Download Substrate-Independent Palladium Atomic Layer Deposition

Full Paper
Substrate-Independent Palladium Atomic Layer Deposition
By Jay J. Senkevich,* Fu Tang, Diana Rogers, Jason T. Drotar, Christopher Jezewski, William A. Lanford, Gwo-Ching Wang,
and Toh-Ming Lu
A novel method is presented for the atomic layer deposition (ALD) of palladium on a tetrasulfide self-assembled monolayer
functionalized SiO2 surface. Additionally, a novel reducing agent (glyoxylic acid) was used to remove the organic ligands from
the chemisorbed palladium(II) hexafluoroacetylacetonate metallorganic. Glyoxylic acid is an effective reducing agent above
200 C, which is not optimal for palladium but it is effective enough to show proof-of-concept and deposit a Pd ªseedº layer.
Palladium was also deposited on iridium at 80 C and 130 C via a hydrogen process or on the Pd seed layer at 80 C. The
60 Š Pd film grown on the tetrasulfide self-assembled monolayer (SAM) showed nearly random texture and higher carbon
and fluorine contamination levels compared to the one grown on Ir. The 55 Š film grown on Ir at 80 C is highly (111) textured with a grain size of ~60 Š, as shown by reflection high energy electron diffraction (RHEED). The higher contamination
levels of the Pd film deposited on the tetrasulfide SAM, as measured by X-ray photoelectron spectroscopy (XPS), is attributed to the high temperatures needed to deposit the Pd seed layer. The higher deposition temperatures cause more dissociation
of the hfac ligand and a higher metallorganic desorption rate. These equate to less Pd being deposited and with higher contamination levels.
1. Introduction
The use of self-limiting chemical reactions to sequentially grow monolayers of transition metals will positively
impact many diverse applications. Among these applications for metal ALD are noble metal catalysts on rough
electrode and mesoporous bulk materials,[1,2] Cu seed
layers for the electrochemical deposition and CVD of
Cu,[3] conformal adhesion layers to Cu metallic overlayers,[4] and alkylthiolate self-assembled monolayers.[5] In
particular, palladium is useful as a catalyst in fuel cells[6]
and for hydrogenation reactions, gas sensors[7], and hydrogen permselective membranes.[8] Palladium, like the other
noble metals, is rather costly. Therefore, a driving force exists to reduce the quantity used as a function of its activity.
ALD is ideal to extract the highest performance out of
noble metal catalysts with least cost, since the growth it
produces is highly conformal and well-controlled. As a result, monolayer-by-monolayer control of growth can take
place over complicated geometries and bulk mesoporous
supports.
To date, metal ALD has had limited success. This is primarily due to the lack of metallorganic chemisorption on
oxide-terminated surfaces and the lack of appropriate re±
[*] Prof. J. J. Senkevich, Prof. F. Tang, Dr. J. T. Drotar, Dr. G.-C. Wang,
Prof. T.-M. Lu
Dept. of Physics, Rensselaer Polytechnic Institute
110 8th St., Troy, NY 12180-3590 (USA)
E-mail: [email protected]
Dr. D. Rogers
Dept. of Chemistry, Rensselaer Polytechnic Institute
110 8th St., Troy, NY 12180-3590 (USA)
Dr. C. Jezewski, Prof. W. A. Lanford
Dept. of Physics, University at Albany
1400 Washington Ave., Albany, NY 12222 (USA)
258
Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
ducing agents. The lack of chemisorption with palladium(II)
hexafluoroacetylacetonate [PdII(hfac)2] on oxidized Ta, Cu,
and P surfaces has been shown previously.[9] The inefficency of hydrogen as a reducing agent on non-metallic
surfaces is well established, and is attributed to the lack of
polarizability and high bond strength (435.99 kJ mol±1) of
molecular hydrogen.[10] Alcohols and aldehydes have been
investigated as possible reducing agents for Cu ALD with
CuII(hfac)2 as a precursor. Unfortunately, Solanki and
Pathangey[3] undertook their depositions above the decomposition temperature of CuII(hfac)2, which is 230 C.[11] Coincidentally, 230 C is also the decomposition temperature
of PdII(hfac)2 used here.[9,12] Blackburn et al. did not show
any growth of Pd via chemical fluid deposition on oxideterminated surfaces below 230 C.[13]
Metal ALD has been successful in limited cases on noble
metal surfaces or with halogenated precursors that interact
preferentially on oxide-terminated surfaces. Martensson
and Carlsson used CuII(tmhd)2 with H2 on Pt/Pd seeded
glass substrates.[14] As has been established,[9] the noble
metals serve two purposes. These surfaces preferential chemisorb CuII(tmhd)2 or PdII(hfac)2, and are known to dissociate molecular hydrogen.[15]
H2 ® 2 H (on noble metals)
(1)
The dissociated hydrogen then serves two purposes. It reduces the Cu2+ or Pd2+ metal ions to the elemental metal
and it cleaves the tmhd or hfac ligand, allowing a second
pulse of the metallorganic to chemisorb onto the newly created surface.
