Spin trimers in Ca3Cu2Ni(PO4)4M. Ghosh and K. Ghoshray

Citation: Low Temperature Physics 38, 645 (2012); doi: 10.1063/1.4734011 View online: http://dx.doi.org/10.1063/1.4734011 View Table of Contents: http://scitation.aip.org/content/aip/journal/ltp/38/7?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Magnetic and dielectric properties of one-dimensional array of S=1/2 linear trimer system Na2Cu3Ge4O12 J. Appl. Phys. 115, 17E125 (2014); 10.1063/1.4865776 Transition from Pauli-paramagnetism to ferromagnetism in CaCu3(Ru4xMnx)O12 (0 x 3) perovskites J. Appl. Phys. 109, 123914 (2011); 10.1063/1.3597783 Size-driven magnetic transitions in La 1 / 3 Ca 2 / 3 MnO 3 nanoparticles J. Appl. Phys. 108, 063918 (2010); 10.1063/1.3488619 Submillimeter and Millimeter Wave ESR Measurements of the S=1/2 Trimer Chain M3Cu3(PO4)4 (M=Sr,Ca) AIP Conf. Proc. 850, 1037 (2006); 10.1063/1.2355057 Magnetic properties and spin dynamics in hole-doped S=1 AF chain: 89 Y NMR and susceptibility in Y 2x Ca xBaNiO 5 J. Appl. Phys. 83, 6605 (1998); 10.1063/1.367608

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 131.217.6.8 On:

Wed, 03 Sep 2014 09:45:23

Spin trimers in Ca3Cu2Ni(PO4)4

M. Ghosh and K. Ghoshraya)

ECMP Division, Saha Institute of Nuclear Physics, 1/AF Bidhannagar, Kolkata 700064, India(Submitted February 29, 2012)

Fiz. Nizk. Temp. 38, 815–820 (July 2012)

The nature of spin trimer and 3d spin dynamics in Ca3Cu2Ni(PO4)4 is studied using 31P NMR

measurements. The data are obtained by replacing one of the Cu ions by a Ni ion in the one

dimensional spin trimer compound Ca3Cu3(PO4)4. The NMR spectrum suggests the presence of

two types of magnetically inequivalent phosphorous atoms, as in the parent compound. This reveals

the presence of only one type of trimer, Cu2-Ni-Cu2, rather than of three types, Cu2-Cu1-Cu2,

Cu2-Cu1-Ni and Ni-Cu1-Ni, as indicated by neutron scattering. Hence, the ground state of

Ca3Cu2Ni(PO4)4 appears to be a quintet. These results also explain the magnetic susceptibility

behavior, which indicates no reduction in the net spin of the trimer in Ca3Cu2Ni(PO4)4, as expected

for JNi-Cu¼�0.85 meV compared to the value JCu-Cu¼�4.74 meV derived from neutron

scattering. The spin lattice relaxation rate suggests that the three magnon mediated scattering

process dominant in Ca3Cu3(PO4)4 is reduced in Ca3Cu2Ni(PO4)4. VC 2012 American Institute ofPhysics. [http://dx.doi.org/10.1063/1.4734011]

1. Introduction

Low-dimensional quantum spin systems have attracted

considerable attention because of the possibility of novel

magnetic effects such as quantum critical phenomena,1 Hal-

dane gap formation,2 and field induced Bose-Einstein con-

densation.3 Enhancement of quantum fluctuations due to low

dimensionality could suppress the magnetic ordering and

give rise to several such unusual phenomena. Recently it has

been suggested that in an S¼ 1/2 antiferromagnetic (AF)

chain with period 3, exchange coupling exhibits a plateau in

the magnetization curve where the magnetization stays con-

stant at a fraction of the saturation level over a finite range

of magnetic fields.4–6 This phenomenon in quantum spin

chains is analogous to the quantum Hall effect-topological

quantization of a physical quantity in a changing magnetic

field. Trimer chains are the first spin systems predicted to

have magnetization plateaux. However, spin trimer systems

are rare in nature. Among the few such systems, La4Cu3-

MoO12 is a 2D AF system in which the three Cu2þ spins

form a triangle, so that AF intra-trimer interactions are

frustrated7; nevertheless this system undergoes an AF

ordering near 2.6 K. Due to strong nearest neighbor (NN)

AF exchange, the net spin of the trimer decreases from 3/2

to 1/2 below 250 K, as revealed by the inverse susceptibility

v�1 curve.

