Novel Ion-Lens Configurations: A Further Step to 2-nm Ion-Beam Lithography

Composite axially symmetric immersion ion lenses are considered that consist of an electrostatic and a magnetic lens. For the first time, their performance is evaluated over the entire range of operating conditions: from the case of a zero magnetic field to the case of a zero ion energy on the target. Operating conditions are characterized in terms of = W _t/W ₀, where W ₀ is the energy of an ion at the boundary of the region in which the trajectories are parallel to the axis and W _t is that on the target. For the first time, simple analytical approximations are derived for C _c/r, C _s/r, f/r, and NI, where C _c is the chromatic-aberration coefficient, C _s is the third-order spherical-aberration coefficient, f is the focal distance, NI is the magnetomotive force of the coil, and r is the outer radius of the coil. The behavior of the four quantities is explored as a function of . The following conclusions are drawn: (i) The aberrations are maximum for a zero magnetic field. (ii) The aberration coefficients decrease monotonically with increasing NIand decreasing , the lens changing from an accelerating to a decelerating one. (iii) If , then C _s/r – ^1/4, C _c/r – ^1/6, f/r– ^1/3, and NI – ^–1/2. (iv) The lenses are suitable for resistless heavy-ion projection lithography and can provide 20 × 10¹¹ pixels of area 2 × 2 nm² for an exposed area of 3 × 3 mm². (v) Used in heavy-ion microprobe systems, the lenses could enable resistless lithography over much larger areas than existing equipment.     

Fast correlation attacks on certain stream ciphers

Suppose that the output of a running key generator employed in a stream cipher is correlated to a linear feedback shift register sequence (LFSR sequence) a with correlation probabilityp>0.5. Then two new correlation attacks (Algorithms A and B) are presented to determine the initial digits of a, provided that the numbert of feedback taps is small (t<10 ifp0.75). The computational complexity of Algorithm A is of orderO(2^ck), wherek denotes the length of the LFSR andc<1 depends on the input parameters of the attack, and Algorithm B is polynomial (in fact, even linear) in the lengthk of the LFSR. These algorithms are much faster than an exhaustive search over all phases of the LFSR, and are demonstrated to be successful against shift registers of considerable lengthk (typically,k=1000). On the other hand, for correlation probabilitiesp0.75 the attacks are proven to be infeasible against long LFSRs if they have a greater number of taps (roughlyk100 andt10).This work was supported in part by GRETAG Ltd., Regensdorf, Switzerland.     

Two signaling schemes for improving the error performance of frequency division duplex (FDD) transmission systems using transmitter antenna diversity

We propose two signaling schemes that exploit the availability of multiple (N) antennas at the transmitter to provide diversity benefit to the receiver. This is typical of cellular radio systems where a mobile is equipped with only one antenna while the base station is equipped with multiple antennas. We further assume that the mobile-to-base and base-to-mobile channel variations are statistically independent and that the base station has no knowledge of the base-to-mobile channel characteristics. In the first scheme, a channel code of lengthN and minimum Hamming distanced _minN is used to encode a group ofK information bits. Channel code symbolc _i is transmitted with thei th antenna. At the receiver, a maximum likelihood decoder for the channel code provides a diversity ofd _min as long as each transmitted code symbol is subjected to independent fading. This can be achieved by spacing the transmit antennas several wavelengths apart. The second scheme introduces deliberate resolvable multipath distortion by transmitting the data-bearing signal with antenna 1, andN–1 delayed versions of it with antennas 2 throughN. The delays are unique to each antenna and are chosen to be multiples of the symbol interval. At the receiver, a maximum likelihood sequence estimator resolves the multipath in an optimal manner to realize a diversity benefit ofN. Both schemes can suppress co-channel interference. We provide code constructions and simulation results for scheme 1 to demonstrate its merit. We derive the receiver structure and provide a bound on the error probability for scheme 2 which we show to be tight, by means of simulations, for the nontrivial and perhaps the most interesting caseN=2 antennas. The second scheme is backward-compatible with two of the proposed digital cellular system standards, viz., GSM for Europe and IS-54 for North America.     

