Controlled Reduction of Bionanodots for Better Charge Storage Characteristics of Bionanodots Flash Memory

This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 140.113.38.11

This content was downloaded on 25/04/2014 at 10:40

Please note that terms and conditions apply.

Controlled Reduction of Bionanodots for Better Charge Storage Characteristics of

Bionanodots Flash Memory

View the table of contents for this issue, or go to the journal homepage for more 2009 Jpn. J. Appl. Phys. 48 04C190

(http://iopscience.iop.org/1347-4065/48/4S/04C190)

Controlled Reduction of Bionanodots for Better Charge Storage

Characteristics of Bionanodots Flash Memory

Yosuke Tojo1, Atsushi Miura1;2, Yukiharu Uraoka1, Takashi Fuyuki1, and Ichiro Yamashita1;3

1_{Nara Institute of Science and Technology, 8916-5 Takayamacho, Ikoma, Nara 630-0192, Japan} 2_{National Chiao Tung University, 1001 Ta-Hsueh Road, Hsinchu 30010, Taiwan}

3_{Panasonic, 3-4 Hikaridai, Seika, Kyoto 619-0237, Japan}

Received October 1, 2008; revised December 2, 2008; accepted December 11, 2008; published online April 20, 2009

We proposed a new process technology, named the ‘‘bio-nano-process’’, in which semiconductor processing technology and biotechnology are conbined. We utilized a ferritin protein cobalt core as a memory node, and succeeded in performing the operation of floating gate memory. In this study, we undertook the reduction control of a cobalt core embedded in silicon oxide by thermal annealing. We also fabricated metal–oxide–semiconductor (MOS) capacitors with using the cobalt core and evaluated their electronic properties. As a result, we could elucidate the contribution of the metallic cobalt in the core by controlling of the ambient and temperature. We found that memory windows become large with increasing contribution of metallic cobalt. #_{2009 The Japan Society of Applied Physics}

DOI: 10.1143/JJAP.48.04C190

1. Introduction

Recently, attention has been focused on semiconductor-nanodot-embedded metal–oxide–semiconductor (MOS) memory devices for use in future high-speed and low-power-consumption logic and memory devices. Conven-tional flash memory uses a plate-type floating gate for the electric charge storage node. Utilization of the nanodot as a floating gate is a promising approach to improving device property of flash memory.1,2)_{We have been demonstrating} the utilization of the bionanodot (BND) as the electric charge storage node of flash memory by choosing a biochemically synthesized metal oxide nanodot formed in the vacant cavity of supramolecular cage-shaped protein ferritin.3–7)_{It is well known that different kinds of inorganic} nanodots, such as iron oxide, cobalt oxide, and compound semiconductor-nanodots, can be formed in the ferritin cavity by biomineralization.8–10)We call this biomolecule-utilizing device fabrication process the bionanoprocess (BNP).3) Here, the BNDs are synthesized as metal oxide because the synthesis process is biomineralization in aqueous solu-tion. Therefore the reduction of embedded BNDs is necessary to achieve effective charge confinement in them in flash memory devices. For example, we have reported a functioning of embedded iron oxide BNDs (Fe-BNDs) in the MOS capacitor memory structure, that was enabled by annealing of embedded Fe-BNDs.6)_{The results suggest that} the conversion of oxide BND to metal BND by reduction improves the charge storage capability of flash memory with oxide BNDs. This strongly suggests the possibility of controlling the phase of the metal–oxygen complex. How-ever, in the case of iron, changing of the reduction states is complicated.3)In contrast, for cobalt, changing the reduction condition from cobalt oxide to metallic cobalt is simple. In this study, we examined the controlled reduction of cobalt oxide BNDs (Co-BNDs) by changing the reduction con-ditions, such as atmosphere and temperature. We observed the memory characteristics of Co-BND flash memory fabricated under different annealing conditions, and the reduction condition dependence of charge storage character-istics and an improvement of the charge storage capacity of Co-BND memory have been revealed.

