Conference presentation of A.Talyzin (ppt file, ~5 mb) at ICSM-2006. Title: "Reaction of Fe with C60 at HPHT conditions: formation of Fe3C and implications for "magnetic carbon"

The original paper on "Magnetic Carbon" was retracted on 30 of March 2006.

1. Who are we?

During the summer 2004 we worked together with Dr.Makarova on the project "Synthesis of magnetic carbon in piston cylinder apparatus".

In the frame of this project we synthesized about 20 samples by High Pressure High Temperature treatment of C₆₀. Despite the strong similarity of structural properties to "magnetic carbon" samples, none of our samples appeared to show bulk ferromagnetism. In order to check possible reasons for poor reproducibility of the "Magnetic carbon" we studied carefully all published papers and some raw experimental data. This study helped us to find many misconnections, mistakes and weak points in publications which claimed intrinsic ferromagnetism of carbon.

This page is related to movie filmed by Swedish TV (UR) about "Magnetic carbon". The page is aimed on the audience of this movie but we hope also that broader range of specialists could read it. We attempted to make presentation of fact understandable for anyone. Please send your comments and questions to our e-mail address: magneticcarbon@yahoo.com .

2. ⌠Extraordinary claims require extraordinary evidence. Carl Sagan

Was the claim of ferromagnetism in pure carbon extraordinary?  Yes, it challenged conventional theory of ferromagnetism.

Was the evidence extraordinary or just sufficient? No. Some of the evidence presented in the original publication was later not confirmed.

When some carbon sample is found to show ferromagnetic properties:

is it enough for conclusion about ferromagnetism of carbon? No!

It is necessary to prove that this sample is not contaminated by ferromagnetic impurities.

Why analysis of impurities is so important? Because there were many occasions when reported intrinsic ferromagnetism in unusual materials was later proved to be a "parasitic effect" from impurities. It is enough to have 400 microgram (0.04% by mass) of iron per gram of C₆₀ to explain all observed ferromagnetism.

3. Main points of our criticism.

1. Impurity analysis presented in "Nature"¹ paper was given only for pristine C₆₀ (22 ppm). High pressure treatment is complicated technological process and there are many ways how the sample could be contaminated during the synthesis (see also Additional information ). The impurity analysis method used for estimation of impurity levels requires at least 20 mg of material while the size of individual samples after high pressure treatment was smaller. Also, each sample was cut on many pieces, for example the magnetic data shown in the original publication were obtained on 3.3 mg piece¹. Analysis of similar piece later showed (by surface related method) about 20 times higher level of contamination (up to 400 ¨gg/g Fe²).

2. The "Nature" paper¹ states that ferromagnetic loops can be reproduced after heating of the sample at 640K for 2 hours. The same sample was in earlier study (Carbon, 2001³) reported to completely depolymerize at 472K. According to several other papers all kinds of HPHT polymers transform back to pristine C₆₀ at temperatures 500-600K. Magnetic properties preserved also even after heating up to 800K as it was later published². The assignment of ferromagnetism to specific polymeric structure and its defects was incorrect.

3. Magnetic data shown in Nature paper were obtained on the sample prepared at 975K, lowest temperature in the studied samples set. The conclusion about narrow interval just before point of C₆₀ collapse" is therefore not valid: Rh-polymers synthesized at lower temperature were not studied prior publication in "Nature".

4. Only three samples out of three sets (over 20 samples) were proved ferromagnetic, only one was pure Rh-polymer (6 GPa and 975K), none of these samples was synthesized in specified temperature interval 1025-1050K and one of three was even made at different pressure. The Corrigendum published in 2005 states that one sample was synthesized at 2.5 GPa (Rh-polymers could form at this pressure only as a small impurity) and temperature 1125K. This sample according to structural data was composed mostly by graphite-like carbon phase which forms as a result of C₆₀ collapse (fullerene cage collapse above ~1075K¹). Two samples showed nearly identical magnetic properties: one was true Rh-polymer (synthesized 6 GPa and 975K) and second was by 80-90% amorphous carbon formed after collapse of C₆₀. This means that ferromagnetism is not connected to fullerenes. Note that detail study of magnetic properties with temperature dependence was performed only for these two samples.

