X-ray photoelectron spectroscopic investigation on Fe geometrical sites of iron nitride thin ﬁlms

(1)

X-ray photoelectron spectroscopic investigation on Fe geometrical sites of iron nitride thin ﬁlms

Yin-Chih Lin¹^†, Jhen-Yong Hong¹^†, Chia-Nan Yen¹, Shi-Yuan Tong², Mean-Jue Tung², Hung-Wei Shiu³, Chia-Hao Chen³, and Minn-Tsong Lin^1,4*

1Department of Physics, National Taiwan University, Taipei 106, Taiwan

2Material and Chemical Engineering Laboratory, Industrial Technology Research Institute (ITRI), Hsinchu 300, Taiwan

3National Synchrotron Radiation Research Center (NSRRC), Hsinchu 300, Taiwan

4Institute of Atomic and Molecular Sciences, Academia Sinica, Taipei 106, Taiwan E-mail: [email protected]

Received October 15, 2014; accepted October 20, 2014; published online February 10, 2015

A partially ordered Fe16N2thinfilm, which exhibits a higher saturation magnetization than a bcc-Fe thin film, was grown on a Au(001) texture on a GaAs(001) substrate for studies of crystalline structure, electronic structure, and magnetic properties. Fe 2p3/2and 2p1/2X-ray photoelectron spectroscopies (XPS) reveal the electronic hybridization between the Fe atoms and the adjacent N atoms, whereas a multipeak analysis suggests the charge-transfer-induced electronic rearrangement of electronic configuration in Fe(8h) and Fe(4e) geometrical sites. These results are consistent with the previous model and help explain the saturation magnetization enhancement in theα-FeN system.

1. Introduction

In contemporary magnetic industries, a material with high saturation magnetization, low coerciveﬁeld, and resistance to oxidation is indispensable for fabricating devices and sensors.

The bcc-Fe has a saturation magnetization of 1600 emu/cm³, but is easily oxidized in air. The nitride state of iron is more chemically stable than its element state.¡-FeN is considered a remarkable material for industrial application because of its high saturation magnetization¹^–⁶⁾(1500 to 2350 emu/cm³).¡- FeN has two phases, ¡A-FeN (Fe8N) and ¡AA-FeN (Fe16N2), which have the same body-centered tetragonal (bct) structure (a = b = 5.72 Å, c = 6.29 Å⁷⁾) and chemical ratio (Fe/N = 8⁷⁾), but different order of Fe6N octahedron. The Fe6N octahedron in Fe₈N is randomly distributed, whereas that in Fe₁₆N₂ shows ordered distribution, as shown in Fig. 1(a).

Depending on the geometrical rule and the distance of the nearest-neighboring N atom, the Fe in both Fe₈N and Fe₁₆N₂ could be classified into Fe(8h), Fe(4e), and Fe(4d) geometrical sites, as shown in Fig. 1(b). In previous years, Wang and coworker presented a reproducible and reliable FeN growth method^8,9) and discussed the effect of the N site.¹⁰⁾ The ordering of the Fe6N octahedron cluster strongly affects the average saturation magnetization¹⁰⁾ (¡A-FeN+¡AA-FeN) and plays a critical role in saturation magnetization enhancement. In addition to magnetic measurement by vibrating sample magnetometry (VSM), polarized neutron reflectom- etry (PNR) and X-ray magnetic circular dichroism (XMCD) were performed to directly carry out the in-depth saturation magnetization distribution^11,12)and element-specific magnetic moment.⁸⁾ Moreover, a theoretical model of “cluster(Fe6N) and atom(Fe)” was presented to discuss the mechanism of saturation magnetization enhancement. The charge transfer between Fe and N in the Fe₆N cluster was predicted to play a critical role in the generation of a high saturation magnetization state, and probably to enhance the saturation magnetization.¹³⁾ However, the electronic structure of ¡-FeN is still unknown. In this study, X-ray photoelectron spectroscopy (XPS) was applied to probe the electronic structure of the

FeN system and in turn clarify its impact on the magnetization properties of the ¡-FeN system.