A3Cu3(PO4)4 (A:Ca, Sr, Pb) is another spin trimer

system, in which Cu2-Cu1-Cu2 trimers are arranged in linear

chain, and the inter-trimer AF exchange interaction J2 is 3 K

while the intra trimer NN AF exchange interaction J1 is

about 126 K.8,9 As a result, below 45 K, the v�1 vs T curve

shows clear signs of a reduction in the net spin of the trimer

from 3/2 to 1/2.10 Thus, the ground state is a doublet, and is

a model system of a linear HAF trimer. In principle, a dou-

blet ground state of this kind can be changed into a singlet

ground state by substituting a Cu2þ (S¼ 1/2) spin in the

trimer by Ni2þ (S¼ 1), provided only one type of trimer

(e.g., Ni-Cu1-Cu2/Cu-Cu1-Ni) is formed. For the first time,

this provides an opportunity of realizing a gapped quantum

spin trimer system, with the possibility of observing Bose-

Einstein condensation (BEC) above a certain critical value

of an external magnetic field. BEC was observed in the spin

dimer system TlCuCl3 by inelastic neutron scattering.10,11

To investigate this in Ca3Cu3(PO4)4, Podlesnyak et al.12

studied inelastic neutron scattering in Ca3Cu3(PO4)4.

Unfortunately the singlet state was not realized. These stud-

ies also indicated that the middle position is exclusively

occupied by Cu2þ, whereas the end positions are statistically

populated by Cu2þ and Ni2þ; thus, three types of trimer,

Cu2-Cu1-Cu2, Cu2-Cu1-Ni and Ni-Cu1-Ni, are created with

doublet, triplet, and quintet ground states, respectively,

instead of only one type, Cu2-Cu1-Ni, with a triplet ground

state. The same result was obtained in an elastic neutron

scattering study.13 From these findings it was concluded that

the possible reason for the failure to observe a singlet ground

state in Ca3Cu2Ni(PO4)4, is the formation of three types of

trimers. Moreover, it was also shown that, JCu-Cu in Ca3Cu2-

Ni(PO4)4 increases to �4.92 meV from the �4.74 meV

observed in A3Cu3(PO4)4 and JNi-Cu¼�0.85 meV. Surpris-

ingly, the magnetic susceptibility data for Ca3Cu2Ni(PO4)4

(Refs. 13 and 14) in the range 2-300 K, did not manifest the

expected reduction in the net spins of the trimers Cu2-Cu1-

Cu2 and Cu2-Cu1-Ni, because of strong NN AF exchange

between the Cu2þ spins, as was seen in Ca3Cu2(PO4)4, where

only one type of trimer, Cu2-Cu1-Cu2, is present, with

weaker NN AF exchange as mentioned above. On the other

hand, the same behavior in Ca3Cu2Mg(PO4)4 indicated the

presence of more than one type of trimer with a clear

reduction in the net spins of certain types of trimers.14 A

comparison of the susceptibility data in Ca3Cu2(PO4)4,

Ca3Cu2Ni(PO4)4 and Ca3Cu2Mg(PO4)4, indicated that the Ni

atom in Ca3Cu2Ni(PO4)4 probably occupies the middle posi-

tion instead of the end positions of a trimer. On the other

hand, the Mg atom in Ca3Cu2Mg(PO4)4 occupies an end

position. In this connection it may be mentioned that, based

on neutron scattering in Ca3Cu2Ni(PO4)4, Pomjakushin

1063-777X/2012/38(7)/6/$32.00 VC 2012 American Institute of Physics645

LOW TEMPERATURE PHYSICS VOLUME 38, NUMBER 7 JULY 2012

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 131.217.6.8 On:

