Solid State Communications, Vol. 11, pp. 1423—1429,1972.PergamonPress. Printedin Great Britain

MAGNETIC MEASUREMENTSON (CH3NH3)2MnCI4,A QUASI TWO-DIMENSIONAL HEISENBERG ANTIFERROMAGNET

W.D. van Arnstel and L.J. de Jongh

NatuurkundigLaboratoriumder Universiteit van Amsterdam,The Netherlands

(Received8 September1972 by G.W. Rathenau)

We havemeasuredthe magneticsusceptibility of the first memberofa seriesof tetragonalcompoundsof generalformula (C~H2~+1NH3)2MnCl4 (n = 1,2,3 ), which may be consideredto consistof nearlyisolatedantiferromagneticlayers. The interlayercoupling J’ is esti-matedto be 10_9-_10_8of the intralayerexchangeJ, a result 102smallerthan maybe realizedwithin the well-known K2NiF4 structure.Quantitatively,we obtain f/k = —5.0 ±0.2K for the intralayer exchangeandHA /HE = 1.1 x 10~for the anisotropy,the latter being derivedfrom the spin flop field.

INTRODUCTION up to the present,indications that it must indeedbe of the order of the dipolar couplingcan be

TWO-DIMENSIONAL (2-d) Heisenbergmagnetism found by consideringthe magneticstructureofhas hitherto mainly beenstudiedin compounds thesecompounds,asdeterminedby neutrondif-with the K2NiF4 structurefor the antiferromag- fraction. For K2NiF4 itself, a ferromagneticnetic,and in the closely relatedseries(C~H~+1 alignmentalong the c-axiswas found,

4as wouldNH.

4)2CuX4 (n = 1,2,3 X Cl, Br) for the be favouredby the dipolar coupling. On the otherferromagneticcase(for a short review andrefer- hand in e.g. Rb2MnF4 a ferro- as well asan anti-encesseee.g.De Jonghet a!.’). In the K2NiF4 ferromagneticalignmenthas beenobserved,

4bothstructure,the 2-d characterarisesmainly from structuresoccurringevenin the samesample.the body-centeredtetragonalsymmetry,which in Clearly, thereis a subtlebalancebetweenthecaseof an antiferromagneticalignment within different types of interlayercoupling,which mightthe NiF

2 sheetsleadsto a cancellationof the be understoodon basis of a competitionbetweeninteractionbetweennearestneighbouringlayers,

2 the dipolar and the superexchangeinterlayerat leastas far as the static propertiesare con- interaction. If this would indeedbe the case,itcerned.An estimateof the superexchangeinter- follows that bothmustbe of the sameorder ofactionbetweennext nearestlayers,J’, may be magnitude,and alsothat the mentionedrule ofobtainedusingthe rule of thumb that eachad- thumb gives a reasonableestimate.ditional interveningnonmagneticanionreducesthe interactionat leastby a factor 102. Since ~‘ On the otherhand,in the seriesof Cu com-and J occur via 4 and 1 ligands,respectively, poundsthe symmetryargumentdoesnot apply,this yields J’ 106J, where J is the exchange sincethe layersare ferromagrieticallyordered.within the layer. The dipolarcoupling between Here the 2-d propertiesoriginatefrom the factthe next nearestlayersis evensmaller;calcu- that the magneticlayersare separatedbe twolations of Colpa3give lO~_lO of J, depen- sheetsof alkyl ammoniumgroups. By varying nding on the particularcompound.Althoughno from 1—10, the separationbetweenthe Cu layersdirect determinationof J’ has beenaccomplished is enlargedby a factor 3. Estimatesof the

1423

1424 MAGNETIC ~VEASUREMENTSON (CH3NH3)2MnCl4 Vol. 11, No. 10

superexchangeinteractionbetweenadjacent We mentionthat independentfrom our work,layers, usingthe sameargumentgiven above, EPR measurementson (CH3NH3)2MnC14andlead to I’ = 10

