Antiferromagnetic Ground State in Organic Quasi-1D Ferromagnet

�-Phase para-Nitrophenyl Nitronyl Nitroxide

Ko-Ichi KAJIYOSHI, Takashi KAMBE�, Masafumi TAMURA1 and Kokichi OSHIMA

Graduate School of Natural Science and Technology, Okayama University, 3-1-1 Tsushimanaka, Okayama 700-85301Condensed Molecular Materials Lab., RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198

(Received March 10, 2006; accepted April 24, 2006; published June 26, 2006)

We have studied the ground state of an organic quasi-one-dimensional (Q1D) ferromagnet, the �-phasepara-nitrophenyl nitronyl nitroxide (�-p-NPNN), by low-frequency ESR and magnetic torque measure-ments, and have found that it is an antiferromagnetically ordered state with an orthorhombic anisotropy.Antiferromagnetic resonance (AFMR) modes are observed below 0.6 K. From the angle dependence ofAFMR, the spin easy axis is found to point the [0�111] direction. The spin-flop (SF) field is estimated to be520 Oe. From the analysis of the AFMR modes in terms of a mean-field approximation, we haveestimated the interchain interactions and the anisotropy field parameters in the ground state. Themagnetic phase diagram along the spin easy axis has been determined using a high sensitivemicrocantilever technique. On the field dependence along the easy axis, the torque response showsanomalies at 520 Oe and around 2100 Oe, corresponding to the spin-flop and the full-polarization ofspins, respectively. The easy axis and anisotropy in the antiferromagnetic state are found to be wellexplained by our calculations of the dipolar energy with respect to the spin structure obtained in thiswork.

KEYWORDS: �-para-nitrophenyl nitronyl nitroxide, quasi-one-dimensional ferromagnet, electron spin resonance,magnetic torque

DOI: 10.1143/JPSJ.75.074702

1. Introduction

Four-type polymorphic structures exist in the para-nitro-phenyl nitronyl nitroxide (p-NPNN) family, which is called�, �, �, and �.1,2) Among them, �-phase is well known as thefirst genuine organic ferromagnet.1) The long-range orderingoccurs around Tc � 0:6 K without a static magnetic field.The stability of this phase allows studying well its structureas well as its magnetism. Generally speaking, the intermo-lecular arrangement of magnetic molecules plays a keyrole in deciding the sign of the magnetic interaction. In thep-NPNN family, NO–NO2 and NO–phenyl contact betweenneighboring molecules induce ferromagnetic interactionsbetween spins on molecules while NO–NO contact in-duces antiferromagnetic interaction due to a gain of directtransfer energy between singly-occupied-molecular-orbitals(SOMOs).1,2) This is because the wave function of theSOMO mainly exists around an ONCNO fragment in themolecule. For the �-phase, the magnetic coupling in theac-plane consists of NO–phenyl contact while the coup-ling along b-axis consists of NO–NO2 contact, leading tothe three-dimensional (3D) ferromagnet. By contrast, it isthought that the �-phase undergoes an antiferromagneticphase transition around TN ¼ 0:65 K.1) Above TN, evidenceof the one-dimensional ferromagnetic fluctuation is presentin the specific heat measurement under magnetic fields.Structurally, molecules along the c-axis are connected byNO–NO2 ( ferromagnetic) contact, forming a one-dimen-sional ferromagnetic chain.3) These chains couple antiferro-magnetically through NO–NO contact along the a-axis.4)

Therefore, the comparison of the spin structure of the �-phase with that of the �-phase is expected to give usefulinformation concerning the magneto-structural correlation in

magnetic molecules. However, the thermal instability of the�-phase have prevented us from examining its ground statein more detail. For example, the spin structure has not beenknown yet.

In this paper, we report ESR absorptions above andbelow TN using low-frequency and low-temperature ESR.We succeeded in observing antiferromagnetic resonance(AFMR) below TN for the first time. This observationdirectly establishes the antiferromagnetic ground state withan orthorhombic anisotropy of �-p-NPNN. We performedmagnetic torque measurements on single crystals using ahigh sensitive cantilever method.5) The torque responseshows clear anomaly around a spin-flop (SF) transition and aphase boundary between the antiferromagnetic phase and thefull-polarization of spins. We determine the magnetic phasediagram.