2. Experimental procedure

The structure of our sample was designed with the following layer sequence: 2 nm Cu/FeN/buffered Au(001) texture/

GaAs(001), in which the Cu thinﬁlm is used as a capping layer. Prior to the deposition, the wafer was cleaned and etched in the sequence of Piranha solution, hydrochloric acid (HCl), and deionized water. Au(001)-textured thinﬁlms were grown on a GaAs(001) substrate at 200 °C by thermal

(a)

(b)

Fig. 1. (Color online) Crystalline structure of Fe16N2with ordered N atom in the lattice. (b) Schematic diagram of Fe6N octahedral cluster in the Fe16N2

structure. According to the geometrical rule, the Fe atom could be classiﬁed into Fe(4e), Fe(8h), and Fe(4d) geometrical sites.

†These authors contributed equally to this work.

Japanese Journal of Applied Physics 54, 033002 (2015)

http://dx.doi.org/10.7567/JJAP.54.033002 REGULAR PAPER

(2)

evaporation with a deposition rate of 0.2 Å/s,¹⁴⁾ whereas iron nitride thinﬁlms were deposited on the Au(001) texture at 100 °C in an Ar and N₂ mixture environment by DC sputtering (at the power of 150 W and a total pressure of 4 mTorr). The thickness was checked using atomic force microscopy (AFM; NPX 200) and a quartz crystal micro- balance (QCM; Inﬁcon XTM/2). The crystalline structure was characterized using grazing incidence X-ray diffrac- tion (GI-XRD; TTRAX-3). The magnetic properties were measured using a vibrating sample magnetometer (VSM;

Lakeshore 735) with the ultimate sensitivity of 10^¹6emu.

Prior to each measurement, the instrument was calibrated using a Ni sphere standard sample. To determine the saturation magnetization per unit volume (Ms), the magnetic moment was measured under a parallel magnetic ﬁeld of 10 KOe. XPS with a high energy resolution of 0.05 eV was performed to examine the electronic structure at beamline 09A1 of the National Synchrotron Radiation Research Center (NSRRC) in Taiwan, followed by XPS data processing using XPSpeak software (version 4.1).¹⁵⁾

3. Results and discussion

FeN thin ﬁlm has various phases, such as £-Fe4N for face- centered tetragonal (fct) structure,^16,17)-Fe3N for hexagonal close-packed (hcp) structure,^18,19) and ¡-FeN(¡AA-Fe16N2/

¡A-Fe8N) for bodied-centered tetragonal (bct) structure.²⁰⁾ During the reactive sputtering process, phase formation is strongly related to the amount of reactive gas (N2) used and must be carefully examined. Figure 2 shows the saturation magnetization of the samples as a function of N2 partial pressure ranging from 0 to 1.0 mTorr. The sample fabricated in a nitrogen-free environment exhibited a saturation magnetization of approximately 1600 emu/cm³ (2.04®B/Fe atom), which is the same as that of the standard bcc-Fe.

The saturation magnetization becomes maximum at a N₂ partial pressure of 0.2 mTorr and decreases again as the N₂ partial pressure decreases. According to previous studies,^8,9) the generation of N-rich phases, such as those of£-Fe4N and

-Fe3N, might reduce saturation magnetization. As a result, the FeN sample fabricated under a N2 partial pressure of 0.2 mTorr exhibited a saturation magnetization of 1750 emu/cm³ (2.43®B/Fe atom), which is clearly greater than

that of bcc-Fe by over 10%. These results led to the fabrication of FeN with a higher saturation magnetization.