3—1020J for n = 1—10. The (C2H5NH3)2MnC14havebeencarried out by

dipolar interlayer couplinghasbeencalculated Boeschet a!. ~ For the ethyl compoundour esti-by Colpa~ to be smaller than 10-’ of J~for all n. matesfor the inzerlayer interactionare 10-14and

2.2 x 10b0 of J for thesuperexchangeand the

Now it occurredto the authorsthat a corn- dipolar part, respectively.The correspondingbination of the separationmechanismand the numbersfor (C3H7NH3)2MnCl4are 1018andsymmetry argumentwould result in valuesof 1.3 x 10h1.J’/J~smaller by someorders of magnitudethan

thosewithin the K2NiF4 structure.We therefore It should be pointed out that all the abovedecidedto try to replaceCu by Mn, in the hope estimatesfor J’/J~apply to the ideal crystal

that the exchangewithin the layer would change structures.In caseof lattice imperfectionsorfrom the ferro- to the antiferromagneticsign. distortions of e.g. magnetostrictiveorigin, theIndeed we succeededin growing single crystals symmetryargumentloses its validity and J’/J~of (CH3NH3)2MnCl4and, as indicatedby the title will becomeone or two orders of magnitudelarger.

of this paper, this expectationwas fulfilled. On the whole oneexpectsthat, due to the presenceSince the crystal structure is of face-centered of theseimperfections as well as of phonons,tetragonalsymmetry,

6~we may use the symmetry valuesof J’/J( smaller than say about10b0 are

argumentso that the interlayer coupling is again not likely to be feasible.betweennext nearestlayer, leadingnow to J’ =

10~°Jas far as the superexchangeis concerned, EXPERIMENTAL DETAILSas J involves two (CH

3NH3) groups and two Clanions.For the dipolar interlayer coupling Colpa

3 Single crystals of (CH3NH3)2MnC14were

finds 2.5 x 10~J. preparedby slow evaporationof a solution of

manganesechloride and methyl ammoniumchlor-Unfortunately, little can be said beforehand ide in ethyl alcohol. It proved to be useful to

aboutthe amount of anisotropy,which is another ensurean as low as possibleconcentrationofdeviation from the ideal 2-d Heisenbergsystem water, in order to avoid the formation of im-

met in the experimentalexamples.Quite gener- purities ase.g. MnCI2. 4H20. In spiteof ourally, one may expectthree sourcesof anisotropy precautionsthe presenceof a small amount ofin thesecompounds,namely the dipolar inter- impurity could not be prevented,as will be seenactions,the crystal field effects andthe super- below.exchangemechanism.Crystal field effectswillof coursebe very small for the Mn

2~ion, in fact The sampleon which measurementswere

observedvalues of the anisotropyoften agree taken, wasa small pink platelet, of dimensionsquite well with thosecalculated from the dipolar 3.6 x 2.7 x 0.8mm3,weighing about 13mg.interaction (a = HA/HE 10~).Yet, in one corn- Chemicalanalysesproved the weight percentages

pound (BaMnF4)

8a considerablylower value has of Cl andMn to be within 1 per centof the cal-beenmeasured,which may arisewhen the various culatedvalues. No tracesof Cu, Fe, Ni, Cr andsourcesof anisotropyhavedifferent preferential Co largerthan 500ppmcould be detected.A higherdirections so that they may partly canceleach resolution could not be obtainedas a consequenceother. Evidently, finding sucha compoundis a of the small samplesize.matterof trial anderror. In the presentcasetheanisotropyis abouttwo thirds of thecalculated An X-ray analysisat room temperaturebydipolar result. But since the anisotropyis now Moleman’°of our laboratory, showed(CH

3NH3)2

about 106 times largerthan the estimated inter- MnC14 to crystallize in the face-centeredtetrag-layer coupling, it is evident that in this compound onal structure,with lattice parametersa0= 7.2353it is by far the most important deviation from (±0.0002)A and co = 19.3994(±0.0007)A.ideality that we have to reckonwith.