2. Experimental Methods

Figure 1(a) depicts a schematic drawing of ferritin (left) and the cross section of Co-BND-accommodated ferritin (right). Ferritin gives several advantages for flash memory fabrication process, such as 1) uniform BND formation (approximately 7 nm) using cage-shaped ferritin protein as biotemplates; 2) high density and selective deposi-tion owing to the self-assemble ability of protein outer surface; 3) the formation of different kinds of metal and semiconductor nanodots by replacing Fe.9–11) _The BNDs, used as charge storage nodes of memory, were synthesized in the vacant cavity of ferritin protein by biomineralization.4,9)

Co-BND-accommodated ferritins were adsorbed on p-Si(100) substrates with 3-nm-thick thermal oxide thin film. To prevent multilayer formation, drop cast of ferritin solution was spun out in a sealed plastic tube. The outer protein shell was removed by oxidation with UV irradiation in ozone atmosphere, and a Co-BND monolayer was obtained. To prevent the re-oxidation of annealed Co-BNDs and achieve the controlled reduction of Co-BNDs, the fabricated Co-BND array was buried in thin SiO2 film. A

schematic drawing of the fabricated sample structure is shown in Fig. 2. The thickness of SiO2 was set to 3 nm

for X-ray photoelectron spectroscopy (XPS) measurements and 20 nm for the MOS capacitor. The fabricated samples were annealed inert (N2, 100%) and reductive (H2: N2¼

4% : 96%) atmospheres. Samples were annealed at 350, 450, 600, 700, and 800_{C for 10 min. Annealed samples were}

measured by XPS and transmission electron microscopy (TEM) to confirm the reduction of embedded Co-BNDs and the morphology after high temperature annealing, respectively.

7 nm 12 nm

Core

Fig. 1. (Color online) Schematic drawing of ferritin protein (left) and cross-sectional view of ferritin with BND (right).



We also fabricated an annealed Co-BND-embedded MOS capacitor by the same fabrication procedure as described elsewhere,5)_{and high-frequency capacitance–voltage (C–V)} characteristics were measured to elucidate the effect of the reduction.

3. Results and Discussion 3.1 Composition estimation

As mentioned above, the Co-BND in the protein shell was synthesized as cobalt oxide, Co3O4. XPS spectra of Co-BND

after annealing under inert, 100% N2and reduction in 4% H2

atmospheres are shown in Figs. 3(a) and 3(b), respectively. XPS spectra of Co-BNDs annealed in inert gas [Fig. 3(a)] showed a main peak at 779.5 eV. It is reported that Co3O4

has a Co 2p3=2 peak at 779.5 eV.12) As we can see in XPS

spectra of N2-annealed samples, the position of the Co 2p3=2

peak shows no dependence on annealing temperature. This means that embedded Co-BNDs were not reduced even at 800_{C under inert atmosphere. In contrast to N}

2 annealing,

XPS spectra of H2-annealed Co-BNDs show a new peak at

778.3 eV. This peak is assigned to 2p3=2 of metallic Co.12)

In H2 atmosphere, the metallic Co peak appeared from

350_{C, and the increase of annealing temperature induced}

the decline of the Co3O4 peak and the increment of the

metallic Co peak. Eventually, only the metallic Co peak was observed at 800_{C. These results indicated that}

the annealing atmosphere is more important than the

annealing temperature in the control of reduction conditions of embedded BNDs. The peak intensity change after H2

annealing suggests that the control of the reduction ratio between Co3O4 and metallic Co is possible by choosing

appropriate annealing temperature and atmosphere. for anneal. Most of the embedded oxide-Co-BNDs were reduced to metal-Co-BNDs above 450C, which is the annealing temperature applied in the device fabrication process. This suggests that we can expect sufficient charge storage in the metal portion of embedded the Co-BND charge-storage node of flash memory.

3.2 Morphology estimation

From the viewpoint of the use of annealed BNDs in flash memory, the morphology of BNDs after high-temperature treatment should be confirmed. High-temperature treatment of a nanodot sometimes causes undesirable deformation and diffusion of the nanodot. Therefore, we performed TEM analysis, using JEM-3100FEF (JEOL), of nonannealed and annealed Co-BND in SiO2 to check the morphology,

deformation, and diffusion of Co-BND. The electron beam spot size and acceleration voltage were 0.5 nm and 300 kV, respectively. Observation positions, which are depicted in Fig. 4, of the bottom oxide, Co-BND, and the upper oxide were separated by more than 8 nm. Figure 4 depicts the cross-sectional TEM image of embedded Co-BNDs before [Fig. 4(a)] and after [Fig. 4(b)] annealing under a reducing atmosphere at 800_C.