2. Experimental

Single crystals of �-phase of p-NPNN were grown bythe slow evaporation method. An acetonitrile dissolution ofp-NPNN molecules had been kept at 60 �C. Several dark-greenish single crystals, having a typical dimension of 0:5�0:5� 0:2 mm3, are obtained after the evaporation. Singlecrystals have plate-like shapes with crystal facets parallelto the bc-plane. Nakazawa et al. reported the instability of�-phase below room temperature (RT).1) To avoid degener-ation to the stable �-phase, we carefully held �-phasecrystals at a constant temperature in a clean atmosphere.Prior to every experiment, we checked the crystals by X-raydiffraction and selected the ones with no twinning andno contamination of other phases. Low-temperature X-raydiffraction measurements using single crystals indicate thatno evidence of any structural phase transition is presentdown to 25 K.6) The crystal data at 25 K are triclinic spacegroup P�11, a ¼ 8:985ð1Þ A, b ¼ 11:984ð2Þ A, c ¼ 6:4167ð5Þ�Corresponding author. E-mail: kambe@science.okayama-u.ac.jp

Journal of the Physical Society of Japan

Vol. 75, No. 7, July, 2006, 074702

#2006 The Physical Society of Japan

074702-1

A, � ¼ 97:015ð7Þ�, � ¼ 104:953ð8Þ�, � ¼ 81:914ð2Þ�, V ¼658:3ð1Þ A3, and Z ¼ 2. Figure 1 shows the NPNN mole-cule, the low-temperature structure and the intermolecularcontacts. NO–NO2 and NO–phenyl contacts are indicatedby the solid and dotted lines, respectively. The right figureshows the schematic map of magnetic interactions in thebc-plane, where p-NPNN molecules are depicted by closedellipses. The one-dimensional ferromagnetic chain (J12) runsparallel to the c-axis. The chains couple with each otherferromagnetically along the bþ c-axis (J13) and antiferro-magnetically along the a-axis (J14). The low-temperaturestructural analysis indicates anisotropic changes of contactsbetween molecules, suggesting the effective enhancement ofthe one-dimensionality of ferromagnetic chains. This resultis consistent with specific heat measurements and magneticsusceptibility measurements.1) Detailed structural data willbe published elsewhere.

The magnetic phase diagram of �-phase was presentedby magnetic susceptibility measurements and specific heatmeasurements (see Fig. 2).1) It has been proposed that thelong-range ordered phase below TN ¼ 0:6 K is antiferro-magnetic, but no other confirmation on the spin structure hasbeen given. To observe ESR absorptions in this orderedphase and in the vicinity of TN, a low-magnetic field (low-frequency) should be desirable. At the usual X-band fre-quency, we could not observe any ESR absorptions whichcan be identified as AFMR or ferromagnetic resonance(FMR) below 0.5 K because the used magnetic field at X-

band is larger than the phase boundary at 0.5 K. In Fig. 2,the EPR field at X-band is defined by an arrow X.

Accordingly, we constructed new ESR cavities availablefor the low-frequencies and low-temperatures below TN.One of them is a LC circuit resonator, where L is aninductance of pick-up coil and C is a capacitance. The LC

circuit resonator is tuned at the resonance frequency of�300 MHz, which corresponds to the EPR resonance fieldaround 100 Oersted, by adjusting the capacitance of theparallel condenser. Samples are placed at the center of thepick-up coil with a diameter around 2 mm. The large fillingfactor of the coil allows us to observe AFMR signals at lowtemperatures with a high accuracy. The oscillating magneticfield (H1) generated by the coil is perpendicular to the staticmagnetic field. All LC resonators including samples aredirectly cooled by liquid 3He.

