CA2377315A1 - Method for the synthesis and characterization of supported metal nanoclusters of controlled size, surface distribution, shape and interfacial adhesion - Google Patents

Abstract

A method of fabrication of the construction of nanostructures on substrate surfaces is provided, which allows the control of the sizes, the surface distributions, the shapes and the stability thereof. The method further enables the characterization of these parameters, as well as the tailoring of the size and distribution of nanostructures on substrate surfaces.

Description

TITLE OF THE INVENTION
Method for the synthesis and characterization of supported metal nanoclusters of controlled size, surface distribution, shape and interfacial adhesion.
FIELD OF THE INVENTION
[0001] The present invention relates to nanoclusters. More specifically, the present invention is concerned with a method for synthesizing and characterizing metal nanoclusters having a controlled size, surface distribution, shape and interfacial adhesion.
BACKGROUND OF THE INVENTION

[0002] in the booming field of nanotechnologies, efforts are being made to provide nanostructures usable in nanoelectronics, under the form of quantum dots, single electron transistors, nanophotonic devices for instance, in the manufacturing of chemical products such as highly specialized catalyzers, and far the purpose of.
superconductivity as well. Such applications require mastering the sizes, distributions and adhesions of nanoaggregates.

[0003] A number of deposition methods have been used in constructing nanoclusters. The e-beam method suffers from law throughput because of serial lithographic methods. Stranski-Krastonov growth requires the initial formation of a strained epitaxy, with subsequent growth progressing through uncontrolled Ostwald ripening, leaving residual metal between clusters. Nanoclusters deposited onto substrates, even under the most precisely controlled mass selection conditions, are subject to size distributions and thermodynamically determined shapes.

[0004] The precise control of nanoclusters properties, such as size, lateral surface position and positional stability, is necessary for applications such as supported heterogeneous catalysts, magnetic and optical data starage media and others where size, lateral position and shape influence the optical, magnetic, electronic and chemical properties. The precise control of nanocfuster alignment and stability is necessary in the fabrication of nanostructures, and in the fabrication of material requiring precise alignment to achieve their optimum material conductivity potential.
OBJECTS OF THE INVENTION

[0005] An object of the present invention is therefore to provide an improved method for synthesizing and characterizing metal nanoclusters having a controlled size, surface distribution, shape and interfacial adhesion.
BRIEF DESCRIPTION OF THE DRAWINGS

[0006] In the appended drawings:

[0007] Figure 1 is a curve illustrating the peak separation of the C1 s XPS
spectrum of the HOPG surface after a 2 keV Ar+-treatment for 10 min;

[0008) Figure 2 is a curve showing angle-resolved C1 s XPS intensities for both the untreated HOPG surface and the same surface after a 2 keV Ar+-treatment for 4 min;

[0009] Figure 3 shows (a) the C2:Cto~, peak intensity ratio as a function of a duration of a 2 keV Ar+-treatment at 0° (perpendicular) and 70°
take-off angles, respectively; (b) the result of a SRIM simulation of a 2 keV Ar+ damage at a beam angle of 57°.

[0010] Figure 4 is a contact AFM image of Cu clusters on a 2 keV Ar+-treated HOPG surface, wherein the dimensions are in microns;

[0011] Figure 5 shows the C2 component relative intensity and the Cu cluster number density on HOPG as functions of the duration of an Ar+-treatment ;

[0012] Figure 6 shows the Cu cluster coalescence kinetics on HOPG having different surface defect densities ;

[0013] Figure 7 illustrates the effect of surface defects on the Cu cluster coalescence coefficient ;

[0014] Figure 8 illustrates the Cu cluster size effects on the Cu cluster coalescence kinetics for Ar+-treated HOPG surfaces having the same defect density (2 keV Ar+ irradiation for 4 mm) ;

[0015] Figure 9 illustrates the dependence of the coalescence parameter on the Cu cluster size for HOPG treated with 2 keV Ar+ for 4 min ;

[0016] Figure 10 illustrates the coalescence coefficient for different initial Cu cluster sizes, as a function of 2 keV Ar+ irradiation time ;

[0017] Figure 11 shows high resolution (a) C 1s, (b) O 1s, (c) Si 2p and (d) N
1 s XPS spectra for a Cyclotene surface treated with a 3 keV NZ+ beam for 20 min;

[0018] Figure 12 is a graph showing atomic concentrations of Cyclotene as a function of exposure to a 3 keV N2+ beam;

[0019] Figure 13 is a graph showing changes in the component peaks of (a) C 1 s, (b) O 1 s and (c) Si 2p XPS spectra of Cyclotene as a function of exposure to a 3 keV N2+ beam;

[0020) Figure 14 is a graph showing relative atomic concentrations as a function of take-off angle for Cyclotene exposed to a 3 keV N2+ beam for 20 min;

[0021) Figure 15 is a graph showing relative concentrations of (a) C 1 s, (b) O
1 s, (c) Si 2p, and (d) N 1 s XPS spectral peak components of Cyciotene, as a function of take-off angle, on exposure to a 3 keV N2+ beam for 20 min; and [0022] Figure 16 illustrates the coalescence kinetics of a nominal 8 h Cu deposit on Cyciotene that was (a) untreated and (b) treated with a 3 keV N2+
beam for 20 min.
DESCRIPTION OF THE PREFERRED EMBODIMENT
[00231 The present description refers to a number of documents, the content of which is herein incorporated by reference in their entirety.
[0024] Generally stated, the present invention provides a method for controlling the deposition of copper aggregates on dielectric surfaces.
[0025] More precisely, the present invention provides a method for controlling the sizes of copper nanoaggregates deposited on dielectric substrate and their distributions on the surface thereof. Additionally, the present invention provides a method allowing enhancing the adhesion of such copper nanoaggregates on the surfaces they are deposited on.
[0026] Therefore, the present invention provides for repeatabilibly controlling the size and the distribution of copper aggregates deposited on dielectric surfaces.
[0027] The present invention stems from the observation that copper, when deposited by vaporization on dielectric surtaces, does not wet the dielectric surtaces, such as, for example, the Dow Cyclotene Polymer and the highly oriented pyrolytic graphite, known as "HOPG". On the contrary, instead of forming smooth layers on such surfaces, vaporized copper forms nanoaggregates. The size of these nanoaggregates is of the order of nanometers. The way these nanoaggregates are arranged on the dielectric surfaces is assessed through their distribution, which relates to the number thereof per unit surtace.
[0028j More specifically, it is found that the copper nanoaggregates are generally spherical in shape, with a diameter of a few manometers.
[0029j Studies on the mechanisms of the formation of such nanoaggregates, of their adhesion properties on a substrate, and on coalescence mechanisms yielding larger aggregates have been undertaken.
[0030] In particular; on the one hand, chemical and physical techniques, such as plasma treatment, ion beams method, are tested to modify the surtaces of substrates in order to act on the interfacial adhesion. On the other hand, methods for modifying the dimensions and distributions of the nanoaggregates by using ion beams are developed. Finally, ion beams are used to reach both targets: to control the nanoaggregates and to modify interfacial adhesion.
[0031] That is, it is shown herein that evaporated and sputtered metals deposit, onto several insulating substrates important to the microelectronics industry, in the form of nanoclusters. While such nanoclusters tend to be spherical, those deposited by gentle landing from the gas phase, using cluster deposition tools (seeded supersonic nozzle sources, gas aggregation sources, etc.), have non-spherical (icosahedral, etc.) shapes. Nanoclusters that do float adhere well to substrates undergo lateral diffusion and coalesce in vacuum, with the coalescence kinetics being inversely related to the extent of cluster-substrate adhesion.

