low-field (0.1 T) magnetic susceptibility measurements confirm TN1≃9.3T_{N1}\simeq9.3 K and TN2≃7.3T_{N2}\simeq7.3 K [Upper inset to Figure 1(c)], consistent with our original reports. 75^{75}As NMR measurements were performed on field-aligned powders. The polycrystalline sample was thoroughly ground into powders in a glove box, and mixed with Stycast 1266 with a weight ratio ≈0.2\approx0.2 [22]. The mixture was placed in a strong magnetic field of 14 T at 300 K in a physical property measurement system (PPMS-14, Quantum Design) and held for 12 h before the Stycast was cured. Because CeNiAsO is an easy-plane (ab-plane) system (see below), the field alignment was carried out on a rotating sample plate that rotates at a slow speed (∼5\sim5 rpm). 75^{75}As (75γn=7.2919^{75}\gamma_n=7.2919 MHz/T, I=3/2I=3/2) NMR spectra were recorded in a stepped frequency sweep spin-echo method at an external field B0∼10B_0\sim10 T. The measurements were made for both B0∥cB_0\parallel c and B0⊥cB_0\perp c. The precise value of B0B_0 was detected by the 63^{63}Cu shift in a second, pre-calibrated coil nearby. Spin-lattice relaxation rate (1/T11/T_1) was measured in a standard inversion recovery method on the central (12↔−12\frac{1}{2}\leftrightarrow-\frac{1}{2}) transition, and T1T_1 was extracted by fitting the recovery curve to

M(t)=M(∞){1−2f[110exp⁡(−tT1)+910exp⁡(−6tT1)]},(1)M(t)=M(\infty)\left\{1-2f\left[\frac{1}{10}\exp\left(-\frac{t}{T_1}\right)+\frac{9}{10}\exp\left(-\frac{6t}{T_1}\right)\right]\right\}, \tag*{(1)}

where M(∞)M(\infty), ff and T1T_1 are fitting parameters. Anisotropic magnetic susceptibility of CeNiAsO was also measured on the aligned sample with B0=10B_0=10 T in a PPMS equipped with a vibrating sample magnetometer (VSM).

Results and Discussion

CEF effect and magnetic anisotropy

Before presenting the NMR results, it is necessary to discuss briefly the crystalline electric field (CEF) effect in CeNiAsO. The magnetic anisotropy of a Ce-containing compound is usually described by the CEF effect whose Hamiltonian in C4vC_{4v} point group is

H^CEF=B20O^20+B40O^40+B44O^44,(2)\hat{H}_{CEF}=B_2^0\hat{O}_2^0+B_4^0\hat{O}_4^0+B_4^4\hat{O}_4^4, \tag*{(2)}

where BlmB_l^m (l=2,4;m=0,4l=2,4; m=0,4) are CEF parameters, and O^lm\hat{O}_l^m are Stevens operators [23]. According to Wu’s inelastic neutron scattering results[21], B20=76.7B_2^0=76.7 K, B40∼−1B_4^0\sim-1 K, B44≈10.4B_4^4\approx10.4 K. It is found that the j=5/2j=5/2 multiplet of Ce3+^{3+} splits into three Kramers doublets, with the first and second excited doublets sitting at ∼226\sim226 and 790 K above the ground states, as shown in Table I and the inset to Figure 1(b). Simulated field dependent magnetizations at 300 K for field parallel and perpendicular to c-axis are provided in the mainframe of Figure 1(b). One clearly sees McM_c much smaller than MabM_{ab}, suggesting ab-plane as the magnetic easy-plane. This confirms the validity of the way we aligned the powdered sample. Based on this simulation, the gg-factor anisotropy for the ground-state doublet is found to be g⊥c/g∥c=3g_{\perp c}/g_{\parallel c}=3.

B20=26.7B_2^0=26.7 K, B40=−1B_4^0=-1 K, B44=10.4B_4^4=10.4 K
EiE_i (K)∣Ψi⟩\lvert\Psi_i\rangle
0∣±1/2⟩\lvert\pm1/2\rangle
226∓0.6621∣±5/2⟩∓0.7494∣∓3/2⟩\mp0.6621\lvert\pm5/2\rangle\mp0.7494\lvert\mp3/2\rangle
790∓0.7494∣±5/2⟩∓0.6621∣∓3/2⟩\mp0.7494\lvert\pm5/2\rangle\mp0.6621\lvert\mp3/2\rangle

Table I. CEF parameters, energy levels and wave functions in CeNiAsO (C4vC_{4v}) at zero magnetic field. The CEF parameters are from Ref. [21].

The temperature dependence of magnetic susceptibility measured under an external field B0=10B_0=10 T is shown in Figure 1(c). More technical detail about the data analysis of magnetic susceptibility in aligned powders can be found in Ref. [24]. Indeed, χc(T)\chi_c(T) is much smaller than χab(T)\chi_{ab}(T) in the full window 2≤T≤3002 \le T \le300 K. The two antiferromagnetic transitions are visible in the d(Tχc)/dTd(T\chi_c)/dT plot as depicted in the inset to Figure 1(c). Note that the transitions defined by such are relatively lower than in our original report[15], because the large field 10 T likely suppresses both AFM orders. These susceptibility results will be compared with the NMR shifts in the following subsections.

