Formation of Ru nanocrystals by plasma enhanced atomic layer deposition for nonvolatile memory applications

Article abstract of DOI:10.1063/1.2338793

The formation of Ru nanocrystals is demonstrated on a SiO₂ substrate by plasma enhanced atomic layer deposition using diethylcyclopentadienyl ruthenium and NH₃ plasma. The island growth of Ru was observed at the initial stages of the film formation up to a nominal thickness of 11.1 nm. A maximum Ru nanocrystal spatial density of 9.7 × 10¹¹ /cm² was achieved with an average size of 3.5 nm and standard deviation of the size of 20%. Electron charging/discharging effect in the Ru nanocrystals is demonstrated by measuring the flatband voltage shift in the capacitance-voltage measurement of metal-oxide-semiconductor memory capacitor structure.

Full text of DOI:10.1063/1.2338793

Formation of Ru nanocrystals by plasma enhanced atomic layer deposition for
nonvolatile memory applications
Sung-Soo Yim, Moon-Sang Lee, Ki-Su Kim, and Ki-Bum Kim
Citation: Applied Physics Letters 89, 093115 (2006); doi: 10.1063/1.2338793
View online: http://dx.doi.org/10.1063/1.2338793
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/89/9?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Charge trapping characteristics of Au nanocrystals embedded in remote plasma atomic layer-deposited Al2O3
film as the tunnel and blocking oxides for nonvolatile memory applications
J. Vac. Sci. Technol. A 30, 01A104 (2012); 10.1116/1.3639131
Nonvolatile memory characteristics of atomic layer deposited Ru nanocrystals with a SiO 2 / Al 2 O 3 bilayered
tunnel barrier
J. Appl. Phys. 107, 013707 (2010); 10.1063/1.3275346
Development of plasma-enhanced atomic layer deposition grown Ru–WCN mixed phase films for nanoscale
diffusion barrier and copper direct-plate applications
J. Vac. Sci. Technol. B 27, 631 (2009); 10.1116/1.3097856
Application of plasma enhanced chemical vapor deposition silicon oxynitride layers in nonvolatile semiconductor
memory devices
J. Vac. Sci. Technol. B 27, 494 (2009); 10.1116/1.3054353
Evaluation of integrity and barrier performance of atomic layer deposited W N x C y films on plasma enhanced
chemical vapor deposited Si O 2 for Cu metallization
Appl. Phys. Lett. 89, 081913 (2006); 10.1063/1.2338768
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
120.117.138.77 On: Thu, 27 Nov 2014 04:04:20

APPLIED PHYSICS LETTERS 89, 093115 ͑2006͒
Formation of Ru nanocrystals by plasma enhanced atomic layer
deposition for nonvolatile memory applications
Sung-Soo Yim, Moon-Sang Lee, Ki-Su Kim, and Ki-Bum Kima͒
School of Materials Science and Engineering, Seoul National University, Seoul 151-742, Korea
͑Received 26 May 2006; accepted 3 July 2006; published online 30 August 2006͒
The formation of Ru nanocrystals is demonstrated on a SiO2 substrate by plasma enhanced atomic
layer deposition using diethylcyclopentadienyl ruthenium and NH3 plasma. The island growth of
Ru was observed at the initial stages of the film formation up to a nominal thickness of 11.1 nm. A
maximum Ru nanocrystal spatial density of 9.7ϫ1011 /cm2 was achieved with an average size of
3.5 nm and standard deviation of the size of 20%. Electron charging/discharging effect in the Ru
nanocrystals is demonstrated by measuring the flatband voltage shift in the capacitance-voltage
measurement of metal-oxide-semiconductor memory capacitor structure. © 2006 American Institute
of Physics. ͓DOI: 10.1063/1.2338793͔
Memory device structures with nanocrystal ͑or quantum
dot͒ floating gates have been considered as a possible solu-
tion for the next generation of nonvolatile memory devices.1
However, for the integration of nanocrystals into the memory
device structure, it is important to develop a reproducible
deposition method which guarantees the simultaneous con-
trol of the spatial density, size, and size distribution of the
nanocrystals. Recently, direct deposition utilizing an initial
stage of chemical vapor deposition ͑CVD͒,2–4 a thermally
induced self-organization process by precipitation of im-
planted ions,5,6 or the agglomeration of physical vapor de-
posited thin metal films7,8 has been widely investigated for
nanocrystal formation. Among these processes, the low pres-
sure CVD ͑LPCVD͒ of Si has been found to offer relatively
good controllability of both the size and spatial density of the
nanocrystals and, thus, is widely employed as one of the
most promising process technologies to form nanocrystals.9
However, the control of nucleation and growth at the initial
stages of CVD appears to remain a very difficult problem,
because complicated chemical reactions such as decomposi-
tion, physisorption, chemisorption, and the surface reactions
of the introduced precursor/reactant molecules occur simul-
taneously within a short deposition time ͑typically less than
1 min in the LPCVD system͒. Also, the film growth which
results from combinations of various process parameters
such as the temperature, precursor/reactant ratio, and pres-
sure of the precursor/reactant is too difficult to control.