Among all FeN phases, only ¡AA-FeN exhibits a higher saturation magnetization than a normal Fe thinﬁlm, whereas other phases, such as-Fe3N and£-Fe4N, show lower saturation magnetization. This implies that the percentage of the ordered Fe₆N octahedron strongly affects the average saturation magnetization and needs to be examined. Figure 3 shows the GI-XRD spectra of the FeN sample fabricated under a N₂partial pressure of 0.2 mTorr. When the scattering vector is aligned in the Au[011] direction, the peak at 28.4°

represents the (002) plane of Fe₁₆N₂, and the peak at 59.2°

reveals the combination of the (002) plane of Fe8N and the (004) plane of Fe16N2. By aligning the scattering vector along the Au[001] direction, the peak at 45.0° shows the (220) plane of Fe16N2. The order degree of Fe6N could be deﬁned as^21,22)

D ¼½I⁰⁰_ð002Þ^obs=½I⁰⁰_ð004Þþ I⁰_ð002Þ^obs

½I⁰⁰ð002Þ^cal=½I⁰⁰ð004Þ^cal ; ð1Þ where½I⁰⁰ð002Þ^obs and½I⁰⁰ð004Þþ I⁰ð002Þ^obs are the integrated intensities from the measurement, and ½I⁰⁰_ð002Þ^cal and

½I⁰⁰_ð002Þ^cal are the integrated intensities calculated for the Fe₁₆N₂ single crystal. The calculation indicates that the sample is a mixture of Fe₈N (random Fe₆N distribution) and Fe₁₆N₂ (ordered Fe₆N distribution) with D = 0.296. In contrast, the XRD spectra of the FeN sample fabricated under a N2partial pressure of 0.5 mTorr only show a peak at 59.2°

(not shown here), which implies that the sample is only the Fe8N phase (D = 0) and a very-high N2 pressure may cause the random distribution of the Fe6N cluster. These results suggest that the N2partial pressure strongly affects the formation of the Fe16N2phase during fabrication.

To understand the enhanced saturation magnetization of ¡-FeN(¡AA-Fe16N₂/¡A-Fe8N), a strong correlation²³⁾ and the charge transfer via the existence of empty nitrogen orbitals²⁴⁾ between the Fe atom and the N atom were proposed. However, in the model of “cluster(Fe6N) and atom(Fe)”, only the ordered Fe6N octahedral cluster contrib- utes to the generation of a high magnetic state.¹³⁾The ordered Fe₆N in Fe₁₆N₂ has strong Coulumb interaction, which results in the charge difference between interior and exterior parts of the Fe6N octahedral cluster.¹³⁾The N atom functions as a medium that transfers the charges between adjacent Fe(8h) and Fe(4e) atoms. The high magnetic state of 4 µBper Fe atom is induced in Fe(4e) and Fe(8h) geometrical sites.¹³⁾

Fig. 2. (Color online) Saturation magnetization of Fe–N thin ﬁlm as a function of N2partial pressure. The inset shows the magnetic hysteresis loops of the bcc-Fe sample (black curve) and FeN sample (red curve) fabricated under a N2partial pressure of 0.2 mTorr.

(a) (b)

Fig. 3. GI-XRD of the FeN sample fabricated under a N2partial pressure of 0.2 mTorr with the scattering vectors along (a) the Au[011] direction, and (b) the Au[001] direction.

Jpn. J. Appl. Phys. 54, 033002 (2015) Y.-C. Lin et al.

(3)

However, in the disordered Fe₆N octahedron, the charge difference between interior and exterior parts disappears because of random distribution.¹³⁾ Therefore, the Fe sites in Fe₈N maintain the normal magnetic state of 2®Bper Fe atom, which is the same as in the case of the pure Fe thinﬁlm. The charge transfer in ordered Fe₁₆N₂ was predicted using this model. To discover the supporting evidence for charge transfers by observing the electronic structures, high-energy resolution XPS was used to probe the details of the electronic structure.

Prior to the measurement, the samples were sputtered using 2 kV Ar ions until the surface signal of Cu disappeared entirely and a Au standard sample was used for energy calibration. The crystalline structure of the sample was checked again by using XRD, which excludes the possibility of structure damage during the Ar-plasma process. Figure 4 shows the Fe 2p and N 1s spectra of the sample fabricated under a N₂partial pressure of 0.2 mTorr. The Fe 2p spectrum resembles the main peak and the neighboring shoulder peak.