75As^{75}\mathrm{As} NMR spectra

The local environment of 75As^{75}\mathrm{As} in CeNiAsO can be viewed also in Figure 1(a). Each As is surrounded by four nearest-neighbor Ni (nonmagnetic, 2.365 Å) and four nearest-neighbor Ce (3.230 Å). Besides, there are two next-nearest-neighbor Ce sitting right above (4.055 Å) and below (4.058 Å) As. We will get back to this later on.

The full 75As^{75}\mathrm{As} NMR spectra of aligned CeNiAsO under B0≃10B_0 \simeq10 T are shown in Figure 2, measured at 10 K (red) and 2 K (blue), respectively. For B0∥cB_0 \parallel c and T=10T=10 K, three equally separated peaks are resolved at 73.68, 65.96 and 81.46 MHz, corresponding to the central (1/2↔−1/21/2 \leftrightarrow-1/2) and satellite (±3/2↔±1/2\pm3/2 \leftrightarrow\pm1/2) transitions of the I=3/2I=3/2 75As^{75}\mathrm{As} nucleus. The nuclear quadrupolar splitting frequency is extracted νQ=7.75\nu_Q=7.75 MHz at 10 K, and remains essentially unchanged down to 2 K, deep inside the AFM ordered phase, manifesting that the principle axis of the electric field gradient (EFG) does not change with AFM transitions. The value of νQ\nu_Q is ∼22%\sim22\% smaller than that in CeFeAsO (9.9 MHz)[25], suggestive of a drastic change in the charge density distribution around As and hence the character of Fe/Ni-As bonds in these two compounds, which was confirmed previously by band structure calculations[14]. Another salient feature of the spectrum at 2 K is that each peak splits into two peaks, and the splittings are the same among central and satellite transitions, consistent with an AFM ground state at low temperature. We should point out that a small peak is seen at 72.08 MHz [denoted by a asterisk in Figure 2(a)], which should be attributed to a small amount of non-aligned powders. This is demonstrated by the central peak for B0⊥cB_0 \perp c observed exactly at the same frequency [Fig. 2(b)].

Image not extracted

Figure 2. (a) The 75As^{75}\mathrm{As} frequency-sweep NMR spectra of CeNiAsO at 10 K (red) and 2 K (blue) with the magnetic field B0≃10B_0 \simeq10 T parallel with cc. The arrows denote all the central and satellite transitions of the I=3/2I=3/2 75As^{75}\mathrm{As} nucleus. The asterisk indicates a small amount of non-aligned powder. (b) Ibid, but for field perpendicular to cc.

Due to the symmetry, we assume VzzV_{zz} is along c-axis, and Vxx=Vyy=−12VzzV_{xx}=V_{yy}=-\frac{1}{2}V_{zz}, where VxxV_{xx}, VyyV_{yy} and VzzV_{zz} represent the EFG for the three principal axes. Therefore, νQ′=νQ/2\nu'_Q=\nu_Q/2 is expected for B0⊥cB_0 \perp c. Experimentally, we get νQ′=3.9(1)\nu'_Q=3.9(1) MHz, as expected [Fig. 2(b)]. At 2 K, we also see two sets of peaks due to the AFM ordered. For both field orientations, the central peak splits; it, therefore, is reasonable to infer that the internal field at As sites has both components ∥c\parallel c and ⊥c\perp c. To show this more directly, we calculated the dipolar fields at each As site exerted by Ce local moments and display them in Figure 1(a) by the blue arrows. In this calculation, the magnetic structure parameters suggested by neutron scattering and μ\muSR measurements were adopted[21]. At 2 K, the calculated dipolar field is (−0.0146,0,0.0252)(-0.0146, 0, 0.0252) T. Since in our sample the powders are randomly oriented in the ab\mathbf{ab} plane, we expect there is some distribution of internal field, and this may explain why the split peaks for B0⊥cB_0 \perp c are broader and with longer tail than for B0∥cB_0 \parallel c. However, we note that the two well separated AFM peaks (B0⊥cB_0 \perp c) can hardly be fit to a powder pattern with Gaussian broadening, cf Figure 3(d). One possibility might be that the in-plane magnetic anisotropy is not strong, so the application of 10 T field can reorient the moments in the AFM order.

Image not extracted

Figure 3. 75As{}^{75}\mathrm{As} frequency-sweep NMR spectra of CeNiAsO at various temperatures under B0≈10B_{0} \approx10 T. (a-b) B0∥cB_{0} \parallel c. (c-d) B0⊥cB_{0} \perp c. The dashed lines indicate the Larmor frequency. For clarity, the spectra for T>TN1T > T_{N1} shown in panels (a) and (c) are amplified by a factor of TT. The solid red line in (d) is a powder-pattern fitting with Gaussian broadening.