Vol. 11, No. 10 MAGNETIC MEASUREMENTS ON (CH3NH3)2MnC14 1425

_______________________________ Although the susceptibility exhibits the

(CH3NH3)2MnCI~ features characteristicof the quasi2-d rnaganese2~l0~ salts,

8’12 deviationsare found at low temperatures,

which may be attributedto the presenceof an~f’ ~ impurity. This is concludedfrom the fact that ~

I / ~ doesnot decayto zero, whereas reaches

1~10~r H13 ku valuesat T = 0 much higher (relative to

~ ~:}~ than thoseobservedin other 2-d Mn salts.6.13 kO. Although the impurity effect is relatively small,

/ : N: 817 ku II~7 N 616k0. in earlier measurementson more contaminated

0L~ samplesimpurity contributions nearly 10 times0 100 TI~)— 200 30 largerweredetected.The reduction of the im-

FIG. 1. Molar magneticsusceptibility of the purity content was realized by the processofsingle crystal of (CH3NH3)2MnCl4 as a function recrystallization. Since the transition temperatureof temperature,measuredfor threedifferent field associatedwith the impurity is below 4.2 K, wevalues. The symbols refer to the measuring . . . 13

points, the full curves representthe true sus- anticipatethe presenceof MnC12(T0 1.9K) orceptibility of (CH3NH3)2MnCl4, obtainedafter MnCl2. 4H20 (T~ 1.6K)’

4 in our samples.correctionfor an impurity contribution. The Evidence for this mayalso be the fact that intransition temperature,identified with the tem- Rb

2MnC14 a similar impurity contribution has

which 3x71/aT reachesits maximum, beendetected(De Jongh,unpublished).In thisC — case, the measurementswere extendedbelow

4.2 K andthe T0 of the impurity was found atSusceptibility experimentsin the temperature 1.9K (in a field of about l0koe).

range4.2—300K havebeenperformedin a pen-dulum magnetometer(force method), the tempera- In orderto separatethe impurity contributionture beingdeterminedwith a thermocouple.The as well as to analysethe data,we havemade use

overall accuracyof the measuringpoints is of two theoreticalpredictionsconcerningthe sus-1—2 per cent. In order to obtain a value for the ceptibility. At high temperaturesthe seriesex-anisotropy,a determinationof the spin flop field pansionfor the susceptibility for thequadraticwascarried out, by measuringthe magnetization lattice hasbeenused.This is an expansionin

curve at T = 4.2K in fields up to 70koe. This powersof J/kT of which 6 termshavebeencal-wasdone in cooperationwith Roeland,Muller culated for the quadratic lattice by Rushbrookeand Breur in the highfield pulsemagnetof our and Wood,~ whereasan additional 7th term haslaboratory.1~ beenobtainedby Stephensonet a!. 16 Due to the

finite numberof terms known in the expansion,EXPERIMENTAL RESULTS AND DISCUSSIONS this theorybreaksdown at low temperatures.

However, it hasbeenshownby Lines that it

In Fig. 1 we haveplotted the magneticsus- just coversthe temperaturerangein which theceptibility as measuredwith the field parallel to broad maximum occurs. We havecalculatedthethe c-axis (perpendicularto the platelet) and seriesprediction for >~with 5, 6 and 7 terms inwithin the planeperpendicularto this axis. In the expansionand used it in the temperaturerangethis plane no measurableanisotropycould be for which the results did not differ by more than

detected.The susceptibility hasbeencorrected one per cent, which amountsto kT/~JI> 12.for a diamagneticcontribution of — 0.58x 1O~cm

3/g, which, following Breed’2 hasbeenesti- On the other hand at T = 0, we mayrely onmatedfrom the known susceptibilities of diamag- the prediction for the perpendicularsusceptibilitynetic compoundscontainingthesameelements from spin-wavetheory. According to Keffer’8that constitute (CH

3NH3)iMnCl4. This is to be >~‘~(0) is given by

comparedwith the value )~~ax= 66 x 106cm3/g, )~° I ExS(a) e(a)

attainedat the roundedmaximum(Fig. 1) occur- )C~(0) = — _____ — ________ . (1)1 ‘ ~!.aI s (2÷a)zSring at about82K. 2 L