Another example where a metal has successfully been
deposited via ALD is tungsten (via WF6 and Si2H6).[16] In
this chemical system, WF6 does not have much difficulty inDOI: 10.1002/cvde.200306246
Chem. Vap. Deposition 2003, 9, No. 5
Full Paper
teracting with the hydroxylated SiO2 surface and Si2H6 acts
here as a reducing agent. However, two drawbacks of this
method exist; HF is generated as a side reaction and Si2H6
is only effective reducing metal fluorides. As a result, there
is still great interest in developing an ALD technique
where metallorganics can be used and deposited on nonnoble metal surfaces, for example, polymers, oxidized metals and oxidized barrier layers.
The work presented here explores the ALD of Pd with
PdII(hfac)2 and H2, on an Ir surface, and with PdII(hfac)2
and glyoxylic acid (both shown in Fig. 1) on a tetrasulfide
self-assembled monolayer. The technique is substrate-indeCF3
sulfide
tetrasulfide
SAM (13 Å)
alkyl
SiO2 (14 Å)
Si(100)
Tetrasulfide SAM
C
OO
- CH
- Pd2+ O
O
HC
C
O O
H-C-C-OH
C
glyoxylic acid
CF3
palladium (II) hexafluoroacetylacetonate
Fig. 1. The chemical structures of the Pd ALD precursor and novel reducing
agent used in this study.
pendent since any surface can be sulfide- or thiol-terminated via self-assembled chemistry, plasma-enhanced surface modification, or sulfur-functionalized CVD oligomers.
The ALD thin films (< 60 Š,) are analyzed via Rutherford
backscattering (RBS) spectrometry, XPS, and RHEED.
2. Results and Discussion
2.1. Palladium ALD on Ir
The use of Ir as an ALD substrate accomplishes two
things. First, it has been established that PdII(hfac)2 preferentially chemisorbs on Ir.[9] Second, Ir is known to dissociate molecular hydrogen,[15] which acts as a reducing agent
and removes the hfac ligands from the PdII(hfac)2 metallorganic. Further, it is important to establish two experimental
findings to make the ALD process viable. First, the precursor (or metallorganic) should be able to be sublimed without decomposing. This can be undertaken with thermogravimetric analysis (TGA). Fortunately, PdII(hfac)2
sublimes without decomposing, as has been shown by Poston and Reisman.[17] Second, the metallorganic should not
decompose at the surface of the substrate until temperatures are reached that exceed the deposition temperature of
the metal. This has been shown previously with PdII(hfac)2
to be 230 C on Ir[9] and on an unspecified surface.[18]
As a result of the above work, the processing window for
the ALD of Pd is between 60 C and 230 C. The lower limit is due to the condensation of the precursor and the upper
limit is due to its thermal decomposition (non-self-limiting
growth). The substrate structure of Ir and the tetrasulfide
SAM is shown in Figure 2, and explained in more detail in
the experimental section. Figure 3 shows the thickness of
Chem. Vap. Deposition 2003, 9, No. 5
http://www.cvd-journal.de
Noble Metal
Fig. 2. The structure of the tetrasulfide SAM and Ir substrates used in this
study.
Thickness (Å)
O O
PdII(hfac)2 + H2
a. at 80 °C
b. at 130 °C
150 Å Ir
Si-O-C
linkage
CF3
C
CF3
Step 1: PdII(hfac)2 + glyoxylic acid
at 210 °C, 120 cycles
Step 2: PdII(hfac)2 + H2
at 80 °C, 120 cycles
65
60
55
50
45
40
35
30
25
20
15
10
5
0
130 °C
80 °C
0
40
80
120
160
200
240
# of Cycles
Fig. 3. Deposition of Pd vs. number of cycles. The lower rate at 130 C is attributed to desorption of PdII(hfac)2 from the Pd surface.
the ALD Pd thin films as a function of the number of cycles. The areal density of the Pd films was obtained by
RBS, and then converted to a thickness value by:
1 Š = 6.80 ” 1014 atoms cm±2 obtained from the density and
molar mass of palladium.
As can be seen from Figure 3, the Pd films grown at
130 C do not exhibit a linear relationship between film
thickness and the number of cycles. This paralinear growth
is attributed to the difference in bond energies between Ir±
PdII(hfac)2 and Pd±PdII(hfac)2. The atomic polarizabilities
of Ir and Pd are (7.6 ± 1.9) Š3 and (4.8 ± 1.2) Š3 respectively.[19,20] The larger polarizability helps to stabilize the van
der Waals forces between the Ir surface and the PdII(hfac)2
metallorganic. The change between the initial growth (Ir±
PdII(hfac)2) and the subsequent growth (Pd±PdII(hfac)2)
does not occur until nearly ~10 Š, due to the growth mechanism of ALD of Pd via PdII(hfac)2. It should be realized
that monolayer-by-monolayer deposition of metallic atoms
does not occur, but instead the monolayer deposition of, for
example, PdII(hfac)2 occurs assuming no desorption. The
primary reason for the lack of monolayer-by-monolayer
growth is due to the large size of the organic ligands associated with the metallorganic precursor.