Some properties of order-Statistics filters

LetX ₁,X ₂,... be a stationary sequence of random variables with Pr{X _t, x}=F(x),t=1, 2,... Also let _i ^n,(t) ,i=1,...,n, denote the ith order statistic (OS) in the moving sample (X _t–N ,...,X _t,...,X _t+N) of odd sizen=2N+1. ThenY _t=a _i X _i ^n(t) with a _i=1 is an order-statistics filter. In practicea _i0,i=1,...,n. Fort>N, the sequence {Y _t} is also stationary. IfX ₁ X ₂, ... are independent, the autocorrelation function (r)=corr(Y _t,Y _t+r) is zero forr >n – 1 and forr n – 1 can be evaluated directly in terms of the means, variances, and covariances of the OS in random samples of sizen +r fromF(x).In special cases several authors have observed that the spectral density functionf() of {Y _t} is initially decreasing for > 0. This result is made more precise and shown to hold generally under white noise. The effect of outliers (impulses) is also discussed.This research was supported by the U.S. Army Research Office.     

Avoiding linear dependencies in LFSR test pattern generators

Linear Feedback Shift Registers (LFSRs) constitute a very efficient mechanism for generating pseudoexhaustive or pseudo-random test sets for the built-in self-testing of digital circuits. However, a well-known problem with the use of LFSRs is the occurrence of linear dependencies in the generated patterns. In this paper, we show for the first time that the amount of linear dependencies can be controlled by selecting appropriate characteristic polynomials and reordering the LFSR cells. We identify two classes of such polynomials which, by appropriate LFSR cell ordering, guarantee that a large ratio of linear dependencies cannot occur. Experimental results show significant enhancements on the fault coverage for pseudo-random testing and support the theoretical relation between minimization of linear dependencies and effective fault coverage.This work was partially supported by NSF grant MIP-9409905, a 1993–94 ACM/IEEE Design Automation Scholarship and a grant from Nissan Corporation. A preliminary version of this work appeared in A Class of Good Characteristic Polynomials for LFSR Test Pattern Generators, in Proc. of IEEE International Conference on Computer Design, Oct. 1994, pp. 292–295, where it received the ICCD'94 Best Paper Award.     

A detailed study of current–voltage characteristics in Au/SiO2/n-GaAs in wide temperature range

The forward and reverse-bias current–voltage (I–V) characteristics of Au/SiO₂/n-GaAs (MIS) type Schottky barrier diode (SBDs) have been investigated in the wide temperature range of 80–400 K. The zero-bias barrier height (_Bo) and ideality factor (n) assuming the thermionic emission (TE) mechanism show strong temperature dependence. While n decreases, _Bo increases with increasing temperature. Such temperature dependence of _Bo is an obvious disagreement with the reported negative temperature coefficient (α_temp) of barrier height. Therefore, we have reported a modification which includes the n and electron-tunneling parameter (αχ^1/2δ) in the expression of reverse-saturation current (I₀). After this modification, the value of α_temp obtained as −4 × 10⁻⁴ eV/K which is very close to α_temp of GaAs band-gap (−5.4 × 10⁻⁴ eV/K). Richardson plot of the ln(I₀/T²) versus 1/T has two linear region; the first region is (200–400 K) and the second region (80–150 K). The values of the activation energy (E_a) and Richardson constant were obtained from this plot and the values of E_a and Richardson constants (A^*) are much lower than the known values. These behaviors of the Au/SiO₂/n-GaAs (MIS) type (SBDs) have been interpreted by the assumption of a double-Gaussian distribution of barrier heights (BHs) at the metal–semiconductor interface giving a mean BHs () of 1.20 and 0.68 eV and standard deviation (σ_s) of 0.1503 and 0.0755 V, respectively. Thus the modified ln versus q/kT for two different temperature ranges (200–400 K and 80–150 K) plot then gives mean barrier heights and A^*, 1.18 and 0.66 eV and 7.08 and 3.81 A/cm² K², respectively. This value of the A^* 7.08 A/cm² K² is very close to the theoretical value of 8.16 A/cm² K² for n-type GaAs. Hence, all these behaviours of the forward-bias I–V characteristics of the Au/SiO₂/n-GaAs (MIS) type SBDs can be successfully explained on the basis of a TE mechanism with a double-Gaussian distribution of the BHs.     