1426 MAGNETIC NEASUREMENTS ON (CH3NH3)2MnC14 Vol. 11, No. 10

/ ~ /_

02k- 2-’ I- / -

l~ 2 — /FIG. 2. The reducedsusceptibility of (CH3NH3)2 / 2MnC14 plotted vs. a relative temperature.Here 6is the Curie—Weisstemperatureand C the Curie _________________________constant.Also shownare the theoreticalpredic- 3 10 20

306kO.’i0

tions from molecularfield theory (M.F.) for thesusceptibility in the paramagneticregionand the FIG. 3. Magnetizationcurve of (CH

3NH3)2MnC14perpendicularsusceptibility of an antiferromag- measuredat T = 4.2K with the field parallel tonet; the prediction from the high-temperature thec-axis. Open triangles are the results for theseriesexpansion(H.T.S.) for the quadratic uncorrected~ at 4.2K measuredwith the pen-Heisenbergantiferromagnetwith S = 5/2, and dulum magnetometer(seeFig. 1). The other sym-the spin wave result for ~ (T = 0) for the same bols refer to three different measurementsin themodel (horizontal broken li’ne). high-field pulse magnet.

Here Ng2~4/4z~J!denotesthe molecular was obtained.In the next stepwe determinedthe

field prediction for ~j (0), a is the anisotropy combinationof i/k and C~that give best agree-parameterHA/HE, S is the spin value, z the mentbetweenthevalues for y.~(0) andXmax and

numberof nearestneighboursand AS(a)and e(a) the theoreticalresults (it hasbeenfound byarecorrectionsarising from the effects of zero- De Jonghand Colpa2°that the value ~max’

point spin deviations. Values for AS(a)as a attainedat the susceptibility maximum, providesfunctionof a havebeengiven by Lines17 and by a more reliable criterion for determining i/k thanColpa et a!. 19 The quantity e(a) hasbeenshown the temperatureTmax at which this maximumoc-by Breed12 to be nearly independentof a, and curs.) From thevalue of C~an impurity contentwe will usethe value e(0) = 0.632 18 in what of 0.7 per centwasderived. For the exchangewe

follows. It hasbeenfound by De Jongh2°that obtain i/k = —5.0 ±0.2 K. Using this, the ulti-the spin-wave prediction for x~(0)is in good matevalue a = HA/HE = 1.1 x iO~wascalculatedagreementwith experimentalresults obtainedon from the spin flop field (seebelow). After cor-

2-d antiferromagnets. recting for the impurity the true susceptibility of(CH

3NH3)2MnCl4 is found, which hasbeenshown

Assumingthe contribution of the impurity to in Fig. 1 by the full curves,and is plotted to-be of the form x~= C~/(T+ O~),andtaking O~= getherwith the theoretical predictionsin Fig. 2.2K, thereare two unknownconstantsto be de- The observeddeviation from the seriesexpansiontermined, namely C~andthe exchangeconstant result below T~6~hasalso beenfound in otherf/k of (CH3NH3)2MnCl4. We took the following 2-d manganesesalts (De Jongh andColpa)

2°and

procedure.A preliminary value for i/k was ob- may be attributedmainly to the small numberoftamed by comparingthe susceptibility dataat termsknown in the series.Note the largedeviationthe highest temperatureswith the series predic- from the molecular field theory, encounteredintion. From the value of the spin flop field (see these2-d antiferromagneticsystems.below), the anisotropyparametera was subse-quently estimated,andwith the aid of this AS(a)

Vol. 11, No. 10 MAGNETIC MEASUREMENTS ON (CH3NH3)2MnCl4 1427

The results of the high-field magnetization Sincethe measuredanisotropy, HA/HEexperimentsat T = 4.2K are given in Fig. 3. 1.1 x iO~, is more than five ordersof magnitude

To obtain a largeenoughsignal in spite of the largerthan the interlayer coupling, therecan besmall dimensionsof the available samplespace, little doubt that the anisotropyis thedeviation

the platelet wascut in two andthe crystals were from the ideal 2-d Heisenbergmodelthat drivesgluedon top of eachother. This additional pos- the system into long rangeorderingat T~ 47 ±sible sourceof misorientation may accountfor 3K. The resulting value for kT~/Ii~= 9.4, al-

the fact that the spin flop transition is not sharp. though subject to a considerableerror margin, isFurthermorethe initial slope of the magnetization in good agreementwith thoseof other 2-d Mn