At 80 C the vapor pressure of PdII(hfac)2 is much lower,
and the Pd±PdII(hfac)2 bond is strong enough that little desorption occurs. However, when Pd is grown on Ir via
PdII(hfac)2/H2, an incubation period is consistently observed. It is known that H2 has a very high reactivity with
Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
259
Full Paper
2.2. Palladium ALD on Ir and the Tetrasulfide SAM (XPS)
Ex-situ XPS was used to investigate the chemical quality
of the Pd ALD films. In particular, comparisons were made
for Pd deposited on Ir versus the tetrasulfide SAM. The
primary difference between the two types of depositions is
that a Pd seed layer is deposited via ALD on the tetrasulfide SAM at temperatures > 200 C, where glyoxylic acid is
active, and then Pd is deposited by the PdII(hfac)2/H2 process at 80 C used with Ir. Glyoxylic acid was chosen as an
effective metallorganic reducing agent due to its use as an
electroless Cu reducing agent,[22] and its thermal dissociation to form CO2 and H2CO at temperatures greater than
200 C.[23]
The positions of both the Pd 3d5/2 peaks in Figure 4 at
335.7 eV are close to what has been previously established
in the literature for elemental Pd. Elemental Pd has been
shown to have a Pd 3d5/2 binding energy of 334.8 eV,[27]
335.3 eV,[28] and 335.4 eV.[29] The slight shift to larger binding energies may be attributed to charge referencing or
slight contamination from the dissociation of the hfac ligand.[30] However, the relationship between the chemical
contamination and the binding energy shift may be less important since the Pd deposited on the tetrasulfide SAM has
higher contamination levels (C and F) in contrast to the Pd
deposited on Ir but still has the same Pd 3d binding energy.
The C 1s and F 1s XPS spectra of Pd deposited on the tetrasulfide SAM and Ir surfaces are shown in Figures 5 and
6. The Pd deposited on the tetrasulfide SAM has significantly higher levels of carbon and fluorine compared to the
Ir substrate. This is likely caused by the high temperatures
(>200 C) needed to deposit the Pd seed layer with
C-C
285.0 eV
Pd on tetrasulfide ML
C/Pd ratio 500
0.16
C-O
286.5 eV
C=O, CFX
287.7-288.3 eV
400
χ 2 = 4.2 300
Pd 3d5/2
335.3 eV
Pd on tetrasulfide SAM
Pd 3d3/2
340.6 eV
C-H
284.6 eV
200
3000
intensity (cps)
PdII(hfac)2, and therefore this observation is likely attributed to carbon residing at the Ir surface.[21] After the prerequisite number of cycles of H2, the carbon can be adequately
cleaned and Pd ALD can commence. The adventitious carbon on the Ir surface was observed with XPS (not shown)
and was significantly greater than on, for example, a SiO2
surface. Further, as the temperature of the substrate is
raised, the ability of hydrogen to clean the Ir surface should
increase.
Pd on Ir
335.7 eV 2000
341.0 eV
shake-up
shake-up
intensity (cps)
0.09
1000
346
344
342
340
338
336
334
0
332
100
Pd on Ir
290
χ = 5.2
2
288
286
284
282
0
280
Binding Energy (eV)
Fig. 5. The C 1s XPS spectra for the Pd ALD films deposited on the tetrasulfide SAM and Ir surfaces. The high binding energy shoulder on the peak
associated with the C±C or adventitious bonding is associated with ±CFx,
C=O, or C±O bonding due to the dissociation of the hfac ligand.
Binding Energy (eV)
Figure 4 shows the Pd 3d XPS spectra for Pd deposited
on Ir and on the tetrasulfide SAM. The Pd 3d5/2 peak for
the tetrasulfide SAM surface is referenced to carbon (C 1s
peak) at 285.0 eV from previous work.[24,25] The Pd 3d5/2
peak for the Ir surface is referenced to adventitious carbon
at 284.6 eV for the C 1s peak. The value for adventitious
carbon on Ir was determined with a separate experiment.
This measurement must be done with care since the tail of
the Ir 4d5/2 peak at 297 eV can interfere with the C 1s peak
at 284.6 eV.[26] As a result, quantitative results are hard to
come by, though the peak position is easily obtainable.