Another type of cavity is a bridged loop-gapped (LG)resonator.7,8) The bridged LG resonator was used at theregion up to 1 GHz. The schematic diagram of the LGresonator is shown in Fig. 3. The LG resonator part in thefigure is a Cu cylinder, and it has a narrow gap. By tuning

O N N

O

N

OO

Fig. 1. NPNN molecule and low-temperature structure of �-p-NPNN. NO–NO2 and NO–phenyl contacts are indicated by the solid and dotted lines,

respectively. The right figure shows the schematic map of magnetic interactions in the bc-plane, where p-NPNN molecules are depicted by the closed

ellipses.

0.4

0.5

0.6

0.7

0.8

0

Tem

pera

ture

(K

)

Magnetic field (Oe)

P

AF

LF

X

300020001000

Fig. 2. Magnetic phase diagram of �-p-NPNN. The closed and open

circles express the phase transition temperatures, which were obtained by

specific heat measurements and ac susceptibility measurements, respec-

tively.1) AF; antiferromagnetic, P; paramagnetic. Arrows indicate EPR

magnetic fields at usual X-band resonator (X) and homemade low-

frequency resonator (LF).

Closed ring

LGR

Coupling loop

coaxial cablerod

INSERT QUARTZ DEWAR

Shield

(With samples)

Static Field

Fig. 3. Schematic diagram of LG resonator. Samples are located in the

double-walled quartz dewar (not shown).

J. Phys. Soc. Jpn., Vol. 75, No. 7 K.-I. KAJIYOSHI et al.

074702-2

the capacitance at the gap, the resonance frequency variesquasi-continuously from 700 MHz to 3 GHz. A microwavecoaxial cable is linked by the coupling loop at the top part ofthe LGR. An ideal coupling is established by adjusting thelength t1 in the figure. The change in length t2 varies theresonance frequency of LG resonator precisely. A small-diameter double-walled quartz dewar is inserted in the LGresonator and samples located inside the dewar are directlyimmersed in liquid 3He. The unloaded quality factor of theresonator is approximately 2300 at 1.4 K and 3 GHz. Thestatic magnetic field was controlled by the Bruker ER-032Mand applied in the horizontal plane. The temperature ofsamples was monitored by the vapor pressure of liquid 3Heand by the Cernox resistance thermometer inside the dewar.The lowest available temperature is about 0.45 K. ESRabsorption signals were directly obtained by monitoring thereflection intensity from the resonator using a scalar networkanalyzer (Agilent Technology, 8719ET).

We measured magnetic torque responses using commer-cial piezo-resistive microcantilever (Seiko instruments).This technique had been developed by Ohmichi et al. fordHvA measurements.5) A sample is attached at the tip ofthe cantilever using an epoxy glue. The inset photo ofFig. 8 shows the cantilever with the NPNN single crystal.A resistance bridge circuit was used and the off-balancedvoltage of the bridge circuit was detected by a lock-in-amplifier (Stanford Research System SR830). The magnetictorque signal is proportional to the off-balanced voltage ofthe bridge circuit. The sample and cantilever are directlycooled by liquid 3He and the sample temperature wasmonitored by a Cernox resistance thermometer. The canti-lever is mounted on the adjustable sample holder, and itsflexing direction is in the horizontal plane. Therefore, thedirection of the applied field can be rotated with respect tothe crystal axis.

3. Results and Discussion

3.1 Antiferromagnetic resonanceFigure 4(a) shows a temperature dependence of electron

paramagnetic resonance (EPR). The rf frequency is about286 MHz, and the static magnetic field is applied parallel tothe c-axis for EPR measurements. EPR signals broaden withdecreasing temperatures, independently from the magnetic

field direction, and completely disappear below 0.6 K. Atthe lowest temperature, no trace of FMR signal, which isobserved in �-phase crystals,9) is present. This implies thatsamples have experienced neither a degeneration to thestable �-phase, nor a degradation into the Curie-like para-magnetic contamination.