Atomic layer deposition ͑ALD͒, on the other hand, uses
a self-limiting film growth mechanism caused by surface
saturated reactions through the sequential exposure of the
precursor and reactant at each ALD cycle.10 One ALD cycle
typically consists of precursor pulsing for precursor adsorp-
tion, purging of the physisorbed precursors, reactant pulsing
for surface reaction, and purging of the by-product. There-
fore, the ALD process appears to have the following advan-
tages over the conventional CVD process in terms of its abil-
ity to control thin film formation processes. First, the
deposition conditions can be precisely controlled by making
independent adjustments of the process parameters at each
subcycle stage. Second, the deposition time can be digitally
controlled by adjusting the number of ALD cycles. Third, the
growth rate ͑or equivalently the adatom flux͒ is relatively
independent of process variables such as the temperature and
precursor/reactant ratio and can be controlled simply by ad-
justing the precursor dose. In particular, the purging stage
can play the role of shifting the deposition process close to
the equilibrium condition by offering more free time for the
surface diffusion of the deposited adatoms. This means that
the deposited adatoms have more chance to seat on the en-
ergetically stable sites without being disturbed by unwanted
species such as the physisorbed precursors and by-products
of the reaction. Therefore, the ALD process can be a more
useful process for depositing uniform nanocrystals in specific
deposition systems such as that involving the formation of
metal nanocrystals on a dielectric substrate, which generally
tends to lead to island growth. Herein, we demonstrate that
the ALD process can be a facile route to uniform nanocrystal
deposition by using the plasma enhanced ALD ͑PEALD͒ of
Ru as an example.
The ALD of Ru has been studied for applications involv-
ing a glue layer for Cu electrodeposition,11 electrodes for
dynamic random access memory capacitors,12 and metal
gates for complementary metal-oxide semiconductor.13 It has
been reported that the ALD of Ru on dielectric substrates
shows a substantial incubation period, which indicates the
possibility of having a large process window for island
growth.14 In addition, the PEALD process can be used to
modify the nucleation stage. Recently, Kwon et al. reported
that the PEALD Ru process on TiN shows no apparent incu-
bation period. This result implies that the nucleation rate is
significantly enhanced by the PEALD process.11 Another
reason for selecting Ru is its high chemical and thermal sta-
bility, which is compatible with current semiconductor tech-
nology. The relatively large work function ͑4.7 eV͒ of Ru is
also advantageous for the application of this material to
nanocrystal floating gate memory devices, because this
would be expected to give rise to better retention and charg-
ing effectiveness.7
Ru nanocrystals were deposited on thermally grown
SiO2 ͑100 nm͒/Si ͑100͒ substrates using a showerhead-
type GENI-MP1000 PEALD system ͑ASM-Genitech, Inc.͒
at a deposition temperature of 300 °C and total pressure of
3 Torr. Ar ͑99.999%͒ gas was used as both a carrier
gas and purging gas. Diethylcyclopentadienyl ruthenium
a͒
Electronic mail: kibum@snu.ac.kr
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
0003-6951/2006/89͑9͒/093115/3/$23.00 89, 093115-1 © 2006 American Institute of Physics
120.117.138.77 On: Thu, 27 Nov 2014 04:04:20

093115-2
Yim et al.