260
Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
250
Pd on tetrasulfide ML
688.9 eV
200
χ 2 = 3.4
150
F/Pd ratio
0.12
100
intensity (cps)
Fig. 4. The Pd 3d XPS spectra for 60 Š of Pd deposited on the tetrasulfide
SAM and for 55 Š of Pd deposited on Ir. The peak position of the Pd 3d5/2
peak is slightly shifted from elemental Pd, which might be attributed to
charge referencing or slight chemical contamination from the dissociation of
the hfac ligand.
689.5 eV
0.02
Pd on Ir
50
χ 2 = 4.6
0
694
692
690
688
686
684
Binding Energy (eV)
Fig. 6. The F 1s XPS spectra for the Pd ALD films deposited on the tetrasulfide SAM and Ir surfaces. The peak position of the spectra is correlated with
the fluorine being associated with carbon.
http://www.cvd-journal.de
Chem. Vap. Deposition 2003, 9, No. 5
Full Paper
Pd-S
162.7 eV
120
90
S2p1/2
60
intensity (cps)
S2p3/2
30
0
168
166
164
162
160
Binding Energy (eV)
Fig. 7. S 2p XPS spectrum showing the Pd±S interfacial reaction between
PdII(hfac)2 and the tetrasulfide SAM.
63.70eV
240 cycles
Ir-Pd
60.72eV
1600
120 cycles
63.89eV
Ir4f5/2
60.91eV
Ir4f7/2
1200
800
intensity (cps)
glyoxylic acid. Higher temperatures favor a greater degree
of ligand dissociation, which is problematic with the b-diketonate metallorganics.[30] Both C 1s spectra show evidence
of CFx, C=O, or C±O bonding due to dissociation from the
original hfac ligand. Much of the carbon evident from Figure 6 is attributed to adventitious sources existing at the
top of the Pd deposit due to ex-situ analysis. Proper chemical analysis of the Pd ALD films would be better undertaken in-situ; however, experimental difficulties preclude this
set up at this time.
Golub et al. have shown that the position of the F 1s peak
is correlated to atomic percent of fluorine that exists in
polymer repeat units.[31] Poly(tetrafluoro ethylene) (PTFE)
has 67 at.-% of fluorine, and a corresponding binding energy of 689.2 eV.[31] The same ±CF2± group has a binding energy in the C 1s spectra of 290.0±291.4 eV. Lin et al. showed
that physisorbed PdII(hfac)2 on a clean Cu surface at 120 K
had a F 1s binding energy of 688.2 eV.[30] After annealing
the physisorbed PdII(hfac)2 at 473 K, they saw the formation of a new peak at 682.9 eV. However, the fluorine
peaks associated with the Pd ALD films in Figure 6 are at
688.9 eV (on tetrasulfide SAM) and 689.5 eV (on Ir), possibly indicating a loss of carbonyl groups from the hfac ligand. Also, it should be noted that carbon-bonded fluorine
typically has larger F 1s binding energies compared to metal fluorides, and the binding energies of the F 1s spectra in
Figure 6 are consistent with the former.
400
15 cycles
0
II
2.3. Interfacial Interactions between Pd (hfac)2 and the
Tetrasulfide SAM and Ir Surfaces
It has been established previously that PdII(hfac)2 interacts weakly, or not at all, with hydroxyl- or oxygen-terminated surfaces.[9] Further, weak van der Waals interactions
are not strong enough to make ALD possible. As a result,
chemisorption needs to exist between PdII(hfac)2 and the
tetrasulfide SAM and Ir surfaces. That chemisorption exists
means that interfacial Pd±S and Pd±Ir bonding should be
evident. In order to see such bonding in the XPS spectra,
for example, the Pd±S signal should be large relative to the
C±S or S±S bonding, if the S 2p spectrum is used for analysis.
Figure 7 shows the S 2p spectrum for a monolayer of
PdII(hfac)2 deposited on the tetrasulfide SAM surface
(Fig. 2). The tetrasulfide SAM has a peak at 164.1 eV in
the S 2p XPS spectrum (previously established).[25] After
pulsing PdII(hfac)2 over the surface, chemisorption is evident and a significant binding energy shift occurred in the
S 2p spectrum, as shown in Figure 7. The composite S 2p
peak exists at 163.1 eV, whereas the deconvoluted spectra
has two peaks, one at 162.7 eV associated with the S 2p3/2
peak, and one at 163.9 eV associated with the S 2p1/2 peak.
The spin orbital splitting is 1.18 eV between the S 2p peaks.
The shift in the Pd 3d spectrum for one monolayer versus a
60 Š film is not significant.
Chem. Vap. Deposition 2003, 9, No. 5
http://www.cvd-journal.de
70
68
66
64
62
60
58
Binding Energy (eV)
Fig. 8. The Ir 4f XPS spectra showing the Pd±Ir interfacial reaction between
PdII(hfac)2 and the Ir surface. The binding energy shift is only obvious for
the 240 cycle Pd film because only the interfacial chemistry contributes to
the Ir signal.