After EPR signals disappeared, new ESR absorptionsignals emerge at different magnetic fields within a specificangular region.10) Figure 5(a) shows the angular dependenceof absorptions at 0.45 K, where the magnetic field is appliedin the bc-plane. Angles defining the applied magnetic fielddirection, � and �, relative to crystal axes are shown in theinset of Fig. 5(b). Note that the observed resonance fields areremarkably different from the corresponding EPR fields. At1.16 GHz, observed ESR absorptions consist of two resonantmodes.11) And these modes can be observed in narrowangular regions: �25 < � < �15� and 0 < � < 60� at 284MHz, and �55 < � < 15� at 1.16 GHz, respectively. Asshown in Fig. 5(b), a typical angular dependence of AFMRaround the spin-easy axis was clearly observed within thebc-plane (� ¼ 0�). At �easy ¼ �20� � 10� and � ¼ 0�, themaximum separation between two AFMR modes is obtainedand the angular dependence shows a symmetric ellipsearound �easy. The resonance fields of lower-field signalsshow a negative slope as a function of the frequency. Theseresults are consistent with our assumption on the spin-easyaxis. Note that the spin-easy axis is approximately in [0�111]direction.

Next, we analyze the AFMR in terms of the mean-fieldapproximation.12) Specific heat measurements under themagnetic field showed the existence of one-dimensionalferromagnetic fluctuation above TN. The overall temperaturedependence of the specific heat is well understood bythe one-dimensional ferromagnetic Heisenberg model withinterchain couplings.1) As shown in Fig. 1, the one-dimen-sional ferromagnetic chain runs along the c-axis. Therefore,on our assumption that intrachain spins are strongly coupled,we applied a four-sublattice model with an orthorhombicanisotropy in order to investigate AFMR modes. Figure 6shows the schematic diagram of the four-sublattice model.Spins along the c-axis form ferromagnetic chains. The sys-tem can also be seen as ferromagnetic planes coupled

200

Ref

lect

ed p

ower

(a.

u.)

1.40 K1.25 K1.09 K0.95 K0.71 K

0.60 K

0.53 K

abc

0

20

40

60

80(a) (b)

Magnetic Field (Oe) Temperature (K)

Line Width (O

e)

0.5 1.510 100

Fig. 4. Temperature dependence of EPR spectra (a) and line width (b) of

�-p-NPNN at 284 MHz. In (a), the magnetic field is applied parallel to the

c-axis.

100 400 700

Sig

nal I

nten

sity

(a.

u.)

Mangetic Field (Oe)

θ=φ=0

2515105

−5−15

−25

−35−45

−50

−55

200 400 600

-50

0

50

Resonance Field (Oe)

(a) (b)

φ=0

a*

bc

H

Fig. 5. (a) Angular dependence of antiferromagnetic resonance (AFMR)

spectra of �-p-NPNN at 1.16 GHz. T ¼ 0:45 K. (b) The inner circle

shows the angular dependence of resonance fields at 284 MHz, and the

outer circle shows the dependence at 1.16 GHz. T ¼ 0:45 K.

J. Phys. Soc. Jpn., Vol. 75, No. 7 K.-I. KAJIYOSHI et al.

074702-3

antiferromagnetically, because the in-plane coupling (J13)along bþ c is ferromagnetic and the inter-plane coupling(J14) along a is antiferromagnetic,1) which are shown bydotted and dot-dashed lines in Fig. 6, respectively. Usingthis model, a total free energy can be formulated within themean-field approximation as follows:

