Table of contents

The experimental infrared photoelastic stress fringe patterns under a thin oxide film edge in silicon substrates were compared with the simulated patterns based on analytic solutions obtained according to a simple concentrated force model (CFM) (Hu S M 1978 Appl. Phys. Lett. 32 5) and an improved CFM (Wong S P et al 2001 Appl. Phys. Lett. 79 1628). The singular property of the stress field in the substrate at a thin film edge was shown to be associated with the concentrated force assumption. The results clearly demonstrated the limitation of the concentrated force assumption widely adopted in the study of thin film edge induced stresses in substrates and posed new fundamental questions on the nature of the singularity of the stress field under a thin film edge.

Vickers microhardness of (0001), (100) and (110) planes of ZrB₂ single crystal prepared by the floating zone method has been investigated at various temperatures and loading times. As the temperature increases from 25°C to 1000°C, hardness drops from ∼20.9 GN m⁻² of all planes to ∼7.85 GN m⁻² for (0001) plane and ∼4.91 GN m⁻² for (100) and (110) planes. The hardness of (100) and (110) planes exhibits almost same tendency and is always lower than that of (0001) plane by about 35%. The thermal softening coefficients of all three planes strongly depends on the temperature range with clear inflections at 400°C and 700°C. The loading time dependence of hardness is used to calculate the activation energy for creep. In addition, a relationship was found that shows the variation of hardness with temperature to be proportional to the variation with the loading time in a specific temperature range.

Large-scale single-crystalline CdO nanowires have been successfully fabricated from the electrochemical deposition mixture film of Cd and Te at 450°C, and characterized by x-ray diffraction (XRD) powder, scanning electron microscopy (SEM) and transmission electron microscopy (TEM). SEM and TEM images show that these nanowires are uniform with diameters of about 40–60 nm, length up to several tens micrometres. XRD and selected area electron diffraction analyses altogether indicate that these CdO nanowires crystallize in a NaCl cubic structure. The growth mechanism of these nanowires is also proposed as vapour–liquid–solid mechanism, in which Te serves as a liquid-forming agent.

Microwave plasma chemical vapour deposition of nanostructured diamond films was carried out on curved surfaces of Ti–6Al–4V alloy machined to simulate the shape of a temporomandibular joint (TMJ) dental implant. Raman spectroscopy shows that the deposited films are uniform in chemical composition along the radius of curvature of the TMJ condyle. Thin film x-ray diffraction reveals an interfacial carbide layer and nanocrystalline diamond grains in this coating. Nanoindentation hardness measurements show an ultra-hard coating with a hardness value of 60±5 GPa averaged over three samples.

Up to now the Preisach model of magnetism is supposed to be the best model that can deal with minor loops and predict qualitatively some their features. In general, two necessary properties of the magnetization process (congruence and wiping-out) are usually not fulfilled. It is believed that the Preisach model describes harder magnetic materials better than softer ones. In this work we summarized all our experience in the application of the Preisach model to minor loops experimental data on industrial steels, soft and hard. The deviations of minor loops behaviour from that predicted by the model were combined into several classes; their influence on the resulting Preisach distribution function was determined. All these deviations were found to have the same tendency for both soft and hard steels leading to appearance of nonphysical regions of negative values on the Preisach distribution plane and delta-functions along the Preisach triangle diagonals. These effects were increased several times by the accommodation process.

The magnetic domain structure in Permalloy (Ni₈₁Fe₁₉) micropatterns (10–100 μm) on NiO has been investigated by means of soft x-ray photoemission electron microscopy. The exchange anisotropy between the Ni₈₁Fe₁₉ patterns and the NiO layer results in the formation of complex domain structures which markedly differ from the simple Landau–Lifshitz configurations. The domain structures reflect the competition between the exchange anisotropy and the dipole–dipole interaction in a weakly coupled system. The observed domain structures change with the feature size, as the domain patterns lose complexity in the smaller structures.

Magnetic properties of the Co/Nd multilayers with a 2 Å probe layer of ⁵⁷Fe evaporated in UHV conditions were studied in the temperature range 4.2 K<T<300 K by means of VSM magnetometry and Mössbauer spectroscopy. The isomer shift, hyperfine field distribution and its direction were obtained from a Mössbauer data computer fitting. The easy magnetization direction was found close to the sample normal for all samples at low temperatures and there was a stronger tendency to the perpendicular direction for the sample when the probe layer was placed in a close vicinity of the Co/Nd interface. For this sample the out-of-plane easy direction was preserved at 300 K. For all other studied samples, in-plane anisotropy was observed at room temperature.

