Magnetism in disordered carbon as a function of the extentof graphitization

K. Govind Raj a,b, P.A. Joy a,b,n

a Physical and Materials Chemistry Division, CSIR-National Chemical Laboratory, Pune 411008, Indiab Academy of Scientific and Innovative Research (AcSIR), CSIR-National Chemical Laboratory, Pune 411008, India

a r t i c l e i n f o

Article history:Received 10 September 2013Accepted 28 September 2013by A.H. MacDonaldAvailable online 12 October 2013

Keywords:A. Amorphous carbonA. Disordered carbonB. GraphitizationD. Magnetic properties

a b s t r a c t

Magnetic properties of disordered carbon have been investigated as a function of the extent ofgraphitization. It is found that the magnetization of the disordered carbon decreases with increasingdegree of graphitization. Treatment with acid modifies the magnetic characteristics considerably and theoriginal magnetic characteristics are retained upon further heat treatment. The results show that theintrinsic magnetic behavior of the disordered carbon depends on the microstructure and that the edgestates play a critical role in deciding the magnetic interactions in the amorphous carbon system.

& 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Carbon based materials containing only s and p electrons andexhibiting ferromagnetism are promising candidates for variousmagnetic applications due to their low cost and light weight.There are many studies reported in the literature on the magneticcharacteristics of carbon derived from various resources [1].Magnetic properties observed in carbon based materials have beenascribed to the electronic properties arising frommixed sp2–sp3 sitesandpeculiar edge states [2].Magnetic responseof carbon structures isfound to be highly dependent on their microstructure and thepresence of heteroatoms in the carbon network or guest moleculesbetween the graphitic layers or inside the pores [3–5]. These observa-tions are in good agreement with theoretical predictions [6,7].Extinction of ferromagnetism and development of diamagnetic char-acter in highly oriented pyrolytic graphite (HOPG) after annealing atveryhigh temperatures,due to the increase ingrain size and reductionof edge states, point towards the importance of grain boundary,defects or vacancies [8]. Irradiation dosage dependent magneticordering in proton irradiated HOPG [9] and nanodiamond [10]irradiated with 15N and 12C also have established the role of defectstowards creating a magnetic microstructure. Moreover, the observa-tion of a spin-glass like state in different carbon nanosystems, due tothe complex interactions among isolated spin clusters, is of particularinterest for studying the development of magnetic interactions in thecarbon based materials [11–13].

Most of the studies reported on the magnetic properties carbonare performed on ordered carbon allotropes with relatively highcrystallinity. Here we report magnetic properties of amorphouscarbon which is more disordered than those studied previously, sothat more clarity can be brought to the magnetism in disorderedcarbon which is easily processable than the ordered allotropes.

2. Experimental section

Amorphous carbon was prepared by the pyrolysis of a coconutshell. A dry coconut shell was crushed into small piecesð � 10 mm� 10 mmÞ and pyrolyzed at different temperatures inthe range 500–1000 1C, in a horizontal programmable tubularfurnace under a flowing nitrogen atmosphere. The temperature ofthe furnace was kept constant at the desired heat treatmenttemperature (HTT) for 4 h and further cooled to ambient tem-perature. The brittle carbon pellets thus obtained were powderedusing an agate mortar and a pestle. The as-pyrolyzed samples weretreated with concentrated HCl at 80 1C for 24 h and all the acidtreated samples were recovered by filtration and washed severaltimes using double distilled water.

The samples heat treated at different temperatures are labeled asHTxxx and the acid treated samples are labeled as HTxxxA, wherexxx is the pyrolysis temperature. To check the efficiency of the acidextraction process, HT500A obtained after washing HT500 with acidwas again subjected to acid treatment and analyzed for impuritiesand the carbon material thus obtained is labeled as HT500A1.