F ¼ �X4

i¼1

Mi �H0

þ AðM1 �M2 þM3 �M4Þ þ BðM1 �M3 þM2 �M4Þ

þK1

2

X4

i¼1

Mxi

M0

� �þ

K2

2

X4

i¼1

Myi

M0

� �; ð1Þ

A ¼ �4

N

4jJ13jðg�BÞ2

; B ¼4

N

4jJ14jðg�BÞ2

; ð2Þ

where Mi ¼ ðN=4Þg�BhSii, H0 is the static magnetic field,A and B are the molecular field coefficients, K1 and K2 arethe orthorhombic anisotropy parameters, and M0 is themagnitude of the sublattice moment. Figure 7 shows theobtained frequency-field diagram of AFMR. Solid linesshow typically observable modes while dotted lines showthe unobservable ones in ESR measurements. The highfrequency modes have no absorption intensity due to theout-of-phase motion of sublattice moments between theferromagnetic layers. Consequently, the four sublatticemodel can be well described by the two-dimensionalferromagnetic layers coupled antiferromagnetically, in otherwords, two sublattice model. To analyze the experiments,we assume the relation between the intra- and inter-planeinteraction, jJ13=kBj þ jJ14=kBj ¼ 0:2 K.1) Considering thatthe !5-mode should soften at the saturation magnetic fieldwhich will be discussed in what follows, we could evaluatefollowing parameters:

J13=kB¼ 0:127 K; J14=kB¼ �0:073 K;

K1=M0¼ 120 Oe; K2=M0¼ 230 Oe:

The SF field is also estimated to be 520 Oe. This valuesatisfactorily agrees with magnetic torque measurements.

3.2 Magnetic torque measurementsFigure 8 shows the magnetic field dependence of mag-

netic torque signals at 0.45 K, where the magnetic field isapplied parallel in the bc-plane. The definition of �, shown inthe inset of Fig. 8, is the same as that in Fig. 5. The torquesignal at 1.45 K is also shown in the inset, and depends

parabolically on the magnetic field. No clear angularvariation and no clear anomaly were found at this temper-ature. On the other hand, at T ¼ 0:45 K, magnetic torquesignals show a peculiar angular dependence. A sharpanomaly around HSF (�520 Oe) is observed in the vicinityof � � �35� while a step like anomaly observed aroundHc (�2:1 kOe) has a weak angular dependence. No otheranomaly is found up to 7 kOe. These angular dependencesresemble well that of antiferromagnetic torque responsesin -(BEDT)2FeCl4, which exhibits an antiferromagneticordering below 8 K.13) In the case of -(BEDT)2FeCl4, thelower field anomaly corresponds to the SF transition. Asshown in the previous section, our low-frequency ESR

J13

J14

J12S1

S1

S2

S2

S3

S3

S4

S4

c

a

b+c

b

Fig. 6. Model for the AF ground state. The suffix, i, of the spin, Si,

indicate the number of the assigned sublattice moment. The strong

ferromagnetic coupling runs along the c-axis as shown by solid lines. The

in-plane (along bþ c) coupling is ferromagnetic and the inter-plane

(along a-axis) coupling is antiferromagnetic, as shown by dotted and dot-

dashed lines.

Magnetic Field (Oe)500 10000

0

500

1000

5000

4500

5500

Spin Flop field

Fig. 7. Frequency (!=�)–field (H) diagram of AFMR for H0 k spin-easy

direction. Closed squares are experimental results at 0.45 K. Lines

show the calculated result with J13=kB ¼ 0:127 K, J14=kB ¼ �0:073 K,

K1=M0 ¼ 120 Oe and K2=M0 ¼ 230 Oe. Solid lines are observable modes

while dotted lines are unobservable ones.

0 1000 2000 3000

Tor

que

Sig

nal I

nten

sity

(a.

u.)

Magnetic Field (Oe)

HCH

SF

10o

-10o

-30o

-35o

-40o

-60o

θ=

0 1000 2000 3000Magnetic Field (Oe)

Cantilever

Samplec

b

1.45 K

Tor

que

Sig

nal I

nten

sity

(a.

u.)

0.45 K

θ

Fig. 8. Magnetic field dependence of torque signals at 0.45 K. HSF and Hc

are the spin-flop (SF) field and the SF–F transition field, respectively. The

inset photo shows the cantilever with a single crystal of �-p-NPNN and

the angle � is also defined. The inset figure shows the field dependence of

torque signal in the paramagnetic phase at 1.45 K.