In this paper, new experimental results of the microwave spectral properties of quasistatic magnetoelectric (ME) particles based on small ferrite resonators with linear-form surface electrodes have been illustrated. Existence of local coupling between electric and magnetic dipolar polarizations, which can be essential for realization of the so-called bianisotropic materials, has been proved. Also, a method of boosting the ME effect in a particle has been suggested.

We study theoretically the dynamics of a system of two magnetizable particles suspended in a non-magnetic fluid subject to a rotating magnetic field when a modulation on the Mason number (ratio of viscous to magnetic forces) is applied. We find, using a periodic modulation, that a resonant-like phenomenon between the periodic modulation of the Mason number and the intrinsic radial oscillation of the system without modulation occurs. For a random perturbation of the Mason number, we obtain an optimum noise strength at which the average interparticle distance reaches the lowest value. When a weak periodic modulation and a noise source are included in the Mason number, stochastic resonance (SR) is found for different frequencies and amplitudes of the modulation. An interpretation of this SR phenomenon is made by means of a threshold crossing mechanism.

The Kane proposal for the construction of a solid state quantum computer calls for the placement of a metallic A-gate above each nuclear qubit. Voltage biases applied to these A-gates are to be used to control the evolution of the qubits. In this paper, we consider the decohering effect of thermal fluctuations, the so-called Johnson noise, in the A-gate biases, on the average evolution of qubits undergoing single qubit rotations, as well as those that are not undergoing any operation at all. We find that although the errors introduced are of a different nature for each case, phase errors for the static qubits and both phase and bit-flip errors for qubits undergoing operations, the net error probability is similar for both, and low enough when compared to the timescale of single qubit operations for them to be corrected efficiently.

It was found that under intense CW excitation of Yb³⁺/Nd³⁺ : YAG nanocrystallite ceramics by 976 nm laser diode (pumping directly Yb³⁺ ion), the visible hot emission occurred. Its intensity exhibited long emission build-up times (of an order of seconds) after turning on the excitation source. The intensities of hot emission of Nd³⁺ ion increased with the excitation power demonstrating the dependence I^N, with the order parameter N demonstrating a linear dependence on the energy distance ΔE = E_em^hot−E(⁴F_3/2).

A simple apparatus dedicated to the calibration of a low-order double-crystal birefringent wave plate is presented. These wave plates are widely used in optical instrumentation and the residual errors associated with these devices can be very important in high-resolution polarimetry. Misalignment between the crystals is considered and simulated. An analytical treatment of such unperfected device is given and is followed by numerical examples. The fit between the experimental values obtained using this apparatus and the theoretical ones proves the interest of the method and enabled us to assess the wave plate optical characteristics.

Excitation of surface plasmons on a metal substrate, via the attenuated total reflection method can theoretically offer preferential absorption of light at one particular wavelength, whilst reflecting the nearby spectrum. Normally this `filtering' action is limited to removal of p-polarized light, and the acceptance angle of such a filtering device is very narrow, which limits practical applications, such as separation of fundamental and laser harmonics. The possibility of avoiding this angular precision is explored by considering the complex permittivity of metal composites. By using a two or more layer structure, as opposed to a single metal substrate, the acceptance angle of the device can be broadened, by a factor of about 15 times. An example is discussed for separation of the fundamental and harmonics from a Nd : YAG laser. Variants of the structure allow the design of an in-line transmission filter for the various wavelengths with sufficient angular tolerance to include focusing lenses. Avoidance of laser ablation of the metal is discussed.

Fractal dimensional analysis was employed to obtain a quantitative measure of the morphology of polymer networks formed by UV irradiation induced polymerization of photo-reactive mesogenic monomers dissolved in a liquid crystal host medium. The fractal dimensions obtained, may be interpreted by polymer network growth following a percolation-like model for monomer concentrations well below the solubility limit. On passing the solubility limit, the polymerization process changes from a (radical chain) solution polymerization to a dispersion polymerization, with fractal dimensions decreasing and suggesting a cluster–cluster aggregation process for monomer concentrations above the solubility limit, similar to the aggregation of colloidal particles.

Variable dose (VD), T_m–T_stop, initial rise (IR), variable heating rate (VHR), peak shape (PS) and computerized glow curve deconvolution (CGCD) methods are used to determine the number of peaks, the order of kinetics (b), the activation energy (E_a) and attempt-to-escape frequency (s) associated with the glow peaks in CaF₂ : Dy (TLD-200) after β-irradiation between the dose level 0.1 and 110 Gy. The T_m–T_stop procedure indicates that the glow curve of this crystal consists of at least nine glow peaks. The dose variation experiment indicates that seven of them, namely peaks 1–6 and 8, are of first-order kinetics and peaks 7 and 9 are of general-order kinetics. However, the T_m–T_stop procedure and the CGCD method have indicated that peak 6 has general-order kinetics too. The activation energy found with the IR, VHR, PS and CGCD methods for peak 4 yield very close values. For all other peaks, there is no agreement between the results of all the applied methods. This work also indicates that the post-irradiation annealing and the heating rate have pronounced effects on the evaluated kinetic parameters of all glow peaks.