In order to get more detail across the magnetic transitions, we now focus on the central peaks at different temperatures in Figure 3. For better clarity, the spectra are vertically offset, and the position of Larmor frequency ν0=75γnB0\nu_0=^{75}\gamma_nB_0 is marked by a dashed line. Magnetic anisotropy can be clearly seen by the opposite movement of peaks at lower temperature, cf Figure 3(a) and (c). For B0⊥cB_0 \perp c, at 300 K, the resonance peak is on the right-hand side of ν0\nu_{0}, and it gradually drifts to the left-hand side upon cooling, indicative that a sign change in NMR shift probably occurs. See below.

Just below TN1T_{N1} and with B0∥c\mathbf{B}_{0} \parallel c, we see a shoulder appear on each side of the central peak [cf 8.5 K in Figure 3(b)]. As TT decreases, the shoulders evolve into two broad peaks, and meanwhile the height of original central peak in the paramagnetic state shrinks. Since NMR is a microscopic local probe, such an evolution probably indicates phase segregation between the emergent AFM phase and the paramagnetic phase. These patterns can be fit to three Gaussian peaks, and one finds that the two AFM peaks are broad, implying distribution of the internal field in this AFM order, which might be understood by an incommensurate SDW order as proposed by neutron scattering and μ\muSR experiments[21]. Near 7 K, the spectra are very complicated, and are hard to be fitted to multi-peak functions.

the spectra are very complicated, and are hard to be fitted to multi-peak functions. Below 6.8 K, two new well separate peaks can be resolved, and this should be associated with the lower-temperature commensurate AFM order suggested by Wu et al [21]. The window 6.8-7.3 K probably is a region where these two AFM phases coexist. Such a phase coexistence is also observed for B0⊥c\mathbf{B}_{0} \perp c, in the similar temperature range [Figure 3(d)]. It is interesting that for this field orientation, we do not see a clear trace of phase segregation just below TN1T_{N1}, the reason for which is still unknown.

In spite of some unclear features, two conclusions can be drawn from the temperature dependent 75As{}^{75}\mathrm{As} spectra. First, two different AFM orders appear at low temperature, an incommensurate one between TN1T_{N1} and TN2T_{N2}, and a lower-temperature phase that seems more likely a commensurate type. Second, for both types of AFM orders, we see peak splittings in both B0∥c\mathbf{B}_{0} \parallel c and B0⊥c\mathbf{B}_{0} \perp c. This suggests that the internal fields at As sites have both in-plane and out-of-plane components. We will revisit this issue later on.

75As NMR shifts and internal field

NMR shifts (KK) of 75As{}^{75}\mathrm{As} can be extracted from the central transition lines. Taking account the correction from the second order quadrupolar effect[26],

ν∥=75γnB0(1+Kc),(3)\nu_{\parallel} = {}^{75}\gamma_{n}B_{0}(1+K_{c}), \tag*{(3)}
ν⊥=75γnB0(1+Kab)+3νQ216⋅75γnB0,(4)\nu_{\perp} = {}^{75}\gamma_{n}B_{0}(1+K_{ab}) + \frac{3\nu_{Q}^{2}}{16\cdot{}^{75}\gamma_{n}B_{0}}, \tag*{(4)}

and the derived KcK_{c} and KabK_{ab} are displayed in Figure 4(a) as functions of TT. Both Kc(T)K_{c}(T) and Kab(T)K_{ab}(T) can be well fit to a Curie-Weiss formula in the TT range 20-300 K, with the resultant Weiss temperatures θWc=−23(3)\theta_{W}^{c} = -23(3) K and θWab=−11(2)\theta_{W}^{ab} = -11(2) K, confirming AFM-type exchange couplings between Ce moments. For both field orientations, 75K(T){}^{75}K(T) gradually deviates from Curie-Weiss’s law below ∼15\sim15 K, probably as a consequence of Kondo effect.

Image not extracted

Figure 4. (a) 75^{75}As NMR shifts in paramagnetic phase as functions of TT, for field parallel and perpendicular to c. The error bars are determined by the half width at half maximum of NMR peaks. The solid lines are the Curie-Weiss fits to 75Kab(T)^{75}K_{ab}(T) and 75Kc(T)^{75}K_c(T) in the paramagnetic phase, leading to θWab=−11(2)\theta_W^{ab}=-11(2) K and θWc=−23(3)\theta_W^c=-23(3) K, respectively. (b) 75K^{75}K vs. χ\chi plots with TT as an implicit parameter.