The Pd ALD films deposited on Ir are slightly different
than the ones deposited on the tetrasulfide SAM because
the Ir film is ~150 Š versus just one monolayer for the sulfur-terminated SAM. As a result, interfacial bonding is
only evident in Figure 8 for the 31 Š Pd film that was deposited with 240 cycles at 130 C. Before that point, the Ir±
Ir bonding overwhelms the Pd±Ir bonding, and this peak is
convoluted in the Ir 4f spectra. The Ir 4f7/2 peak is centered
at 60.91 eV, the literature value.[24] The Pd±Ir interfacial
bond is centered at 60.72 eV. No literature value exists for
this bonding, but Ir±Si bonding has been reported at
60.70 eV, which may be similar to Pd±Ir.[32]
2.4. Reflection High-Energy Electron Diffraction
(RHEED)
The structure of the Pd ALD films is most easily obtained by using a RHEED camera under ultra high vacuum
(UHV) conditions. RHEED is sensitive to the structure of
angstrom thick metallic films, whereas conventional X-ray
Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
261
Full Paper
ferred (111) texture. The lattice constant is (3.8 ± 0.1) Š,
based on the (111) d-spacing. The grain size is ~ 60 Š,
which is very similar to the thickness of the Pd film. Also, it
has a texture spread of ~ 70 at the surface and ~ 40 in the
bulk of the film. A high-quality Pd ALD film would be expected for the deposition of a metal on a higher surface
free energy metal that is relatively smooth, ~ 10 Š, root
mean square (rms) (from atomic force microscope (AFM)
measurements).
The Pd ALD film grown on the tetrasulfide SAM possesses very small grains and has nearly random texture.
This is uncharacteristic of a metal film that exhibits monolayer wetting, as previously shown.[25] However, the difficulty that arises with the use of
PdII(hfac)2 is the high volatility it
(111)
Pd on Ir
possesses at low temperatures (e.g.,
lattice constant=3.8 Å ± 0.1 Å
1 torr at 60 C).[12] Further, the hfac
ligand shows some instabilities even
(200)
at room temperature,[30] and as the
(220)
(222)
dosing or substrate temperature is
raised the dissociation of the hfac li(331)
gand becomes worse. As previously
explained, glyoxylic acid is only effective as a reducing agent at temperatures greater than 200 C. Therefore,
2
4
6
8
10
the processing method developed
-1
Ring Position (Å )
here is more suitable for lower vapor
Pd on tetrasulfide SAM
pressure metallorganics that do not
(111)
dissociate at the deposition temperatures needed to undertake ALD for
(220)
their respective metal.
diffraction (XRD) is not appropriate due to the small
X-ray cross-section.
Figure 9 shows the RHEED images and the corresponding spectra of Pd on Ir, Pd on the tetrasulfide SAM, and
the Ir substrate itself obtained from digitizing the images.
The ~150 Š Ir thin film thermally evaporated on hydrogen-terminated Si(100) shows some slight (111) orientation
but exhibits nearly random texture. Based on a d-spacing
of (2.86 ± 0.04) Š for the (111) plane, the calculated lattice
constant is (3.81 ± 0.06) Š, close to the literature value of
3.839 Š.[33]
Despite the nearly random texture of the Ir substrate,
the 55 Š Pd ALD film deposited on it has the much pre-
Intensity (arb.)
8000
6000
4000
2000
0
0
Intensity (arb.)
6000
(222)
4000
2.5. Chemical Analysis of Glyoxylic
Acid
2000
0
0
2
4
6
8
10
-1
Ring Position (Å )
Ir substrate
(111)
12000
-1
(200)
10000
Intensity (arb.)
-1
(111) postion = 2.856 Å ± 0.042 Å
lattice constant = 3.81 Å ± 0.06 Å
8000
(220)
(113)
(222)
6000
4000
2000
0
0
2
4
6
8
10
-1
Ring Position (Å )
Fig. 9. RHEED images and spectra for 55 Š of Pd on Ir (top), 60 Š of Pd on tetrasulfide SAM (middle),
and the 150 Š Ir substrate (bottom). The Pd on Ir exhibits a higher quality texture than the Ir substrate
itself. However, the Pd on the tetrasulfide SAM is nearly random, which is attributed to the higher temperatures needed to achieve the Pd seed layer.
262
Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
http://www.cvd-journal.de
Glyoxylic acid is used here to deposit a Pd seed layer via ALD on top
of the tetrasulfide SAM. After the
seed layer is deposited, Pd can be deposited via H2 at low temperatures
where the growth rate per cycle is
higher. The preliminary experiments
showed that glyoxylic acid is only effective as a reducing agent above
200 C, the temperature where it
thermally decomposes (according to
previous work by Back and Yamamoto).[23] They showed that, at lower
temperatures (~ 200 C), a 5:1 ratio
exists between CO2 and formaldehyde for the decomposition products.