Wed, 03 Sep 2014 09:45:23

et al.13 also suggested that the valance of Ni has a reasonable

value if it occupies the middle position of the Cu2-Cu1-Cu2

trimer (according to a bond valence sum (BVS) calculation),

so that the Ni could occupy the middle position. Thus the

possibility of forming only one type of trimer, Cu2-Ni-Cu2,

with a quintet ground state cannot be neglected in the case of

Ca3Cu2Ni(PO4)4.

To understand this further, we have made a 31P NMR

study of polycrystalline samples of Ca3Cu2Ni(PO4)4. Fig. 1

illustrates the existence of a single type of trimer, Cu2-Cu1-

Cu2, in the unit cell of Ca3Cu2(PO4)4, when projected in the

ab plane, along with bonding of the two types of phospho-

rous (P) atoms, via the oxygen atom. P1 is bonded with two

nearest neighbor magnetic ions within a trimer and P2 is

bonded with the magnetic ions between the two trimers.

Now, if in Ca3Cu2Ni(PO4)4, the Ni ion occupies the Cu2 site,

resulting in three types of trimers as mentioned above, then

it follows from Fig. 1 that the local magnetic environments

of all the P1 atoms in the unit cell will not be identical and

that same would be true for P2. As a result, the resonance

line positions of all the P1 atoms would not superimpose at a

particular frequency, and the same would be true for P2.

This would produce a more complicated 31P NMR spectrum

than that observed for Ca3Cu3(PO4)4,15 where the spectrum

is a superposition of two resonance lines corresponding to

P1 and P2 indicating that all the P1 nuclei experience the

same local magnetic field, and that the same is true for P2.

On the other hand if the Ni atom occupies the Cu1 position,

then there will be only one type of trimer, i.e., Cu2-Ni-Cu2.

In this case, Fig. 1 implies that all the P1 atoms will be

linked with Cu2 on one side and with Ni on the other side

within a trimer, and therefore, will have identical magnetic

environments, while all the P2 atoms will be bonded on both

sides with Cu2 atoms of the trimers of the two consecutive

chains. In this case the 31P resonance line for Ca3Cu2-

Ni(PO4)4 would be a superposition of two lines similar to

that in the pure compound. Thus NMR studies should pro-

vide valuable information about the possible type of trimers

formed in Ca3Cu2Ni(PO4)4, and, therefore, on the magnetic

ground state of this system. In addition, the 31P nuclear spin

lattice relaxation rate should provide additional information

about the effect of Ni substitution on the one-dimensional

spin dynamics, which is currently a topic of great interest.

2. Experiment

A polycrystalline sample of Ca3Cu2Ni(PO4)4 was synthe-

sized using the solid state reaction method, by heating a stoi-

chiometric mixture of CuO (99.99%), NiO (99.99V), CaCO3

(99.99%) and NH4H2PO4 (99.99%) very slowly up to 600 �Cand then annealing at 900 �C for 120 h, with several interme-

diate grindings. The sample is characterized at room tempera-

ture using x-ray diffraction (XRD) with CuKa radiation

(k¼ 1.5406 A). All the peaks can be indexed on the basis of

monoclinic structure with space group P21/a and cell parame-

ters a¼ 17.718(1) A, b¼ 4.888(2) A, c¼ 8.84(1) A, and

b¼ 123.8(1)�, in satisfactory agreement with reported values.13

The NMR measurements were carried out at fields H¼ 1.39 T

(vR¼ 24.005 MHz) and H¼ 7.04 T (vR¼ 121.423 MHz) using

a conventional phase-coherent Thamway PROT4103 pulse

spectrometer. The spectrum was recorded by changing the fre-

quency step by step and recording the spin-echo intensity by

applying a p/2-s-p/2 solid echo sequence. Observations at tem-

peratures in the range 4-300 K were made in an Oxford contin-

uous flow dynamic cryostat with a ITC503 controller. The shift

is measured with respect to the position of the 31P NMR line

(vR) in H3PO4 solution.