J. Phys. Soc. Jpn., Vol. 75, No. 7 K.-I. KAJIYOSHI et al.

074702-4

measurements support this scenario and typical AFMRsignals along the spin easy axis are clearly observed aroundthis direction. Accordingly, we conclude that the lower fieldanomaly at HSF is the SF transition and that � � �35� � 10�

direction is parallel to the spin-easy axis. On the other hand,the anomaly around Hc should correspond to the full-polarization (F) state as evidenced by the weak angulardependence around Hc. The weak field dependence above Hc

may be due to the thermal fluctuation of spins.Figure 9 shows the angular dependence of magnetic

torque signals at 0.45 K. The case of two magnetic fields(H ¼ 200 Oe < HSF and H ¼ 2:5 kOe > Hc) are shown. Theangular dependence, except for the signal amplitudes,remains unchanged above HSF. The magnitude of the torquesignals I follows, sin 2ð� � �0Þ, in each case. Contributionof higher-order anisotropic terms is negligible. We estimate�0 in each case to be �AF

0 ¼ �33� in the AF phase and �F0 ¼�20� in the F state, respectively. The inset of Fig. 9 showsthe magnetic field dependence of the differential of �0,�d�0=dH. �0 shows a rapid change around 2 kOe, which islower than Hc. �

AF0 in the AF phase corresponds well to the

direction where the anomaly of the torque signal is observedaround HSF, as shown in Fig. 8.

Figure 10 shows the magnetic phase diagram of �-p-NPNN along the spin-easy direction. The inset of the figureshows the temperature dependence of torque signals, wherethe magnetic field direction is parallel to the spin-easy axis.Rapid decreases of the torque signals, which are shown byarrows, indicate the onset temperatures of the Neel orderedstate. The onset temperatures are indicated by the closedcircles in the main panel of the figure. The closed trianglesand squares indicate transition fields obtained by the fielddependence of torque signals at the lowest temperature.Together with the present torque experiment, the phaseboundary determined by specific heat and ac susceptibilitymeasurements are also indicated.1) The phase boundary byour torque measurements shows slightly different temper-ature dependence compared with the specific heat and thesusceptibility measurements. This slight difference plays arole in estimating the antiferromagnetic interaction betweenferromagnetic layers. Thus, it is necessary to measure thephase boundary at the lower temperatures for more precise

discussions on exchange interactions. Moreover, we pointout that the phase boundary as well as the anomaly in �0below Hc may be associated with an anomaly of dipole–dipole energy between the spins, which will be discussed inthe next section.

3.3 Dipolar energy calculationsBoth AFMR and magnetic torque experiments have

revealed that the ground state of quasi-one-dimensional(Q1D) ferromagnet �-p-NPNN is antiferromagnetic with theorthorhombic anisotropy. In order to elucidate a major con-tribution on the spin anisotropy, we calculate the mag-netic dipole–dipole energy and compare it with experimentsin both phases. Since the SOMO wave function mainlyspreads around NO groups of NPNN molecule, we assumedthat the spin density is distributed equivalently on ONatoms.14) This model calculation has reproduced the pre-vious calculation on the dipole energy in the �-p-NPNN.15)

As denoted previously, the sign of magnetic interactionsbetween molecules could be expected by the type of contactbetween the molecular fragments. NO–NO2 contact alongthe c-axis and NO–phenyl contact along the bþ c directionshould induce ferromagnetic interactions. In the AF phase,we assume that spins in the bc-plane align parallel to eachother, and antiparallel alignment between the neighboringlayers.16)

Figures 11 and 12 show the calculated dipolar energyin the AF and F phases, respectively. In the maps displayedin the lower panels, the abscissae are the spin direction inthe bc-plane and the ordinates are the spin direction alongthe a�-axis. Sinusoidal curves in the upper panels indicatethe dipole energies as a function of the spin direction in thebc-plane. In the AF phase, the in-plane anisotropy is smalland, thus, the spin direction is well confined in the bc-plane.The lowest dipole energy is obtained at � ¼ �30� in thebc-plane. This result satisfactorily reproduces torque experi-ments, where the spin-easy axis is obtained around �AF,

-100 -50 0 50 100

2500 Oe 200 Oe

Tor

que

Sig

nal I

nten

sity

(a.

u.)