As seen in the TEM image of embedded Co-BNDs before annealing [Fig. 4(a)], Co-BNDs were aligned Co-BNDs with a spacing of 3 – 5 nm from the interface of the Si substrate and SiO2. Co-BNDs were deposited on thermally grown

SiO2 with a thickness of 3 nm. Therefore, the observed

spacing corresponds to the thin SiO2 layer. The observed

diameter of the BNDs are almost 7 nm and they retain the spherical shape after SiO2 deposition by plasma enhanced

CVD. We have obtained a very similar TEM image for the annealed sample. Annealed Co-BNDs retain their monolayer arrangements and spherical shape even after annealing. From the analysis of the high-resolution TEM image, a slight change in the size was found after annealing.

To check the diffusion of cobalt after annealing, we performed energy dispersive X-ray spectroscopy (EDS) and elemental mapping of the same sample as depicted in Fig. 4. Figures 5(a) and 5(b) show the obtained EDS data measured before and after annealing, respectively. In the EDS data before annealing [Fig. 5(a)], we can detect the signal from Co-BNDs : 7 nm XPS : 3 nm Capacitor : 20 nm SiO₂ Substrate : p-Si ( 100 ) 3 nm

Fig. 2. (Color online) Schematic drawing of BND-embedded stacked sample structure for XPS measurement and MOS capacitor fabrication.

770

780

790

800

Intensity ( arb

.unit )

Binding energy ( eV )

350

°

C

450

°

C

600

°

C

700

°

C

800

°

C

770

780

790

800

Intensity ( arb

.unit )

Binding energy ( eV )

350

°

C

450

°

C

600

°

C

700

°

C

800

°

C

Co

O

Co

Fig. 3. (Color online) XPS spectra of annealed samples for different annealing atmospheres and temperatures. (a) N2(100%) and (b) H2 (H2: N2¼4% : 96%). 20 nm ( a ) ( b ) Upper SiO2 Co-BNDs Bottom SiO2

Fig. 4. (Color online) TEM images of Co-BNDs (a) before and (b) after annealing in reducing atmosphere at 800_C.

cobalt at 6.8 keV in the region of the Co-BND monolayer, but not in the upper and bottom oxide areas. We merely observed the signals of Si from p-Si(100) and SiO2 and the

signal of O from SiO2in the upper and bottom oxide areas.

Cobalt is not diffused to the surrounding SiO2 before

annealing. Similar to the EDS spectra of non annealed sample, that of the annealed sample did not show any Co signal from the upper and bottom SiO2, as shown in

Fig. 5(b). This clearly indicates that Co does not diffuse to the surrounding SiO2.

The elemental mappings shown in Fig. 6 indicate that Co does not diffuse after annealing. The mapping images of oxygen, silicon, and cobalt are depicted in Fig. 6. Figures 6(a) and 6(e) show the TEM images of the non annealed and annealed samples, respectively. Corresponding elemental mappings of O, Si, and Co before annealing are shown in Figs. 6(b) to (d), and those of after annealing are shown in Figs. 6(f) to (h), respectively. As shown in the mappings of Co in Figs. 6(d) and 6(h), no change in the distribution of Co in the stacked structure before and after annealing was observed. This indicates that Co did not diffuse even after annealing.

In general, a nano sized particle has inferior heat-resistance properties and is relatively easily decomposed by high-temperature treatment. It has been reported that BND is also decomposed by heat treatment at 700_C.13)_In the reported case, Fe-BND and bare BND arrays were exposed to nitrogen atmosphere, i.e., without the protective film of SiO2 applied in this study. We observed slight

shrinkage of embedded Co-BNDs after high-temperature annealing. We reported a similar reduction in the size in the case of the annealing Fe-BNDs.6)_{This behavior is due to the} desorption of oxygen after annealing, and the shrinkage of the annealed Co-BNDs is considered to be due to a similar reason. In summary, the results of TEM measurements revealed that the Co-BNDs embedded in SiO2 are not

diffused to the surrounding SiO2 nor collapsed, because of

the protection provided by the deposited SiO2 thin film

against the applied high temperature. 3.3 Capacitance–voltage characteristics

Figure 7 shows the observed capacitance–voltage (C–V) characteristics of the fabricated Co-BND embedded MOS capacitors. The curves in Fig. 7 depict the C–V obtained for (a) the reference sample without Co-BND, (b) the sample annealed in N2: 100% at 800C and (c) the sample annealed