3. Results and discussion

3.1. NMR spectra and shift parameters

Figure 2 shows the 31P NMR spectrum of Ca3Cu2-

Ni(PO4)4 at 300 K in a field of 7 T along with the same for

Ca3Cu3(PO)4 (Ref. 14) and Ca3CuNi2(PO4)4 (Ref. 14) as a

comparison. The spectrum for Ca3Cu2Ni(PO4)4 shows a typ-

ical overlap of two powder patterns for nuclear spin I¼ 1/2

having different component of shift parameters, similar to

that observed in the other two compounds. It is to be noted

that the position of the resonance line for P2 in Ca3Cu3(PO)4

FIG. 1. Three trimers (A, B, and C) in Ca3Cu2(PO4)4, viewed from the zaxis showing the possible superexchange paths between the trimers in the abplane. Cu1 and Cu2 atoms are indicated by blue and yellow spheres, respec-

tively, and O and P atoms by red and green spheres. The Ca atoms are not

shown. The structure has been drawn using the coordinates given in Ref. 13.

FIG. 2. 31P NMR spectra of polycrystalline Ca3CuNi2(PO4)4,14 Ca3Cu2-

Ni(PO4)4 and Ca3Cu3(PO4)4 (Ref. 14) at 300 K recorded at H¼ 7 T (open-

circles). The theoretical curve (dashed) calculated using Eq. (1), along with

two constituent lines (continuous) corresponding to resonance lines of two

inequivalent phosphorous atoms are also shown. The vertical line indicates

the reference position at frequency 121.42 MHz.

646 Low Temp. Phys. 38 (7), July 2012 M. Ghosh and K. Ghoshray

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 131.217.6.8 On:

Wed, 03 Sep 2014 09:45:23

and in Ca3Cu2Ni(PO4)4 are close to each other, whereas, in

Ca3CuNi(PO4)4 the shift of the line for P2 is higher. In the

case of Ca3CuNi2(PO4)4, a higher value of the shift is

expected, as the P2 atom is bonded on both sides with Ni2þ

ions with S¼ 1. On the other hand, in Ca3Cu3(PO4)4, P2 is

bonded on both sides with Cu2þ ions with S¼ 1/2. In

Ca3Cu2Ni(PO4)4, if the Ni atom occupies the Cu1 position,

with the end positions of the trimer being occupied by Cu2

atoms, then P2 in this compound will also be bonded with

Cu2 atoms on both sides, so that the local magnetic field

experienced by P2 in this compound would also be similar to

that of the pure compound. Moreover, if only Cu2-Ni-Cu2

trimer is present in Ca3Cu2Ni(PO4)4, then P1 atoms would

be bonded with Cu2 on one side and Ni on the other side,

similar to the situation in Ca3CuNi2(PO4)4, where the only

the trimer Ni- Cu2-Ni is present. In this case, the local mag-

netic field at a P1 site in Ca3CuNi2(PO4)4 and Ca3Cu2-

Ni(PO4)4 would be almost the same. Interestingly, it can be

seen from Fig. 2 that the positions of the resonance lines of

P1 in Ca3Cu2Ni(PO4)4 and Ca3CuNi2(PO4)4 are very close

to each other. These observations indicate that the Ni atom

in Ca3Cu2Ni(PO4)4 most probably occupies the Cu1 position

of the trimer in Ca3Cu2Ni(PO4)4; this means that only one

type of trimer, Cu2-Ni-Cu2, is present in this compound. In

order to determine the shift parameters, the experimental

spectra for Ca3Cu2Ni(PO4)4 were fitted to the following

equation over the entire temperature range,

v ¼ vR½1þ Kiso þ Kaxðcos2 h� 1Þ þ Kansin2h cos 2/�; (1)

where Kiso, Kax and Kan are the isotropic, axial and anisotropic

parts of the shift, respectively, arising from Hlocal at the site of

the 31P nucleus through the electron nuclear hyperfine and the

dipolar interactions. �R is the reference frequency as men-

tioned earlier. h and u are the Euler angles between the princi-

pal axes of the hyperfine coupling tensor and the direction of

the external magnetic field (Zeeman field).