2curveis fairly large, which is mainly due to the salts, in which the J’/iI values may be expectedimpurity contribution. In spite of theseshort- to be at least a hundredtimes larger. This sub-comings the spin flop transition is clearly in- stantiatestheexperimentalevidence,compileddicated.We havedeterminedthe value of the by De Jonghet a!., 1 for the fact that in thesespin flop field HSF by extrapolatingthe low- and systemsthe long range order is establishedby

high-field curves, taking for HSF the field at the anisotropyat approximatelythe sametern-which the jump in the magnetizationis halfway perature,which is itself very nearto that atcompleted(demagnetizationeffects are negligible which the ferromagneticandthe staggeredanti-at thesefield values). More sophisticatedtreat- ferromagneticsusceptibility are predictedto

ments,taking into accountmisorientation, lead diverge.21 This result may be understoodby can-to resultsnot different within the experimental sideringthat nearto this temperaturethe corre-error. In this way we obtain HSF = 36 ±3 koe, lations havebecomeof such a long rangethatfrom which we calculate a = 1.1 x iO~,using evena small deviation from ideality may ‘trigger’the formula H~F= 2HEHA/(l — ~<I//X ) and the onsetof long rangeorder. We point out that

f/k = —5.0 K. That we haveattributed the ob- for the ideal 2-d Heisenbergmodel the transitionserveddiscontinuity in the magnetizationcurve to a state with infinite susceptibility would notto a spin flop processin (CH

3NH3)2MnC14,is be accompaniedby the appearanceof long rangejustified by the fact that the temperatureof order, sincethe presenceof the latter hasbeen4.2 K is well abovethe transition temperatures ruled out for any T> 0 by Mermin and Wagner.22

associatedwith the anticipated impurities.Furthermore,the obtainedanisotropyfield HA = As a last remark we note that for the tran-850 (±8%)oe is of the orderof the dipolar ani- sition temperatureT~= 47 ±3K, we took thesotropy,as calculatedby Colpa,

3 who obtained temperatureat which the tangentof the ~,,1-curve

HAdIP = 1275oe. Apparently,part of the dipolar reachesits maximumvalue. The susceptibilityanisotropyis cancelledby the other anisotropy becomesanisotropicat a higher temperature,i.e.mechanisms. 53 ±1 K: Sucha situation is commonly observed

in the 2-d antiferromagnets,and since the iden-tification of T~with the temperatureof the max-

CONCLUSIONS .

imum in 9>~,,/aThasbeencheckedby independentFromthe argumentsgiven in the introduction determinationsof T~using other techniques,e.g.

we may expectthe interlayer coupling in (CH3 for K2NiF4,4~we should take it as an exper-

NH3)2MnCI4 to be about l0~—.lO~of the intra- imental fact that the anisotropypersists in a

layer exchange,unless for instance a magneto- small regionabove T~.Apparently, thereis alsostrictive distortion at T~or lattice imperfections an anisotropypresentin the short rangeinter-would destroythe symmetryargumentfor the actions,becomingmanifestat a higher temperaturecancellationof the interaction betweennearest thanthat at which the long rangeorder is estab-neighbouringlayers. Thus the expected JYJI, is lished.at least two ordersof magnitudesmaller thanwhat may be realized within the K2NiF4 structure. In conclusionwe may saythat the here men-

It can furthermorebe expectedthat the interlayer tioned series of Mn salts, offers a nice way ofcouplingis mainly of dipolar origin, studying2-d antiferromagnetism.Since the Mn

canalso be replacedby other magneticions

1428 MAGNETIC MEASUREMENTS ON (CH3NH3)2MnC14 Vol. 11, No. 10

(e.g. Fe2, reference7) there are certainly many Prof. F.A. Muller, Drs. Y. TammingaandW. Breur

new possibilitieswithin this crystal structure. for assistancewith the measurements.We aregrateful for a critical readingof the manuscriptbyDr. A.R. Miedemaand Prof. G. de Vries. This work

Acknowledgements— We would like to expressour is part of the researchprogramof the ‘Stichtingthanksto Dr. J.H.P. Colpa for providing us with F.O.M.’ andwasmadepossible by financial sup-thecalculationsof the variousdipolar fields men- port from the ‘NederlandseOrganisatievoortioned in this paper,and to Dr. L.W. Roeland, z.w.o.’