Powder X-ray diffraction (XRD) studies were performed on aPhillips X'pert pro diffractometer using Cu Kα radiation. Raman

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/ssc

Solid State Communications

0038-1098/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.ssc.2013.09.037

n Corresponding author. Tel.: þ91 20 2590 2273; fax: þ91 20 2590 2636.E-mail address: pa.joy@ncl.res.in (P.A. Joy).

Solid State Communications 177 (2014) 89–94

spectra were recorded on a Horiba JY labraman HR 800 microRaman spectrometer using 633 nm He–Ne laser. Quantitativedetermination of impurities were performed using inductivelycoupled plasma optical emission spectrometry (ICP-OES) analysison a Spectro Arcos FH-12 analyzer. Magnetic measurements wereperformed using a Quantum Design MPMS 7T SQUID VSM and thedata were corrected for diamagnetic background signal from thesample holder.

3. Results and discussion

Pyrolysis of the coconut shell above 400 1C is known to inducecarbonization and further increase in the heat treatment tem-perature leads to ordering of amorphous carbon by removal ofdangling bonds and heteroatom containing terminal groups[14,15]. XRD patterns (Fig. 1) of the pyrolyzed samples consist oftwo broad peaks corresponding to the (002) and (100) Braggreflections of amorphous carbon. The in-plane (La) and out-of-plane (Lc) coherence lengths were calculated from the FWHM ofthe (100) and (002) peaks respectively [16,17]. With the increasein HTT from 500 to 1000 1C, La is found to increase from 15 to 20 Å,and Lc increases from 12 to 16 Å.

Raman spectra (Fig. 2) of the different samples consist of twopeaks due to graphitic G peak at � 1575 cm�1 corresponding tothe in-plane bond stretching motion of sp2 carbon atoms and theD peak at � 1335 cm�1 which will be absent in a perfect graphiticstructure and appears only in the presence of structural disorder.For carbon based materials, depending on the extent of graphiti-zation, intensity ratio of D-peak to G-peak, I(D)/I(G), is reportedto follow different trends with La, where IðDÞ=IðGÞpL�1

a andIðDÞ=IðGÞpL2a for graphitic and amorphous carbon, respectively[18–21]. This change over in the dependency of I(D)/I(G) with Laoccurs through a broad maximum. For the present pyrolyzedsamples (Fig. 3) the Raman G peak position initially increasessharply and then decreases slowly after reaching a maximumvalue, with the increase in HTT. Similarly, as shown in the inset ofFig. 3, I(D)/I(G) ratio initially increases sharply as La is increased

and reaches a broad maximum at higher La values close to 20 Å.These observations indicate the transition from amorphous carbonto nanocrystalline graphite. Thus, the present samples are in theboarder region separating the two carbon forms and hence doesnot obey the above given relations.

Quantitative determination of impurities in pyrolyzed samplesperformed by ICP-OES after acid extraction showed that apartfrom the ferromagnetic elements Fe, Co and Ni in trace amountsðo100 ppmÞ, many non-magnetic elements are also presentwhich include K ð � 600 ppmÞ, Ca ð � 350 ppmÞ, Na ð � 300 ppmÞ,Al ð � 200 ppmÞ and B ð � 100 ppmÞ. The efficiency of the acidextraction process in removing the magnetic impurities wasverified by a second acid treatment on HT500A and further ICPanalysis. The ICP analysis showed o5 ppm each of Fe, Co and Ni,indicating complete removal of the magnetic impurities after thefirst stage extraction itself. There is not much change in the Ramanspectra and the XRD pattern of HT500A compared to that of HT500(Fig. 4) which rules out any major changes in the carbon micro-structure on acid treatment.Fig. 1. Powder XRD patterns of the as-pyrolyzed carbon samples.

Fig. 2. Raman spectra of the as-pyrolyzed carbon samples.

Fig. 3. Variation of the Raman G peak position as a function of heat treatmenttemperature. Inset: variation of the I(D)/I(G) ratio as a function of La.