Figure 3 shows the measured 31P NMR spectra at differ-

ent temperatures in the range 4-300 K for Ca3Cu2Ni(PO4)4

along with theoretical spectra. The experimental curves

below 300 K are also well fitted with two constituent lines.

This also suggests that the internal magnetic fields at each

type of P site vary identically throughout the range 4-300 K.

Therefore, based on the spectral features over the whole tem-

perature range, it also appears that the Ni atom occupies the

Cu1 position in a Cu2-Cu1-Cu2 trimer, as discussed above.

Figure 4 shows the variation of Kiso, Kax and Kan with T, and

the insets illustrate the variation of these parameters with the

bulk susceptibility.

We also did the NMR experiment at a much lower reso-

nance frequency of 24 MHz with H¼ 1.39 T. The shift param-

eters obtained by fitting the spectra recorded at H¼ 1.39 T for

160–300 K, are also included in Fig. 4. The close agreement

between the values obtained by analyzing the spectra at two

widely different resonance frequencies validates the accuracy

of the results.

The shift is related to vspin by the relation,

K ¼ K0 þHhf

NlB

vspinðTÞ; (2)

where Hhf is the hyperfine field, K0 is the chemical shift, N is

the Avogadro number and vspin(T) is the contribution of

FIG. 3. 31P NMR spectra of polycrystalline Ca3Cu2Ni(PO4)4 at different

temperatures recorded with H¼ 7 T. The notation is as in Fig. 2.

FIG. 4. Temperature variation in Kiso (a), Kax (b), and Kan (c), measured for

H¼ 7 T: the filled squares are for P1 sites and the stars are for P2 sites. For

H¼ 1.39 T: the triangles are for P1 sites and the filled circles are for P2 sites.

The inset shows the dependences of Kiso (a), Kax (b), and Kan (c) for both P1

and P2 sites on vM with temperature as an implicit parameter.

Low Temp. Phys. 38 (7), July 2012 M. Ghosh and K. Ghoshray 647

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 131.217.6.8 On:

Wed, 03 Sep 2014 09:45:23

electronic spin to the magnetic susceptibility. As long as the

hyperfine field remains constant, K should follow v(T). The

contribution to Hhf arises from the transferred hyperfine

interaction, which is a property of the electronic structure,

and the dipolar interaction, neither of which is temperature

dependent. Although K(T) need not have the same symmetry

as v(T), the temperature dependence of K(T) should reflect

that of v(T). The inset to Figs. 4(a)–4(c) shows that Kiso, Kax

and Kan vary linearly with v(T) in the range 15–300 K. The

linearity suggests a unique hyperfine coupling constant over

this temperature range. Below 15 K, there is a deviation

from linearity. By using Eq. (1), we determine the values of

isotropic, axial and the anisotropic parts of the hyperfine

field for these two sites. Table 1 lists the values of Hisohf , Hax

hf ,

and Hanhf for the P1 and P2 sites for Ca3Cu2Ni(PO4)4, together

with those for Ca3Cu3(PO4)4 (from our measurements),14 for

comparison.

3.2. Nuclear spin-lattice relaxation rates

In order to investigate the effect on the electron spin dy-

namics of replacing one of the Cu2þ ions by a Ni2þ ion in a

trimer of Ca3Cu3(PO4)4, we have measured the 31P nuclear

spin-lattice relaxation rate 1/T1, which is a very sensitive

probe of the low energy electron spin fluctuations. 1/T1 is

measured by the saturation recovery method using a single

p/2 pulse. The measurements were made in an external mag-

netic field of 7 T at temperatures of 4–300 K and in a field of

1.39 T for 160–300 K, in order to see whether the strong field

dependence of 1/T1 observed in case of Ca3Cu3(PO4)4,16 as

well as in the isostructural Sr3Cu3(PO4)4 (Ref. 16) in the

high temperature range, is also present in Ca3Cu2Ni(PO4)4.