REFERENCES

1. DE JONGH L.J., BLOEMBERGEN P. and COLPA J.H.P.,Physica58, 305 (1972).

2. PLUMIER R., i. app!. Phys.35, 950S(1964).

3. COLPA J.H.P.,Private communication.Seealso: COLPA J.H.P.,Physica56, 185 (1971).

4. BIRGENAU R.J., GUGGENHEIM H.J. andSHIRANE G., Phys.Rev. B~,2211 (1970).

5. COLPA J.H.P.,Physica57, 347 (1972).

6. FOSTER J. and GILL N.S., i. Chem.Soc.A 2625 (1968).

7. MOSTAFA M.F. and WILLETT R.D., Phys.Rev. B4, 2213 (1971).

8. HOLMES L., EIBSCHUTZ M. and GUGGENHEIMH.J., SolidState Commun.7, 973 (1969).

9. BOESCH H.R., SCHMOCKER U., WALDNER F., EMERSON K. and DRUMHELLER J.E., Phys.Lett. 36A, 461 (1972).

10. MOLEMAN A.C., Private communication.

11. ROELAND L.W., MULLER F.A. and GERSDORFR., p. 153. Proc. mt. Conf. on High MagneticFields, Nottingham(1969).

12. BREED D.J., Thesis,University of Amsterdam(1969); Physica37, 35 (1967).

13. DOUGLASS D. andSTRANDBERG M., Physica27, 1 (1961).

14. LASHEEN M.A., VAN DEN BROEK J. andGORTER C.J., Physica24, 1061 (1958).

15. RUSHBROOKEG.S. and WOOD P.J.,Mo!. Phys. 1, 257 (1958).

16. STEVENSON R.L., PIRNE K., WOOD P.J. and EVE J., Phys.Lett. 27A, 2 (1968).

17. LINES ME., i. Phys. Chem.Solids3’i, 101 (1970).

18. KEFFER F., Spinwaves,Handbuchder Physik,Band 18, Teil II, SpringerBerlin (1966).

19. COLPA J.H.P., SIEVERTS E.G. andVAN DER LINDE R.H., Physica51, 573 (197t).

20. DE JONGH L.J., Phys.Rev.40A, 33 (1972); seealso DE JONGH L.J. andCOLPA J.H.P., to bepublished.

21. STANLEY H.E. and KAPLAN T.A., Phys. Rev.Lett. 17, 913 (1966); STANLEY H.E. Phys. Rev.158, 546 (1967); i. app!. Phys. 40, 1546 (1969).

22. MERMIN N.D. andWAGNER H., Phys.Rev. Lett. 17, 1133 (1966).

23. MAARSCHALL E.P., BOTTERMANA.C., VEGA S. and MIEDEMA A.R., Physica41, 473 (1969).

Vol. 11, No. 10 MAGNETIC MEASUREMENTS ON (CH3NH3)2MnC14 1429

On a mesuréIa susceptibilité magnétiquedu membrepremier de Iasériedes compos~sde formule gén~rale(C~H2~+1NH3)2MnCl4(n =

1,2,3 ), qui peuventétre regardésde consisterde couchesanti-ferromagnétiquespresqueisolées.Le couplemententreles couchesJ’ est estiméd’étre 10~—10~de la constanted’échangeJ dans leplan, un résultatcentfois plus faible que peutêtreréalisé dans lastructurede K2NiF4 bienconnue.Quantitativementon a obtenuJ/k = —5.0 ±0.2K pour la constanted’échangedans le plan etHA/HE = 1.1 x iO~pour l’anisotropie, celui-ci suivantde Ia valeurde champseuil de spin—flop.