High-resistivity nanogranular Co–Al–O films for high-frequency applications

(1)

High-resistivity nanogranular Co–Al–O films for high-frequency applications

Pedram Khalili Amiria兲 and Yan Zhuang

HiTeC/DIMES, Delft University of Technology, LB 2.380, Mekelweg 4, NL-2628 CD Delft, The Netherlands Hugo Schellevis

ECTM/DIMES, Delft University of Technology, Delft, The Netherlands Behzad Rejaei, Marina Vroubel, and Yue Ma

HiTeC/DIMES, Delft University of Technology, LB 2.380, Mekelweg 4, NL-2628 CD Delft, The Netherlands Joachim N. Burghartz

Institute for Microelectronics Stuttgart (IMS CHIPS), 70569 Stuttgart, Germany

共Presented on 9 January 2007; received 30 October 2006; accepted 30 November 2006; published online 16 April 2007兲

This work presents a series of high-resistivity nanogranular Co–Al–O films with maximum resistivity of ⬃110 m⍀ cm. The films were deposited using pulsed dc reactive sputtering of a Co72Al28target in an oxygen/argon ambient. The samples were characterized by scanning electron

microscopy 共SEM兲, M-H loop measurements, and s-parameter measurements on microstrip transmission lines with Co–Al–O magnetic cores. The high-frequency magnetic permeability profile was extracted from the microstrip measurements. Reduction of deposition power resulted in resistivity enhancement, as well as reduction of coercivity and permeability. SEM images reveal an average grain size of ⬃80 nm for films with the highest resistivity. © 2007 American Institute of

Physics.关DOI:10.1063/1.2710235兴

I. INTRODUCTION

Integration of ferromagnetic materials in standard silicon technology holds great promise for performance enhance-ment and size reduction of integrated passive components.1–6 Proposed applications include integrated magnetic inductors, transformers, circulators, isolators, filters, and interconnects. Major limitations of these devices are eddy current losses within the magnetic films, arising due to their conductivity, as well as ferromagnetic resonance 共FMR兲 losses. Besides, the films usually need to have a high magnetic permeability to create the desired beneficial effects. The result is an ongo-ing quest for ferromagnetic films with high resistivity, high FMR frequency, and acceptably large dc permeability.

One approach to realize high-resistivity magnetic films in recent years has been the investigation of films with a granular structure.7–11Nano/micro granular Ni–Fe films have been shown to yield high resistivities,7but this comes at the costs of a rather low permeability and poor control of granu-lar nucleation during film deposition. Other granugranu-lar films reported in the literature mostly consist of complicated pro-cessing involving nonconductive elements. Their deposition either demands sophisticated multiple-target sputtering equipment,9 making them less attractive for mainstream in-tegrated circuit use due to the high cost, or relies on less accurately controlled processing, e.g., deposition with varied gas flow, impeding the realization of films with very high resistivity.8

In this paper we present a study on pulsed dc reactive sputtering of Co72Al28in an ambient with fixed concentration

of 20% oxygen. Deposition power is used as the control

parameter defining the oxygen content共and thus resistivity兲 of the sputtered film. Power control is advantageous com-pared to the more common approach of using gas flow as the control parameter 共e.g., in Ref. 8兲, allowing for changes in

the deposition conditions in steps of ⬍50 W. This enables one to perform a systematic study of the changes in resistiv-ity, permeabilresistiv-ity, coercivresistiv-ity, and granularity of the resulting films, as well as the realization of ultrahigh-resistivity films. Practically, we could only obtain the latter共with resistivity of ⬃110 m⍀ cm兲 by power control, as the transition region be-tween 共high-resistivity兲 magnetic and nonmagnetic films is too narrow for gas flow control to yield reproducible results. To characterize the high-frequency permeability profile of the films, microstrips were fabricated using Co–Al–O as the magnetic core. High-resistivity films reported in this study showed good reproducibility, a result of their simple struc-ture and deposition method.