Appl. Phys. Lett. 89, 093115 ͑2006͒
FIG. 1. Number of deposited Ru on SiO2 as a function of the ALD cycles.
͑Ru͑EtCp͒ ͒ ͑99.60%͒ vapor was generated in a bubbler at
2
80 °C, at which the saturated vapor pressure of Ru͑EtCp͒
2
is 0.18 Torr, and carried into the process chamber by Ar
at a flow rate of 100 sccm ͑standard cubic centimeter per
minute͒. An rf plasma of NH3 ͑99.999%͒ diluted by Ar was
used as a reducing agent. The flow rates of NH3 during puls-
ing and Ar during purging were both 150 sccm. One ALD
FIG. 2. Plan-view TEM micrographs of Ru nanocrystals deposited for 100
͑a͒, 200 ͑b͒, 300 ͑c͒, and 500 ͑d͒ ALD cycles. The inset figures are HRTEM
images of the corresponding cycles.
cycle consisted of Ru͑EtCp͒ pulsing, purging, NH3 plasma
2
treatment at a rf power of 100 W, and purging for durations
of 5, 5, 15, and 3 s, respectively. To determine the optimal
pulsing times affording self-limited growth, the pulsing times
n−1
f͑n͒ = k␣ − k͑␣ − ␤͒͑1 − ␤͒
,
͑1͒
of Ru͑EtCp͒ and NH3 were independently varied. The
2
nominal growth rate of Ru was saturated at 0.025 nm/cycle
where f͑n͒ is the number of atoms deposited at the nth ALD
cycle and k is the conversion factor from the surface cover-
age to the number of atoms. Using this model, it is found that
the adsorption probability of the Ru precursor on Ru is about
270 times larger than that on the SiO2 substrate. The line in
Fig. 1 shows the fitting lines derived from this model, which
shows a good match with the experimental results.
when the Ru͑EtCp͒ and NH3 pulsing times exceeded 5 and
2
10 s, respectively. The spatial density, average size, size dis-
tribution, and average height of Ru as a function of the num-
ber of ALD cycles were measured using plan-view and
cross-sectional transmission electron microscopy ͑TEM͒
͑JEOL JEM-3000F with a field emission gun operated at
300 kV͒ and scanning electron microscopy ͑JEOL JSM
7401F͒. The resistivity of the 25 nm thick Ru film was about
35 ␮⍀ cm. Capacitance-voltage ͑C-V͒ measurement was
conducted using an Agilent 4284A precision LCR meter.
Figure 1 shows the number of Ru atoms deposited as a
function of the number of ALD cycles. The number of Ru
atoms deposited for less than 300 ALD cycles was calculated
from the molar volume of bulk Ru ͑8.17 mole/cm3͒ and the
average volume of the nanocrystals, which was estimated
from their average size and height in the plan-view and
cross-sectional TEM micrographs assuming that they have a
spherical shape. The number of Ru atoms corresponding to 1
ML is defined as 1.76ϫ1015 atoms/cm2, which is calculated
from the molar volume to the power 2/3. From 300 ALD
cycles, where the Ru islands start to agglomerate, the number
of deposited Ru atoms is calculated from the surface cover-
age and nominal thickness. As shown in Fig. 1, the trend
between the amount of Ru deposited and the number of ALD
cycles shows a nonlinear relationship below 500 ALD
cycles, unlike ideal ALD growth behavior. Recent studies
have revealed that the difference in the probabilities of the
precursor being adsorbed on the deposited material and sub-
strate is the reason for the nonlinear growth behavior in the
initial stages of ALD.15 From the analyses of Puruunen15 and
Lim et al.16 the number of atoms deposited per cycle can be
calculated from the probabilities of the Ru precursor being
adsorbed on Ru, ␣, and SiO2, ␤. In our case, the number of
Figure 2 shows the bright-field plan-view TEM micro-
graphs of the Ru nanocrystals deposited on SiO2 after 100
͑a͒, 200 ͑b͒, 300 ͑c͒, and 500 ͑d͒ ALD cycles. The inset
shows the high resolution TEM ͑HRTEM͒ micrographs of
the corresponding samples. Below 200 ALD cycles, the for-
mation of discrete and circular nanocrystals on SiO2 with
good spatial and size uniformity is observed ͓Figs. 2͑a͒ and
2͑b͔͒. The coalescence of the Ru nanocrystals occurs after
200 ALD cycles, as shown in Figs. 2͑c͒ and 2͑d͒. It is noted
that Ru forms islands even at 500 ALD cycles ͓Fig. 2͑d͔͒,
which corresponds to a nominal thickness of 11.1 nm. It is
considered that this wide process window for the island
growth of PEALD Ru on SiO2 comes from the preferential
deposition of Ru on the preformed nuclei. In addition, the
wetting angle between Ru and SiO2 is expected to be close to
180°, due to the large difference in surface energies
͓3.0 J/m2 for Ru ͑Ref. 17͒ and 0.2 J/m2 for SiO2 ͑Ref. 18͔͒
and negligible interface energy between Ru and SiO2. Thus,
this large wetting angle causes the ALD system of Ru on
SiO2 to have a greater tendency toward island growth.