By using the multi-peak curve ﬁtting process, the Fe 2p_3/2 spectrum can be decomposed into three contributions:

(1) 707.0, (2) 708.5, and (3) 710.5 eV. The main peak at a binding energy (BE) of 707.0 eV represents the Fe in Fe8N and Fe(4d) geometrical site in Fe16N2, which is located at the same BE as standard Fe (707.0 eV).²⁵⁾ The remaining peaks at BEs of 708.5 and 710.5 eV, which are attributed to the geometrical sites of Fe(8h) and Fe(4e), exhibit an energy shift toward the higher binding energy. The energy shift of the Fe(4e) peak is larger than that of the Fe(8h) peak because the distance between the Fe(4e) atom and N atom is much smaller than that in the case of Fe(8h). Quantitatively, the intensities of Fe(8h) and Fe(4e) are estimated as 30 and 15%, respectively, as listed in Table I. Taking into account that the intensity of Fe(4d) in Fe₁₆N₂ is 15% according to the proportion¹⁰⁾ of Feð4dÞ : Feð8hÞ : Feð4eÞ ¼ 1 : 2 : 1, Fe16N₂ is estimated to be approximately 60% in the sample, which is larger than the estimation (D = 29.6%) from the XRD result.

Because the scattering coefﬁcients of I⁰⁰_ð004Þ and I⁰_ð002Þ are different, the composition of Fe16N2 could not be deduced directly from the D value. This behavior is not only shown in

the Fe 2p3/2spectrum but also in the Fe 2p1/2spectrum. The inset in Fig. 4 shows the clearly different features at the shoulder parts in the Fe 2p spectrum of the FeN samples, which were fabricated under N₂ pressures of 0.2 (red curve) and 0.5 (black curve) mTorr. Moreover, the peak in the N 1s spectrum at a BE of 397.0 eV exhibits a shift toward the lower energy as compared with the single-atom state (409.9 eV)²⁶⁾ and the adsorption state (403.9 eV).²⁷⁾The energy shift in the XPS spectrum implies that the electronic hybridization may be caused by the rearrangement of electronic conﬁguration in the Fe and N core levels, which proves the existence of charge transfers in the Fe and N atoms in Fe₆N octahedral clusters.

4. Summary

In summary, the partially ordered Fe16N2thinﬁlm exhibits a higher magnetization than the bcc-Fe thinﬁlm sample. Fe 2p and N 1s spectra, along with the multipeak analysis, indicate the electronic hybridization between the N atoms and the adjacent Fe atoms in Fe₁₆N₂, which in turn supports the charge transfer phenomenon in the proposed cluster(Fe₆N) and atom(Fe) model. The study suggests that understanding of the electronic interaction between Fe and N atoms could help in the investigation of the saturation magnetization enhancement in the partially ordered Fe₁₆N₂ system.

Acknowledgments

The work was supported by the Industrial Technology Research Institute (ITRI) of Taiwan under the FY101-102 program and the National Science Council (NSC) of Taiwan under Grant Nos. 101-2112-M-002-024-MY3 and 102-2120- M-002-005.

1) T. K. Kim and M. Takahashi,Appl. Phys. Lett.20, 492 (1972).

2) Y. Sugita, K. Mitsuoka, M. Komuro, H. Hoshiya, Y. Kozono, and M.

Hanazono,J. Appl. Phys.70, 5977 (1991).

3) D. C. Sun, E. Y. Jiang, M. B. Tian, C. Lin, and X. X. Zhang,J. Appl. Phys.

79, 5440 (1996).

4) C. Ortiz, G. Dumpich, and A. H. Morrish,Appl. Phys. Lett.65, 2737 (1994).

5) M. Takahashi, H. Shoji, H. Takahashi, H. Nashi, T. Wakiyamam, M. Doi, and M. Matsui,J. Appl. Phys.76, 6642 (1994).

6) K. Nakajima and S. Okamoto,Appl. Phys. Lett.54, 2536 (1989).

7) K. H. Jack,Proc. R. Soc. London, Ser. A208, 200 (1951).

8) J. P. Wang, N. Ji, X. Liu, Y. Xu, C. Sanchez-Hanke, Y. Wu, F. M. F.

de Groot, L. F. Allard, and E. Lara-Curzio,IEEE Trans. Magn.48, 1710 (2012).