Ar+ Induced Surface Defects on HOPG and Their Effect on the Nucleation, Coalescence and Growth of Evaporated Copper L
[0032] In a first aspect of the present invention, it is shown that a component peak of the Class XPS spectrum in Ar+-irradiated HOPG, at 285.6 eV, is due to surface defects. This is done by demonstrating a strong correlation of the component with the size and number density of the Cu cluster.
[0033] Further, the coalescence behavior of Cu clusters on HOPG having different surface defect densities caused by Ar+ ion irradiation is investigated. It is found that, on the one hand, the growth of larger clusters takes place through cluster coalescence, and, one the other hand, that, as the cluster size approaches atomic dimensions, the growth takes place through Ostwald-like ripening.
[0034] Finally, a model for nucleation and growth, in which deposited atoms and small clusters interact with defect sites and are immobilized to some extent, is described.
[0035] Firstly, the motivation for the works yielding the method of the.present invention will be described.
[0036] An understanding of the nucleation, initial growth and coalescence of metallic clusters on thin films at the initial stage of deposition is of prime importance in the production of nanoscale structures and devices, and it has been motivating many theoretical and experimental recently. Due to their importance in industrial chemical catalysis, metal clusters or nanoparticles on oxide substrates have been extensively investigated and documented over the last two decades [1-8]. In addition, HOPG, due to its well-defined surtace structure and weak interactions with metals, has been widely used in the study of supported metal clusters deposited from the vapor [9-23J
or as size-selected clusters [24-26J.
(0037] Much progress in the understanding of the nucleation, initial growth and coalescence of metallic clusters on thin films at the initial stage. of deposition has recently been made through the use of high-resolution in-situ STM [3-7, 26J.
An interesting aspect of metal cluster diffusion on the HOPG surface is that the value of the diffusion coefficient, "D", is surprisingly found to be greater than expected, sometimes reaching 10-a cm2/s [27-28J
(0038] Although scanning transmission electron microscopy, or "STEM", was initially used to study nucleation, growth and coalescence [29, 32), the limitation of in-situ, time-dependent observations led researchers to combine STEM with molecular dynamics simulations [7,27,28]. While STM/AFM has an obvious potential in the study of in-situ cluster coalescence, growth and surface diffusion behavior, these techniques may suffer from tip convolution effects [33-37J, even for tips giving atomic resolution on flat terraces. This has caused most researchers to use clusters heights instead of cluster diameters. In fact, there are stilt tip limited-size effects from height measurement results [37J if the separation between clusters is not large enough compare to the tip dimension.
(0039] It was found eisewere by the present applicants [38, 39J, that XPS can be used in the estimation of the average cluster diameter and has been used in-situ to determined time-resolved coalescence and growth without the limitations of surface' topography and surface conductivity. Additional interfacial chemical and compositional information is also provided with high sensitivity.
[0040j Although surface defects are generally considered to be one of the most important factors for nucleation, growth and coalescence [7, 24, 25, 40-44J, there is still little quantitative, systematic description of their effect on these properties. Those surface defects created by low energy Ar+ radiation on HOPG have been investigated by Raman spectroscopy. This was done through monitoring the ordered Raman-active E2g peak at 1580 em-' (the so-called "G mode"), and that at 1356 cm's ("D mode") due to the onset of disorder effects [45-48j. It was previously shown that the D:G peak intensity ratio was influenced by disorder and related to its length (see Nakamura and al., [47-48j). AES was used to evaluate surface damage effects induced by low energy rare gas ion sputtering by analyzing the C(KLL) peak shape change due to the appearance of a high-energy shoulder; this method is limited by the SIN of the spectrum (See Steffen and al. [49,50]).
[0041] Surface probe microscopy, particularly STM with its superior space resolution, has been extensively used to study surface damage induced by low energy ion bombardment [51-66j. The general results of these studies are the following:
(1 )the surface exhibits protrusions or hillocks, separated by flat terraces on which normal atomic order is observed;
(2) the number of hillocks depends on the incident ion dose and energy; and (3) the average dimensions of the hillocks are insensitive to the ion incident energy and species.
[0042] Generally, there are some difficulties in using STM to distinguish surface point defects [56-57], because STM images of vacancy defects ("VD") and interstitial defects ("ID") have contributions from both the states and the geometry of a focal electron. It was suggested in the past [60, 64j that one type of defect could be distinguished from the other by measuring the local tunneling barrier height (~) and tunneling spectroscopy (trough an I-V curve). However, a question remains concerning how to relate the defect density with hillock density. Indeed, it is difficult to separate the number point defects in a STM image of a single hillock. In addition, the observed surface defect states may also be influenced by the surface state of the probe tip [44j.
[0043] XPS, one of the most sensitive surface techniques, has also been used to analyze HOPG surface structure and HOPG surface damage [67-70j. It was previously suggested [68] that the X-ray excited Auger line referred to as "C(KLL)" may be used to evaluate the relative concentrations of sp2 and spa hybridization in a-C thin films. The increase of asymmetry on the high energy side of the C(KLL) spectrum was then attributed, upon Ar~ ion irradiation, to the onset of disorder. It was further suggested in the past [69] that the change of C1s peak asymmetry under irradiation is due to the production of a spa component (Jackson and Nuzzo). Such an explanation however is in conflict with the previous AES results [49,50].
[0044) For their part, the present applicants have previously proposed [70]
that the change in asymmetry on the high binding energy side of the C1s spectrum of HOPG under Ar+ bombardment is due to an homolytic bond scission and to the creation of a less delocalized sp2 network. A new peak located at 285.6 eV, which is 1.0 eV
higher than the main C1s peak, is associated with a free radical site resulting from such a bond scission and sp2 network. This conclusion was confirmed by confocal Raman analysis data. fn fact, it appears that these free radical sites are the surface defects that are to be considered. The. purpose here is therefore to discuss these surface defects in terms of their effect on the nucleation, growth and coalescence Cu clusters deposited by evaporation.
[0045] Having introduced the general background surrounding the work yielding to a method according to a first aspect of the present invention, a summary of the experimental steps at the origin of this aspect will now be given.
[0046 A component peak of the C1 s XPS spectrum was used to evaluate surface defects induced by keV Ar+ radiation on HOPG. As explained hereinabove, it was previously showed that a component 1.0 eV higher than the main C1s peak at 284.6 eV, whose intensity is strongly correlated with the extent of Ar+ irradiation, is due to less extensive sp2 electron delocalization caused by the breaking of surtace bonds.
Here, it is shown that, for evaporated Cu deposited onto the Ar+-treated HOPG surface, both the number density and average size of the Cu clusters formed are correlated with the surface defect density. These results also indicate that the nucleation of Cu clusters takes place at these defect sites, and that the Cu cluster must further overcome an energy barrier to diffuse and coalesce. The coalescence process follows the universal equation, d = kt", where "d" is the average cluster size, and "k" and "a" are two coalescence parameters influenced by the interaction between a cluster and a substrate surface. The Cu cluster coalescence coefficient "DS" is strongly dependent on the surface defect density and on the initial size of the Cu cluster. This is used to suggest electrostatic interactions between deposited atoms and small clusters with defect sites.
[0047] The detailed experimental set ups will now be briefly described.
[0048] The samples of the type ZYA HOPG, having dimensions 10 mm x 10 mm and 2 mm thick, were obtained from SPI fnc: they were cleaved with adhesive tape just prior to each experiment and immediately inserted into a spectrometer.
[0049] XPS was carried out in a VG ESCALAB 3 Mark II, using non-monochromated Mg Ka X-rays (1253.6 eV). The base pressure in the analysis chamber irvas less than 10 ''° tort. Spectral peaks were separated using an in-house, non-linear least mean squares program. Ar+ treatment took place in the instrument preparation chamber at a pressure less than 10 -'° tort, using an ion energy of 2 keV and a current density of about 10'3 ionslcm2.s. The angle between the Ar+ beam and the surface of a sample was about 57°. Following the Ar+ treatment, the samples were immediately transferred to the analysis chamber without exposure to the atmosphere.
[0050] Copper (Cu) was also evaporated, as described in the art, at a deposition rate of about 4 Almin, by electron beam. The nominal Cu thickness was monitored with a quartz crystal oscillator placed near the sample.
[0051] The results obtained using the above-described experimental set up are now described in detail.

[0052] First, the results related to the C1s XPS spectra of Ar+-induced surface defects and disorder will be enumerated.
[0053] As was recently shown by the present applicants [70], the C1 s XPS
spectrum undergoes the following changes under Ar+ irradiation:
(1 ) the n* ~-- ~ shakeup, at about 291.2 eV, rapidly disappears;
(2) there is an increase in asymmetry on the high binding energy side of the main C1 s peak; and (3) there is an increased FWHM of the C1 s peaks, suggesting that the surface defect formation decreases the lifetime of a hole.
[0054] Following the Ar+ irradiation, the C1s spectrum 10 shown in Figure 1 can be separated into four components, as detailed by the present applicants [70].
Figure 1 illustrates the peak separation of the C1 s XPS spectrum of the HOPG
surface after a 2 keV Ar+-treatment for 10 min. Briefly, a C1 peak 12, seen at about 284.6 eV, is attributed to extensively delocafized sp2 bonding; a C2 peak 14, at about 285.6 eV, is attributed to more localized sp2 banding as a result of bond scission; a C3 peak 16, at about 286.5 eV, is attributed to spa bonding; and a C4 peak 16, at about 288.0 eV, is a shake-up of the C2 peak 14.
[0055] Figure 2 shows an angle resolution of the intensity of a C1s spectrum for HOPG samples before (curve 20) and after (curve labeled 22) Ar+
irradiation. The results agree with previous results in the art [67]. In particular, the observed loss of diffraction peaks indicates that the surface structure is destroyed to some depth under Ar+ irradiation. A SRFM simulation [130], at an angle of 57° gives a 2 keV Ar+ collision event (damage) depth extending down to about 4 nm.
[0056] Similar angle resolution of the C2:C1 peak intensity ratio [70] leaves no doubt that the C2 component occurs at or near the surface, as is confirmed by the time dependence of the C1 : C2 ratio at perpendicular (0°) (curve labeled 32 in Figure 3a) and grazing (70°) (curve labeled 34 in Figure 3a) take-off angles.
It appears that the C4 : C2 intensity ratio remains unchanged during irradiation, as does the C3 peak intensity [70].
[0057] As determined from a SRIM simulation of 2 keV Ar+collision events as a function of depth illustrated in Figure 3b, the averaged damage at 0°
(3 ?~ cos 8 = 4 nm) is slightly more than half the averaged damage at 70° (3 ~, cos 8 =
1.35 nm). This is in exact agreement with the results of XPS displayed in Figure 3a.
(0058] Secondly, the results related to the initial nucleation of evaporated Cu on the Ar+-treated HOPG surface will now be enumerated.
[0059] A typical surface morphological AFM image of Cu clusters on the Ar+-treated HOPG surface is illustrated in Figure 4. The clusters 40 are spherical in shape, as confirmed by TEM [38], and uniformly distributed on the surface 42. There is no preferential decoration on terrace steps, such as seen in the art for most relatively unreactive metal clusters on the untreated HOPG surtace [21, 24, 25; 71, 72].
[0060] Cu was evaporated on HOPG surfaces with different defect densities, produced by different Ar+ irradiation times. The initial Cu cluster number density was estimated from intensity ratios in the following way:
_I" __ Ia ~~1- a d ~~°~
IS I° t-O 1-a d ~~_ (1 ) where "la~ and "IS" are XPS peak intensities from Cu cluster and substrate respectively, "d" is the Cu cluster average size, and "~." is the inelastic mean free path, best referred to as the attenuation length. "8" is the Cu coverage given by the following relation:
O = d (2) where "w~ is the Cu effective thickness. Therefore, substituting (2) into (1 ), the following relation is found:
Ia __ I° ~1- a d l ~°~ (3) IS I°d-wl-eal~=
Since the Cu cluster average size d and the Cu effective thickness w are further related to the cluster density ~n" by the following relation:
n d 3 = w or h = w (4) d the following can be written:
_Ia __ Ia n d2~1-a d l~°~ (5) IS 1° d -~ °d2~1-a d l~'S~
[0061] The cluster density n may be estimated from Equation (5) by using the cluster average size d obtained from the cluster: substrate peak intensity ratio. Such a method avoids using the effective thickness and the problems related therewith since the effective thickness may vary with treatment.
[0062] The result is shown in Figure 5 as a function of Ar+ irradiation. It should be noted that all the XPS data were acquired within a few minutes after sample deposition. The similarity of the time-dependent surface defect concentration and the time-dependent initial Cu cluster number density is clearly seen. This implies that the Cu nucleation site is located at the surface defect site.
[0063] Thirdly, the results related to the. dependence of the Cu cluster coalescence on surface defects at room temperature will now be detailed.
[0064] Figure 6 shows the dependence of the Cu cluster coalescence on surface defects is shown by the coalescence kinetics of Cu_ clusters on HOPG
surtaces exposed to Ar+ treatment for different times. Although the same amount of Cu was deposited in all cases, the average cluster size decreases with treatment time. Indeed;
while the coalescence is invariably given by the following the power law described ealier hereinabove:
d = k to (6) where "k" and "a" are constants that depend on the defect density and on the deposition condition, the higher the relative surface defect concentration, the smaller the initial Cu cluster size.
[0065] For convenience in describing the coalescence behavior of the Cu cluster the coalescence coefficient "DS" is used [39]. It is assumed that surface diffusion occurs by Brownian motion, as formalized by the following relation:

DS . 1 ~x ~ (7) 4 t and that:
/x2' _ /x\2 =1/n (~) where <x2> is the average square displacement of the cluster over time t.
Combining Equations. (4) and (8), the following equation is obtained:
DS - 4 w3 t ('-3qt (9) [0066] This indicates that the cluster coalescence rate depends on the cluster size a (when a ~ 113) and on time. The time dependence of DS, for different surface defect densities, has been calculated from equation (9) and is illustrated in Figure 7. As expected, the higher the surface defect density, the lower the value of the coalescence coefficient. It is to be noted that the data in Figure 7 are in reasonable agreement with reported diffusion coefficient values determined by STM in the art [73, 74].

(0067] The effect of the Cu cluster size on the coalescence is of great concern to people in the art [27, 28, 40-44, 73]. Such effect has here been evaluated by the present applicants by depositing different sizes of Cu cluster on HOPG
having the same surface defect density, through controlling the thickness of the Cu deposition.
Figure 8 shows that there are only minor differences when the initial cluster size is greater than 1 nm. However, Cu clusters whose initial size is less than 1 nm experience a faster coalescence, as indicated by the 0.2 nm data in the Figure 8. The dependence of a on the cluster size is shown Figure 9, where an inverse logarithmic relation indicates that the larger the cluster size, the smaller the value of a.
(0068] The time dependence of DS on the cluster size .is found in Figure 10, where all the HOPG surfaces were treated for 4 min by a 2 keV Ar+ beam, for comparison purposes. It should be understood that a comparison of the relative positions of the different treatments would lead to serious error, since the time between deposition and XPS analysis varies with the amount of Cu being deposited.
Clearly, the value of a is essentially constant with the thickness of the deposited layer until it decreases to the diameter of a Cu atom.
(0069] The various results reported hereinabove were analyzed along the following fines.
(0070] Concerning the surface defects created on HOPG by Ar+ irradiation, it is believed that the results demonstrate that threshold energy exists for ion-induced defects in HOPG [49, 50]. More precisely, in the case of rare gas ions, the threshold energy increases linearly with ionic radius, from 22.5 eV for He to 47.5 eV
for Kr for example. An energy of 2 keV for Ar+ was used in the above-described experiments, which is therefore much higher than the threshold energy for creating surface defects.
Other possible processes for the present irradiation conditions include sputtering, bond scission and ion penetration. Ar+ penetration will create point defects in the subsurface region. As mentioned earlier, a SKIM simulation (see Figure 3b) indicates a damage depth of about 4nm for 2 keV Art, which is about equal to the C1 s photoelectron probe depth of about 4nm in HOPG. Normal collision processes lead to a non-uniform depth distribution for the defect density, as noted hereinabove and as confirmed by the SRIM
simulation.
[0071] As noted by the present applicants [70], such defects are electrophilic and; therefore, appear at higher binding energies than expected. The following Egelfoff's suggestion [75] is used, that states that:
~ Ea = kvI n~ (1 O) j0072] where "n~" is a defect atom co-ordination number and "~Eb" is a binding energy increase due to a co-ordination number reduction. The change in the number "Nd" of defects per unit volume induced by ion irradiation per unit time can be written as follows:
d Nd U - Sn Nd 11 dt r ~No - Nd~~ ( ) where "N" is the density of HOPG (known in the art as 1.25 x 1023 atoms/cm3 [48]), "s" is a displacement cross section, "~" is an incident ion flux, "v" is a mean number of displaced atoms in a cascade per primary impact (i. e., the damage function), and "s"" is the sputtering yield of the carbon atoms at regular HOPG sites. This gives the following dependence of defect density on time:
No a~t~
Nd = Cl - a ~n+6~~ ~t~ (12) Sn -~ Sd -f- 6(~U
[0073] This result indicates that the volume defect density depends on the ion flux ~, the damage function v, the displacement cross-section a and also on the sputtering coefficient s". The dashed Line in Figure 5 represents the fit to the experimental data.

[0074] Moreover, the saturation, which represents a maximum defect density, is given by the following equation:
Nd (max. = No 6~U (13) s" + 6~U
[0075] It is also possible to determine the C2 : C1 peak intensity ratio, which is given by the following relation:
~~u (1. a ~sR+a~u ) rJ
Nd __ Na . (14) N No - Nd ' s" + a~v a "+~U r [0076] Such a non-linear dependence of surface defects upon ion irradiation was previously found in the art by using STM image analysis [22, 53]. The hillock number density, which represents the surface defects, so obtained was then also approximately proportional to the number of ion impacts (with the exception of C+
bombardment) [53, 65] under low dose ion irradiation. This correspondence will not hold true when a single hillock does not correspond to a single point defect.
[0077) Turning now to the nucleation of Cu at the initial stage of deposition on the HOPG surface, it is concluded that the cluster nucleation takes place at surtace defect sites. The surtace defect density induced by Ar+ ion irradiation may be estimated and compared with the number of clusters. As seen in Figure 5, the cluster density for 8 A of deposited Cu varies from about 5x10"/cm2, for untreated HOPG, to about 6 x 10'31cm2 at saturation.
[0078] Two conclusions can immediately be derived from Figure 6:
(1) the maximum Cu cluster density is about 1-2 % of the surface carbon density;
and (2) the cluster density saturation is reached after 3 min of Ar+ irradiation, which is consistent with the time dependence of surface defect production.
[0079] This should be compared to Figure 3a, where, as the take-off angle is lowered, resulting in the analysis becoming more surface-sensitive, a plot of C2/C~°,~,, as a function of irradiation time, shows that C2 is about 20 %, still greater than the Cu cluster density. This difference in cluster density and defect density magnitudes may be attributed to one or more of the following:
(1) a Cu cluster may sit on several defect sites; and (2) a spectrum of cluster-defect site interaction energies may exist, with the cluster held only at sites where the interaction is high enough.
[0080] Indeed, metal cluster escape, also called detachment, from nucleation sites bas been seen in the art by TEM in the case of Au clusters on AI203 [7, 43]. As suggested by the equation (1), the defects may be single-point or mufti-point.
[0081) From the time dependence of the Cu cluster size shown in equation (6), the time dependence of the Cu cluster number density may be determined as follows:
n = k3 t 3" (15) [0082] Since there is a finite time necessary between initial deposition and XPS observation, the initial cluster number density measured must always be less than the nucleation density. This is another possible reasan for the difference in cluster density and defect density magnitudes:
[0083) In relation to the coalescence and growth of Cu clusters on HOPG, Cu cluster coalescence on HOPG with different extents of surface defects is illustrated in Figure 7 and, with different amounts of initial Cu thickness on the same surface defect density, in Figure 8. Both cases follow equation (6), where the value of exponent a is often used as an indication of the growth mechanism.
[0084] In one limiting scenario, mass transport between clusters may occur atom by atom, in which case an atom detaches from one cluster and moves to another.
Known in the art, this process is called Ostwald ripening, with experimental values of a ranging from 0.25 to 0.33 [77, 78). In another limiting scenario considered in the art, cluster growth occurs through cluster diffusion and coalescence, with a found to be 0.20 or less [44, 79, 80).
[0085 Despite many theoretical and experimental studies devoted to the meaning of the value of a, there is still no generally satisfactory understanding among people in the art [4). Because of the importance of the value of a in ascertaining the cluster coalescence mechanism, a short discussion will now be presented.
(0086] The present applicants recently studied cluster growth not only following (static) but, also, during (dynamic) Cu deposition [81]. A
comparison of a values for the two conditions, on variously treated Cyclotene substrates, is seen in Table I, in which are listed values of a for static and dynamic Cu cluster growth under various conditions.
Surface treatmentDynamic growth Static growthAdhesion (N) Untreated 0.32 0.14 1.8 Ar+-treated 0.34 0.12 2.6 Table I
In Table I, the specific surface treatment processes used are detailed elsewhere [90, 91) and the growth conditions used are explained in [81). In the case of the static growth (column 3), the total Cu thickness used was 8 A, deposited at 0. 1 A/s.
Finally, the adhesion (column 4) is measured by a MicroScratch TesterT"", according to experimental details described in [92).

[0087j Cu is known to coalesce, in the static case, by the movement of whole clusters [39], and this is reflected in the a values. However, in the dynamic case, the a values are substantially larger. Higher a values attributed to "Ostwald-like"
ripening were also reported iri the art from experimental [78, 82-86] and theoretical [87-86] studies on both static and dynamic cluster growth. The phrase "Ostwald-like" is used here since there is a contribution from an atomic deposition process, as well as from an atomic surface diffusion. The higher a value during dynamic cluster growth may be due to the process domination of adatoms striking the clusters or the substrate surface between, and diffusing to the clusters [88]. That is, during dynamic growth, cluster coalescence only plays a minor rote because of its smaller a value.
[0088] As can be further seen in Table I, interfacial interaction between cluster and substrate appears to be another factor that affects the value of a. Stronger interfacial interactions during static growth cause Cu retention, increasing the density of nucleation sites. This interaction also retards the motion of clusters across the substrate surface, giving slower cluster growth and a correspondingly smaller value of a. Such a model is consistent with the experimental data of the present applicants [38, 39, 70, 81, 90-92) and with some others reported in the art [84). Combining the static and dynamic results of Table I, it can be seen that, even in the case of Ostwald ripening, cluster-substrate interaction plays an important role.
[0089 The initial cluster size is another important factor affecting the value of a, as is illustrated in Figure 9. It should particularly be noted that there is an increase of a to 0.30 as the cluster size approaches that of an adatom, and as the cluster migration becomes atom migration. At the other extreme of Figure 9, clusters larger than 6 nm do not appear to grow. A similar phenomenon was also found in the art for Pd clusters on Ti02 [4, 44J. Considering Figures 5 and 9, it can be seen that the surface defect density, through its influence on cluster size, influences a [4, 24, 25, 44j. In a recent finding by the present applicants, it is found that ion beams can significantly increase the value of a [93).

[0090] Therefore, it can be concluded conclude that the value of a is strongly affected by the following parameters:
(1 ) the interaction between cluster and substrate surface;
(2) the initial cluster size; and (3) the surface defect density.
[0091] From Equations {6) and {9), it is further found the following relation:
1-3a D$ = 4 w d a (16) indicating that the coalescence coefficient D5 decreases as the cluster size increases, for a less than 1/3. For static coalescence, a is less than 1/3 unless the cluster size approaches that of an adatom. In the case under consideration here case, a ranges from 0.05 to 0.3, giving values of (1-3a)la in the range of 17 to 0.33.
[0092] In confirmation of Equation. 16, several experimental and theoretical studies [73, 78, 94-102] known in the art have found a similar dependence. For example, it was reported that Pd clusters on Ti02 substrates show a DS ~c d'~
dependence [4, 44], which is consistent with others' prediction [100], while it was elsewhere suggested that the value of a ranged from DS x d'3 to DS ~ d'' depending on the specifics of diffusion [94]. Still another theory indicated that DS ~c d'' [101] and a further one predicted that DS
~ d'2 [102]. A Lennard-Jones simulation [73] indicated that the rate of diffusion of a cluster varies roughly as the inverse the contact area between the cluster and the substrate: DS ~c d-2. It was also found that Ag clusters on Ag (100) diffuse roughly as DS
~ d''~6 indicating that the dependence of coalescence coefficient on cluster size is dominated by interaction between cluster and substrate [95].
[0093] The dependence of the coalescence coefficient on the initial size, as illustrated in Figure 10, indicates that, when the size of the cluster is atomic, the growth occurs by Ostwald-like ripening, while, when the size is greater, the growth takes places by cluster coalescence.
[0094] Finally, a relationship between defect site and cluster nucleation is presented. As previously demonstrated by the present applicants [70), Cu deposition onto Ar+-treated HOPG gave no evidence of chemical reaction (i. e., no carbide was formed) nor were there any other changes manifested in the XPS and Raman spectra that could be used to indicate any other sort of interaction. Thus, since some sort of interaction undeniably takes place, it must be very weak. Further, it must take place between an electropositive Cu atom and a positively charged defect site. Here it is suggested that this is an induced dipole-charge interaction. In such a case, the interaction energy "U" is U =_ -a ez~2 r4 (16) where a is the polarizability of a Cu atom, "e" is the charge of a site and "r" is the site-dipole distance. Such an interaction is very weak and short-range.
0095 Interestingly, it was recently found in the art [103), in relation on the calculation of the properties of linear, planar and 3-dimensional Cu clusters by density functional theory, that all clusters a few atoms in size and greater, no matter what their shape, experience charge separation, meaning that atoms having the greatest co-ordination numbers have the greatest negative charge. Thus, as the cluster begins to grow, it changes from an induced dipole to a permanent dipole (quadruple, etc.) and its adhesion to the site actually increases slightly for a short time. This may be the reason behind the present finding that, in some cases, larger clusters move more easily than atoms.
The Surface Modification of Dow Cyclotene by Low Energ~N~+ Beams and its Effect on the Adhesion of Evaporated Cu Films (0096] In a secand aspect of the present invention, attention is drawn to the surface modification of Dow Cyclotene by low energy NZ+ beams and its effect on the adhesion of evaporated Cu films.
0097 Basically stated, low energy (3-6 keV) N2+ beams are used to modify Dow Cyclotene for the purpose of grafting N-containing groups onto the Cyclotene surface. In-situ XPS analysis demonstrates an extensive loss of aromaticity due to bond breaking by the beam, white angle resolution shows that implantation occurred substantially below the Cyclotene surface. The paucity of N-containing groups at the outer surtace and the resultant poor adhesion of the Cu clusters permit extensive cluster coalescence.
[0098] As an introduction to the matter of the second aspect of the present invention, it is reminded that Dow Cyclotene 3022, also known as "BCB", a low permittivity insulator, is one of several candidates for near-future "ULSI"
(for Ultra Large Scale Integration) and "GS1" (for Giga-Scale Integration) technologies, especially in combination with copper metallurgy. However, when Cu is deposited, either by evaporation or sputtering, as a base layer for subsequent electrochemical deposition, its adhesion is generally found to be weak. There are several methods that can be used to promote adhesion in this case, among which the chemical modification of the Cyclotene surface through the use of plasmas, ion beams and lasers, in order to graft functional groups onto the surface. Here it is found that N2 plasma modification achieves the highest adhesion of evaporated Cu on the Cyclotene surface. This low energy (below 15 eV) technique modifies the surface layer but does not penetrate below.
However, such thin modified layers may not be mechanically adequate for device mechanical stability.
In order to increase the modified layer thickness, low energy (in the range of keV) ion beam modification may be a better choice due to its implantation effects. Such beams have already been used in the art for polymer metallization and the surface modification of metals. The present aspect of the invention deals with the comparison of law energy N2+ beams with the previously used N2 plasma technique as a method of modifying the Dow Cyclotene surface. Cu cluster coalescence dynamics are used to evaluate adhesion to the modified surface.
[0099] The experimentally steps at the foundation of the method of the present aspect of the invention are described as follows.
[00100] Cyclotene 3022 samples were prepared as is well known to people in the art [91,105-107, 121J. Briefly, cleaned Si wafers were treated with 1%
(w/w) aqueous y-aminopropyl triethoxysilane followed by a 46% (w/w) solution of B-staged Cyclotene 3022 in mesitylene. After spin deposition, the wafers were linearly heated at a rate of 1 °C/min to 250°C, under an N2 atmosphere, and were permitted to cool down to room temperature, still under an N2 atmosphere, before removal. Cyclotene layers were about 1 micron thick.
[00101] X-ray photoelectron spectroscopy was carried out as known in the art [91, 105-107J: a VG ESCALAB 3 Mk II, operating at a pressure below 2 x 10'x' torr, used non-monochromated Mg Ka radiation at 1253.6 eV. High-resolution spectra were obtained at a perpendicular take-off angle, using pass energy of 20 eV and 0.05 eV
steps. After Shirley background removal, the peaks were separated using an in-house non-linear least squares program, using peak shapes and widths previously available in the art for this material [91, 105-107, 121J. It is to be noted that, with the exception of angle-resolved data, the take-off angle was always perpendicular to the Cyclotene surface.
[00102] N2+ beam treatment of the Cyclotene took place in the instrument preparation chamber at a pressure inferior to 10'9 torr, using a VG AG21 cold cathode gun, with 3-6 keV kinetic energy beams and under a working pressure of 4x 10'5 torr.
The angle between the beam and the surface was about 57°.

~~i~se~~E~t ~~ ~v~~c~r~ti~r~ ~4 ~~r~~c~rs~~d ~s ~r~~i~~si~ d~sc~ib~a~ ire the art ~~, ~~~-1~~~, a~ ~ ~~t~ ~~ ~.~~.,~s. ~ h~ s~ trey ~d s~r~~b~s ~r~r~~
i~r~~di~i:~i~
tr~~~s~~rr~d t~ the: ~r~~iysis ~~~~r~~s~r ~~~i~h~~"~ ~~ir~g ~tr~~s~h~ri~ e~r~
~s~re.
~irsv, the r~s~its ~i~t~ir~~d i~~ refuting t.~ s~d°f~c~
r~~~diti~~~~i~r°
~y~~i~t~r~~ dy ~3~- ~~~r~s r~aii r~~~~~ L~ ds~ri~d.
~~'i~~~ ~ r~~ti~~~t~i~ div~~r~r~~~s ~~~r~ ~~u~c~ ~~r P~2f ~~~r~s i~~ th~3 3-~
~~~~
r~r~~. ~~~r this r~~s~r~, the ~~ii~v~rirr~ ciis~s~ssi~~ ~~ili ~~ ~~r~xi~~~d v~
the ~~r~~~ts ~~" ~ '=~e'~J
~~~~s.
~~~'~~~~ ~si~a the ~~~J, ~~~~~~~~r~t r~s~its ;~r~~i~a~ss~ d~t~rr~i~~d ~~ ih~
~r~s~~t ~~~fi~~r~ts ~~ . , ~ ~~-r ~)7~, ~h~ i~r~s~r~t ~~ s, ~~ s, ~i~i~ ~3~d ~
s s~~gutr~ a~~~r~
s~~~ratad ir~t~ t~'~ir ~~~~~~~:~lt ~~~i~s. ~h~s~ ~~~ sh~~r~ ~ ~ ;=~r< ire ~,i~i~r~ ~ '. i~i~~r~ ~ ~
sh~~.~s high r~s$~i~ri~~ ~Gj ::~~ s, ~~;~ ~ ~ s, ~~~i~~ ~r~d ~dy i~3'i s ~~~
~~~~c;tr~ ~~~~ ~
~jl~~~~~re~ ~3a'~~~C~ ~~'~~t~d'e~Jfi:~"3 c'$ ~ 'K~'~!° .''~2' ~y'F-'~rs'i ~~~r ~~ ~Y"i~, ii~Jil~r~lr'~ ~h~ ~;~'~-°~~~i-<3~"~s°~r~~:~d ~,~r~g~r~~r~ts ~~~~~r e~rith des'c~:~~d lira's. ~ h~ ~t~ri~~tior~s ~rf th~s~
p~~iss ~r~ gi~~r~ ire ~~l~l~
ii, which ~~~c~~r~~s the pr~s~or~t i~zø ~a~~s~ tre~t~~~f r~skaits pith th~s~
~r~ ~r~- ~~~~rr~ ~r~d i'~a gl~s~~ tr~~tr~e~t ~s ~~t~ir~~d i~ the ~~t.

~ abf~ ff [Oaf~1~ hero are some differences, =~rhich ~fviff bo discussed hur~inbef~~;w.
s hose peak fits ropresent i:he gr cat~st statistical significance obtair~abfe, and rn~ore c ,nsistont ever fho soon angfos ~tsvd to ob:_ain the angle-resof~rod data eve ~rifl shortf~°
disc~sss.
~~~19~~ ~-ligh-resoi~aticr ~ °i s, C3 1 s, ~i ~p and ~ 1 s ?~~~
spoctra, E~o~°
Cycfotene oxpos~;d to a ~ ko'~~ d~2~. boam vor ~~ mrn, are sfnovvn ire f=ig.
fa. ~har~ges fo~~~d in tg a ~ is spo~;~e°ui~ ir~cl~do tf~o rapid riisappearanco ~~ tf~e r~' a re shako-up, arid ar as~mrr~ot~,3 on ti'~o high bir~;~ing s~s:er~~ silo cv ti~o pock as r~ev~~
co~,ponors~s aro introduced. ~ new pock vvas ~-~fso ir~~rod~ced in ~otf~ th~r ~ 1s spocvr~r~ in i=igurr ~'ib and the ~i ~p sp~;ctr ~ar~ in t~~~ ~i~s~ro 1 ~ c, vy~~ile sorr~o peaks proa~ioL~sf~ foE Bnd ~~ other troatrrsonts X91, 1 ~~~~ pro~ic~d fir tried b~~ the ~resor~t appf icar~~s wore absent. oho P~ r s spectrum, shogun in Fig~sre 11d, indicates that tvvo ~-containing peaf~,s v~~oro present, .which suggests that. t~.ro groups may ha~ro been iratrflduced, as pre~rio~asfy found en ~~2 pfasr~~a treatment X91]. ,~itorr»iti~efy, it mad indicate that two ~-containing fragments vverP implanted, as v~iif be ft~~°ti~er discussed f~oreinbofovv.

[00109] Figure 12 illustrates the evolution of the surface composition on beam exposure time as determined by using XPS peak areas and sensitive factors. It can be seen that he near-surface relative N concentration increased with beam exposure time, while those for Si, O and C decreased.
[00110] Component evolution as a function of beam exposure time is illustrated in Figures 13. As seen in Figure 13a, the C1 component decreases as the C2 component increases, while the C3 and C4 components remain effectively constant.
Thus, it can be understood that bonds are broken and electron delocalization becomes more localized during exposure to the beam. The -Si-O-Si- bond breaking appears to stabilize after 3 mm, as seen in the O 1s and Si 2p peak evolutions illustrated in Figures 13b and 13e. While the N1 s component peak evolutions (not shown) continue to increase with beam exposure time, the N1 : N2 ratio remains constant at 2 : 1.
[00111] The elemental depth distribution induced by a 20 min beam exposure has been determined by angle-resolved XPS, and the results are shown in Figure 14. It appears that both Si and O concentrations are enriched at the outer surface while N and C concentrations are depleted. Such changes are similar to those found for C, Si and O
on Ar+ beam exposure [107) and indicate that, in the present experiment also, the surface is being sputtered by the beam. The N profile shown in Figure 14 indicates that little N has remained at the surface. Whether this is also due to sputtering or to implantation will be considered hereinbelow.
[00112] Angle-resolved XPS permitted the determination of the peak component profiles. An example is seen in Figures 15 far Cyclotene exposed to a 3 keV
beam for 20 min. Both Figures 15a and 15d indicate that the relative concentrations of the C and N components are maintained as a function of depth. However, Figures 15b and 15c indicate a disagreement, since 01 and Si1, which were previously attributed to the same structure (-Si-O-Si-), move in opposite directions. This will be further discussed hereinbelow.

[00113] Turning now to results related to Cu deposition on the modified Cyclotene surface, it appears that no obvious changes occurred for C 1 s, O 1 s, Si 2p and N 1 s on the evaporation of about 3ML Cu onto the treated Cyclotene surface. The deposited Cu formed clusters, as in previous studies made by the present applicants [91, 105, 106]. The initial cluster coalescence and growth was followed by XPS. The coalescence and growth dynamics of Cu clusters were found to be strongly correlated with the interaction between Cu and the Cyclotene surface.
[00114] Typical Cu cluster coalescence processes, for a nominal 8 A thick Cu, on both untreated Cyclotene and Cyctotene pre-treated for 20 min with 3 keV
N2+, are illustrated in Figure 16. As noted in previous studies by the present applicants [81, 122]
on copper coalescence, cluster coalescence is made possible when interaction between Cu cluster and substrate surface is weak. The clusters are then able to diffuse, and on contact, coalesce and grow in size. It is reminded that the change in the average size d of Cu cluster with time t follows the power-law (6).
[00115] The present applicants previously showed that the adhesion of Cu film, as measured by the critical load "L~" is directly related to the coalescence parameters k and a through the coalescence. coefficient, Dg [81, 122, 123].
The critical load is the lowest load, measured during a microscratch test as detailed in the art [92, 123], at which a film delamination in initially observed. The Cu film critical load may be estimated, in the present case, from the k and a values. !t is found to be around 6 N.
Although this is slightly higher than that for Cu on untreated and Ar+-treated Cyclotene surfaces [92], it is much lower than the 1..~ value of about 17 N obtained for Cu on the N2 plasma-treated Cyclotene surface.
[00116] The above reported results are analyzed and yield the following points.

[00117) First, the Peak Components are considered. As was found in previous XPS studies on treated Cyclotene and HOPG surfaces by the present applicants [81, 91, 106-107, 122], peak attributions are difficult and, often, mufti component, as can be comprehended from Table II. For example, in the presence of N, the C2 component is a combination of the free radical produced on bond breaking and the formation of C-NHCHO, etc. [91j. In this particular case, the free radical contribution to this peak may be estimated by assuming it to be the same as that formed on Ar+ treatment.
Then, subtraction of this component gives a 10% contribution from C-NHCHO, etc. This is consistent with what is found for N in Figure 12.
[00118) The peak separation of the N 1s spectrum presented hereinabove gave peaks in positions identical to those found on N2 plasma treatment. For those N
fragments that have reacted, it is believed that the attributions are correct here, as well.
However, some of the fragments may not have reacted and may exist interstitially. The fact that only two N 1s peaks exist implies that any unreacted fragments also contribute to those same two peaks. This is consistent with data obtained on HOPG treated with N2+ beams [124, 125j.
[00119) The different peak ratios, namely N1 : N2 = 1 : 1 for N2 plasma treatment and 2 : 1 in the present case, indicate that the energy difference (N2+ is the major component in our N2 plasma generator [111]) plays a role in the reaction. The role played by the beam energy is also revealed in the free radical data in the O 1 s and Si 2p component spectra. For example, the absence of the 01 s peak at 532.8 eV
suggests [107J that the -Si-O free radical does not exist although the Si 2p peak at 103.0 eV
suggests that it does. One reason for this confusion is that N has a lower electronegativity than O' and, when replacing it, as in -Si-N- and -C-N-bonds, causes electron density changes at photo-emitting atoms that are reflected in binding energy shifts. It is shown in the art for example that a trigonal N has a Pauling electronegativity of 3.91 while a digonal O has a Pauling electronegativity of 6.21 [126j. Using efectronegativity arguments, group electronegativity values of 3.27 for O-CH3 and O-C2H5, and 2.66 for NH-CH3 and NH-C2H5 are also calculated in the art [126).
Such ratio is in reasonable agreement with the ratio of measured inductive substituent parameters elsewhere tabulated in the art [127). Thus, the apparent absence of some component peaks, when compared to the present N2 plasma study and the different component ratios may signal different reaction products due to energy differences in the treatment processes.
(00120] Concerning nitrogen profile and Cu adhesion, as stated hereinabove, it is known a threshold exists for ion-induced defects in HOPG [112-114). It is reminded that for rare gas ions, this threshold increases linearly with ionic radius from 22.5 eV for He to 47.5 eV for Kr for example. There is little doubt that the threshold is in this range for N2+ on Cyclotene. That is, below , this threshold, N2+ is expected to react at the substrate surface since there would be no reason for N2+ to remain at the surface if it has not reacted. Above this threshold, it is expected to undergo a Coulomb explosion, producing N+, which penetrates into the subsurface. Moreover, some of the fragments may react and some may form interstitials.
(00121] A short digression on Coulomb explosions is now in order. This phenomenon is also known as collision-induced dissociation in the art (128).
It occurs when a projectile, such as a molecule, a molecular ion or a cluster for example, strikes an object, such as a gas molecule or a surface for example, with a force sufficient to strip one or more of its electrons. This highly charged projectile, which is excited to a vibrational continuum, then undergoes dissociation. In the case of 4-10 keV
N2+ ions for instance interact with a He target to lose an electron. The N22+ tnus formed dissociates into 2N+.
(00122] fn the case of the present plasma system, ion energies lie below 15 eV [112,129), and little or no subsurface penetration is expected. However, for N2+
beams in the 3-6 keV range, substantial penetration of the fragments of the Coulomb explosion is expected, and whether any N will remain at the surface to react is questionable.

[00123] In order to answer such a question, ion penetration simulations were performed using "SRJM" (for Stopping and Range of Ions in Matter, described in the art [130J). Since N2+ separates into N+ that penetrates, the simulation used N+ at 1.5 keV, resulting in each of the fragments retaining half the energy of the NZ+ ion, with none lost to the substrate. This is believed to be a reasonable approximation for a massive substrate in the art [128, 131], These ions then simulated penetrating polymeric substrates of densities similar to that of Cyclotene (about 0.95 g/cm3).
[00124] It is found that the penetration maximum (at around 5-6 nn) was invariably beyond the XPS probe depth, which is about 4 nm for N. However, up to 4 nm in depth, the simulation profile was surprisingly similar to the profile generated from Figure 14. Such a simulation profile (not shown) indicates little, if any, N
at the sample surface. Since the atomic displacement threshold is only slightly higher in energy than the threshold for ion-induced defects [112-114], most of the penetrating N+
may have reacted. For those penetrating N+ that have not reacted and that therefore lie interstitially, the possibility exists for orbital overlap with the Cyclotene structure. For example, N'', whose configuration is [He] 2s2 2PX 2py, may have interacted with one side of a 1 Egg aromatic HOMO of a benzenoid ring, in a fashion similar to what was found for the reaction of Cu with untreated Cyclotene by the present applicants [105].
This would serve to reduce the formal charge on the N, reducing its binding energy.
Indeed, this may be the reason for the higher than expected N1 : N2 ratio.
[00125] Considering that only the N2 component, at about 1l3 of the total N
concentration (see Figure 15d) represents N-containing groups capable of interaction with Cu [91 J, and that only those few at the surface can react, it can be anticipated a little increase in Cu adhesion occurs, as compared to the untreated surface. This is indeed found in the present case, by measuring the critical load of Cu evaporated onto Cyclotene, both treated and untreated [92J. Both untreated and Ar+ ion-treated Cyclotene have L~ values of about 2 N, while N2 plasma-treated Cyclotene has an L~
values of about 17 N. In the present case, the estimated L~ values is about 6 N, as determined from a relationship between Cu coalescence and L~ demonstrated by the present applicants.
[00126] For the purpose of a direct comparison with samples treated by N2 plasma, which were exposed to atmosphere on transfer from the plasma chamber to the XPS, these samples were also intentionally exposed to atmosphere for 1 min.
Such exposure did not change the coalescence coefficient within experimental error.
That is, the adhesion remains the same. This indicates that interfacial interactions, rather than beam-induced surface roughness effects, dominate the adhesion of Cu film on Cyclotene. This is identical to what was previously found for the Ar+ ion treatment of Cyclotene [107].
[00127] Summarizing this second aspect of the present invention, it can be said that XPS analysis of Cyclotene treated with 3-6 keV N2+, when compared with N2 plasma treatment, shows that the reaction path followed depends on the energy of the treatment. The higher energy beam treatment resulted in penetration of the N-containing species into the subsurface, with few N-containing groups remaining at the outer surface. This results in a lower adhesion of Cu when compared to the plasma treatment, where N-containing groups are limited to the outer surface.
[00128] From the foregoing, it will be apparent to people in the art that the present invention provides a method to deposit strongly adhering metallic nanoclusters, less than 10 nm in size, whose diameter, lateral surface positioning and positional stability (adhesion) are closely controlled, onto substrates of potential interest to industry, in the construction of nanostructures.
[00129] It will be further apparent that the present invention provides a method to characterize the sizes, the surface distributions, the shapes and the stability of such nanoclusters through the use of appropriately modified angle-resolved XPS, supported by Monte-Carlo simulations.

[00130] It will be also apparent that the present invention provides a method to vary the size and distribution of nanoclusters on substrate surfaces, such as HOPG and low permittivity polymers, through ion beam irradiation, and also by laser irradiation.
[00131] It will be also apparent that the present invention provides a method enabling the use of laser, ion beam and plasma surface treatments to chemically modify the substrate surface so as to react with the nanoclusters, binding them strongly.
[00132] It will be also apparent that the present invention provides a method enabling to determine the size, the distribution, the shape and the positional stability of such nanoclusters. Such a determination of some of these properties is often difficult by TEM, when the cluster size is at the spatial resolution limit of the instrument. Similarly, AFM/STM tip effects make such determinations difficult.
[00133] The present invention provides for a method to determine, nondestructively and simply, nanocluster dimensions and surface densities by using XPS intensity ratios at a fixed electron emission angle.
[00134] The present invention also provides for a method to follow coalescence kinetics, a measure of the stability of nanoclusters as well as to identify substrate defect sites at which cluster nucleation and growths occurs, and to quantify their relationship.
[00135] The present invention also provides for a method for using angle-resolved XPS, which is an accepted technique for the non-destructive, in situ characterization of the thickness of uniform films deposited onto surfaces, to obtain XPS
data on clusters that, as expected, do not fit the standard model of a uniform film with an abrupt interface at the substrate, because, when applied to (discontinuous) nanoclusters, the use of angle-resolved XPS requires modification. This was achieved by introducing modifications of the model that permit the determination of the sizes of nanoclusters, as well as their shapes and spatial configurations on the substrate surfiace.
[00136] The present invention also provides for a method adapting Monte-Carlo simulations to the study of nanoaggregates, based on the realization that preliminary Monte-Carlo simulations of the angle-resolved XPS data, which take into account nanocluster size, shape, and both size and number density distributions, qualitatively account for the experimental results. Therefore, the present invention provided for a method allowing such calculations, by demonstrating the necessity of including such variables in the' calculations. It is clear from these results that a Monte-Carlo simulation section should be included with the experimental section, each furnishing feedback for the other.
[00137] As will be appreciated by people in the art, the present invention permits the construction of nanostructures and devices to be used in nanoelectronics, such as single electron transistors and high-density data storage for example.
[00138] Furthermore, people in the art will foresee that the present method may allow to precisely control metal nanocluster size (less than 10 nm) and lateral surface positioning on substrates, through the use of laser, ion beam and plasma treatments. It may also allow to fix them, at specified locations, by increasing their adhesion through the chemical modification of selected areas on the substrate surface, using laser machining.
[00139) Moreover, obviously the present invention paves the way to modified and adapted the angle-resolved XPS technique, by allowing development of Monte-Carlo simulations of the behavior of nanoclusters on substrates; as functions of size, shape, surface distribution and stability. This in turn has the potential to permit such simulations to be used for the characterization of nanoclusters determined by angle-resolved XPS intensities, as well as feedback for our experimental procedures.

[00140] Finally, people in the art will appreciate that the method of the present invention, while enabling to control nanoaggregates in a simple way, in a cluster tool for example, without exposure to the atmosphere, can be easily extended to semiconducting substrates, permitting the growth of nanoclusters while avoiding the restrictions imposed by Stranski-Krastonov growth.
(00141] As people skilled in the art may also consider, the precise control of nanocluster alignment and stability achieved by the method of the present invention may be used for fabricating superconductive devices, of various shape and form, to conduct power.
(00142] Although the present invention has been described hereinabove by way of preferred embodiments thereof, it can be modified, without departing from the spirit and nature of the subject invention as defined in the appended claims.

REFERENCES
1. C.T. Campbefl, Surf. Sci. Rept. 27(1997) 1.
2. C.E. Pofrer, B.K. Hance, J.M. White, J. Phys. Chem. 97 (1993} 5965.
3. C.R. Henry, Surf. Sci. Rept. 31 (1998)231.
4. M.J.J. Jak, C. Konstapel, A. van Kreuningen, J. Verhoeven, J.W.M. Frenken, Surf. Sci. 457 (2000) 295.
5. 5. Gan, Y. Liang, D.R. Baer, A.W. Grant; Surf. Sci. 475 (2001 ) 159.
6. C.E.J. MitcheII,A. Howard, M. Camey, R.G. Egdell, Surf. Sci. 490 (2001 ) 196.
7. J. Carrey, J.-L. Maurie, F. Petroif, A. Vaures, Phys. Rev. Lett. 86 (2001 ) 4600.
8. J.E. Reddic, J. Zhou, D.A. Chen, Surf Scf. Lett. 494 (2001 ) L767.
9. A.R. Arthur, A.Y. Cho, Surf. Sci. 36 (1973} 641.
10. J.L. Sacedon, C.S. Martin, Thin Sohid Films 10 (1972) 99.
11. J.S. Maa, T.E. Hutchinson, J. Vac. Sci. Technol. 14(1977) 116.
12. I. Jirka, Surf. Sci. 232 (1990) 307.
13. W.F. Egefhoff, Jr., G.G Tibbetts, Phys. Rev. B 19 (1979) 5028.
14. V. Vijayakrishnan, C.N.R. Rao, Surf. Sci. 255 (1991 ) L516.
15. A. Humbert, M. Dayez, S: Granjeaud, P. Ricci, C.R. Henry, J. Vac. Sci.
Technol. B 9 (1991 ) 804.
16. P. Marcus, C. Hinnen, Surf Scf. 392 (1997) 134.
17. Y. Kawashima, G. Katagiri, Phys: Rev. B 59 {1999) 62.
18. G.K. Wertheim, S.B. Di Cenzo, Phys. Rev. B 37 (1988) 84.
19. Q. Ma, R.A. Rosenberg, Appl. Surf Sci. 140 {1999) 83.
20. H. Brune, Surt Sci. Rept. 31 (1998)121.
21. C.M. Whelan, C.J. Bames,Apph. Surf Sci. 119 (1997) 288.

22. H.-A. Durand, K. Sekine, K. Etoh, K. Ito, I. Kataoka, Thin Sohid Films 336 (1998) 42.
23. C. Binns, S.H. Baker, C. Demangeat, J.C. Pariebas, Surf. Sci. Rept. 34 (1999) 105.
24. C. Brechignac, Ph. Cahuzac, F. Carhier, M. de Frutos, A. Masson, C. Mory, C. Colhiex, B. Yoon; Phys. Rev. B 57 (1998) 82084.

25. B. Yoon, V.M. Akuhin, Ph. Cahuzac, F. Carlier, M. de Frutos, A. Masson, C.
Mory, C. Colliex, C. Brechignac, Surf. Sci. 443 (1999) 76.

26. C. Binns, Surf. Sci. Rept. 44 (2001 ) 1.

27. L. Bardotti, P. Jensen, A. Hoareau, M. Treilleux, B. Cabaud, Phys. Rev.
Lett.
74 (1995) 4694.

28. P. Jensen, Rev. Mod. Phys. 71 (1999)1695.

29. C. Chapon,C.R.Henry, Surf. Sci. 106 (1981) 152.

30. M. Fliiehi, P.A. Buffat, J.-P. Bore!, Surf. Sci. 202 (1988) 343.

31. L. Bardotti, P. Jensen,A. Hoareau, M. Treilleux, B. Cabaud, Phys. Rev.
Lett.
74(1995)4694.

32. W.D. Luedtke, U. Landman, Phys. Rev. Lett. 82 (1999) 3835.

33. E.C.W. Leung, P. Markiewicz, M.C. Goh, J. Vac. Sci. Technol. B 15 (1997) 181.

34. L. Lai, E.A. Irene, J. Vac. Sci. Technol. B 17 (1999) 33.

35. S. Kitching, P.M. Williams, C.J. Roberts, M.C. Davies, S.J.B. Tendler, J.
Vac.
Sci. Technol. B 17 (1999) 273.

36. V.J. Garcia, L. Martinez, J.M. Briceno-Valero, C.H. Schilling, Probe Microsc. 1 (1997) 107.

37. D.-Q. Yang, Y.-Q. Xiong, Y. Guo, D.-A. Da, W.-G. Lu, J. Mater. Sci. 36 (2001 ) 263.

38. D.-Q. Yang, M. Meunier, E. Sacher, Appl. Surf Sci. 173(2001 ) 134.

39. D.-Q. Yang, E. Sacher, J. Appl. Phys. 90 (2001 ) 4657.

40. C.H. Herry, C. Chapton, B. Mutaftschiev, Thin Solid Films 33 (1976) Li.

41. A.D. Gates, J.L. Robins, Thin Solid Films 149 (1987) 113.

42. K.R..Heim, S.T. Coyfe, G.G. Hembree, J.A. Venabies, M.R. Scheinfein, J.
Appl. Phys. 80 (1996) 1161.

43. U. Diebohd, J.-M. Pan, T.E. Madey, Surf. Sci. 33 1-333 (1995) 845.

44. M.JaJ. Jak, C. Konstapel, A. van Kreuningen, J. Verhoeven, J.W.M. Frenken, Surf Sci. 474 (2001 ) 28.

45. G.F. Feng, R. Zallen, J.M. Epp, J.G.. Dillard, Phys. Rev. B 43 (1991 ) 8678.

46. K.K. Tiong, P.M. Amirtharaj, F.H. Pollak, D.E. Aspnes, Appl. Phys. Lett.

(1984) 122.

47. K. Nakamura, M. Kitajima, Appl. Phys. Lett. 59 (1991 ) 1550.

48. K. Nakamura, M. Kitajima, Surf. Sci.; 283 (1993) 255.

49. H.J. Steffen, C.D. Roux, D. Marton, J.W. Rabalais, Phys.Rev. B44 (1991 ) 3981.

50. H.J. Steffen, D. Marton, J.W. Rabalais, Phys. Rev. Lett. 68 (1992) 1726;
Nucl.
Instrum. Methods Phys. Res. B 67 (1992) 308; D. Marton, K.J. Boyd, T. Lytle, J.W. Rabalais, Phys. Rev. B 48 (1993) 6757.

51. L. Porte, M. Phaner, C.H. de Vihleneuve, N. Mobncoffre, J. Tousett, Nuci.
Instr. Methods Phys. Res.B44 (1989) 116.

52. L. Porte, C.H. de Villeneuve, M. Phaner, J. Vac. Sci. Technol. B 9 (1991) 1064.

53. R. Coratger, A. Claverie, F. Ajustron, J. Beauviilain, Surf Sci. 227 {1990) 74;
R. Coratger, A. Claverie, A. Chahboun, V. Landry, F. Ajustron, J.Beauvillain, Surf Sci., 262 (1992) 208.

54. F. Thibaudafl, J. Cousty, E. Balanzat, S. Bouffard, Phys. Rev. Lett.
67(1991 )1582.

55. K.P. Reimann; W. Boise, U. Geyer, K.P. Leib, Europhys. Lett. 67 (1995) 463.

56. T. Li, B.V. King, R.J. MacDonald, G.R. Cotterill, D.J. OConnor, Q. Yang, Surf.
Sci. 312 (1994) 399; Q. Yang, T. Li, B.V. King, R:J. MacDonald, Phys. Rev.
B53 (1996) 3032.

57. J.8. Hahn, H. Kang, S. Song, LC. Jeon, Phys. Rev. B 53 (1996) 81725; J.R.
Habn, H. Kang, Surf Sci. 357-358 (1996) 165.

58. M. Kappel, M. Steidi, J.Biener, J.Kuppers, Surf Sci., 387 (1997) L1062.

59. K. Mochiji, S. Yamamoto, H. Shimizu, S. Ohtani, T. Seguchi, N. Kobayashi, J.
Appl. Phys. 82 (1997) 6037.

60. H. Ogiso, W. Mizutani, S. Nakano, H. Tokumoto, K. Yamanaka, Appl. Phys. A
66 (1998) S 1155.

61. B. An, S. Fukuyama, K. Yokogawa, M. Yoshimura, Jpn. J. Appl. Phys. 39 (2000) 3732.

62. S. 1-labenicht, Phys. Rev. B 63 (2001) 125419.

63: F. Forst, D. Hirsch,A. Schindler,Appl. Surf Sci. 179(2001)8.

64. J.8. Hahn, H. Kang, Phys. Rev. B 60 (1999) 6007.

65. K. Shimada, T. Ishimaru, T. Yamawaki, M. Uchigasaki, K. Tomiki, T.
Matsukawa, 1. Ohdomari, J. Vac. Sci. Technol. B 19 (2001 ) 1989.

66. F. Forst,A. Schindier, F. Bigl, Phys. Rev. Lett. 85(2000)4116.

67. A. Hoffman, G.L. Nyberg, J. Liesegang, Phys. Rev. B 45 (1992) 5679.

68. J.C. Laskovich, S. Scaglione,Appl. Surf Sci. 78(1994)17.

69. S.L. Jackson, R.G. Nuzzo, Appl. Surf Sci. 90 (1995) 195.

70. D.-Q. Yang, E. Sacher, Surf Sci., in press.

71. G M Francis, L Kuipers, J.R.R.A. Cleaver, R E Palmer, J. App!. Phys. 79 (1996) 2924.

72. A. Humbert, M. Dayez, S. Granjeaud, P. Ricci, C. Chapon, C.R. Henry, J.
Vac. Sci. Technol. B 9(1991 ) 804.

73. P. Deltour, J.-L. Barrat, P. Jensen, Phys. Rev. Lett. 78 (1997) 4597; L.J.
Lewis, P. Jensen, N. Combe, J.-L. Barrat, Phys. Rev. B 61(2000)16084; J.-M.
Wen, S.-L. Chang, J. W. Burnett, J. W. Evans, P. A. Thiel, Phys. Rev. Lett. 73 ( 1994) 2591.

74. E. Ganz, K. Sattler, J. Clarke, Suit Sci. 219 (1989) 33.

75. W.F. Egelhoff, Jr., Surf. Sci. Rept. 6 (186) 253.

76. T. Meguro, A. Hida, M. Suzuki, Y. Koguchi, H. Takai, K. Yamamoto, K.
Maeda, Y. Aoyagi, J. Vac. Sci. Technol. B 19 (2001 ) 2745.

77. M. von Smoiuchowski, Phys. Z. 17 (1916) 585.

78. G. Rosenfeid, K. Morgenstern, M. Esser, G. Comsa, Appf. Phys. A 69 (1999) 489.

79. D. Kandel, Phys. Rev. Lett. 79 (1997) 4238.

80. J. Eggers, Phys. Rev. Lett. 80 (1998) 2634.

81. D.-Q. Yang, E. Sacher, in Metallization of Pplymer 2, E. Sacher, ed., Kluwer, New York, in press.

82. M. Gautier-Soyer, S. Gota, L. Douillard, J.P. Duraud, P. Le Fevre, Phys.
Rev.
B 54 (1996) 10366.

83. S. Xu, B.L. Evans, D.I. Flynn, Cao En, Thin Solid Films 238 (1994) 54.

84. R.L.W. Smithson, D.J. McCiure, D.F. Evans, Thin Sohid Films 307 (1997) 110.

85. C. Winkler, A. Carew, R. Rayai, J. Ledieu, R. McGarth, Surf Rev. Lett. 8 (2001 ) 693.

86. L. Zhang, F. Cosandey, R. Persaud, T.E. Madey, Surt Sci. 439 (1999) 73.

87. J.A. Blackman, A. Wihding, Europhys. Lett. 16(1991)115.

88. J.A. Venables, Surf Sci. 299f300 (1994) 798.

89. H. Schmeisser, Thin Sohid Films 22 (1974) 83.

90. D.-Q. Yang, E. Sacher,Appl. Surf Sci. 173 (2001 ) 30.

91. D.-Q. Yang, L. Martinu, E. Sacher, A. Sadough-Vanini, Appl. Surf. Sci. 177 (2001 ) 85.

92. A. Sadough-Vanini, D.-Q. Yang, L. Martinu, E. Sacher, J. Adhesion 77 (2001 ) 309.

93. D.-Q Yang, E. Sacher, in preparation.

94. C. De W. van Siclen, Phys. Rev. Lett. 75 (1995) 1574.

95. J:W. Wen, S.-L. Chang, J.W. Burnett, J.W. Evans, P.A. Thiel, Phys. Rev.
Lett.
73 (1994)2591.

96. H.C. Kang, P.A. Thief, J.W. Evans, J. Chem. Phys. 93 (1990) 9018.

97. A.F. Voter, Phys. Rev. B 34 (1986) 6819.

98. A.F. Voter, SPIE Proc. 821 (1987) 214.

99. M. Zinke-Aifmang, L.C Feldman, M.H. Grabow, Surt. Sci. Rept. 16 (1992) 377.

100. E.E. Grude, J. Appl. Phys. 38(1967)243.

101. P. Meakin, in Phase Transitions and Critical Phenomena, Eds. C. Domb and J.L. Lebowits, Academic Press, New York, 1988, p. 335.

102. E. Ruckenstein, B. Pulvermacher, J. Catai. 29(1973)224.

103. P.B. Balbuena, P.A. Derosa, J.M, Seminario, J. Phys. Chem. B 103 {1999) 2830.

104. J. Carrey, J.-L. Maurice, F. Petroif, A.Vaures, Surf Sci., in press 105. D.-Q. Yang, S. Poulin, E. Sacher, C. Hyett, Appl. Surf Sci., 165 (2000) 116.

106. D-Q. Yang, E. Sacher, E.M. Griswold, G. Smith, Appl. Surf. Sci., 180 (2001 }
200.

107. D.-Q. Yang, E. Sacher, Appl. Surt. Sci., 173 (2001 ) 30.

108. K.W. Paik, H.S. Cole, R.J. Saia, J.J. Chera, J. Adh. Sci. Technol., 5 (1993) 403.

109. N. Schiihler, P. Oeihafen, J. Vac. Sci. Technol., AIS (1997) 2529.

110. F.S. Shieu, M.H. Shiao, J. Adh. Sci. Technol., 12(1998)19.

111. A. Hallil, O. Zabeida, M.R. Wertheimer, L. Martinu, J. Vac. Sci.
Technol., A18 (2000) 882.

112. M. Collaud, S. Nowak, O.M. Kuttel, P. Groning, L. Schlapbach, Appl. Surf.
Sci., 72 (1993) 113. H.J. Steffen, D. Marton, J.W. Rabelais, Nucl. Instrum. Methods, B67 {1992) 308.

114. D. Marton, K.J. Boyd, T. Lytle, J.W. Rabelais, Phys. Rev., B48 (1993) 6757.

115. L. Zhang, K. Takata, T. Yasui, H. Tahara, T. Yoshikawa, Mater. Chem.
Phys., 54 (1998) 98.

116. J-S Cho, W-K Choi, S-K Koh, K-H Yoon, J. Vac. Sci. Technol., B16 (1998) 1110.

117. G. Mesyats, Y. Klyachkin, N. Gavrilov, A. Kondyurin, Vacuum, 52(1999)285.

118. A. Caballero, D. Leinen, J.P. Espinos, A. Eernandez, A.R. Gonzalez-Elipe, Surt. Interface Anaf., 21 (1994) 418.

119. M. Guemmaz, A. Mosser, J.-J. Grob, R. Stuck, Surf. Coat. Technol., 100-IOl (1997) 353.

120. D.L. Williamson, J.A. Davis, P.J. Wilbur, Surf. Coat. Technol., 103-104 (1998) 178.

121: S. Poulin, D.-Q. Yang, E. Sacker, C. Hyett, T.H. Ellis, Appi. Surf. Sci., (2000) 15.

122. D.-Q. Yang, E. Sacker, J. Appl. Phys., 90 (2001 ) 4768.

123. D.-Q. Yang, E. Sacker, in Proceedings of the 25t" Annual Meeting of the Adhesion Society, Adhesion Society, Blacksburg, VA, 2002, in press.

124. J.A. Taylor, G.M. Lancaster, J.W. Rabelais, J. Amer. Chem. Soc., 100 (1978) 4441.

125. Z.W. Deng, R. Souda, Surf. Sci., 288 (2001 ) 393.

126. D. Bergmann, J. Hinze, in Electronegativity, edited by K.D. Sen, C.K.
Jrargensen, SpringerVerlag, New York, 1987, p. 145.

127. C. Hansch, A. Leo, R.W. Taft, Chem. Rev., 91(1991)165.

128. J. Los, T.R. Govers, in Collision Spectroscopy, edited by R. G. Cooks, Plenum, New York,1978, Chapter 6.

129. 0. Zabeida, L. Martinu, J. Appi. Phys., 85 (1999) 6366.

130. http://www.srim.ora.

131. D.S. Gemmeil, Chem. Rev., 80(1980)301.

Claims (10)

1. A method to deposit strongly adhering metallic nanoclusters as described hereinabove.

2. A method to characterize the sizes, the surface distributions, the shapes and the stability of nanoclusters as described hereinabove.

3. A method to vary the size and distribution of nanoclusters on substrate surfaces, such as HOPG and low permittivity polymers, as described hereinabove.

4. A method enabling the use of laser, ion beam and plasma surface treatments to chemically modify the substrate surface so as to react with nanoclusters deposited thereon, as described hereinabove.

5. A method enabling to nondestructively determine nanocluster dimensions and surface densities by using XPS intensity as described hereinabove.

6. A method to follow coalescence kinetics, as described hereinabove.

7. A method to identify substrate defect sites at which cluster nucleation and growths occurs, as described hereinabove.

8. A method for using angle-resolved XPS to obtain XPS data on clusters, as described hereinabove.

9. A method for adapting Monte-Carlo simulations to the study of nanoaggregates, as described hereinabove.

10. A method of fabrication of the construction of nanostructures, as described hereinabove.