in H2: 4% at 800C. The observed C–V curves are

normalized with maximum capacitance to enable compar-ison. In the case of capacitors without Co-BND [curve (a)], no hysteresis was observed. In contrast, we observed anticlockwise hysteresis due to charge injection to the embedded Co-BND in curves (b) and (c). We observed a larger memory window in the case of the treatment under a reducing atmosphere at 800_{C [i.e., curve (c)]. It is of}

interest that, the Co-BND memory annealed in N2 showed

charge confinement in the embedded ‘‘cobalt oxide’’ BNDs, although the memory window size is not large compare with the MOS capacitors annealed in H2. This indicates that

cobalt oxide BND can also confine the charge in the MOS stacked structure. The relationship between the annealing temperature and observed memory window size (V) is summarized in Fig. 8 for a memory window size extracted from the C–V characteristics of N2- and H2-annealed

samples at different temperatures. N2-annealed samples

(depicted with circle) showed similar V independent of

0

Counts ( arb

.unit )

Energy ( keV )

Co-BNDs

Upper SiO

Bottom SiO

O

Co

( a )

Counts ( arb

.unit )

Energy ( keV )

Co-BNDs

Upper SiO

Bottom SiO

O

Co

( b )

10

8

6

4

2

0

2

4

6

8

10

Si

Si

Fig. 5. (Color online) EDS spectra of embedded BND (a) before and (b) after annealing.

( a )

( b )

( c )

( d )

( e )

( f )

( g )

( h )

20 nm

Fig. 6. (Color online) TEM images of (a) before and (e) after annealing. EDS mapping images of (b) O, (c) Si, and (d) Co before the annealing, and (f), (g), and (h) after annealing in reducing atmosphere, respectively.

the annealing temperature. Instead, the V of H2 annealed

MOS capacitors increased with annealing temperature. On the basis of the spectral change in XPS, the observed widening of the memory window is considered to be due to the increment of the metallic portion of Co-BND. This result suggests that the reduction of oxide BND to metallic Co-BND can improve the charge storage capacity of Co-BND flash memory.

The observed relationship between the memory window size and the annealing temperature can be explained using the band alignment diagrams under the flat band condition depicted in Fig. 9. As indicated by XPS data, Co-BNDs exist as cobalt oxide, Co3O4, before annealing and as metallic

cobalt, Co, after annealing. Potential barriers between Si and oxide Co-BND for EC(EC) and EV(EV) are depicted in

Fig. 9(a), respectively. The deff(e) and deff(h) are effective

potential well depths of metal Co-BND for electrons and holes with respect to the ECand EVof Si, respectively. In the

case of Co3O4, the conduction band energy (EC) of Co3O4

is positioned at 0.4 eV higher than that of Si substrate. Therefore, the injected electrons in Co-BND will easily back-tunnel to the Si substrate when the applied gate voltage returns to the flat band condition. Therefore, the charge capacity should be poor and the memory window cannot be large. In contrast, in the case of metallic Co, the Fermi level of Co (Ef 5:0 eV4)) is positioned in the band gap of the Si

substrate and can form the deep potential well of 0.95 eV between the ECof Si and Efof Co. The advantage of using a

metallic nanoparticle in floating gate memory is the deeper potential well in comparison with the use of a Si

nano-crystal.14) We should note that the electron back-tunneling owing to the increase in the potential energy of charged electrons by Coulomb charging energy can be ignored in this case. The potential of metallic Co-BND increases when a single electron injection to the embedded BND is calculated on the basis of the self-capacitance of BND (Cself ¼2"oxdÞ.14)Here, d denotes the diameter of the dot.

Coulomb charging energy gap (W) is obtained from W ¼ ðneÞ2=Cself; ð1Þ

where n denotes the number of electrons and e denotes the elementary charge. From the calculation of eq. (1) with d ¼ 7 nm, the charging energy gap of 0.1 eV is obtained. From the analysis of the C–V curve, the observed memory window size 2.5 – 3.0 V with a dot density of 6:5  1011cm2, which suggests that there are less than two stored electrons for each dot. Even after a potential increase of 0.2 eV after electron charging, the potential well depth is more than 0.7 eV. This value is sufficiently large to prevent the electron back-transfer from metallic Co-BND to the Si substrate. Therefore, better electron confinement can be achieved in BND storage node even under the flat band condition. Along with the above interpretation, the increase of the memory window size is explained as being due to the metallic Co instead of oxide Co3O4 in the BND stacked

memory structure. The obtained results indicate that the reduction of oxide BND to metallic BND can improve the charge storage properties. The reduction of oxide BND resulted in longer charge retention in which the stored electrons were not back-transferred to the substrate under the flat band voltage condition and a larger charge storage capacity, that is obviously reflected in the memory window width of the C–V characteristics.