Figure 5 shows the 1/T1 vs T curves for one type of phospho-

rous site (P2), measured in magnetic fields of 7 T and 1.39 T

for Ca3Cu2Ni(PO4)4, together with those for Sr3Cu3(PO4)4

(Ref. 16) and Ca3CuNi2(PO4)4 (Ref. 17) as a comparison. It

is seen that 1/T1 for Sr3Cu3(PO4)4 is strongly field depend-

ent, as reported for the isostructural Ca3Cu3(PO4)4, while the

same curve for Ca3CuNi2(PO4)4 shows negligible field de-

pendence. On the other hand 1/T1 for Ca3Cu2Ni(PO4)4, still

remains field dependent with the influence of H smaller than

for the pure compound. This shows that the substitution of

Ni2þ (S¼ 1) in place of Cu2þ (S¼ 1/2) strongly affects the31P nuclear relaxation mechanism. 1/T1 in such compounds

is governed by the thermal excitations of the 3 d spins, and

can be written as

1

T1

¼ 2c2nkBTRjAðqÞj2v00ðq;xnÞ=xn; (3)

where cn is the nuclear gyromagnetic ratio, A(q) is the (wave

vector dependent) hyperfine coupling, and v00(q, xn), is the

dissipative component of the chain dynamic susceptibility

evaluated at the nuclear Larmor frequency xn.18

The values of 1/T1 for Ca3Cu2Ni(PO4)4 remain almost

constant in the temperature range 6.5–300 K for H¼ 7 T.

Below 6.5 K it shows a decreasing trend which is also con-

firmed by the change in the 31P nuclear magnetization recov-

ery curves from 6.5 to 4 K (Fig. 6). In the pure compound,

however, 1/T1 remains constant over 50–300 K and decreases

continuously at lower temperatures. In Ca3CuNi2(PO4)4, 1/T1

remains constant over 26–300 K and has a peak at 16 K owing

to an antiferromagnetic transition. The T independent behav-

ior of 1/T1 is typical feature of fast spin fluctuation (xe�xn)

of the paramagnetic moments. A comparison of this behavior

in the three compounds suggests that substitution of one cop-

per ion in a trimer by a Ni ion reduces the extent of the field

dependence and also extends the range over which 1/T1 is in-

dependent of T. This indicates a reduction in the magnetic

field dependent three-magnon mediated relaxation process,

with an enhancement of the two-magnon contribution, in

TABLE 1. Values of 31P hyperfine fields.

Parameters, kOe/lB Ca3Cu3(PO4)4 Ca3Cu3Ni(PO4)4

Hisohf for P1 1.28 6 0.03 1.32 6 0.02

Hisohf for P2 0.43 6 0.03 0.3 6 0.03

Haxhf for P1 0.44 6 0.01 0.41 6 0.01

Haxhf for P2 0.16 6 0.01 0.20 6 0.02

Hanhf for P1 0.50 6 0.02 0.32 6 0.02

Hanhf for P2 0.15 6 0.01 0.13 6 0.01

FIG. 5. 1/T1 as a function of T for Sr3Cu3(PO4)4 (Ref. 16) (a), Ca3Cu2-

Ni(PO4)4 (b) and Ca3CuNi2(PO4)4 (Ref. 17) (c) for H¼ 7.045 T (solid trian-

gles) and H¼ 1.39 T (solid circles). The error bar for each point corresponds

to the size of the symbol.

648 Low Temp. Phys. 38 (7), July 2012 M. Ghosh and K. Ghoshray

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 131.217.6.8 On:

Wed, 03 Sep 2014 09:45:23

Ca3Cu2Ni(PO4)4, while the former contribution is still present.