K. Govind Raj, P.A. Joy / Solid State Communications 177 (2014) 89–9490

For the as-pyrolyzed (Fig. 5) as well as the acid treated samples(Fig. 6), magnetic measurements at 2 K showed S-shaped magneti-zation curves, similar to the Brillouin type saturating trend withJ¼1/2 as reported in the case of graphene nanoribbons [13] and themagnetization is not saturated even at 6 T. However, the magnetiza-tion curves for both sets of samples show a significant coercivity,which indicates that the observed magnetism cannot be explainedby localized electron spins and that the samples are not simplyparamagnetic. Indication for this kind of superparamagnetism-likedisordered magnetic behavior has been recently reported in gra-phene nanoribbons [13]. The magnetization at the highest fielddecreases with increasing HTT. Magnetization at 6 T remains almostconstant for the samples pyrolyzed at 500 and 600 1C and drasticallydrops above HTT¼600 1C. A similar trend is observed in the case of

the intensity ratio of the D and G Raman bands, I(D)/I(G) (Fig. 3). Thedecreasing magnetization with increasing HTT is similar to thatreported previously for nanographite based carbon materials [11],where the results are explained in terms of disordered magnetismcaused by random strengths of inter-nanographite antiferromag-netic interactions mediated by π-conduction carriers.

A large decrease in the magnetization at low temperatures isobserved when HTT is increased from 600 1C to 800 1C, and above800 1C magnetization decreases gradually. This behavior can berelated to the carbon structure, where the samples pyrolyzed at500 1C, 600 1C and 700 1C show gradual ordering of amorphouscarbon with La values 15.3 Å, 16.3 Å, and 16.9 Å, respectively.In this region, the samples are more disordered, and to induce asmall increase in La requires removal of defects in large numberswhich leads to the large decrease in the magnetization. As HTT isincreased to 700 1C, I(D)/I(G) is increased beyond unity showingthe ordering of amorphous carbon to nanocrystalline graphite.Heat treatment beyond 800 1C induces much more rapid increasein La giving a clear indication that there is a fast relaxation tonanocrystalline graphite in this region. This indicates that thechanges in the magnetization are directly correlated with thedefect structure, pointing out the possible intrinsic origin ofmagnetism in the pyrolyzed samples.

Magnetization measurements on the acid treated samples(Fig. 6) indicate that the nature of the magnetization curves andthe decrease in the magnitude of the magnetization at 6 T issimilar to that observed for the as-pyrolyzed samples (Fig. 5). Onlya decrease in the value of the magnetization is observed after acidtreatment. The lower magnetization of the as-pyrolyzed samplescan be due to the extra contribution from some magnetic impu-rities present in these samples or due to the reduced defects.To quantify the maximum possible contribution from impuri-ties towards the measured magnetization, contribution from themagnetization of ferromagnetic elements was calculated usingtheir concentration obtained from ICP-OES and standard massmagnetization values for bulk ferromagnetic Fe (221.9 emu/g),Co (162.5 emu/g) and Ni (57.50 emu/g) at 0 K [22]. For theas-pyrolyzed samples, Fe, Co, and Ni concentrations varied inthe range 66–78 ppm, 19–22 ppm, and 19–28 ppm, respectively.Considering the extreme case where all the three elements behaveas bulk ferromagnetic metals at 0 K, their contribution towardsmagnetization is calculated as 0.02 emu/g of carbon. Thus, theimpurity contribution towards magnetization is 3% and 28% forHT500 and HT1000, respectively, even though nearly the sameamount of ferromagnetic impurities are present in both samples.

Fig. 4. Comparison of the Raman spectra (a) and XRD patterns (b) of HT500 andHT500A.

Fig. 5. (Color online) Magnetization curves of the as-pyrolyzed samples at 2 K.Inset: enlarged curves showing magnetic hysteresis.

Fig. 6. (Color online) Magnetization curves of the acid treated samples at 2 K. Inset:enlarged curves showing magnetic hysteresis.