As the temperature is raised, a lower
ratio is evident but CO2 is always the
dominant
species.
Nevertheless,
glyoxylic acid is an effective reducing
Chem. Vap. Deposition 2003, 9, No. 5
Full Paper
agent since it decomposes cleanly to yield oxidation products and electrons available for reduction reactions with
the metallorganic.
The as-received glyoxylic acid is a hydrate, and water is
known to dissociate b-diketonate metallorganics.[34] Proton
nuclear magnetic resonance (1H NMR) spectroscopy was
used to analyze the glyoxylic acid as-received and after dehydration to ensure that this glyoxylic acid was utilized as a
reducing agent rather than effects caused by the presence
of water. Figure 10 shows the 1H NMR spectra for the
glyoxylic acid before and after dehydration under vacuum
heating to 60 C for 60 min. There are four peaks in the
glyoxylic
acid
TMS
acetone
intensity (arb. units)
dehydrated
acid-H2O
proton exchange
as-received
(hydrated)
9
8
7
6
5
4
3
2
1
0
ppm
Fig. 10. 1H NMR spectroscopy of the glyoxylic acid reducing agent before
(bottom spectrum) and after (top spectrum) dehydration in vacuum
(10 sccm Ar carrier and 10±4 torr base pressure) at 60 C for 60 min. The
broad peaks at 7.2 ppm and 8.5 ppm are attributed to a proton exchange
mechanism between glyoxylic acid and water. When the peak is at higher
chemical shifts it is more associated with the acid than free water.
proton spectra for glyoxylic acid dissolved in d6-acetoneÐ
the tetramethylsilane reference peak at 0 ppm, 2.1 ppm acetone (impurity in the d6-acetone), 5.3±6.1 ppm glyoxylic
acid, and a broad peak attributed to proton exchange
between the glyoxylic acid and water.[35] Two items are
apparent from the 1H NMR spectra. First, the integrated
area between the broad peak at 7.2 ppm for the as-received
sample and the group of peaks at 5.3±6.1 ppm associated
with glyoxylic acid is 2.9:1. In contrast, the ratio for the
dehydrated sample is 0.8:1. Second, the peak associated
with proton exchange shifts from 7.2 ppm (full width at
half
maximum
(FWHM) = 17.1 Hz)
to
8.5 ppm
(FWHM = 113 Hz). The conclusions to be made from these
observations are that much less water exists in a glyoxylic
acid sample without taking into consideration the water in
the d6-acetone and the water adsorbed after the sample
was dehydrated. Further, the protons are much more associated with the acid than free' water after dehydration.
This last conclusion comes from the larger chemical shift
for the dehydrated sample. As a result, glyoxylic acid is effective as a reducing agent, there is little associated water,
and therefore water plays a minor role in the removal of
the hfac ligands from the PdII(hfac)2 metallorganic.
Chem. Vap. Deposition 2003, 9, No. 5
http://www.cvd-journal.de
3. Conclusions
To date there has been very little success with the ALD
of metals via metallorganics. The method presented here is
a novel one, which allows ALD metals to be deposited onto
dielectric and barrier layer surfaces. The method is
achieved by activating the dielectric surface with sulfide
terminated silanes, then depositing a Pd seed layer by the
sequential pulsing of PdII(hfac)2 and glyoxylic acid above
200 C. Once the seed layer is deposited then Pd ALD is
achieved by the sequential pulsing of PdII(hfac)2 with H2 at
low temperature (80 C), where PdII(hfac)2 has a long transient time on the newly created Pd surface. The method is
appropriate to all metals where the metallorganics can sublime without decomposing and chemisorb on an appropriate substrate without dissociating. Since glyoxylic acid is
only effective above 200 C, precursors that have a low
vapor pressure at those temperatures are required, or the
development of reducing agents that are effective at lower
temperatures is needed.
The thicknesses of the Pd ALD films were measured as a
function of the number of ALD cycles via RBS. At a deposition temperature of 130 C, the relationship was not
linear due to the high volatility of PdII(hfac)2, or in other
words, its high desorption rate from the freshly created Pd
surface. However, at lower temperatures (80 C) the desorption rate is not significant and linear growth is observed. The Pd ALD films were also characterized by XPS
and RHEED. The films grown on the tetrasulfide SAM
had a higher levels of carbon and fluorine contamination
attributed to the high temperature need to deposit the Pd
seed layer. This was related to the dissociation of the hfac
ligand. When the films were grown on Ir the contamination
levels were lower due to the temperature dependence of
the hfac dissociation. Further, the films grown on Ir exhibited (111) texture with a grain size of ~ 60 Š for a 55 Š
thick film. In contrast, the Pd films grown on the tetrasulfide SAM showed very small grains with a nearly random
texture. This relatively poor structure, as compared to the
Pd film grown on Ir, was attributed to the dissociation of
the hfac ligand, thus allowing fluorine and carbon contamination at the Pd seeded surface.