Fig. 4. (Color online) Fe 2p_3/2and 2p_1/2XPS curves of FeN sample fabricated under a N₂partial pressure of 0.2 mTorr. The open circles and red curve represent the original data and the bestﬁtted result, respectively. The insets show the N 1s spectrum of the FeN sample fabricated under 0.2 mTorr and the normalized Fe 2p spectrum of the FeN sample fabricated under 0.2 (red curve) and 0.5 (black curve) mTorr. (Photon energies are 900 eV for the Fe spectrum, and 620 eV for the N spectrum.)

Table I. Fitting result of Fe 2p spectrum by the analysis of XPSpeak software (version 4.1).

Binding energy (eV)

FWHM (eV)

Intensity

(%) Source

Fe 2p_3/2 707.0 1.1 53.7 Fe in Fe8N, Fe(4d) in Fe16N2

708.5 1.8 30.9 Fe(8h) in Fe16N2

710.5 2.7 15.4 Fe(4e) in Fe16N2

Fe 2p_1/2 719.8 1.2 55.4 Fe in Fe8N, Fe(4d) in Fe16N2

721.1 1.8 29.5 Fe(8h) in Fe16N2

723.5 2.6 15.1 Fe(4e) in Fe₁₆N₂

(4)

9) N. Ji, Y. Wu, and J. P. Wang,J. Appl. Phys.109, 07B767 (2011).

10) N. Ji, L. F. Allard, E. Lara-Curzio, and J. P. Wang,Appl. Phys. Lett.98, 092506 (2011).

11) N. Ji, V. Lauter, H. Ambaye, and J.-P. Wang,SPIN2, 1250004 (2012).

12) N. Ji, V. Lauter, X. Zhang, H. Ambaye, and J. P. Wang,Appl. Phys. Lett.

102, 072411 (2013).

13) N. Ji, X. Liu, and J. P. Wang,New J. Phys.12, 063032 (2010).

14) D. Y. Noh, Y. Hwu, H. K. Kim, and M. Hong,Phys. Rev. B51, 4441 (1995).

15) Web [http://www.uksaf.org/software.html].

16) K. Ito, G. H. Lee, K. Harada, M. Suzuno, T. Suemasu, Y. Takeda, Y. Saitoh, M. Ye, A. Kimura, and H. Akinaga,Appl. Phys. Lett.98, 102507 (2011).

17) B. C. Frazer,Phys. Rev.112, 751 (1958).

18) A. Leineweber, H. Jacobs, F. Huning, H. Lueken, H. Schilder, and W.

Kockelmann,J. Alloys Compd.288, 79 (1999).

19) K. H. Eickel and W. Pitsch,Phys. Status Solidi39, 121 (1970).

20) N. Ji, M. S. Osofsky, V. Lauter, L. F. Allard, X. Li, K. L. Jensen, H.

Ambaye, E. Lara-Curzio, and J. P. Wang,Phys. Rev. B84, 245310 (2011).

21) S. Okamoto, O. Kitakami, and Y. Shimada,J. Magn. Magn. Mater.208, 102 (2000).

22) S. Okamoto, O. Kitakami, and Y. Shimada,J. Appl. Phys.79, 5250 (1996).

23) W. Y. Lai, Q. Q. Zheng, and W. Y. Hu,J. Phys.: Condens. Matter6, L259 (1994).

24) A. Sakuma,J. Appl. Phys.79, 5570 (1996).

25) C. D. Wager, M. W. Riggs, L. E. Davis, J. F. Moulder, and G. E.

Mullenberg, Handbook of X-ray Photoelectron Spectroscopy (Perkin-Elmer, Eden Prairie, MN, 1979) Chap. 1, p. 80.

26) Photoemission in Solids I: General Principles, ed. M. Cardona and L. Ley (Springer, Berlin, 1978) Chap. 6, p. 266.

27) H. Tillborg, A. Nilsson, B. Hernnas, N. Martensson, and R. E. Palmer,Surf.

Sci.295, 1 (1993).