In metals, NMR shifts usually contain multiple contributions,

K(T)=Ko+KS(T),(5)K(T) = K^{o} + K^{S}(T), \tag*{(5)}

where the temperature independent KoK^{o} arises from the orbital shift, and KS(T)=Ahfχ(T)K^{S}(T) = A_{hf}\chi(T) is the spin (Knight) shift with AhfA_{hf} being the hyperfine coupling constant. A K−χK-\chi plot is given in Figure 4(b). For B0∥c\mathbf{B}_{0} \parallel c, this plot gives Ahf,c=0.86(2)A_{hf,c} = 0.86(2) T/μB\mu_{B}. Below ∼15\sim15 K, the linear relation between K(T)K(T) and χ(T)\chi(T) gradually collapses. In the Kondo lattice compounds, this is usually termed a Knight shift anomaly[27, 28]. This provides an estimate to the coherent Kondo scale T∗∼15T^{*} \sim15 K. We should mention that such a Knight shift anomaly has been observed in a large variety of heavy-fermion materials, and multiple theoretical models have been proposed to attribute the anomalous hyperfine coupling either to a temperature dependence due to Kondo screening[29] or to different occupations of CEF levels[30]. A more recent hypothesis is a phenomenological two-fluid model[31, 32, 33, 34] in which the hyperfine couplings to localized ff-electron (Fluid 1) spins and itinerant heavy-electron (Fluid 2) spins are of different values. As the weight of the latter starts to increase significantly below T∗T^*, K(T)K(T) deviates from its linear relation to χ(T)\chi(T). A similar trend is also seen for B0⊥c\mathbf{B}_0 \perp c near the same temperature, but the feature is relatively weaker. In fact, well above T∗T^*, another more prominent shift anomaly is also observed in this field direction at around 110 K. Since this temperature scale is about 50% of the first CEF splitting energy, it therefore is likely to be caused by CEF depopulation effect[30]. An average value of the hyperfine coupling constant in this field direction is estimated Ahf,ab=−0.42(9)A_{\mathrm{hf},ab}=-0.42(9) T/μB\mu_B. It should also be noted that in CeNiAsO, the coherent temperature T∗T^* is rather close to the single-ion Kondo scale TKT_K estimated from our previous entropy analysis[15].

Internal fields determined from the AFM-induced splitting (Figure 3) are summarized in Figure 5. Two prominent features are identified. (i) For both orientations, discontin-

Image not extracted

Figure 5. Internal field at 75^{75}As sites as a function of TT. (a), B0∥c\mathbf{B}_0 \parallel c. (b), B0⊥c\mathbf{B}_0 \perp c. The hatched areas indicate phase coexistence. The abbreviations are: CA = commensurate antiferromagnetic, ICA = incommensurate antiferromagnetic, PM = paramagnetic.

uities are clearly visible in Bint(T)B_{\mathrm{int}}(T) around TN2T_{N2}, confirming a modification in magnetic order takes place. (ii) The magnitude of BintB_{\mathrm{int}} is about 6–8 times that of the dipolar field mentioned above (at 2 K). This manifests that the internal field at As sites is dominated by the transferred hyperfine field. Unlike the long-range dipolar interaction, the transferred hyperfine interaction is short-range and relies on orbital hybridization[35]. The ratio Ahf,c/Ahf,abA_{\mathrm{hf},c}/A_{\mathrm{hf},ab} now ∼2\sim2 reminds us that such a hybridization is through the As-4pz4p_z orbital. A tentative try is to consider the four nearest-neighbor Ce sites only, which has the similar symmetry as in the case of LaFeAsO and BaFe2_2As2_2 (there each As has four nearest neighbor Fe moments). Following the symmetry analysis in Ref. [35],

Btrh=∑i=14Ai⋅mi,(6)\mathbf{B}_{\mathrm{trh}}=\sum_{i=1}^{4}\mathbf{A}_i\cdot\mathbf{m}_i, \tag*{(6)}

where Btrh\mathbf{B}_{\mathrm{trh}} is the transferred hyperfine field, mi\mathbf{m}_i is the moment at the ii-th Ce site, and Ai\mathbf{A}_i is the nearest-neighbor hyperfine coupling tensor that takes the form

Ai=[AaaAabAacAbaAbbAbcAcaAcbAcc].(7)\mathbf{A}_i= \begin{bmatrix} A_{aa} & A_{ab} & A_{ac}\\ A_{ba} & A_{bb} & A_{bc}\\ A_{ca} & A_{cb} & A_{cc} \end{bmatrix}. \tag*{(7)}