4. Experimental
Two surfaces were used for the ALD of palladium, as shown in Figure 2.
150 Š of Ir (99.95 %, Alfa Aesar, Ward Hill, MA) was electron beam-deposited onto a Si(100) wafer. The other surface was a tetrasulfide-terminated self-assembled monolayer solution grown on the native oxide of
Si(100) via bis[3-(triethoxysilyl)propyl] tetrasulfide (>97 %, Gelest Inc.,
Morrisville, PA). The details of the growth and characterization of this
monolayer has been published previously [17].
A custom-built vacuum chamber with computer controlled gas flow was
used for the ALD of palladium via PdII(hfac)2. The base pressure of the vacuum system was 5.0 ” 10±4 torr with the use of a roots blower/direct drive
roughing pump (Leybold RUVAC WS/WSU 151 and TRIVAC D25 BCS
(hydrocarbon prepped)) and a 3.7 L foreline trap (bronze gauze) to prevent
oil backstreaming. The vacuum chamber was warm walled to prevent con-
Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
263
Full Paper
densation of the PdII(hfac)2 precursor. Further, pneumatic valves (that turn
off or turn on the gas flow and mass flow meters) can control the flow of
gases into the system under steady state conditions.
Two types of deposition were undertaken, one on a noble metal surface
(Ir) and one on a tetrasulfide SAM. The depositions (outlined in Fig. 2) on
Ir were undertaken at a deposition temperature of (80 ± 4) C or (130 ±
4) C, and a PdII(hfac)2 sublimation temperature of (49.8 ± 0.2) C. The deposition chamber walls were kept at 60±65 C, and the lines between the
sublimation tube and the deposition chamber at 90±95 C to ensure no cold
spots existed to condense PdII(hfac)2. During each experiment, 30 sccm Ar
(99.999 %, Air Products) was flowed as a purge gas, 10 sccm Ar as a carrier
gas for PdII(hfac)2, and 8.0 sccm H2 (99.999 %, Air Products) as a reducing
gas. The pulse sequence was 5 s of PdII(hfac)2 followed by 10 s of dead
time' (with just 30 sccm Ar flowing as a purge gas), and then 20 s of H2. The
number of cycles per deposition varied from 8 to 240 for Pd ALD on Ir.
The depositions on the tetrasulfide SAM were slightly different due to
the inefficacy of glyoxylic acid (Alfa Aesar, MA) as a reducing agent below
200 C [31], and the high rate of desorption of PdII(hfac)2 on Pd at temperatures above 200 C due to its high vapor pressure and low interaction energy
with Pd. This behavior may be followed in terms of Langmuir kinetics [36].
As a result, these depositions consisted of two parts. The first part was the
deposition of a seed layer of Pd, with PdII(hfac)2 pulsed for 5 s alternatively
with 60 s pulses of glyoxylic acid (sublimation temperature (60 ± 2) C and
15 sccm Ar carrier). As with all of the depositions, a 10 s dead time was
used. It took 120 cycles to deposit the seed layer of Pd, after that point the
PdII(hfac)2/H2 process was used at 80 C for a further 120 cycles.
The RHEED images were taken with a 15 keV electron beam source that
produces a 100 lm diameter spot size. The pressure in the stainless steel
chamber was ~1 ” 10±9 torr. The high-energy electron beam was incident at
1 from the sample plane. The diffracted electrons pass through two metal
retarding grids before striking the phosphor-coated screen, imaged using a
16-bit Princeton Instruments CCD camera. The NMR work on the analysis
of the water in the glyoxylic acid was undertaken on a Varian Unity INOVA
300 (300 MHz). 1H NMR spectra were undertaken on dehydrated and as-received glyoxylic acid with eight scans with a pulse width of 8 ls at 90 in d6acetone with 1 v/v-% tetramethylsilane as a proton standard.
Received: September 27, 2002
Final version: February 25, 2003
±
[1] H. P. Hsieh, Inorganic Membranes for Separation and Reaction, Elsevier, New York 1996.
[2] G. Xomeritakis, Y.-S. Lin, in Chemical Vapor Deposition±Proc. 14th
Int. CVD Conf. and EUROCVD-11 (Eds: M. Allendorf, C. Bernard),
Vols. 97±25, The Electrochemical Soc., Pennington, NJ 1997, p. 1604.
[3] R. Solanki, B. Pathangey, Electrochem. Solid-State Lett. 2000, 3, 479.
[4] J. Chae, H.-S. Park, S.-W. Kang, Electrochem. Solid-State Lett. 2002, 5,
C64.