Construction of Time-Relaxed c-Broadcast Networks

C-broadcasting is an information dissemination task where a message, originated at any node in a network, is transmitted to all other nodes with the restriction that each node having the message can transmit it to almost c neighbors simultaneously. If the transmission time of the message is set to be one time unit, a minimal c-broadcast network (c-MBN) is a communication network in which c-broadcasting from any node can be accomplished in log _c+1 n time units, where n is the number of nodes and log _c+1 n is the fastest possible broadcast time. If networks are sparse, additional time units may be required to perform c-broadcasting. A time-relaxed c-broadcast network, denoted as (t,c)-RBN, is a network where c-broadcasting from any node can be completed in log _c+1 n+t time units. In this paper, a network compounding algorithm is proposed to construct large sparse (t,c)-RBNs by linking multiple copies of a time-relaxed network of small size using the structure of another time-relaxed network. Computational results are presented to show the effectiveness of the proposed algorithm.     

Eyring acceleration model in thick nitride/oxide dielectrics

A kinetic model is developed, which can predict failure times in thick nitride/oxide dielectric stacks at elevated temperatures. It is shown that failure time depends on the applied bias stress and temperature and, in the general case, obeys a two-stress Arrhenius-type relationship known as an Eyring acceleration model:

where t_f is the failure time before charge saturation condition occurs, k is the Boltzmann constant, T is the absolute temperature, ΔH is the activation energy, M and N are the voltage and time dependent parameters, respectively, V_STRESS is the externally applied voltage stress and B is a proportionality constant. Furthermore it is demonstrated that the obtained kinetic equation is physically related to the nitride conduction, described by Frenkel–Poole equation.     

Mid- and Long-Wave IR Detectors Using an Hg1 – xCdxTe Heteroepitaxial Layer

A complete production technology is developed for single-cell and array IR detectors using an MBE-grown Hg_{1 – x} Cd _x Te heteroepitaxial layer. The array detectors implement a bump-bonded flip-chip hybrid architecture. The arrays are constructed in photodiodes and have a size of 128 × 128 or 4 × 288. The single-cell detectors are built around a photoconductivity cell. The detectors are operated in the wavelength range 3–5 or 8–12 m and at 78–80 or 200–220 K. Some performance data on the detectors are presented.     

Permittivity spectra and characteristic energy losses of electrons in ZnO at 100 K

For the first time, full sets of fundamental optical functions have been obtained for zinc oxide in the range 0–30 eV at 100 K for E ⊥ c and E ‖ c polarizations. Spectra of the transverse and longitudinal components of transitions and their basic parameters (peak energies E _i , half-widths H _i of transition bands, band areas S _i , and oscillator strengths f _i ) have also been determined for the first time. The calculations are performed using synchrotron experimental reflectance spectra. The main features of spectra of the optical functions and components of transitions are established. These features are compared to the results of known theoretical calculations of the bands and spectra of optical functions.     

On the Security of Modular Exponentiation with Application to the Construction of Pseudorandom Generators

   Abstract. Assuming the intractability of factoring, we show that the output of the exponentiation modulo a composite function f _N,g (x)=g ^x mod N (where N=P⋅ Q ) is pseudorandom, even when its input is restricted to being half the size (i.e. x<

Density functional theory applied to the calculation of dielectric constant of low-k materials

The interest of low-k dielectric materials to reduce capacitance in multilevel metal interconnects of integrated circuits is well known in the semiconductor industry. The use of these materials (especially hydrogen silsesquioxane (HSQ) and methyl silsesquioxane (MSQ): intermetal dielectric applications in the back end of line fabrication) leads to a reduction of the dielectric constant from k4 in a traditional intermetal dielectric material of silicon dioxide to a value of 2.5–3. The physical difference between HSQ or MSQ and a-SiO₂ is the presence of Si–H bonds (for HSQ) or Si–CH₃ bonds (for MSQ) and the density of the material. A theoretical calculation of bond polarizability (Si–H or Si–CH₃) associated to experimental values of electric dipole densities can lead, using the Clausius-Mossotti relationship, to the calculation of the dielectric constant. After validation of the calculation methods both on simulation and experimental values, it is shown that for a constant density, the difference between the materials could be due to the bond polarizability and furthermore that this difference accounts, in part, for the value of dielectric constant. Consequently, even if densification remains the main parameter explaining low-k values, the polarizability of building units of these materials is not negligible.     