K. Govind Raj, P.A. Joy / Solid State Communications 177 (2014) 89–94 91

In the case of thermally induced graphitization, the effect of HTTon amorphous carbon samples is to develop thermodynamicallymore stable nanocrystalline graphite structure through the heattreatment induced migration of the sp3 defects which tend toextend the π-conjugated areas at the expense of strain energy[23,24]. This relaxation to the graphitic structure with increase inHTT causes the reduction of edge states and dangling bonds in thesystemwhich are otherwise contributing to the magnetic behavior

of disordered carbons [25,26]. Thus, for higher HTT samples, sincethe edge states contributing to magnetism being considerablyreduced and the samples being more graphitic, there is anapparent higher impurity contribution towards magnetizationfor HT1000 than for HT500.

As mentioned, the magnetization of all samples decreased afteracid treatment. As shown in Fig. 7, upon first acid treatment,magnetization of HT500 is decreased by 0.242 emu/g. However,the second acid treatment on HT500A reduced the magnetizationonly by 0.017 emu/g. This indicates that impurity phases are notcontributing to the magnetization after the first acid treatment.However, the maximum possible contribution to magnetizationfrom the ferromagnetic impurities (assuming contribution fromferromagnetic metals) which was leached out from respectivesamples is only 0.02 emu/g. Since HT500 is having highest degreeof disorder, the large decrease in magnetization upon the first acidtreatment can be due to the suppression of magnetic contributionfrom magnetic π-edge states due to the chemical modification andstabilization of the dangling bonds by hydrogen or chlorine. Thismodification decreases the localized spin concentration, similar tothe effect of fluorination on activated carbon fibres as reported byKiguchi et al. [27]. For the higher HTT samples, since they are moregraphitic, the acid treatment does not considerably modify theedge states and dangling bonds compared to lower HTT samples.

Temperature variation of the magnetization of all the samples(as-pyrolyzed and acid treated samples) was measured in thetemperature range 2–300 K in a magnetic field of 1 T to gain moreinsight into the magnetic nature of the samples. The zero fieldcooled (ZFC) and field cooled (FC) magnetization curves showdifferent behaviors for the pyrolyzed and acid treated samples.

Fig. 7. (Color online) Magnetization of HT500 before and after two consecutiveacid treatments.

Fig. 8. (Color online) ZFC (black) and FC (red) magnetization curves for (a) HT500, (b) HT500A, (c) HT1000 and (d) HT1000A. Insets show the temperature region where ahump is observed in the magnetization curves.

K. Govind Raj, P.A. Joy / Solid State Communications 177 (2014) 89–9492

The ZFC and FC magnetization curves of the pyrolyzed samplesHT500 and HT1000 as well as the corresponding acid treatedsamples HT500A and HT1000A (samples with the highest andlowest magnetization) are shown in Fig. 8. The enlarged portionsin the figures show the difference between FC and ZFC magnetiza-tion curves very clearly. For all the pyrolyzed samples, an increasein the magnetization is observed below 200 K with a broad humparound 150 K, which is more pronounced in the FC curve than in

the ZFC curve as shown in the insets of Fig. 8. After the acidtreatment process, for all the samples, the broad transition-likefeature and the humps are disappeared. This kind of magnetictransition-like feature can originate due to complex magneticinteractions which compete among themselves forming a spin-glass state or indicates the low dimensionality of the magneticsystem [11–13,28,29].