For the lower-temperature AFM phase with wave vector q=(0.5, 0, 0)\mathbf{q}=(0.5,\ 0,\ 0) [21], the moments m1=−m2=m3=−m4=m\mathbf{m}_1=-\mathbf{m}_2=\mathbf{m}_3=-\mathbf{m}_4=\mathbf{m} [35], and we get Btrh=(4Aacmz, 0, 4Acamx)\mathbf{B}_{\mathrm{trh}}=(4A_{ac}m_z,\ 0,\ 4A_{ca}m_x). It is interesting to note that in order to have a non-vanishing in-plane hyperfine field as suggested by our NMR measurements (cf Figure 5), the zz component of Ce moment should be non-zero. This is different from the coplanar AFM order proposed by neutron and μ\muSR experiments[21]. Potentially, this bifurcation can be fixed by introducing the next-nearest-neighbor hyperfine coupling tensor, Ci\mathbf{C}_i (i=1,2i=1,2). Since the Ce atoms right above and below are of nearly the same distance to As, for simplicity, we treat them as mirror reflection with respect to abab-plane. By adopting the magnetic structure suggested by Ref. [21], we get Btrh=(4Aacmz+2Caamx−2Cabmy,2Cbamx−2Cbbmy,4Acamx)\mathbf{B}_{trh} = (4A_{ac}m_z + 2C_{aa}m_x - 2C_{ab}m_y, 2C_{ba}m_x - 2C_{bb}m_y, 4A_{ca}m_x). In this case, the in-plane hyperfine field is non-vanishing even mzm_z is zero. However, we find it difficult to determine the full A\mathbf{A} and C\mathbf{C} tensors only by the present measurements.

The situation for the incommensurate SDW phase at TN2<T<TN1T_{N2} < T < T_{N1} proposed by Wu et al. [21] is too complicated for this analysis, and is not discussed here.

Spin dynamics

We also investigated the spin dynamics of CeNiAsO by spin-lattice relaxation rate measurements, and the results are summarized in Figure 6. As is well known, for conventional metals (like Cu[36]), the major mechanism to transfer the spin temperature of a nuclear ensemble into equilibrium with the lattice is via spin-flip scattering with conduction electrons[37]; 1/T1T1/T_1T, therefore, is a constant proportional to N2(EF)N^2(E_F) [so-called Korringa’s relation, where N(EF)N(E_F) is the density of states at the Fermi level]. In CeNiAsO, however, for both field orientations, 1/T11/T_1 is essentially constant for TT well above TN1T_{N1}, or in other words, 1/T1T∝T−11/T_1T \propto T^{-1}. This is better expressed in the log-log plot shown in Figure 6(a-Figure 6b). Such a Curie-Weiss-like behavior has been reported in many other Ce compounds like CeIn3_3[38], CePt2_2In7_7[39], Ce2_2CoAl7_7Ge4_4[40], etc[8, 26], and can be ascribed to relaxation dominated by spin fluctuations of Ce local moments. Near 9 K, 1/T11/T_1 peaks for both field orientations, consistent with critical slowing down at the magnetic transitions. The two transition temperatures TN1T_{N1} and TN2T_{N2} are discernible in 1/T1T1/T_1T, as shown in the inset to Figure 6(a-Figure 6b). Below TN2T_{N2}, 1/T11/T_1 decreases rapidly, characteristic of long-range magnetic ordering.

Image not extracted

Figure 6. (a-b) TT dependence of 1/T1T1/T_1T at 10 T, for B0∥cB_0 \parallel c and B0⊥cB_0 \perp c, respectively. The insets show 1/T1T1/T_1T near the AFM transitions. (c) Hyperfine-coupling normalized spin fluctuation rate Ri/Ahf,i2R_i/A_{\mathrm{hf},i}^2 as functions of TT for 75^{75}As. (d) Plots of 75K/T1/2^{75}K/T^{1/2} vs. 1/T1T1/T_1T in the paramagnetic phase for both field orientations.

In general, the spin-lattice relaxation rate provides a measure of spin fluctuations perpendicular to an applied B0B_0, as described by[41]

1/T1T∝γn2∑q⊥B0Ahf2(q,ω)χ′′(q,ω)ω,(8)1/T_1T \propto\gamma_n^2 \sum_{\mathbf{q}\perp B_0} A_{\mathrm{hf}}^2(\mathbf{q},\omega)\frac{\chi''(\mathbf{q},\omega)}{\omega}, \tag*{(8)}

where χ′′(q,ω)\chi''(\mathbf{q},\omega) is the imaginary part of the dynamic susceptibility at wave vector q\mathbf{q} and frequency ω\omega. Considering the magnetic anisotropy of the sample, 1/T1T1/T_1T is decomposed for further study:

(1/T1T)c=2Rab,(9)(1/T_1T)_c = 2R_{ab}, \tag*{(9)}
(1/T1T)ab=Rab+Rc,(10)(1/T_1T)_{ab} = R_{ab} + R_c, \tag*{(10)}

where the spin fluctuation rate

Ri∝Ahf,i2∑q⊥χi′′(q,ω)ω,(11)R_i \propto A_{\mathrm{hf},i}^2 \sum_{\mathbf{q}\perp} \frac{\chi''_i(\mathbf{q},\omega)}{\omega}, \tag*{(11)}

with i=ab,ci = ab, c. Note that here the q\mathbf{q} and ω\omega dependencies of AhfA_{\mathrm{hf}} are neglected. The temperature dependence of the hyperfine-coupling normalized spin fluctuation rates Ri/Ahf,i2R_i/A_{\mathrm{hf},i}^2 are shown in Figure 6(c). Rab/Ahf,ab2R_{ab}/A_{\mathrm{hf},ab}^2 near TN1T_{N1} is about 5 times as large as Rc/Ahf,c2R_c/A_{\mathrm{hf},c}^2. This suggests quasi-2D spin fluctuations in CeNiAsO.