On the convergence rate ofs-numbers of compact Hankel operators

It is well known that the sequence ofs-numbers {s_n},n=0, 1,..., of a compact operator, and particularly a compact Hankel operator =[h _j+k –1], converges monotonically to zero. Since the (n + 1)sts-number s_n measures the error ofL (¦z¦=1) approximation, modulo an additive H function, by nth degree proper rational functions whose poles are restricted to ¦z¦ < 1, it is very important to study how fast {s _n } converges to zero. It is not difficult to see that ifh _n =O(n ^–), for some > 1, thens _n =O(n ^–). In this paper we construct, for any given sequence _n 0, a compact Hankel operator such thats _{n n} for alln.The research of C. K. Chui was supported by the SDIO/IST managed by the U.S. Army under Contract No. DAAL-03-87-K-0025, and by the NSF under Grant No. DMS-8901345. X. Li's research was supported by the SDIO/IST managed by the U.S. Army under Contract No. DAAL-03-87-K-0025. The research of J. D. Ward was supported by the NSF under Grant No. DMS-8901345.     

New challenges to the modelling and electrical characterisation of ohmic contacts for ULSI devices

Continuing developments in semiconductor process and materials technology have enabled significant reductions to be achieved in the contact resistance R_c of devices. This reduction is commonly assessed in terms of the specific contact resistance (SCR) parameter ρ_c (Ω cm²) of the metal–semiconductor interface. Such a reduction in SCR is essential, for as device dimensions decrease, then so also must ρ_c and the corresponding contact resistance in order not to compromise the down-scaled ULSI device performance. Thus the ability to accurately model contacts and measure ρ_c is essential to ohmic contact development. The cross kelvin resistor (CKR) test structure is commonly used to experimentally measure the Kelvin resistance of an ohmic contact and obtain the specific contact resistance ρ_c. The error correction curves generated from computer modelling of the CKR test structure are used to compensate for the semiconductor parasitic resistance, thus giving the SCR value. In this paper the increased difficulty in measuring lower ρ_c values, due to trends in technology, is discussed. The challenges presented by the presence of two interfaces in silicided contacts (metal-silicide–silicon) is also discussed. Experimental values of the SCR of an aluminium–titanium silicide interface is determined using multiple CKR test structures.     

Robust Low Power CMOS Single-Ended to Fully-Balanced Converter Circuits

In this paper two novel single-ended-input fully-balanced-output circuits (SFC), namely unbuffered and buffered SFCs, are proposed for input interface to fully balanced signal processing systems. The unbuffered SFC overcomes the drawback of uncontrollable process variations of resistors and generate well-controlled process invariant common mode output voltage, V _o,com . The adopted active current common mode feedback compensation makes this possible. Simulations using MOSIS 2 m N-well process and a 3 V supply, show that with ±100% variation, V _o,com only varies by less than ±2%. In addition, it is shown that V _o,com is accurately controlled by a preset DC voltage. On the other hand, the buffered SFC adopts a novel body effect-free class AB buffer so as to have low standby power consumption but high current driving capability. It is implemented without resistors and common-mode feedback circuitry. Measurement results from a 1.2 m N-well CMOS chip indicate a bandwidth of 5.5 MHz while driving a 40 pF load with a supply voltage of ±1.5 V. The circuit is capable of supplying more than 3 mA of output current while consuming 1.1 mW of standby power. The THD is less than –55 dB at 1 KHz and the phase error is less than 2° for frequencies up to 1 MHz.     