However, replacement of two Cu ions by Ni ions reduces the

range over which 1/T1 is independent of T, because of the

appearance of magnetic ordering at 16 K, with the H depend-

ence being completely suppressed, so that 1/T1 is governed

only by the two-magnon mediated Raman process. According

to the localized spin model for a high temperature limit,18 1/

T1 at a non magnetic nucleus can be written as

1

T1

¼ffiffiffiffiffiffi

2ppð2cnHhf Þ2 z0SðSþ 1Þ=ð3xexÞ; (4)

where the local spin exchange frequency xex¼ (2/3)� (J2/

�h2)� [zS(Sþ 1)]. The constants z¼ 2 and z0 ¼ 2 correspond

to the exchange-coupled local spin numbers and those of the

local spins interacting with the probing nucleus, and J is the

exchange constant respectively.18

We have calculated the value of 1/T1 for Ca3Cu2Ni(PO4)4

and Ca3CuNi2(PO4)4 taking cn¼ 1.083�108 s�1�T�1 for the

phosphorous nucleus and the values of Hisohf from Table 1. In

the case of Ca3Cu2Ni(PO4)4, if we assume that a Ni atom

occupies the middle position of the trimer, then P1 will be

affected by an intra-trimer exchange interaction between Cu2

and Ni. The strength of this exchange interaction JNi-Cu has

been determined by neutron scattering.12 We have used this

value for calculating T1 for P1 in Ca3Cu2Ni(PO4)4. In this

compound, P2 will be bonded with Cu2 ions of adjacent

trimers. Since the inter-trimer Cu2-Cu2 exchange constant

was not reported by Podlesnyak et al.12 we have used a value

of JCu-Cu (�0.258 meV) derived from magnetic susceptibility

data9 for Ca3Cu3(PO4)4 to calculate T1 for the P2 in Ca3Cu2-

Ni(PO4)4. The resulting value (Table 2) is in close agreement

with experiment. However, in the case of Ca3CuNi2(PO4)4,

since P2 is bonded on both sides with Ni ions and the inter-

trimer JNi-Ni is not known from either neutron scattering or

magnetic susceptibility measurements, we could calculate T1

only for the P1 (Table 2). In this case the agreement of the ex-

perimental and theoretical values is also quite satisfactory. So

our conclusions based on spin-lattice relaxation data further

support our suggestion based on analyzing the 31P line

shape in Ca3Cu2Ni(PO4)4, that a Ni atom most probably

occupies the middle position of the trimer, so that there is

only one type of trimer, Cu2-Ni-Cu2, rather than the three

types, Cu2-Cu1-Cu2, Cu2-Cu1-Ni, and Ni-Cu2-Ni, indicated

by neutron scattering. So the ground state of this compound is

a quintet.

4. Conclusion

We have studied the static and dynamic magnetic prop-

erties of the trimer spin chain compound Ca3Cu2Ni(PO4)4,

which remains in a paramagnetic phase at temperatures of

4–300 K, using 31P NMR studies in the temperature range

4–300 K with an external magnetic field of 7 T and in the

temperature range of 160–300 K with H¼ 1.39 T. At all tem-

peratures the spectrum corresponds to a typical overlap of

two powder patterns, indicating the presence of two types of

P atoms within the unit cell similar to that in Ca3Cu3(PO4)4

and Ca3CuNi2(PO4)4. This finding, together with data on

spin lattice relaxation, indicates that most probably only one

type of trimer, Cu1-Ni-Cu2 exists in Ca3Cu2Ni(PO4)4; this

could also explain satisfactorily the magnetic susceptibility

behavior, which gives no indication of a reduction in the net

spin of any trimer, as the NN interaction JNi-Cu is an order of

magnitude smaller than that of JCu-Cu.12 Our NMR data,

therefore, indicate that the ground state of Ca3Cu2Ni(PO4)4

is a quintet. The behavior of the spin lattice relaxation rate

suggests that replacement of one of the Cu ions by a Ni ion

in the trimer, partially attenuates the relaxation mechanism

attenuated by three-magnon scattering which is the dominant

contribution in the case of Ca3Cu3(PO4)4.