Finally, it is necessary to discuss the influence of the Kondo effect on spin dynamics. Previous work on Kondo lattice systems suggested that the 4f4f-electron dominated 1/T1T1/T_1T is proportional to the ratio of the static susceptibility χ(T)\chi(T) and the dynamical relaxation rate Γ(T)\Gamma(T) of the 4f4f electrons[8, 41, 42, 43], where Γ∝T1/2\Gamma\propto T^{1/2} and T∗T^*[44]. Since K∝χK \propto\chi in this temperature range (Figure 4), then

(1/T1T)4f∝K(T)/T1/2.(12)(1/T_1T)_{4f} \propto K(T)/T^{1/2}. \tag*{(12)}

This relation was found in many Ce-based Kondo compounds including CeFePO[8] that is isostructural to CeNiAsO. In Figure 6(d), we plot K/T1/2K/T^{1/2} vs. 1/T1T1/T_1T in the paramagnetic phase. For both field directions, Eq. (10) is validated at high temperature, but is gradually deviated for TT below 15 K. This provides additional evidence that a coherent Kondo liquid gradually forms below T∗∼15T^* \sim15 K in CeNiAsO.

Discussion and outlook

Relevant issues remain: (1) It is still unknown whether the “phase segregation” observed at TN2<T<TN1T_{N2} < T < T_{N1} is intrinsic or not. First, it is seen only in B0∥c\mathbf{B}_0 \parallel c, but not in B0⊥c\mathbf{B}_0 \perp c; second, it disappears when entering the lower-temperature ordered phase; and third, an XRD pattern (not shown here, but equally good as in our original report[15]) shows very high sample quality without any noticeable impurity phase. Note that such a phase segregation might not be seen with the resolution of neutron scattering or μ\muSR. If this phenomenon is intrinsic, considering the heat-capacity results [see in Figure 1(c)] we may infer a weakly first-order transition at TN1T_{N1}; it, therefore, is interesting to ask how this order is tuned to a QCP under pressure. However, since the temperature over which the high-TT AFM exists is only ∼2\sim2 K, comparable to the width for typical phase transitions, and furthermore, within this range, the AFM seems to barely saturate, it, therefore, is premature to make such a conclusion.

(2) A previous study[16] has manifested the realization of QCP in CeNiAsO either by P-doping or pressure. μ\muSR experiments on CeNiAs1−x_{1-x}Px_xO suggested that the lower-temperature AFM phase is suppressed rapidly by P-doping, and the QCP is between the incommensurate SDW and paramagnetic phases[21]. It is unknown whether the same situation also appears when QCP is accessed by pressure that is a cleaner control parameter than chemical doping. Moreover, an unconventional Kondo-breakdown type QCP was suggested in CeNiAsO[16]; however, the nature of quantum fluctuations (e.g., reflected in scaling laws of Knight shift and spin-lattice relaxation rate[17] and whether critical fermionic degrees of freedom should be involved[45, 46], etc) near the QCP remains unclear, and microscopic evidence for a Fermi-surface reconstruction[47] is also lacking. Hopefully, 75As{}^{75}\mathrm{As} NMR under pressure will be helpful to clarify these open questions.

(3) It is also interesting to mention that neither neutron scattering nor 75As{}^{75}\mathrm{As} NMR experiment sees any clear trace of structural transition in the AFM ordered states. Of particular interest is that below TN2T_{N2}, the wave vector q=(0.5,0,0)\mathbf{q} = (0.5, 0, 0) is exactly the same as those observed in the parent compounds of iron-based superconductors, whereas in the latter, ubiquitous tetragonal-to-orthorhombic phase transitions were observed[5, 35, 48]. The reason for this difference is unclear. A possibility might be that the magnetoelastic coupling in CeNiAsO is much weaker. All these require further investigations in the future.

Conclusions

By 75As{}^{75}\mathrm{As} NMR experiments, we investigated static and dynamic spin susceptibilities of the AFM Kondo lattice material CeNiAsO. The NMR spectra confirm that there are two kinds of AFM ordering that appear below TN1=9.0(3)T_{N1} = 9.0(3) K and TN2=7.0(3)T_{N2} = 7.0(3) K, respectively. Quadrupolar splitting frequency νQ∼7.75\nu_Q \sim7.75 MHz remains essentially unchanged across the AFM transitions. A Knight shift anomaly is observed below T∗∼15T^* \sim15 K, which gives a measure of the onset of coherent cc–ff correlations. Quasi-2D character of spin fluctuations is revealed by the highly anisotropic hyperfine-coupling normalized fluctuation rates. The hyperfine field at As sites transferred from Ce local moments contains both in-plane and out-of-plane components, which highlights the important role played by hybridization to next-nearest-neighbor Ce.