To confirm the intrinsic origin of the disordered magnetic statein the carbon samples, acid treated sample HT500A, where FCmagnetization exactly follows ZFC magnetization without anyhump, was subjected to a further heat treatment at 500 1Cfollowed by acid treatment, which was repeated in multiple cycles.The nature of the FC and ZFC magnetization curves of HT500A(Fig. 9(a)) heat treated again at 500 1C under identical conditionsas in the case of HT500 is similar to that of HT500, showing thebroad hump between 100 and 200 K (Fig. 9(b)). The magnetictransition and the hump disappear on further acid treatment ofthe same sample (Fig. 9(c)), and again reappear on heat treatmentat 500 1C (Fig. 9(d)). As the contribution from magnetic impuritiesis found to be negligible in the samples after the first heattreatment itself, it is clear that the magnetic transition-like featureand the broad hump are due to magnetic contribution from carbonand not from any other impurities.

Fig. 10 shows the magnetization curves of HT500 measuredat 2 K after repeated acid and heat treatments at 500 1C. It is observedthat the magnitude of magnetization at low temperatures is reducedafter acid treatment and then increased after heat treatment againand the trend is repeated for further acid and heat treatments. Asshown in the inset of Fig. 10, the coercivity continuously increasesafter each treatment. Highest coercivity of 280 Oe is obtained for thefinal sample after two subsequent acid treatments and each followed

Fig. 9. (Color online) ZFC (black) and FC (red) magnetization curves of (a) HT500A, (b) HT500A after reheating at 500 1C under identical conditions as in the case of HT500,(c) reheated sample after a second acid treatment and (d) after reheating the second acid treated sample at 500 1C under identical conditions as for previous samples.

Fig. 10. (Color online) Magnetization curves of HT500 after repeated heat and acidtreatments, measured at 2 K. (1) HT500, (2) HT500A, (3) HT500A heat treated at500 1C, (4) sample 3 acid treated, and (5) sample 4 heat treated at 500 1C. Inset:enlarged curves showing magnetic hysteresis.

K. Govind Raj, P.A. Joy / Solid State Communications 177 (2014) 89–94 93

by further heat treatment. The overall results show that the magneticcharacteristics are reproducible after heat/acid treatments underidentical conditions. The only exception is the small decrease in themagnitude of magnetization after each acid treatment and the fieldcooling effects. This decrease in the magnetization is likely to be dueto the increasing chemical modification of the edge states. Thecontrast in ZFC-FC magnetization behavior among heat treated andacid treated carbon samples is quite similar to the observations madeon graphene nanoribbons (GNRs) and chemically converted graphenenanoribbons (CCGNRs) by Rao et al. [13]. GNRs showed field coolingeffect which was absent in CCGNRs which was ascribed to the edgereconstruction involved during the chemical treatment. The presentresults clearly prove the critical dependency of the edge statestowards magnetic interactions in disordered carbon.

4. Conclusions

Amorphous carbon samples derived from a coconut shell withincreasing degree of graphitization show weak ferromagnetic likemagnetization curves at low temperatures with significant coercivity.With the increase in the heat treatment temperature, the materialbecomes more graphitic and the magnetization decreases. On acidtreatment of the pyrolyzed samples, a decrease in the magnetizationand difference in the magnetic characteristics are observed, due to thechemical modification of edge states and dangling bonds on acidtreatment. Proof for this is obtained from studies on repeated heat/acid treatment of the initially acid treated sample. Thus, the presentstudy shows that a disordered magnetic state with random strengthspreviously observed in nanographite and graphene nanoribbons alsoexists in amorphous carbon. This observation in amorphous carboncan trigger further work to understand the magnetism in detail due tothe large possibilities in preparing magnetic amorphous carbon withdifferent microstructures.

Acknowledgment

K. Govind Raj is grateful to Council of Scientific and IndustrialResearch (CSIR), India, for financial assistance in the form of aresearch fellowship.

References

[1] T.L. Makarova, Semiconductors 38 (2004) 641.[2] T. Enoki, K. Takai, Solid State Commun. 149 (2009) 1144.[3] K. Takahara, S. Hao, H. Tanaka, T. Kadono, M. Hara, K. Takai, et al., Phys. Rev.