Low-Voltage Complementary Push-Pull Inverters for ULSI: Simulation and Prospects

Analysis and numerical circuit-design modeling of the speed of current-mode logic devices, as well as complementary push-pull inverters built on scaled 0.2- to 0.1-m transistor structures with the thin heavily doped base, were performed. The speed was analyzed both on the device and system levels. The objects of investigation were also novel low-voltage push-pull inverters based on 0.1-m complementary symmetric bipolar–field-effect structures with the undoped (lightly doped) base. These converters offer the following parameters: delay time t _d = 100–150 ps at P = 5–10 mW and C _l = 100 fF (inverters of type 1) and t _d = 100–150 ps at P = 30–100 nW and C _l = 1 fF (type 2). It is shown that the high speed of these devices makes them candidates for next-generation ULSI.     

Electrical and structural properties of hafnium silicate thin films

Thin (4 nm) hafnium silicate (HfO₂)_x(SiO₂)_1−x/SiO₂ gate stacks (0 < x < 1) grown by metal organic chemical vapour deposition (MOCVD) are investigated in this study. The focus is on extracting the optical constants, and hence bandgaps as well as dielectric constants. The VUV (vacuum ultraviolet) spectroscopic ellipsometry (VUV-SE) technique in the spectral range 140–1700 nm, together with current–voltage and capacitance–voltage techniques were used for studying the optical and electrical properties of the layers, respectively. The bandgap was found to increase from 5.24 eV for HfO₂ to 6 eV for Hf-silicate with 30% Hf. The permittivity was reduced from 21 for HfO₂ layers to 8 for Hf-silicate with x = 0.3. The results suggest that the optimal Hf content is above 0.6, for which the permittivity higher than 10 can be achieved.     

Factorizations in the elementary Abelian p-group and their cryptographic significance

Let G be a finite group and let A _i 1 i s, be subsets of G where ¦A _i ¦ 2, 1 i s and s 2. We say that (A₁, A₂,..., A₃) is a factorization of G if and only if for each g G there is exactly one way to express g = a ₁ a ₁ a ₂··· a ₃, where a _j A _i , 1 i s.The problem of finding factorizations of this type was first introduced by Hajos [3] in 1941. Since then a number of papers have appeared on the subject. More recently, Magliveras [6] has applied factorization of permutation groups to cryptography to obtain a private-key cryptosystem. Factorizations in the elementary abelian p-group were exploited (but not explicitly stated in these terms) by Webb [13] to produce a public-key cryptosystem conceptually similar to cryptosystems based on the knapsack problem.Using the result that certain types of factorizations in the elementary abelian p-group are necessarily transversal (a term introduced by Magliveras), this paper shows that the public-key system proposed by Webb is insecure.     

Quantitative analysis for linear hybrid cellular automata and LFSR as built-in self-test generators for sequential faults

This paper presents a combinatorial method of evaluating the effectiveness of linear hybrid cellular automata (LHCA) and linear feedback shift registers (LFSR) as generators for stimulating faults requiring a pair of vectors. We provide a theoretical analysis and empirical comparisons to see why the LHCA are better than the LFSRs as generators for sequential-type faults in a built-in self-test environment. Based on the concept of a partner set, the method derives the number of distinctk-cell substate vectors which have 2^2k , 1k[n/2], transition capability for ann-cell LHCA and ann-cell LFSR with maximum length cycles. Simulation studies of the ISCAS85 benchmark circuits provide evidence of the effectiveness of the theoretrical metric.This work was supported in part by Reserach Grants No. 5711 and No. 39409 and a Strategic Grant from the Natural Sciences and Engineering Research Council of Canada and by an equipments loan from the Canadian Microelectronics Corporation.A preliminary version of this paper is partially presented at theIEEE ISCAS'94, May 1994.     

Dominating the maximum likelihood estimator in predicting reliability

The reliability of a series system with n independent components are estimated. The failure distributions of these components are assumed to be Weibull with parameters α_i's and β_i's. It is assumed that the shape parameters β_i's are known. This reduces the reliability of the system to be a function of α = (α₁, ..., α_n), ξ(α), say. The MLE of ξ(α) is derived and an estimator which dominates the MLE in terms of risk, under squared error loss, is also derived. The predicted reliability using these two estimators are computed and compared.