θ (o)

0

0.1

0.2

0 1000 2000 3000

−d θ

0/ d

H

Magnetic Field (Oe)

Fig. 9. Angular dependence of torque signals at 0.45 K. Solid lines are

defined in the text. The inset shows the magnetic field dependence of the

differential of �0.

0

1000

2000

3000

Temperature (K)

Mag

netic

Fie

ld (

Oe)

AF

SF

P

Sig

nal I

nten

sity

(a.

u.)

Temperature (K)

400 Oe

1000Oe

1500Oe

0.4 0.70.60.5

10.4 0.6 0.8

Fig. 10. Magnetic phase diagram of �-p-NPNN along the spin-easy axis

determined from magnetic torque experiments. Closed circles show the

on-set temperature of the Neel ordered state under the magnetic field and

the closed triangle and square show the transition fields along the spin-

easy axis. Open squares and open circles were determined by the specific

heat and the ac susceptibility measurements, respectively.1) Solid and

dotted lines are guide for the eye. The inset shows the temperature

dependences of torque signals at various magnetic fields.

J. Phys. Soc. Jpn., Vol. 75, No. 7 K.-I. KAJIYOSHI et al.

074702-5

as shown in Fig. 9. Accordingly, we conclude that theanisotropy of spins in the AF phase is mainly determined bythe dipole–dipole interaction.

In contrast to the AF case, the F phase has a stronganisotropy in the in-plane dipole energy. The lowest dipoleenergy is obtained at � ¼ �33� in the bc-plane. This resultis slightly different from that of torque experiments; �F ��20�. This significant difference could be resolved byassuming that the dihedral angle between the phenyl groupand the ONCNO fragment decreases, or that the spin densitydistribution becomes concentrated on N atoms in the Fphase. It may suggest the existence of molecular magneto-striction. The molecular orbital calculation suggests thata flat arrangement between the phenyl group and theONCNO fragment is energetically favorable for the isolatedNPNN molecule.17) Taking account of this, we can expectsome molecular distortion under high magnetic field nearthe saturation (see Fig. 9), which can affect the dipolaranisotropy such that expected from the present results. Ourpreliminary magnetization measurements indicate that themagnetization saturates around 7 T at 5 K.3) Though theX-ray measurement below TN is difficult, it seems alsouseful to examine the possibility of structural change nearthe saturation above TN, because there should be noboundary between the F state and the saturated paramagneticstate. We expect that the same structural re-arrangementunder the influence of the magnetic field may exist in theparamagnetic phase near the saturation.

4. Conclusion

We have observed the antiferromagnetic resonance inthe low-temperature phase below TN of �-p-NPNN for thefirst time. The frequency–field relation proves the collinearspin arrangement with an orthorhombic anisotropy. Low-temperature magnetic torque measurements had been per-formed using the microcantilever technique. The spin-easy

axis aligns approximately parallel to the [0�111] directionin the antiferromagnetic phase. This is almost consistentwith the result of AFMR measurement. However, the torquecurve in the fully-polarized phase has a different anisotropy.The calculated magnetic dipole–dipole energy indicatesits main contribution on the anisotropy in the AF phase.However, the calculation in the F phase fails to reproduceexperimental results. This is an open question for the futureresearch.

Acknowledgments

We are grateful to Professor Minoru Kinoshita forconsiderate encouragements. They would like to thankProfessor Mitsutaka Okumura for stimulating discussionson the quantum chemical calculations. We also thank thestaff of the Equipment Development Center of Institute forMolecular Science for the technical support for the loop-gapresonator.

1) Y. Nakazawa, M. Tamura, N. Shirakawa, D. Shiomi, M. Takahashi,

M. Kinoshita and M. Ishikawa: Phys. Rev. B 46 (1992) 8906.

2) M. Tamura, Y. Hosokoshi, D. Shiomi, M. Kinoshita, Y. Nakazawa,

M. Ishikawa, H. Sawa, T. Kitazawa, A. Eguchi, Y. Nishio and K.

Kajita: J. Phys. Soc. Jpn. 72 (2003) 1735.