5. All samples studied in Nature paper were synthesized prior to discovery of ferromagnetic properties which mean that no care was taken to avoid contamination. Why all special precautions could be done if samples were not intended for study of ferromagnetism (as it is written in the paper)? No care was taken even after "discovery", the samples were additionally contaminated by improper handling. Precise description of handling methods can be found in the latest review paper (Makarova, 2004⁴): "The main impurities were two elements: calcium that was transferred from the hands of the experimentalist and iron that was introduced from the forceps and tools used in splitting the samples". See also Additional information.

6. Pressure medium which was used in high pressure experiments, Catlinite (or ⌠Katlenite)⁵, is highly contaminated with iron compounds⁶ and was likely a main source of contamination. Cracking of capsules at high pressure conditions (and contamination with material of pressure media) is especially likely at high temperatures when C₆₀ collapses into graphite-like phase with large volume decrease. Catlinite is also partly composed by CaCO₃ which explains why calcium impurity was one of the major in ferromagnetic samples.

7. All properties of ferromagnetic samples can be explained by carbide formation in high pressure experiments. Iron reacts with C₆₀ at high temperature with formation of Fe₃C, which has the same Curie temperature as "magnetic carbon".

8. Contamination of some samples with magnetic impurities is directly shown in supporting materials of Nature paper as a video. Below we show that ANY soft material will become ferromagnetic if touched by unprotected magnet in similar way. The video which we recorded shows how to make "Magnetic rubber" by the same treatment.

Conclusion:

The claim of ferromagnetism in Rhombohedral polymer of C₆₀ synthesized at temperatures "just before fullerene molecules collapse" was based on one single sample which was never studied on impurity levels. No precautions were taken against contamination during synthesis of this sample and subsequent handling.

4. Magnetic rubber.

What everyone can observe in the video provided as on-line supporting material to "Nature" paper and shown in the end of UR movie is very hard proof that no care was taken to avoid contamination. We can prove that ANY soft material will start to react to magnet if treated the same way. The SmCo magnet used for demonstration in this video is extremely strong (hundred times stronger than standard magnet) and always have a lot of metallic dust hanging on. What we see in this remarkable video is that samples are touched repeatedly by unprotected magnet (it was good idea to put a piece of thin paper in between of magnet and sample). In fact, one of small pieces did not wanted in the beginning to jump. Only repeated touching ensured desirable effect. The samples treated this way are not suitable anymore for any magnetic measurements due to contamination with magnetic impurities.

Here you can find video which we made using the same SmCo magnet. It shows how small piece of rubber jumps to magnet after one single touch with magnet. Video of "magnetic rubber" (about 2.7 Mb).

It is clear that strong contamination was introduced during synthesis of samples from pristine C₆₀. One of the samples (not mentioned in any papers) with tetragonal structure was found ferromagnetic with saturation magnetization about 30 times higher than magnetic carbon". Why it was not published? Because analysis showed about 1% of iron in this sample and some clear features typical for small particles. This sample had to jump to magnet best of all!

One more sample was found to be contaminated by material of capsule (Ta) on the level of about 3%, enough to detect impurity-induced superconductivity⁷ .

5. Additional information.

Iron around us. Metallic iron is very strongly ferromagnetic: saturation magnetization of Fe is about 2000 times higher compared to reported value for polymeric C₆₀. It is enough to have 0.04% (by mass) or ~400 μg/g of iron in studied sample to explain observed magnetization. The samples prepared by high pressure methods were rather small, only about 15 mg. Each sample was split on many pieces: some pieces were used for conductivity studies, others sent for measurements of magnetic properties, others for ESR measurements, XRD et cet. The list of people who received small pieces of samples includes 9 person. Why is that important? Because the method of impurity analysis described in "Nature" paper require at least 20 mg of material and this material is destroyed during analysis. Individual high pressure samples were impossible to analyze on impurities by this method.