4. Conclusions

We demonstrated the controlled reduction of Co-BNDs synthesized in ferritin protein for the application to floating gate memory. We investigated the morphology, composition and memory window sizes after annealing. We confirmed that Co-BNDs were reduced to metallic Co-BND under H2

atmosphere at relatively low temperature (450_{C) without}

deforming or diffusing. We also observed that metallic Co-BND strongly contributed to the increase of the charge storage capacitance.

0

0.5

1

1.5

2

2.5

0

Memor

y windo

w siz

e

( V

)

Annealing temperature (

°

C )

N

: 100 %

H

: 4 %

1000

800

600

400

200

Fig. 8. (Color online) Memory window size-annealing temperature characteristics. ( a ) 1.1 eV 3.1 eV ΔEV: 0.6 eV EC Vacuum Level 0.95 eV ΔEC: 0.4 eV EC EV p-Si 3O4 ( b ) 5.0 eV deff(e): 0.95 eV deff(h): 0.15 eV Co p-Si E_V 3.65 eV 2.1 eV EF Co

Fig. 9. (Color online) Band diagram under the flat band condition (a) Co3O4; (b) Co.

0

0.2

0.4

0.6

0.8

1

-8

Nor

maliz

ed Capacitance

( arb

.unit )

Bias Voltage ( V )

8

6

4

2

0

-2

-4

-6

Fig. 7. (Color online) Normalized C–V characteristics of samples (a) without Co-BNDs, (b) after 800_{C annealing in inert atmosphere,} and (c) after 800_{C annealing in reducing atmosphere.}

Acknowledgements

This work was partially supported by the Ministry of Education, Culture, Sports, Science and Technology (MEXT), ‘‘Priority Area’’. This work was also supported by the Leading Project by MEXT, Japan.

1) S. Tiwari, F. Rana, H. Hanafi, A. Harstein, E. F. Crabbe, and K. Chan:

Appl. Phys. Lett. 68 (1996) 1377.

2) S. Tiwari, F. Rana, K. Chan, L. Shi, and H. Hanafi:Appl. Phys. Lett. 69(1996) 1232.

3) I. Yamashita: IEDM Tech. Dig., 2006, p. 447.

4) A. Miura, Y. Uraoka, T. Fuyuki, S. Yoshii, and I. Yamashita:J. Appl. Phys. 103 (2008) 074503.

5) A. Miura, T. Hikono, T. Matsumura, H. Yano, T. Hatayama, Y.

Uraoka, T. Fuyuki, S. Yoshii, and I. Yamashita:Jpn. J. Appl. Phys. 45 (2006) L1.

6) T. Hikono, T. Matsumura, A. Miura, Y. Uraoka, T. Fuyuki, M. Takeguchi, S. Yoshii, and I. Yamashita:Appl. Phys. Lett. 88 (2006) 023108.

7) K. Yamada, S. Yoshii, S. Kumagai, A. Miura, Y. Uraoka, T. Fuyuki, and I. Yamashita:Jpn. J. Appl. Phys. 46 (2007) 7549.

8) T. Dougras and M. Young:Nature 393 (1998) 152.

9) R. Tsukamoto, K. Iwahori, M. Muraoka, and I. Yamashita: Bull. Chem. Soc. Jpn. 78 (2005) 2075.

10) I. Yamashita, J. Hayashi, and M. Hara:Chem. Lett. 33 (2004) 1158. 11) K. Yoshizawa, K. Iwahori, K. Sugimoto, and I. Yamashita:Chem.

Lett. 35 (2006) 1192.

12) J. F. Moulder, W. F. Stickle, P. E. Sobol, and K. D. Bomben: Handbook of X-ray Photoelectron Spectroscopy (Physical Electronics Division, 1979) p. 82.

13) I. Yamashita:Thin Solid Films 393 (2001) 12.

14) C. Lee, J. Meteer, V. Narayanan, and E. C. Kan:J. Electron. Mater. 34 (2005) 1.