Acknowledgments

We thank Shan Wu, Stuart E. Brown, Joe D. Thompson, Yifeng Yang, Rui Zhou, and Long Ma for insightful discussions and technical suggestions. This work is supported by the open research fund of Songshan Lake Materials Laboratory (2022SLABFN27), National Key R&D Program of China (2022YFA1602602), and Guangdong Basic and Applied Basic Research Foundation (2022B1515120020).

References

  1. [1]Y. Kamihara, T. Watanabe, M. Hirano, and H. Hosono, J. Am. Chem. Soc. 130, 3296 (2008).
  2. [2]G. F. Chen, Z. Li, D. Wu, G. Li, W. Z. Hu, J. Dong, P. Zheng, J. L. Luo, and N. L. Wang, Phys. Rev. Lett. 100, 247002 (2008).
  3. [3]X. H. Chen, T. Wu, G. Wu, R. H. Liu, H. Chen, and D. F. Fang, Nature 453, 761 (2008).
  4. [4]C. Wang, L. Li, S. Chi, Z. Zhu, Z. Ren, Y. Li, Y. Wang, X. Lin, Y. Luo, S. Jiang, X. Xu, G. Cao, and Z.-a. Xu, EPL 83, 67006 (2008).
  5. [5]J. Zhao, Q. Huang, C. de la Cruz, S. Li, J. W. Lynn, Y. Chen, M. A. Green, G. F. Chen, G. Li, Z. Li, J. L. Luo, N. L. Wang, and P. Dai, Nat. Mater. 7, 953 (2008).
  6. [6]C. de la Cruz, W. Z. Hu, S. Li, Q. Huang, J. W. Lynn, M. A. Green, G. F. Chen, N. L. Wang, H. A. Mook, Q. Si, and P. Dai, Phys. Rev. Lett. 104, 017204 (2010).
  7. [7]Y. Luo, Y. Li, S. Jiang, J. Dai, G. Cao, and Z.-a. Xu, Phys. Rev. B 81, 134422 (2010).
  8. [8]E. M. Brüning, C. Krellner, M. Baenitz, A. Jesche, F. Steglich, and C. Geibel, Phys. Rev. Lett. 101, 117206 (2008).
  9. [9]S. Kitagawa, H. Ikeda, Y. Nakai, T. Hattori, K. Ishida, Y. Kamihara, M. Hirano, and H. Hosono, Phys. Rev. Lett. 107, 277002 (2011).
  10. [10]S. Lausberg, J. Spehling, A. Steppke, A. Jesche, H. Luetkens, A. Amato, C. Baines, C. Krellner, M. Brando, C. Geibel, H.-H. Klauss, and F. Steglich, Phys. Rev. Lett. 109, 216402 (2012).
  11. [11]A. Jesche, T. Balle, K. Kliemt, C. Geibel, M. Brando, and C. Krellner, Phys. Status Solidi B 254, 1600169 (2017).
  12. [12]C. Krellner, U. Burkhardt, and C. Geibel, Physica B 404, 3206 (2009).
  13. [13]R. Sarkar, A. Jesche, C. Krellner, M. Baenitz, C. Geibel, C. Mazumdar, and A. Poddar, Phys. Rev. B 82, 054423 (2010).
  14. [14]G. Xu, W. Ming, Y. Yao, X. Dai, S.-C. Zhang, and Z. Fang, EPL 82, 67002 (2008).
  15. [15]Y. Luo, H. Han, H. Tan, X. Lin, Y. Li, S. Jiang, C. Feng, J. Dai, G. Cao, Z. Xu, and S. Li, J. Phys.: Condens. Matter 23, 175701 (2011).
  16. [16]Y. Luo, P. Pouvrovskii, S. E. Rowley, Y. Li, C. Feng, A. Georges, J. Dai, G. Cao, Z. Xu, Q. Si, and N. P. Ong, Nat. Mater. 13, 777 (2014).
  17. [17]Q. Si, S. Rabello, K. Ingersent, and J. L. Smith, Nature 413, 804 (2001).
  18. [18]P. Coleman and A. J. Schofield, Nature 433, 226 (2005).
  19. [19]P. Gegenwart, Q. Si, and F. Steglich, Nat. Phys. 4, 186 (2008).
  20. [20]S. Paschen, T. Luhmann, S. Wirth, P. Gegenwart, O. Trovarelli, C. Geibel, F. Steglich, P. Coleman, and Q. Si, Nature 432, 881 (2012).
  21. [21]S. Wu, W. A. Phelan, L. Liu, J. R. Morey, J. A. Tutmaher, J. C. Neuefeind, A. Huq, M. B. Stone, M. Feygenson, D. W. Tam, B. A. Frandsen, B. Trump, C. Wan, S. R. Dunsiger, T. M. McQueen, Y. J. Uemura, and C. L. Broholm, Phys. Rev. Lett. 122, 197203 (2019).