3) P. Turek, K. Nozawa, D. Shiomi, K. Awaga, T. Inabe, Y. Maruyama

and M. Kinoshita: Chem. Phys. Lett. 180 (1991) 327.

4) These consideration on the magnetic interactions has been discussed

on the basis of the room temperature structure, because no low-

temperature structural study was reported. We have analyzed low-

temperature structure of �-phase at 25 K and will report in the

separature paper.

5) E. Ohmichi and T. Osada: Rev. Sci. Instrum. 73 (2002) 3022.

6) T. Kambe, K. Kajiyoshi, K. Oshima, M. Tamura and M. Kinoshita:

Synth. Met. 154 (2005) 301.

7) W. N. Hardy and L. A. Whitehead: Rev. Sci. Instrum. 52 (1981) 213.

8) H. Hirata and M. Ono: Rev. Sci. Instrum. 67 (1996) 73.

9) T. Kambe, K. Kajiyoshi, K. Oshima, M. Tamura and M. Kinoshita:

Polyhedron 24 (2005) 2468.

0

-250 -200 -150 -100 -50 0 50

-80

-60

-40

-20

0

20

40

60

80

θ [o]

a*

-a*

Ene

rgy

[erg

/ m

olec

ule]

-30 deg.

bc-plane

[erg / molecule]

c-b b-cb

Fig. 11. Calculation of the dipole–dipole energy as a function of spin

direction in the AF-phase. The abscissae are the spin direction in the

bc-plane. The ordinate in the lower panel is the spin direction along

a�-axis. The sinusoidal curve in the upper panel indicates the change

of the dipole energy in the bc-plane.

-250 -200 -150 -100 -50 0 50

0

-80

-60

-40

-20

20

40

60

80

a*

-a*

0

[erg / molecule]

Ene

rgy

[erg

/ m

olec

ule]

θ [o]

bc-plane

-33 deg.

c-b b-cb

Fig. 12. Calculation of the dipole–dipole energy in the full-polarization

(F) state. The abscissae are the spin direction in the bc-plane. The

ordinate in the lower panel is the spin direction along a�-axis. The

sinusoidal curve in the upper panel indicates the change of the dipole

energy in the bc-plane.

J. Phys. Soc. Jpn., Vol. 75, No. 7 K.-I. KAJIYOSHI et al.

074702-6

10) K. Kajiyoshi, T. Kambe, K. Oshima, M. Tamura and M. Kinoshita:

Polyhedron 24 (2005) 2464.

11) The asymmetric line shapes of AFMR signals remind us powder

pattern. It is known that the FMR signals for the �-p-NPNN also have

asymmetric line shapes. Thus, it may be common characteristic on the

molecular spin system and may be associated with the non-localization

of spins on isolated molecules.

12) T. Nagamiya, K. Yosida and R. Kubo: Adv. Phys. 4 (1955) 1.

13) T. Sasaki, H. Uozaki, S. Endo and N. Tokyota: Synth. Met. 120 (2001)

759.

14) A. Zheludev, M. Bonnet, E. Ressouche, J. Schweizer, M. Wan and

H. Wang: J. Magn. Magn. Mater. 135 (1994) 147.

15) M. Kinoshita: Jpn. J. Appl. Phys. 33 (1994) 5718.

16) M. Okumura, W. Mori and K. Yamaguchi: Chem. Phys. Lett. 219

(1994) 36.

17) M. Okumura: private communication.

18) K. Takeda, M. Mito, T. Kawae, H. Deguchi, S. Takagi, M. Okumura,

T. Kawakami, K. Yamaguchi and M. Kinoshita: Chem. Phys. Lett. 308

(1999) 181.

J. Phys. Soc. Jpn., Vol. 75, No. 7 K.-I. KAJIYOSHI et al.

074702-7