B 82 (2010) 121417.[4] K. Takai, T. Suzuki, T. Enoki, et al., Phys. Rev. B 81 (2010) 205420.[5] D.W. Boukhvalov, S. Moehlecke, R.R. da Silva, Y. Kopelevich, Phys. Rev. B 83

(2011) 233408.[6] O.V. Yazyev, L. Helm, Phys. Rev. B 75 (2007) 125408.[7] J.O. Sofo, G. Usaj, P.S. Cornaglia, A.M. Suarez, A.D. Hernandez-Nieves,

C.A. Balseiro, Phys. Rev. B 85 (2012) 115405.[8] X. Miao, S. Tongay, A.F. Hebard, Carbon 50 (2012) 1614.[9] P. Esquinazi, D. Spemann, R. Hohne, A. Setzer, K.H. Han, T. Butz, Phys. Rev. Lett.

91 (2003) 227201.[10] S. Talapatra, P.G. Ganesan, T. Kim, R. Vajtai, M. Huang, M. Shima, et al., Phys.

Rev. Lett. 95 (2005) 097201.[11] Y. Shibayama, H. Sato, T. Enoki, M. Endo, Phys. Rev. Lett. 84 (2000) 1744.[12] J. Cernak, G. Helgesen, A.T. Skjeltorp, J. Kovac, J. Voltr, E. Cizmar, Phys. Rev. B 87

(2013) 014434.[13] S.S. Rao, S.N. Jammalamadaka, A. Stesmans, V.V. Moshchalkov, et al., Nano Lett.

12 (2012) 1210.[14] Z. Hu, M.P. Srinivasan, Y. Ni, Carbon 39 (2001) 877.[15] A. Oberlin, Carbon 22 (1984) 521.[16] J. Biscoe, B.E. Warren, J. Appl. Phys. 13 (1942) 364.[17] N. Iwashita, CR. Park, H. Fujimoto, M. Shiraishi, M. Inagaki, Carbon 42 (2004)

701.[18] F. Tuinstra, J.L. Koening, J. Chem. Phys. 53 (1970) 1126.[19] M.A. Pimenta, G. Dresselhaus, M.S. Dresselhaus, L.G. Cançado, A. Jorio, R. Saito,

Phys. Chem. Chem. Phys. 9 (2007) 1276.[20] L.G. Cançado, K. Takai, T. Enoki, M. Endo, Y.A. Kim, H. Mizusaki, et al., Appl.

Phys. Lett. 88 (2006) 163106.[21] A.C. Ferrari, J. Robertson, Phys. Rev. B 61 (2000) 14095.[22] B.D. Cullity, Introduction to Magnetic Materials, 2nd Ed., John Wiley & Sons,

New Jersey (2009) 2009; 531.[23] F.G. Emmerich, Carbon 33 (1995) 1709.[24] K. Takai, M. Oga, H. Sato, T. Enoki, Y. Ohki, A. Taomoto, et al., Phys. Rev. B 67

(2003) 214202.[25] V.L. Joseph Joly, M. Kiguchi, S. Hao, K. Takai, T. Enoki, R. Sumii, et al., Phys. Rev.

B 81 (2010) 245428.[26] K. Takai, M. Oga, T. Enoki, A. Taomoto, Diamond Relat. Mater. 13 (2004) 1469.[27] M. Kiguchi, K. Takai, V.L. Joseph Joly, T. Enoki, R. Sumii, K. Amemiya, Phys. Rev.

B 84 (2011) 045421.[28] A.W. Mombru, H. Pardo, R. Faccio, O.F. de Lima, E.R. Leite, et al., Phys. Rev. B 71

(2005) 100404.[29] D. Arcon, Z. Jaglicic, A. Zorko, A.V. Rode, A.G. Christy, N.R. Madsen,

E.G. Gamaly, B. Luther-Davies, Phys. Rev. B 74 (2006) 014438.

K. Govind Raj, P.A. Joy / Solid State Communications 177 (2014) 89–9494