Figure 3 shows the spatial density and average size of
the Ru nanocrystals deposited on SiO2 as a function of the
number of ALD cycles. The spatial density of the nanocrys-
tals was measured by counting the number of Ru nanocrys-
tals in the plan-view TEM images. The spatial density of the
Ru nanocrystals increases in the early stages of ALD, shows
a maximum density of 9.7ϫ1011 /cm2 at 200 ALD cycles,
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
atoms deposited per cycle is defined as
and then decreases due to the coalescence of the Ru nano-
120.117.138.77 On: Thu, 27 Nov 2014 04:04:20

093115-3
Yim et al.
Appl. Phys. Lett. 89, 093115 ͑2006͒
ϯ10 V, while negligible ⌬VFB ͑Ͻ0.01 V͒ was observed for
MOS capacitor without Ru nanocrystals. It is worth noting
that more enhanced retention characteristics of the present
Ru nanocrystals could be expected from the asymmetric
charging/discharging behavior because electrons should
overcome large tunnel barrier for discharging process. This
asymmetric charging process can also be expected from the
large work function of Ru ͑4.7 eV͒.
In summary, we demonstrate that a uniform nanocrystal
array with a high spatial density and narrow size distribution
can be deposited using the Ru PEALD process. The large
difference in adsorption probabilities of Ru precursors on
between Ru and SiO2 and the large difference in surface
energies of Ru and SiO2 are thought to be responsible for the
large process windows for island growth of the presented
method. Ru nanocrystals with an average size of 3.5 nm and
size deviation of 20% are incorporated into MOS-type
memory structures.
FIG. 3. Spatial density and average size of Ru nanocrystals as a function of
the ALD cycles.
crystals. The average size of the nanocrystals increases lin-
early from 1.5 nm at 60 ALD cycles to 3.5 nm at 200 ALD
cycles, before the coalescence of the Ru nanocrystals occurs.
After 200 ALD cycles, the average size and its deviation
increase steeply due to the coalescence of the Ru nanocrys-
tals. While it is not shown in the data, it is noted that the
standard deviation of the nanocrystal size is measured to be
about 20%, below 200 ALD cycles.
This work was supported by the National Program for
Tera-Level Nano Devices, one of the Frontier R&D Pro-
grams funded by the Ministry of Science and Technology of
Korea.. The authors also acknowledge that part of the fund-
ing came from the National Core Research Center ͑NCRC͒
program through the NANO Systems Institute and BK-21
program through the Ministry of Education of Korea.
We made metal-oxide-semiconductor ͑MOS͒ capacitor
embedded with Ru nanocrystals deposited for 200 ALD
cycles for the confirmation of electron charging/discharging
effects in Ru nanocrystals. Tunnel oxide of 5 nm thickness
was thermally grown on n-type Si substrate using a rapid
thermal process. After Ru deposition, 25 nm thick control
oxide was deposited by plasma enhanced chemical vapor
deposition. Finally, Al top gate was deposited by thermal
evaporation and patterned by photolithography. Figure 4
shows the high frequency C-V characteristics of MOS ca-
pacitor with Ru nanocrystals after voltage sweeping between
1A. Thean and J.-P. Leburton, IEEE Potentials 21, 35 ͑2002͒.