  22. [22]B.-L. Young, M. S. Rose, D. E. MacLaughlin, K. Ishida, O. O. Bernal, H. G. Lukefahr, K. Heuser, E. J. Freeman, and M. B. Maple, Rev. Sci. Instrum. 73, 3038 (2002).
  23. [23]M. T. Hutchings, Solid State Physics: Advances in Research and Applications, edited by F. Seitz and D. Turnbull (Academic, New York, 1964), Vol. 16, p. 227.
  24. [24]K. Cheng, X. He, H. Yang, B. Zhou, Y. Li, and Y. Luo, Phys. Rev. B 100, 205121 (2019).
  25. [25]A. Ghoshray, B. Pahari, M. Majumder, M. Ghosh, K. Ghoshray, B. Bandyopadhyay, P. Dasgupta, A. Poddar, and C. Mazumdar, Phys. Rev. B 79, 144512 (2009).
  26. [26]D. Rybicki, T. Meissner, G. V. M. Williams, S. V. Chong, M. Lux, and J. Haase, J. Phys.: Condens. Matter 25, 315701 (2013).
  27. [27]M. Jiang, N. J. Curro, and R. T. Scalettar, Phys. Rev. B 90, 241109 (2014).
  28. [28]N. J. Curro, Rept. Prog. Phys. 79, 064501 (2016).
  29. [29]E. Kim, M. Makivic, and D. L. Cox, Phys. Rev. Lett. 75, 2015 (1995).
  30. [30]T. Ohama, H. Yasuoka, D. Mandrus, Z. Fisk, and J. L. Smith, J. Phys. Soc. Jpn. 64, 2628 (1995).
  31. [31]S. Nakatsuji, D. Pines, and Z. Fisk, Phys. Rev. Lett. 92, 016401 (2004).
  32. [32]Y.-f. Yang and D. Pines, Phys. Rev. Lett. 100, 096404 (2008).
  33. [33]Y.-f. Yang and D. Pines, Proc. Natl. Acad. Sci. USA 109, E3060 (2012).
  34. [34]K. R. Shirer, A. C. Shockley, A. P. Dioguardi, J. Crocker, C. H. Lin, N. apRoberts Warren, D. M. Nisson, P. Klavins, J. C. Cooley, Y.-f. Yang, and N. J. Curro, Proc. Natl. Acad. Sci. USA 109, E3067 (2012).
  35. [35]K. Kitagawa, N. Katayama, K. Ohgushi, M. Yoshida, and M. Takigawa, J. Phys. Soc. Jpn. 77, 114709 (2008).
  36. [36]A. Abraham, Principles of Nuclear Magnetism (Oxford University Press, London, 1961).
  37. [37]C. P. Slichter, Principles of Magnetis Resonance (Springer-Verlag, Berlin Heidelberg, 1990).
  38. [38]Y. Kohori, Y. Inoue, T. Kohara, G. Tomka, and P. Riedi, Physica B 259-261, 103 (1999).
  39. [39]H. Sakai, Y. Tokunaga, S. Kambe, F. Ronning, E. D. Bauer, and J. D. Thompson, Phys. Rev. Lett. 112, 206401 (2014).
  40. [40]A. P. Dioguardi, P. Guzman, P. F. S. Rosa, N. J. Ghimire, S. Eley, S. E. Brown, J. D. Thompson, E. D. Bauer, and F. Ronning, Phys. Rev. B 96, 245132 (2017).
  41. [41]T. Moriya and K. Ueda, Solid State Commun. 15, 169 (1974).
  42. [42]Y. Kuramoto and Y. Kitaoka, Dynamics of heavy electrons, Vol. 105 (Oxford University Press, London, 2000).
  43. [43]D. MacLaughlin, Hyperfine Interactions 49, 43 (1989).
  44. [44]D. D. Cox, N. E. Bickers, and J. W. Wilkins, J. Appl. Phys. 57, 3166 (1985).
  45. [45]L. Prochaska, X. Li, D. C. MacFarland, A. M. Andrews, M. Bonta, E. F. Bianco, S. Yazdi, W. Schrenk, H. Detz, A. Limbek, Q. Si, E. Ringe, G. Strasser, J. Kono, and S. Paschen, Science 367, 285 (2020).
  46. [46]A. Kandala, H. Hu, Q. Si, and K. Ingersent, arXiv: 2206.01174 (2022).arxiv.org/abs/2206.01174
  47. [47]S. Kawasaki, T. Oka, A. Sorime, Y. Kogame, K. Uemoto, K. Matano, J. Guo, S. Cai, L. Sun, J. L. Sarrao, J. D. Thompson, and G.-q. Zheng, Commun. Phys. 3, 148 (2020).
  48. [48]C. Cruz, Q. Huang, J. Lynn, J. Li, W. Ratcliff, J. Zarestky, H. Mook, G. Chen, J. Luo, N. Wang, and P. Dai, Nature 453, 899 (2008).

Paper details

vlnc:W_xpx6zfrz
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents