Electrochemical Science and Technology TECHNICAL PAPERS
2771
Zhang Hong, Chen Zhenhai and XiaXi
2774
Bruno Scrosati
Polymer‐based lithium batteries have attracted considerable interest in recent years. Presently, both ionically and electronically conducting polymers are available, and thus fully polymeric batteries are possible. The characteristics and properties of polymer electrolytes and polymer electrodes are examined here, and an evaluation of their applicability in electrochemical devices of practical interest is discussed.
2782
Hiroki Tamura, Tatsuya Oda, Masaichi Nagayama and Ryusaburo Furuichi
2787
R. W. Powers and B. R. Karas
Operation of the sulfur electrode in sodium‐sulfur cells is limited by the transport rates of sodium ions as shown by concentration polarization, circulating currents, and the formation of isolated pockets of sulfur and sodium polysulfides. Previous attempts to increase these rates by improved wicking of sodium polysulfides through alumina addition to the carbon mat led to increases in cell resistance. Consequently, the authors' efforts were concentrated mainly on methods to introduce alumina uniformly throughout the carbon mat and on understanding the principal factors affecting mat resistance. Alumina can be introduced uniformly into carbon mat from sulfate solutions. Such additions improve sodium polysulfide wicking without affecting the mat resistance. The effects of packing density, fiber length, the presence of sulfur, and alumina loading on mat resistance were studied along with factors affecting measurement reliability. The current collector‐carbon fiber contact resistance was usually found to be substantially larger than that of the intrafiber component or the fiber‐fiber contacts. In confirmation of these findings, sodium‐sulfur cells containing alumina‐treated carbon mat displayed less concentration polarization and exhibited cell resistances comparable to control cells containing untreated carbon mat.
2794
Jean Boyd Bailey
2798
T. I. Evans and R. E. White
A one‐dimensional mathematical model for the lithium/thionyl chloride primary cell is used in conjunction with a parameter estimation technique, in order to estimate the electrode kinetic parameters of this electrochemical system. The electrode kinetic parameters include the anodic transfer coefficient and exchange current density of the lithium oxidation,
and
; the cathodic transfer coefficient and the effective exchange current density of the thionyl chloride reduction,
and
, and a morphology parameter, ξ. The parameter estimation is performed on simulated data first in order to gain confidence in the method. Data reported in the literature for a high‐rate discharge of an experimental lithium/thionyl chloride cell is used for an analysis.
2806
Kiumars Ghowsi and Robert J. Gale
2811
C. C. Streinz, J. S. Steckenrider, J. W. Wagner and P. J. Moran
Laser holographic interferometry was utilized to monitor the volume change in lithium/iodine poly(2‐vinylpyridine) (P2VP) batteries for the entire discharge at high discharge rates. Two cathode chemistries were used, one containing P2VP with a molecular weight average of 22,000 and the other containing P2VP with a molecular weight average of 54,000. One of each cell type was discharged across constant 1 kΩ (approximately 300 μA/cm2) and 5 Ω (approximately 120 μA/cm2) loads. In all cases, the cells initially expanded nonuniformly. This was followed by more uniform expansion and then a transition to uniform contraction which continued until end‐of‐life. The batteries manufactured with the cathodes containing P2VP with a molecular weight of 22,000 exhibited net contraction relative to initial dimensions and experienced a smaller initial expansion and an earlier transition to contraction, as compared to the cells manufactured with cathodes containing P2VP with a molecular weight of 54,000, which exhibited a net expansion with respect to initial dimensions. The differences observed between the two cell types are explained by a dependence of the mechanical properties of the cathode on the molecular weight of the P2VP. Both cathode types become more pliable with discharge, enabling a mechanical restructuring of material within the cathode to occur. This results in battery contraction. No distinct electro‐chemical events were noted concurrent with the expansion, transition, or contraction stages. Normal high rate behavior and end‐of‐life behavior were observed in all cases.
2816
Z. Nagy, P. J. Hernes, M. Minkoff, G. K. Leaf and R. H. Land
The boundary conditions usually used for solving the differential equations in the theory of electrochemical transient measurements are not applicable to measurements with corroding metals. In an electrode kinetic investigation the measurement starts from equilibrium, where no net current is flowing and no net chemical change is taking place before the application of the measuring signal. The situation is completely different for a corroding electrode because a corroding metal is not in equilibrium. Under these conditions, while there is no net current flowing through the system, there is a net chemical change taking place: the corrosion of the metal. The relaxation curves calculated with the modified boundary conditions can be considerably different from those calculated the conventional way; therefore, the corrosion rates calculated from polarization data corrected for mass‐transport effects using dc transient techniques may be in error. A series of model calculations was carried out to determine this error. It was concluded that classical dc transient techniques can be used to correct polarization data taken with corroding metal electrodes for mass‐transport effects, as long as the system is not completely under mass‐transport control and the negative polarization from the corrosion potential does not exceed 30 mV. Under these conditions the error of the corrosion rate determination will generally be negligible, and will not exceed 15% under any practical condition. However, the dc transient techniques can introduce a serious error when the polarization is more negative than 30 mV. The dc transient techniques cannot be used for fully mass transport‐controlled systems either, but, under this condition, there is no need for the separation of kinetic and mass‐transport effects, and the classical polarization equations are satisfactory.
2820
Alan K. Hauser and John Newman
A singular perturbation analysis has been developed for the dissolution of a copper rotating disk in
that is based on an assumed
reaction mechanism. The underlying objective of the paper is to explore the effect that finite rates of a homogeneous chemical reaction can have on the preceding heterogeneous electrode reaction. Calculated complex‐plane and Bode plots, for a wide range of values of the homogeneous‐reaction rate constant
, demonstrate the significant effect that the chloride complexing reaction can have on the frequency dependence of the faradaic impedance. An additional semicircle results in the complex plane due to the chemical reaction, and provides a basis for determining the homogeneous‐reaction rate constant from frequency‐response data. Other reaction mechanisms that possibly describe the copper dissolution process, as well as ways to distinguish experimentally between the different cases, are discussed. Comparison of the available experimental copper dissolution data with the dimensionless faradaic impedance complex‐plane plot yields a value of the order of 107 s−1 for
. Also, results from the perturbation formulation of the problem are compared to the numerically‐calculated results of the Stefan‐Maxwell impedance model. For certain values of the rate constant
, numerical inaccuracies arise in the generalized multicomponent transport/kinetic algorithm, and the perturbation analysis is useful for quantifying the finite‐difference errors. Additionally, low‐to‐moderate‐frequency and high‐frequency approximations of the faradaic impedance are given, and the regions of validity of the different analytic solutions are illustrated with a reaction zone diagram.
2831
D. Remppel and H. E. Exner
2835
Son Van Nguyen and Kevin Albaugh
The physical and bonding properties of electron cyclotron resonance (ECR) plasma‐deposited silicon nitride and oxide films deposited in many commercial ECR deposition systems were analyzed using various analytical techniques. ECR films deposited at room temperature showed better qualities than those of plasma‐deposited films at higher temperature (350°C). The oxide film qualities are almost comparable to thermal CVD films. However, the nitride film qualities are slightly poorer. For both films, the step coverage is very nonconformal, and sidewall films are much more porous under most deposition conditions. This is due to the anisotropic ion bombardment properties of ECR processing. RF biasing can reduce this effect, fill shallow trenches and planarize the surface structure with appropriate processing conditions.
2841
D. J. De Smet and J. L. Ord
The Simplex algorithm provides a numerically efficient method for the analysis of ellipsometric data on film growth in electrochemical systems. The algorithm finds the complex index of refraction which provides the best fit to a set of data consisting of either Δ and ψ or
and
measurements at a series of thicknesses during the growth of a homogeneous film. The way in which the algorithm operates is demonstrated using data on the electrodeposition of polyanaline and the anodization of titanium. Results obtained with and without weighting factors are compared, and a procedure for estimating accuracy is described along with ways of using electrochemical data to aid in the evaluation of goodness of fit.
2845
Georges G. Perrault
The gallium arsenide‐gallium‐arsenic‐water electrochemical equilibrium diagram is calculated and compared with some results obtained in electrodeposition experiments of gallium‐arsenide.
2850
C. J. Chen and C. C. Wan
This investigation is concerned with the variation of current efficiency under pulse current conditions. The parameters involve the pulse current density, the on‐time, and the off‐time. When the on‐time is shorter than a critical pulse time, the efficiency drops noticeably. A model based on successive charge transfer steps and a side reaction is proposed to explain the experimental observations.
2856
The following article is Free article
S. Chandra and Neeraj Khare
2858
Nancy J. Moreland and James M. Fenton
The effect of a two‐phase electrolyte on diffusion of reactants to a horizontal planar electrode in quiet solution was investigated. Three systems were studied; ferricyanide reduction from basic aqueous solution, ferricyanide reduction from basic aqueous solution with an inert drop of methylene chloride on the electrode, and iodine reduction from aqueous solution with a drop of methylene chloride containing a high concentration of iodine on the electrode. Experiments were conducted with controlled variation of reactant concentration, drop size, number of drops, and electrode area. It was found that with the addition of an inert drop, the reaction rate decreased proportionally to the drop contact area, while the reaction rate increased substantially with the addition of a drop containing a high concentration of iodine.
2862
M. L. Orfield and D. A. Shores
2867
Jong‐Hee Park and Robert N. Blumenthal
By means of a gas‐tight electrochemical cell, oxygen permeation measurements have been performed on the oxygen ion conductor 8 mole percent (m/o) yttria‐stabilized zirconia, as a function of
(800°–1050°C) and
(0.21–10−17 atm). Ionic conductivity was measured by the conventional four‐probe method. The following empirical equations were derived for the ion, electron, and hole conductivities, in ohm−1 cm−1



The electronic diffusivity,
, and mobility, μ, for holes and electrons were determined by nonsteady‐state oxygen permeation measurements. A gas‐switching method was employed. Results indicate that the holes and electrons move by a thermally activated hopping‐type mechanism. The equations for
and μ are as follows For holes


For electrons


The electronic semiconducting properties obey Boltzmann‐type statistics at temperatures investigated. An analysis was carried out by combining results from electronic transport and solid‐state coulometric titration measurements under the assumption of trapping of electrons and holes on the appropriate sublattices. The results indicate that the transport of electrons and holes is due to a thermal excitation process and that electrons and holes are trapped on the appropriate sublattices.
2876
Scott W. Capeci, Peter N. Pintauro and Douglas N. Bennion
The idealized “cluster‐network” structure for Nafion® cation exchange membranes has been used to generate molecular‐level models for membrane salt solubility and anion transport. Equilibrium salt uptake by Nafion was found to be dependent on the size and wall charge of the membrane's inverted micelle clusters. The anion/membrane binary interaction transport parameter was dependent on membrane tortuosity and the electrostatic repulsion forces in the narrow pores which interconnect the inverted micelles. The salt solubility model predicted membrane cation concentrations to within 8% for external
concentrations of 1.0–5.0M. The transport analysis accurately predicted the concentration dependence of the Cl−/membrane binary interaction parameters.
2882
D. Schlettwein, D. Wöhrle and N. I. Jaeger
The electrochemical properties of Cu‐tetra(pyrazino [2,3])‐[b,g,l,q]‐5,10,15,20‐tetraazaporphyrine, 4, and Cu‐tetra(5,6‐dicyanopyrazino [2,3]) [b,g,l,q]‐5,10,15,20‐tetraazaporphyrine, 5, as thin films in contact with an aqueous acid electrolyte were investigated by cyclic voltammetry. The electrodes show stable electrochromic reduction and reoxidation. A strong shift of the redox potentials into the anodic direction could be accomplished in comparison with the analogue compounds Cu‐phthalocyanine, 1, and Cu‐2,3,9,10,16,17,23,24‐octacyanophthalocyanine, 2. There is strong evidence for an irreversible structural change of 4 during the first reduction cycle.
2887
J. F. Wolf, C. E. Forbes, S. Gould and L. W. Shacklette
The redox chemistry of a model compound of polyaniline, N,N′‐diphenyl‐p‐phenylenediamine, has been studied in acidic media. The redox processes of this “dimer” have been shown to be highly reversible and pH‐dependent. The dimer undergoes a two‐step oxidation from amine to radical cation to imine forms. An equilibrium constant was determined for the conversion of a 50:50 mixture of amine and imine forms to the radical cation as a function of proton concentration. A 50% conversion to the radical form occurs at a pH of 2.2. This pH value corresponds very well with the midpoint of the metal‐to‐insulator transition in polyaniline as a function of pH. ESR measurements were used to study the radical cation in solution. Spin density was found to be highest on the nitrogens, but considerable delocalization into central and end phenyls was found.
2892
Xia‐Qin Wu, Yasuhiro Shimizu and Makoto Egashira
2896
Alan K. Hauser and John Newman
Frequency‐response results, as calculated by the Stefan‐Maxwell macroscopic impedance model, are presented for a copper rotating disk in chloride solutions. The working algorithm uses concentrated‐solution theory and accounts for multicomponent diffusion, migration, and homogeneous and hetereogeneous reactions, as well as a finite Schmidt number and interfacial velocity. The validity of the general program was checked by comparing the concentrated‐solution model in the limit of dilute solutions, excess of supporting electrolyte, and infinite Schmidt number to known analytic solutions. Excellent agreement was obtained. Results for copper dissolution are plotted in various forms, enabling the effect of the Schmidt number (Sc) on the frequency‐response of the faradaic impedance to be studied. Specifically, plotting the low‐frequency data according to
vs.pIm
and the high‐frequency data as
vs.
allows the entire frequency spectrum to be utilized in the determination of unknown diffusion coefficients for a given electrochemical system. Additionally, the effect of electrode kinetics on the faradaic impedance as applied to the Sc‐determination procedure has been studied. Finally, new ways of plotting the dimensionless convective‐diffusion impedance reduce the Schmidt number dependence of the frequency response nearly to one curve by stretching the abscissa using
and
. These plots confirm that the slope of the low‐frequency linear region is proportional to
and that the slope of the high‐frequency linear region is proportional to
.
2902
R. Aroca, D. Battisti, G. J. Kovacs and R. O. Loutfy
Unenhanced Raman spectra of two
metallated phthalocyanine (Pc) dyes evaporated onto smooth metal surfaces of Ag, Au, and Cu, were recorded with different excitation laser lines in resonance, and off resonance with the red electronic absorption of Pc molecules. An attempt was made to relate the observed relative intensities obtained with S‐ and P‐polarized light to molecular orientation in the thin solid film. The molecular dyes deposited on the smooth metal surfaces were then coated with 10 nm of silver at 25° and 200°C. The silver island films deposited allowed the observation of SERS and SERRS with enhancement factors up to 750. The polarization properties of SERS observed for each sample are presented and discussed.
2906
S. Salama and John H. Kennedy
2909
C. A. C. Sequeira
The oxidation processes on platinum and gold electrodes in molten
and
containing
were studied in the presence of oxygen (1.0 atm) at 900°C by chronopotentiometry. Well‐defined and reproducible chronopotentiograms could not be obtained at gold electrodes in basic melts. The anodic discharge of oxide ion was shown to take place by a consecutive two‐step electrochemical reaction of a single substance. This process is identified as


Experimental data indicate that oxide ions react slowly with oxygen to form superoxide ions in accordance with the equilibrium reaction

Corrosion effects are observed when
remains in the melt for long periods of time. In addition, it is illustrated that the results are affected by the length of time the electrodes are left unpolarized in the melts, the pretreatment of the electrodes, and the geometry of the salt/noble metal electrodes.
2914
J. W. M. Jacobs, F. W. H. Kampers, J. M. G. Rikken, C. W. T. Bulle‐Lieuwma and D. C. Koningsberger
2924
R. J. Orth, R. S. Parikh and K. C. Liddell
An analytical technique, using the rotating ring‐disk electrode (RRDE), has been developed for the determination of the second‐order rate constant for the homogeneous reaction between Cu(I) and Fe(III) in 1.0 mol/dm3 hydrochloric acid at
. This was accomplished by combining experimental results with solutions to the partial differential equations pertaining to the RRDE. These equations were solved by orthogonal collocation. A value of
has been determined for this rate constant. In addition, an equilibrium constant

has been determined at
. Thus, by combining these results, a second‐order rate constant of
was calculated for the reaction between Cu(II) and Fe(II). Equilibrium constants were also determined for this system in 0.01, 0.05, 0.1, 0.25, and 0.5 mol/dm3 hydrochloric acid at
.
2931
G. Tourillon, R. Côté, D. Guay and J. P. Dodelet
2935
Alan C. West and John Newman
When ohmic resistances dominate over electrode kinetic resistances, the distribution of current density on an electrode can be highly nonuniform. In the limit of zero kinetic resistance (i.e., a primary current distribution) the current density at the edge of an electrode will be infinite if the interior angle of intersection between the electrode and insulator is obtuse. In practical cases, nonzero kinetic resistances exist, and the current density remains finite. Previous results demonstrate that, when the ohmic resistance is large compared to the kinetic resistance, the current distribution can be described by the primary distribution everywhere except the edge region. The purpose of this paper is to describe the deviations from the primary distribution and to show explicitly the manner in which the current density at the edge of the electrode approaches extreme values as the ohmic resistance becomes large. The analysis is valid for any angle of intersection and can be applied in the edge region (for a large ohmic resistance) regardless of the geometric details of the rest of the electrochemical cell. Results are given for linear and Tafel kinetics.
2940
Li Qingfeng, H. A. Hjuler, R. W. Berg and N. J. Bjerrum
The spontaneous growth of aluminum dendrites after deposition was observed and examined in sodium chloride‐aluminum chloride melts. The concentration gradient of
in the vicinity of the cathode surface resulting from electrolysis constitutes a type of concentration cell with aluminum dendrites as electrodes. The short‐circuit discharge of the cell is found to be the driving force for the growth of aluminum dendrites. Such a concentration gradient is proposed to be one of the causes for dendrite formation in the case of metal deposition.
2944
Paul A. Flowers and Gleb Mamantov
TECHNICAL NOTES
2948
J. C. Nelson, C. Kang and A. Bronson
2950
A. A. Sarabi and V. B. Singh
2953
Tetsuo Saji and Yoshiyuki Ishii
Solid‐State Science and Technology TECHNICAL PAPERS
2957
Y. Robach, J. Joseph, E. Bergignat and G. Hollinger
2960
R. J. Bogaert, T. W. F. Russell, M. T. Klein, R. E. Rocheleau and B. N. Baron
2968
Hon‐Sum Wong, Yin Hu and Marvin H. White
Silicon nitride has been widely used as a pH sensing insulator on the gate of the ion sensitive field effect transistor (ISFET) employed in CMOS integrated chemical sensors. In this paper, we report experimental results on the H+ sensitivity of heavily boron‐implanted silicon nitride
. Ion implantation provides a microelectronics technology compatible approach to modify the surface of insulators and tailor the ion sensitivity of ISFETs. This work has suggested new possibilities to study the multiple‐site site‐binding model for blocking insulators. The composition of the surface sites (nature and number of sites), can be conveniently controlled by implant ion species and dose.
2973
T. H. Daubenspeck, E. J. White and P. C. Sukanek
2979
P. V. Nagarkar and E. K. Sichel
Insulating Kapton® H‐film can be made electrically conducting by heat‐treatment in vacuum or by intense laser irradiation in air. The nearly metallic behavior of Kapton after heat‐treatment at 700°C has attracted scientific interest in the material. We report an XPS study of the polyimide after heat‐treatment and laser processing. The carbon, oxygen, and nitrogen binding energies and their evolution with increasing temperature of heat‐treatment have been examined. Comparison is made with pristine and laser‐irradiated polyimide. The carbon, oxygen, and nitrogen spectra undergo significant changes at different processing temperatures. We find that the ether oxygen and the carbonyl carbon are affected at a lower processing temperature than the nitrogen bonds. The laser and heat‐treated products are not identical and they are not pure carbon residues.
2982
Masato Ikegawa and Junichi Kobayashi
A deposition simulator using the direct simulation Monte Carlo method has been developed to investigate deposition profile characteristics on small trenches in semiconductor manufacturing processes from vacuum to atmospheric pressure conditions. This simulator is applied to several deposition processes such as sputter deposition, plasma chemical vapor deposition, and atmospheric‐pressure chemical vapor deposition. Experimental results agree with simulator results concerning molecular collision, sticking coefficient, and appropriate surface‐reaction probability.
2986
R. K. Bhan, P. K. Chaudhury, P. K. Basu and K. C. Chhabra
2989
V. Ilderem and R. Reif
We report for the first time the effects of very low pressure chemical vapor deposited (VLPCVD) titanium silicide on device electrical characteristics. The compatibility of this material for VLSI technology is examined through careful characterization of the shallow junction and gate oxide integrities, sheet resistance, and contact resistivity. It is shown that the integrity of the shallow junctions is preserved if the silicon consumption during the silicide deposition is controlled. If this consumption is not controlled, a Schottky diode behavior is observed for the source/drain junctions. It is shown that titanium silicide lowers the specific contact resistivity of metal/source‐drain and metal/polysilicon structures by an order of magnitude and a factor of 4–5, respectively. The sheet resistance of the doped polysilicon (gate conductor) is reduced by more than two orders of magnitude through using a
structure. In addition, it is shown that VLPCVD titanium silicide has no significant effect on the quality of the gate oxide. Slightly lower breakdown voltages for the
capacitors are measured, which could be due to the stress induced by siliciding the gate. The results presented in this paper indicate the suitability of VLPCVD titanium silicide films for high quality device fabrication.
2993
Jonathan D. Chapple‐Sokol, Carmen J. Giunta and Roy G. Gordon
3003
L. Y. Tsou
There are few plasma etching processes, if any, that meet all the basic requirements for pattern transfer: anisotropic profile, good photoresist integrity, high selectivity to the underlayer, and lack of polymer formation. Chlorine plasma is commonly employed to anisotropically etch polysilicon with good selectivity to gate oxide, but it erodes photoresist and results in a narrower etched polysilicon line. This paper discloses a new plasma etching gas for silicon, hydrogen bromide neat, that etches polysilicon anisotropically at low power with extremely high selectivity to both gate oxide (100/1) and photoresist (60/1). As a result of the low erosion rate of the photoresist, a ten micrometer deep trench in silicon is accomplished with a photoresist mask as opposed to an oxide mask. Preliminary result on suicide etching with hydrogen bromide plasma is also reported.
3006
J. ‐C. Lin, K. J. Schulz, K. ‐C. Hsieh and Y. A. Chang
3011
Hsiung‐Kuang Tsai, Wen‐Jer Tzeng and Si‐Chen Lee
Boron contamination of the intrinsic amorphous silicon hydrogen alloys deposited in a single chamber system using both low (29 kHz) and high (10 MHz) frequency glow discharge has been studied in detail. The contamination process was simulated by flowing the p‐type deposition gas through the system with or without turning on the plasma. Several methods were applied to reduce the contamination and their effects were investigated by measuring the change of the light conductivity and the activation energy of the contaminated i layer. A model that emphasizes the role of plasma in the contamination processes was proposed to explain the observed results. It was found that when using low frequency plasma deposition a dummy cathode method and “soft” deposition of a thin a‐Si:H can be applied to reduce the boron contamination problem.
3016
L. D. Dyer, G. J. Grant, C. M. Tipton and A. E. Stephens
3019
Peter C. Sukanek
A theoretical model for spin coating on a substrate with topography is presented. The equations assume a thin film, with uniform properties in the direction normal to the substrate. By assuming a small amplitude for the substrate variations, two sets of coupled equations are derived, one for the mean value of the thickness and its composition, and another for the fluctuations. The viscosity and surface tension of the film, together with their variation with solvent content, play an important role in determining whether planar or conformal surfaces are achieved. Surface tension tends to planarize the film, whereas surface tension gradients tend to make the film more conformal. If viscosity rises very quickly as a result of solvent evaporation, the film topography becomes “frozen‐in,” and surface forces cannot planarize.
3026
H. ‐J. Herzog, H. Jorke, E. Kasper and S. Mantl
3030
A. Yasuoka, K. Hosono, H. Morimoto, Y. Sasaki, Y. Watakabe and T. Kato
Hybrid lithography of Si and Be ion beams has been adopted to fabricate a mushroom gate of
HEMT. In this paper, registration accuracy of focused ion beam (FIB) lithography will be discussed, and its influence on an overlay accuracy between FIB‐exposed patterns, and other levels exposed by optical lithography, will be considered. The registration repeatability of 0.04 μm is obtained by averaging the registration signals even if a deformation of the signals is taken into consideration. The overlay accuracy (0.11–0.13 μm) seems to be determined by the registration accuracy and a distortion of the FIB and optical lithography systems.
3033
Frank S. Becker, Helmut Treichel and Siegfried Röhl
Arsenosilicate glass (AsSG) with dopant levels up to 15% was prepared using TEOS and triethoxiarsine (TEOA) or triethylarsenate (TEAs). The optimum deposition parameters and film properties like step coverage, uniformity, and particle densities correspond to the results obtained for undoped
. The different dissociation kinetics influence the way in which the arsenic is incorporated into the glass. Films prepared with TEAs are better diffusion sources because they do not tend to produce arsenic pile‐up at the silicon surface. By heat‐treating AsSG films with 6% As for 30–60 min at 1000°C, surface concentrations of >1019 At cm−3 with junction depths below 200 nm are obtained on planar substrates. Slightly lower As concentrations are measured in trenches, an effect correlating with the different crystal orientations. Gate oxides grown on As diffused or undoped substrates show the same electrical qualities. By incorporating the AsSG trench doping into a complete process sequence, trench cell capacitors were produced and evaluated. The AsSG doping principle was found to be a safe and production‐oriented process.
3043
F. Gaspard, A. Bsiesy, M. Ligeon, F. Muller and R. Herino
3047
Zhi‐Min Ling, Luc H. Dupas and Kristin M. De Meyer
3057
Tom Orent
3061
Ch. M. Baas, H. P. Lengkeek and M. Hendriks
The performance of three commercially available surface particle counters (SPCs)—a Tencor Surfscan 4000, an Aeronca WIS 200, and an Inspex EX20/20—was evaluated using latex spheres of 0.3–10 μm in diameter. The relationship between scattering cross section and sphere size was determined for each of the SPCs. In addition, the ratio between the numbers of particles detected by the Inspex and the Tencor was shown to decrease from 1 to 0.4 for increasing sphere size whereas the ratio for the Aeronca and the Tencor was determined to be 1.09(5). Interference effects for latex spheres of various diameters and significant variations of the apparent sphere size over the wafer surface have also been observed. Both phenomena are caused by the spherical shape and transparency of the spheres in combination with the specific designs of the SPC detection systems.
3070
H. M. Manasevit, W. B. Hewitt, A. J. Nelson and A. R. Mason
Boron‐arsenic and boron‐phosphorus films have been grown on Si, sapphire, and silicon‐on‐sapphire (SOS) by pyrolyzing Group III alkyls of boron, i.e., trimethylborane (TMB) and triethylborane (TEB), in the presence of
and
, respectively, in a
atmosphere. No evidence for reaction between the alkyls and the hydrides on mixing at room temperature was found. The films were predominantly amorphous. The film growth rate was found to depend on the concentration of alkyl boron compound and was essentially constant when TEB and
were pyrolyzed over the temperature range 550°–900°C. The films were found to contain mainly carbon impurities (the amount varying with growth temperature), some oxygen, and were highly stressed and bowed on Si substrates, with some crazing evident in thin (2 μm) B‐P and thick (5 μm) B‐As films. The carbon level was generally higher in films grown using TEB as the boron source. Films grown from
and TMB showed a higher carbon content than those grown from
and TMB. Based on their B/As and B/P ratios, films with nominal compositions
and
were grown using TMB as the boron source.
3077
J. S. Papanu, D. W. Hess, D. S. Soane and A. T. Bell
Dissolution of poly(methyl methacrylate) thin films in ketones, alcohol/ketone mixtures, and hydroxy ketones was investigated using an in situ ellipsometer. The dissolution rate in pure ketones was limited by the penetration of solvent molecules into the polymer, as indicated by minimal swelling of the film during dissolution and significant solvent size effects. Activation energies for ketone dissolution were indicative of Case II penetration. Dissolution rate decreased with polymer molecular weight up to 105 g/mol, above which the rate was constant. The thickness of the swollen surface layer was estimated from shifts in the ellipsometric data. No surface layer was detected at a polymer molecular weight of
, but above this value, the surface layer thickness increased with increasing molecular weight. The effects of polymer molecular weight on dissolution rate and thickness of the swollen layer indicated that at lower molecular weights, dissolution occurred by stress cracking. Case II transport model parameters were determined from the ketone dissolution data. Acetone/isopropanol mixtures at 20.4°C showed a transition from swelling to dissolution near 50% by volume acetone. Acetol, a hydroxy ketone, caused only swelling (34°–49°C), whereas diacetone alcohol dissolved the films (20.0°–36.9°C) at approximately one‐fourth the rate of methyl isobutyl ketone.
3083
A. J. Pidduck, D. J. Robbins, A. G. Cullis, D. B. Gasson and J. L. Glasper
In situ elastic laser light scattering (LLS) is a sensitive and noninvasive method for following changes in surface roughness during layer processing. LLS during (001) silicon MBE at 750°–850°C shows features which relate to the formation of extended defects. An intense peak is observed during initial growth when epilayer dislocation densities are high. SIMS and TEM data show that this peak arises from regions of locally high carbon concentration at the layer‐substrate interface. During pre‐epitaxial bake, radial thermal decomposition of thin passivating surface oxide and subsequent “thermal roughening” of exposed silicon both give rise to LLS features, and are considered as possible sources of inhomogeneity in the distribution of carbon existing on the substrate surface prior to deposition. During continued layer growth, a steady rise in LLS intensity is punctuated by abrupt increases due to the impact of silicon dust particles and consequent growth of stacking faults.
3088
A. J. Pidduck, D. J. Robbins, D. B. Gasson, C. Pickering and J. L. Glasper
This paper reports ex situ angle‐resolved light scattering (ARLS) measurements from silicon wafers, prepared to various stages during pre‐epitaxial bake and silicon molecular beam epitaxial (MBE) growth, in a study of the topographic sensitivity of the in situ laser light scattering (LLS) technique described in Part I . Comparison with stylus profilometer traces and transmission electron micrographs confirms that LLS is a sensitive probe of surface topography, on a lateral scale of ∼0.5–10 wavelengths depending on scattering geometry, in general accordance with smooth surface scattering theory. A vertical sensitivity to uniformly distributed roughness of subnanometer amplitude is demonstrated. The ARLS patterns are also shown to be related to differential phase contrast images (analogous to conventional Nomarski images but obtained using a laser scanning optical microscope) of the silicon surfaces.
3094
D. T. C. Huo, M. F. Yan, J. D. Wynn and D. P. Wilt
3098
H. Yamamoto, O. Oda, M. Seiwa, M. Taniguchi, H. Nakata and M. Ejima
3103
Todd A. Cleland and D. W. Hess
In situ diagnostic measurements and reactor modeling have been used to study
radio‐frequency (RF) discharges. Gas‐phase rotational temperatures and
and
concentrations were estimated from in situ Fourier transform infrared (FTIR) absorption measurements; electron densities were determined using a Langmuir probe. Measurements were conducted at 500 mtorr, flow rates from 20 to 60 sccm, and discharge powers of 5, 10, and 20W at 13.56 MHz. The rotational temperatures and electron densities were used as inputs to a reactor model. The model includes reactions among five molecular (
,
,
,
, and
) and three atomic [N, O(3P), O(1D)] species. Calculated values of
and
concentrations were in good agreement with those derived from FTIR data. In addition to providing estimates of the concentration of species not detected with FTIR, the modeling permitted identification of important reaction mechanisms for the formation and destruction of species. Model results revealed
and
to be the most abundant products of
plasma decomposition;
and O(3P) were present at 1 to 7 mole percent (m/o).
, O(1D), and N were present at much lower concentrations (109 to 1011 cm−3), but played an important role in the reactor chemistry.
3111
T. Sugii, T. Aoyama and T. Ito
3115
Z. H. Zhu, Y. H. Lo, M. C. Wu, C. L. Pan, S. Y. Wang, T. K. Gustafson and S. Wang
3123
Yicheng Lu, T. S. Kalkur and C. A. Paz de Araujo
A systematic study of Au/Zn/Au ohmic contacts to Be‐implanted p‐type
by rapid thermal alloying is presented. The processing conditions, such as zinc composition and temperature‐time cycle, are optimized. For the peak hole concentration, about
, contact resistivity as low as
is obtained. The microstructure of the contact is investigated using SEM, EDXA, AES, and XRD measurements in order to explore the correlation of the structural information with electrical characteristics represented by contact resistivity. It is found that during alloying, the
layer near the interface is decomposed due to Ga and As outdiffusion, enhanced by Au interdiffusion. Ga diffuses to the contact surface in a much larger extent than As does. The interaction of Au and Ga, which produces
phase, plays a key role in the formation of alloyed ohmic contacts to
. The rapid thermal processing technique provides better control of Au‐Ga reaction, as well as limited dissolution of the
surface, which makes the optimum ohmic contact possible. The present work has demonstrated that rapid thermal alloying offers obvious advantages over the conventional furnace alloying process, including lower contact resistivity, better interface morphology by limiting interfacial liquid phase reaction, and controlled shallow penetration depth.
3130
H. Donker, W. M. A. Smit and G. Blasse
3135
G. F. Neumark
Donor‐acceptor pair (DAP) luminescence is a powerful technique for impurity characterization in general. Recent work has used it for extensive characterization of
, as well as extended it to obtain relative DAP transition probabilities. The present paper includes a systematic review of the technique, as well as of its application to
. This application has shown two unexpected material properties, and primary emphasis here is on these. One involves relative transition probabilities, showing that alkali interstitial donors have a far lower DAP recombination rate (up to several orders of magnitude) than substitutional donors, and thus effectively act as “killers.” Means of minimizing this problem are discussed. The second emphasized aspect is that resultant donor energies and radii show a strong disagreement with effective mass theory, which would have been expected to apply (the donors are shallow).
TECHNICAL NOTE
3139
Junji Saraie, Katsuyuki Ono and Sadahiro Takeuchi
ACCELERATED BRIEF COMMUNICATIONS
3143
Daniel Mandler and Allen J. Bard
3145
R. S. Robinson