2R. Muralidhar, R. F. Steimle, M. Sadd, R. Rao, C. T. Swift, E. J. Prinz, J.
Yater, L. Grieve, K. Harber, B. Hradsky, S. Straub, B. Acred, W. Paulson,
W. Chen, L. Parker, S. G. H. Anderson, M. Rossow, T. Merchant, M.
Paransky, T. Huynh, D. Hadad, Ko-Min Chang, and B. E. White, Jr., Tech.
Dig. - Int. Electron Devices Meet. 2003, 26.2.1.
3S. J. Baik, S. Choi, U.-I. Chung, and J. T. Moon, Tech. Dig. - Int. Electron
Devices Meet. 2003, 22.3.1.
4B. De Salvo, C. Gerardi, S. Lombardo, T. Baron, L. Perniola, D. Mariolle,
P. Mur, A. Toffoli, M. Gely, M. N. Semeria, S. Deleonibus, G. Ammen-
dola, V. Ancarani, M. Melanotte, R. Bez, L. Baldi, D. Corso, I. Crupi, R.
A. Puglisi, G. Nicotra, E. Rimini, F. Mazen, G. Ghibaudo, G. Pananakakis,
C. M. Compagnoni, D. Ielmini, A. Lacaita, A. Spinelli, Y. M. Wan, and K.
van der Jeugd, Tech. Dig. - Int. Electron Devices Meet. 2003, 26.1.1.
5H. I. Hanafi, S. Tiwari, and I. Khan, IEEE Trans. Electron Devices 43,
1553 ͑1996͒.
ϯ2, ϯ6, and ϯ10 V. The flatband voltage shift ͑⌬VFB
͒
gradually increases as the voltage sweep range increases.
Large ⌬VFB of 3.6 V during backward sweep and 3.1 V dur-
ing forward sweep are achieved at the sweep range between
6E. Kapetanakis, P. Normand, D. Tsoukalas, K. Beltsios, J. Stoemenos, S.
Zhang, and J. van den Berg, Appl. Phys. Lett. 77, 3450 ͑2000͒.
7Z. Liu, C. Lee, V. Narayanan, G. Pei, and E. C. Kan, IEEE Trans. Electron
Devices 49, 1606 ͑2002͒.
8J. J. Lee, Y. Harada, J. W. Pyun, and D.-L. Kwong, Appl. Phys. Lett. 86,
103505 ͑2005͒.
9F. Mazen, T. Baron, J. M. Hartmann, M. N. Semeria, and G. Brémond,
Mater. Res. Soc. Symp. Proc. 737, F.1.9.1 ͑2003͒.
10M. Leskela and M. Ritala, Thin Solid Films 409, 138 ͑2002͒.
11O.-K. Kwon, S.-H. Kwon, H.-S. Park, and S.-W. Kang, J. Electrochem.
Soc. 151, C753 ͑2004͒.
12J. Lin, T. Suzuki, D. Matsunaga, and K. Eguchi, Appl. Phys. Lett. 88,
062902 ͑2006͒.
13M. Tapajna, P. Pısecny, R. Luptaka, K. Husekova, K. Frohlich, L. Har-
matha, J. C. Hooker, F. Roozeboom, and J. Jergel, Mater. Sci. Semicond.
Process. 7, 271 ͑2004͒.
14T. Aaltonen, P. Alen, M. Ritala, and M. Leskela, Chem. Vap. Deposition
9, 45 ͑2003͒.
15R. L. Puruunen, J. Appl. Phys. 95, 4777 ͑2004͒.
16J.-W. Lim, H.-S. Park, and S.-W. Kang, J. Appl. Phys. 88,6327 ͑2000͒.
17W. R. Tyson and W. A. Miller, Surf. Sci. 62, 267 ͑1977͒.
18R. K. Iler, The Chemistry of Silica ͑Wiley, New York, 1979͒, pp. 645–647.
FIG. 4. High frequency C-V characteristics for a MOS capacitor with the Ru
nanocrystal floating gate.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
120.117.138.77 On: Thu, 27 Nov 2014 04:04:20