II. EXPERIMENTS AND DISCUSSION

Granular Co–Al–O films were deposited by pulsed dc reactive sputtering of a Co₇₂Al₂₈ target in an ambient of 80% Ar: 20% O₂. No external magnetic field was applied during sputtering. Deposition power for different films ranged from 0.5 to 1.75 kW. Resistivity measurements were carried out using a conventional four-point-probe setup and

M-H loop measurements were made using a Princeton

Mi-croMag 2900 alternating gradient magnetometer, from which coercivity and dc permeability values were extracted. Reduc-ing deposition power resulted in rapid increase of film resis-tivity, accompanied by a similar decrease in coercivity, as shown in Fig.1. Reducing the power also resulted in rapid decrease of deposition rate. Deposition times for 100 nm

a兲_{Electronic mail: p.khalili@ dimes.tudelft.nl}

JOURNAL OF APPLIED PHYSICS 101, 09M508共2007兲

(2)

films sputtered at 1.75 and 0.55 kW were 32 and 105 s, re-spectively. The highest resistivity 共⬃110 m⍀ cm兲 was ob-tained at 0.55 kW. Reducing the power further to 0.5 kW resulted in a glassy film with no magnetic properties. This is presumably due to complete oxidation of Co. At higher deposition powers共thus higher deposition rates and less re-action time兲, however, only the more easily oxidizing Al at-oms共and possibly part of the Co surface兲 oxidize to produce Co grains surrounded by aluminum oxide. The different composition of the films is also evident from their etching behavior: Granular films deposited at⬎0.55 kW were etched by diluted HNO3, with the etching rate decreasing for films

with higher resistivities. Films deposited at⬍0.55 kW, how-ever, were not properly removed by such an etchant. The increase in resistivity is accompanied by a reduction in the dc permeability 关extracted from initial-magnetization M-H curves,共Fig.1, inset兲兴. Similar to Ref.11, an in-plane radial internal anisotropy field was observed in the M-H loop mea-surements, the magnitude of which 共for films deposited at 0.55 kW兲 increased from ⬃15 to ⬃50 Oe after annealing at 500 ° C for 40 min. The anisotropy is also evident from dif-ferences in the behavior of microstrips made at different ori-entations with respect to the radial direction. It is probably related to the structure of the deposition setup, but further study is needed to fully understand its origin. High-frequency measurements presented in this paper共see below兲 were performed on microstrips patterned along this radial easy axis. M-H loops for high-resistivity granular films show that these films are difficult to saturate, as evident from a long tail in the obtained magnetization curve共Fig.1, inset兲.

A similar behavior was observed in nano-microgranular elec-troplated Ni–Fe 共Ref. 7兲 and is very likely related to the

random orientation of magnetizations and shapes of the in-dividual grains. While a purely metallic film deposited with-out oxygen does not show this behavior 共Fig.1, inset兲, the

tail length steadily increases with increasing film resistivity, an indication of formation of separate grains with thicker spacer walls. Figures2共a兲and2共b兲show SEM pictures of the films deposited at 1.5 and 0.55 kW, respectively. While the thickness of the insulating walls is generally nonuniform, it is evident that the film deposited at lower power is composed

of more distinct grains separated by thicker walls. The com-paratively low resistivity for granular films deposited at ⬎0.7 kW is probably due to the presence of conducting paths as a result of the nonuniformity of the grain wall thick-ness. A tail in the M-H loop, however, is already observed as early as 1.5 kW deposition power, which is consistent with the onset of grain appearance as observed in Fig. 2共a兲. The magnetic effects of grain appearance 共tail in M-H loop and reduction of dc permeability兲 thus appear at an earlier stage than the desired resistivity increase.

Figure 2共c兲 shows one of the microstrip lines used for high-frequency characterization. Microstrip lengths ranged from 1 to 4 mm, and signal-linewidths ranged from 20 to 100␮m. Co–Al–O cores were 100 nm thick and pat-terned into stripes of 200 or 400␮m width. s-parameter measurements on the lines were carried out using an HP 8510 network analyzer, from which the real and imaginary parts of film permeability were extracted6共Fig.3兲. Reduction

of permeability with increasing resistivity is also evident from Fig.3. From this plot, the low-frequency permeabilities for films deposited at 1.75 and 0.55 kW are seen to be⬃50 and⬃10, in reasonable agreement with values obtained from

M-H loops共Fig.1, inset兲. The lack of a clear FMR peak for

the 0.55 kW sample共for which the real part of permeability never reaches zero兲 is an indication of very high magnetic damping. High resistivity in the granular layer thus comes at the price of a reduced permeability and enhanced magnetic relaxation loss. Nevertheless, the permeability of⬃10 for the FIG. 1. Increase in magnetic film resistivity with reducing deposition power

during reactive sputtering of Co–Al–O film. Inset:共top兲 Decrease in coer-civity and low-frequency permeability共obtained from M-H loop measure-ments兲 with decreasing deposition power. 共Bottom兲 Granularity of the film results in a long tail in the M-H loop, which makes the high-resistivity nanogranular layer difficult to saturate. Higher resistivities are accompanied by longer tails in the M-H loop.