[5] S. Busse, J. Kashammer, S. Kramer, S. Mittler, Sens. Actuators B 1999,
60, 148.
[6] A. Wieckowski, H. Kim, C. Rice, P. Waszczuk, Proc. Electrochem. Soc.
2001, 2000-20, 91.
[7] J. R. Pitts, P. Liu, S.-H. Lee, C. E. Tracey, R. D. Smith, C. Salter, Proc.
2000 DOE Hydrogen Program Rev. National Renewable Energy
Laboratory, US Department of Energy, Golden, CO 2000, Vol. 1;
http://www.eere.energy.gov/hydrogenanafuelcells/hydrogen/annual-review2000.html
[8] R. C. Hughes, W. K. Schubert, R. J. Buss, J. Electrochem. Soc. 1995,
142, 249.
[9] J. J. Senkevich, F. Tang, G.-C. Wang, T.-M. Lu, T. S. Cale, C. Jezewski,
W. A. Lanford, Chem. Vap. Deposition 2002, 8, 189.
[10] K. P. Huber, G. Herzberg, Molecular Spectra and Molecular Structure
Constants of Diatomic Molecules, Van Nostrand, New York 1979.
[11] C. Jezewski, J. J. Senkevich, unpublished.
[12] Gmelin Handbook of Inorganic Chemistry, Pt Suppl Vol. A1 (Eds:
G. J. K. Acres, K. Swars), Springer-Verlag, Berlin 1986.
[13] J. M. Blackburn, D. P. Long, J. J. Watkins, Chem. Mater. 2000, 12,
2625.
[14] P. Martensson, J.-O. Carlsson, J. Electrochem. Soc. 1998, 145, 2926.
[15] C. J. Hagedorn, M. J. Weiss, W. H. Weinberg, Phys. Rev. B 1999, 60,
R14016.
[16] J. W. Klaus, S. J. Ferro, S. M. George, Thin Solid Films 2000, 360, 145.
[17] S. Poston, A. Reisman, J. Electron. Mater. 1989, 18, 553.
[18] Gmelin Handbook of Inorganic Chemistry, Pt Suppl. Vol. A1 (Eds:
G. J. K. Acres, K. Swars), Springer-Verlag, Berlin 1986, p. 45.
[19] G. D. Doolen, Los Alamos National Laboratory, unpublished.
[20] A. Zangwill, P. Soven, Phys. Rev. A 1980, 21, 1561.
[21] V. Bhaskaran, M. J. Hampden-Smith, T. T. Kodas, Chem. Vap. Deposition 1997, 3, 85.
[22] Y. Y. Shacham-Diamand, Electrochem. Solid-State Lett. 2000, 3, 279.
[23] R. A. Back, S. Yamamoto, Can. J. Chem. 1985, 63, 542.
[24] R. Lenigk, M. Carles, N. Y. Ip, N. J. Sucher, Langmuir 2001, 17, 2497.
[25] J. J. Senkevich, B. Li, G.-R. Yang, G.-C. Wang, T.-M. Lu, Electrochem.
Solid-State Lett. 2002, 5, C94.
[26] J. F. Moulder, W. F. Stickle, P. E. Sobol, K. D. Bomben, Handbook of
X-ray Photoelectron Spectroscopy (Eds: J. Chastain, R. C. King Jr.),
Physical Electronics, Eden Prairie, MN 1995.
[27] M. Schildenberger, R. Prins, Y. C. Bonetti, J. Phys. Chem. B 2000, 104,
3250.
[28] M. Peuckert, J. Phys. Chem. 1985, 89, 2481.
[29] M. Brun, A. Berthet, J. C. Bertolini, J. Electron Spectrosc. Relat. Phenom. 1999, 104, 55.
[30] W. Lin, B. C. Wiegand, R. G. Nuzzo, G. S. Girolami, J. Am. Chem.
Soc. 1996, 118, 5977.
[31] M. A. Golub, T. Wydeven, R. D. Cormia, J. Polym. Sci., Part A:
Polym. Chem. 1992, 30, 2683.
[32] M. Wittmer, P. Oelhafen, K. N. Tu, Phys. Rev. B 1986, 33, 5391.
[33] B. D. Cullity, Elements of X-ray Diffraction, 2nd ed., Addison-Wesley,
Reading, MA 1978.
[34] E. L. Hahn, D. E. Maxwell, Phys. Rev. 1952, 88, 1070.
[35] D. Yang, J. Hong, D. F. Richards, T. S. Cale, J. Vac. Sci. Technol. B
2002, 20, 495.
[36] A. W. Adamson, Physical Chemistry of Surfaces, 5th ed., John Wiley &
Sons, New York 1990.
______________________
264
Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
http://www.cvd-journal.de
Chem. Vap. Deposition 2003, 9, No. 5