a)Email: kajal.ghoshray@saha.ac.in

1S. Sachdev, Quantum Phase Transition (Cambridge University Press,

New York, 1999).2F. D. M. Haldane, Phys. Rev. Lett. 50, 1153 (1983).3T. Giamarchi, C. Ruegg, and O. Tchenyshyov, Nat. Phys. B 4, 198 (2008).4K. Hida, J. Phys. Soc. Jpn. 63, 2359 (1994).5H. Kageyama, K. Yoshimura, R. Stern, N. V. Mushnikov, K. Onizuka,

M. Kato, K. Kosuge, C. P. Slichter, T. Goto, and Y. Ueda, Phys. Rev. Lett.

82, 3168 (1999).6T. Ono, H. Tanaka, H.A. Katori, F. Ishikawa, H. Mitamura, and T. Goto,

Phys. Rev. 67, 104431 (2003).7M. Azuma, T. Odaka, M. Takano, D. A. Vander Griend, K. R. Poeppelme-

ier, Y. Narumi, K. Kindo, Y. Mizuno, and S. Maekawa, Phys. Rev. B 62,

R3588 (2000).8M. Drillon, M. Belaiche, P. Legoll, J. Aride, A. Boukhari, and A. Moqine,

J. Magn. Magn. Mater. 128, 83 (1993).9A. A. Belik, A. Matsuo, M. Azuma, K. Kindo, and M. Takano, J. Solid

State Chem. 178, 709 (2005).10M. Azuma, Z. Hori, M. Takano, K. Ishida, and Y. Kitaoka, Phys. Rev.

Lett. 73, 3463 (1994).11H. Kageyama, M. Nishi, N. Aso, K. Onizuka, T. Yosihama, K. Nukui,

K. Kodama, K. Kakurai, and Y. Ueda, Phys. Rev. Lett. 84, 5876 (2000).

FIG. 6. The nuclear magnetization recovery curves at 4 and 6.5 K. The inset

shows the same at 12 and 6.5 K.

TABLE 2. Values of T1 at room temperature for two compounds.

Parameters, ms Ca3Cu2Ni(PO4)4 Ca3CuNi2(PO4)4

Ttheo11 for P1 0.69 1.61

Texp11 for P1 0.85 1.82

Ttheo11 for P2 5.94

Texp11 for P2 2.0 3.2

Low Temp. Phys. 38 (7), July 2012 M. Ghosh and K. Ghoshray 649

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 131.217.6.8 On:

Wed, 03 Sep 2014 09:45:23

12A. Podlesnyak, V. Pomjakushin, E. Pomjakushina, K. Conder, and A. Fur-

rer, Phys. Rev. B 76, 064420 (2007).13V. Yu. Pomjakushin, A. Furrer, D. V. Sheptyakov, E. V. Pomjakushina,

and K. Conder, Phys. Rev. B 76, 174433 (2007).14M. Ghosh, M. Majumder, K. Ghoshray, and S. Banerjee, Phys. Rev. B 81,

094401 (2010).15S. Yamamoto, H. Hori, Y. Furukawa, Y. Nishisaka, Y. Sumida, K.

Yamada, K. Kumagai, T. Asano, and Y. Inagaki, J. Phys. Soc. Jpn. 75,

074703 (2006).

16M. Ghosh, K. Ghoshray, B. Pahari, R. Sarkar, and A. Ghoshray, J. Phys.

Chem. Solids 68, 2183 (2007).17M. Ghosh, K. Ghoshray, M. Majumder, B. Bandyopadhyay, and A. Gho-

shray, Phys. Rev. B 81, 064409 (2010).18T. Moriya, Prog. Theor. Phys. 16, 23 (1956).

This article was published in English in the original Russian journal. Repro-

duced here with stylistic changes by AIP.

650 Low Temp. Phys. 38 (7), July 2012 M. Ghosh and K. Ghoshray

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 131.217.6.8 On:

Wed, 03 Sep 2014 09:45:23