FIG. 2. SEM micrographs of Co–Al–O films reactively sputtered at 共a兲 1.5 kW and共b兲 0.55 kW. 共c兲 Top view microphotograph of a microstrip transmission line with Co–Al–O core. The 50␮m wide signal line共middle兲, patterned Co–Al–O core共dark兲, and ground plane 共background of the pic-ture兲 are seen, as well as the measurement pads and vias to the ground.

FIG. 3. Real and imaginary parts of permeability for Co–Al–O films depos-ited at 1.75 kW共low resistivity, solid兲 and 0.55 kW 共very high resistivity, dashed兲, extracted from microstrip measurements. The magnetic film dimen-sions are 2 mm⫻400␮m, elongated in the radial direction of the wafer. Microstrip inductance enhancement at 500 MHz compared to a control de-vice without magnetic core is⬃⫻2 and ⬃25% for 1.75 and 0.55 kW films, respectively.

09M508-2 Khalili Amiri et al. J. Appl. Phys. 101, 09M508共2007兲

(3)

high-resistivity sample is still enough to create ⬃25% in-crease in microstrip inductance. This is much smaller than the appoximately twofold increase brought about by the lower resistivity film in the same configuration. It is worth mentioning, however, that one can choose the high-resistivity 共0.55 kW兲 sample to be much thicker than 100 nm to en-hance its effect. This is not possible for a low-resistivity film, however, because increasing the thickness in that case will cause a dramatic increase in eddy currents.

III. SUMMARY

High-resistivity granular Co–Al–O layers were prepared by pulsed dc reactive sputtering. High-frequency character-ization was performed by microstrip measurements. The ef-fect of deposition power on resistivity and permeability of the films was studied, revealing that resistivity increase is accompanied by coercivity decrease but is only achieved at the price of lower permeability and higher magnetic relax-ation loss.

1_{Y. Zhuang, M. Vroubel, B. Rejaei, and J. N. Burghartz, Proceedings of the} BCTM 2005, Santa Barbara, CA, 9–11 October 2005共IEEE, New York, 2005兲, pp. 26–32.

2_{M. Yamaguchi et al., J. Appl. Phys. 85, 7919}_共1999兲.

3_{V. Korenivski and R. B. Dover, IEEE Trans. Magn. 34, 1375}_共1998兲. 4_{M. Vroubel, Y. Zhuang, B. Rejaei, and J. N. Burghartz, IEEE Electron}

Device Lett. 25, 787共2004兲.

5_{Y. Zhuang, M. Vroubel, B. Rejaei, and J. N. Burghartz, Tech. Dig. - Int.} Electron Devices Meet. 18.06/1, 475共2002兲.

6_{M. Vroubel, Y. Zhuang, B. Rejaei, and J. N. Burghartz, Trans. Magn. Soc.} Jpn. 2, 371共2002兲.

7_{Y. Zhuang, M. Vroubel, B. Rejaei, J. N. Burghartz, and K. Attenborough,} J. Appl. Phys. 97, 10N305–1–3共2005兲.

8_{L. Li, A. M. Crawford, S. X. Wang, A. F. Marshall, M. Mao, T. Schneider,} and R. Bubber, J. Appl. Phys. 97, 10F907–1–3共2005兲.

9_{M. Munakata, M. Namikawa, M. Motoyama, M. Yagi, Y. Shimada, M.} Yamaguchi, and K.-I. Arai, Trans. Magn. Soc. Jpn. 2, 388共2002兲. 10_{S. Ohnuma, H. Fujimori, S. Furukawa, S. Mitani, and T. Masumoto, J.}

Alloys Compd. 222, 167共1995兲.

11_{S. Ikeda, I. Tagawa, Y. Uehara, T. Kubomiya, J. Kane, A. Kakehi, and A.} Chikazawa, IEEE Trans. Magn. 38, 2219共2002兲.

09M508-3 Khalili Amiri et al. J. Appl. Phys. 101, 09M508共2007兲