The impurity concentration in the paper is described in following way: "We have paid great attention to chemical analysis of the pristine material as well as of the polymerized phase. The total amount of magnetic (Fe,Ni, Co) impurities is 22 p.p.m. in the pristine phase." Well, but pristine phase was not ferromagnetic.  The impurity level of studied polymeric sample is not reported. The reader may think that 22 ppm (or 0.0022% by mass) is also concentration of impurities in the studied sample. But it is not. The later studies of some other samples (not the one shown in Nature, this sample was never analyzed) showed that iron concentration is about 10-20 times higher than reported².

The high pressure experiment is complicated technological process which involves many different steps and many risks for contamination. So small is measured effect that real extra-care need to be taken:

-the initial powder and prepared sample should be operated by special tools: cutting tools, spoons and even tweezers need to be plastic

- can not be stored on-air because of metallic dust which is all around us (just leave on the table any magnet and look next day under microscope how much dust it collected)

-steel press-forms can not be used to make pellets (it was preliminary step before high pressure treatment), standard lubricants can not be used for these press forms.

-materials of high pressure assemblage , those which surround sample, all need to be of the same very high purity level. The purity of Nb (material of capsule), graphite (used as a heater), boron nitride (material used to separate them) is unknown.

-the sample was wrapped to Nb foil, it is very difficult to open such a capsule without metallic tools.

It was written that extra care was taken to avoid contamination. But why would anyone care about ppm concentrations of impurities? It is stated in the paper very clearly that the samples were synthesized before discovery of ferromagnetism. Dr.Makarova was searching for superconductivity but instead she found ferromagnetism. Why anyone would take care to avoid metallic tools, to use extra-pure materials in high pressure device etc. if the samples were not intended for study of ferromagnetism?

Interesting to note that the only paper which independently confirms ferromagnetism in polymeric C₆₀ was published by group of Bennington et al. very soon after original paper in "Nature". The paper do not mention Fe impurities at all. It looks like the impurity analysis was not performed. Also, it looks from the paper (and from the movie as well) that they studied some samples synthesized prior publication in "Nature" which mean that precautions against impurities were unlikely. Finally, the review paper published in 2004⁴ also shows that they touched samples with unprotected magnet. Which sample they possibly tested more intensively? The one which was synthesized on boundary to C₆₀ collapse as reported in "Nature". Temperature dependence of magnetization was not published (not measured?), Curie temperature is unknown.

References:

1.      Makarova, T. L. et al. Nature 413, 716-718 (2001).

2.      Höhne, R. & Esquinazi, P. Adv. Mater. 14, 753-756 (2002).

3.      Makarova, T. L. et al. Carbon 39, 2203-2209 (2001).

4.      Makarova, T.L. Semiconductors 38, 615-638 (2004).

5.      Davydov, V.A., A. Carbon 35, 735-743 (1997).

6.      Hall, H.T. Science 128, 445-449 (1958). (http://www.htracyhall.org/pdf/19580097.pdf) See also : http://www.pt-magazine.com/backissues/winter2003/story1.asp

7.      T.Makarova and B.Sundqvist, in E.Buzaneva and Scharff (eds.) in Frontiers of multifunctional materials., 2002.

8.      Makarova, T.L. Sundqvist, B. Kopelevich,Y. Synthetic Metals 137, 1335¡Ì1337 (2003).

9.      Spemann, D et.al. Nucl. Instr. and Meth. in Phys. Res. B 210, 531¡Ì536(2003)

10.  Han, K.-H. et al. Carbon 41, 785¡Ì795(2003).