The recombination of positive ions and atoms during an afterglow period, created during previous electrical breakdown and gas discharge, at the molybdenum glass and copper is investigated using the time delay method. It has been shown that the recombination of the positive ions at the copper wall, while the atoms at the molybdenum wall, at room temperature, is more intensive. The temperature influence on the recombination of these particles at copper has also been monitored by the same method, showing that the positive ion recombination rate increases and atom recombination rate decreases with temperature.

We have investigated the properties of hotspots formed in low energy plasma focus (PF) discharges operating in hydrogen–argon mixtures, at 140 kA current level. A combination of filtered pinhole and slit-wire camera is used to measure the hotspot size and temperature. The results show that the best conditions for reproducible and localized hotspot formation are obtained by adjusting the base pressure in such a way that the mass load allows the time of first radial collapse to coincide with peak current. When the PF is operated with 20% argon content, rather uniform hotspots, of 115 μm characteristic size and 300 eV characteristic temperature, are produced with a better than 80% reproducibility in their axial localization. The electron density in the hotspots is estimated to be ∼10²⁰ cm⁻³. Calculations performed with a CRE code indicate that a significant fraction of the radiation is emitted in the 3.2 to 3.88 keV region, corresponding to K_α emission from highly ionized argon.

The relative importance of atmospheric negative ions in corona formation and breakdown in a radial electric field has been investigated, with special reference to effects occurring at an insulating surface of a cycloaliphatic polyurethane with dolomite filler. The inception times, corona charge, light emission (by UV photography) and sparkover voltages under lightning impulses have been recorded under natural atmospheric conditions and under the influence of negative ions introduced from an auxiliary corona. The presence of excess negative ions on the surface is shown to increase both the charge injected and the mean radius of the corona, attributable to an augmentation of streamer discharges. In similar experiments in air, the excess ions cause a transition to a `glow' discharge. Effects of negative ions on sparkover are not significant on the insulator surface, but they cause a small increase in air. Comparisons between the two cases lead to the conclusion that the most important effect of the ions on the surface is to provide `seed' electrons for streamer propagation, following ion detachment in the field of advancing streamer tips.

The present paper is a continuation of the investigations on the sparking-induced decomposition of SF₆ and SF₆/N₂ (10 : 90) mixtures which have already been carried out in our laboratory, both experimentally and numerically. It concerns the decomposition of SF₆/N₂ mixtures (100 kPa) containing 100%, 10% or 5% of SF₆, under high-energy sparks (3.6 J spark⁻¹) generated in a 340 cm³ experimental cell between a stainless steel point and a stainless steel plane. Our attention was focused on the following main by-products: (SF₄+SOF₂), (SOF₄+SO₂F₂), S₂F₁₀, CF₄, CO and CO₂ which were studied by varying the concentration of the impurities added H₂O, O₂ (0–0.2%), in the presence of atoms such as H and C released from vaporized solid insulators (polyethylene [C₂H₄]_n, polypropylene [C₃H₆]_n, Teflon [CF₂]_n) or from gaseous additives (methane CH₄ (0–4%), ethylene C₂H₄ (0–2%), octofluoropropane C₃F₈ (0–5%)), with the aim of simulating the occurrence of sparking in electrical devices, especially along spacers. As SF₆/CO₂ and SF₆/N₂/CO₂ mixtures are reported to be able to constitute promising SF₆ substitutes for industrial applications, we also studied the chemical stability of SF₆ and SF₆/N₂ (5 : 95) mixtures in the presence of 0–20% CO₂.

The presence of additives CH₄, C₂H₄, C₃F₈ or solid insulator (polyethylene, polypropylene, Teflon) leads to lower production of (SF₄+SOF₂) and S₂F₁₀ in dilute SF₆ than in pure SF₆ when the percentages of additives or the amounts of solid insulator vaporized are high.

Concerning the additive CO₂, we observe an increased production of (SOF₄+SO₂F₂) and a formation of large quantities of CO, more pronounced in SF₆/N₂ (5 : 95) mixtures than in pure SF₆. In contrast, the presence of CO leads to a lesser degree of decomposition of diluted than undiluted SF₆.

Sequences of photographic frames with resolution down to 10 ns and 2 μm pixel⁻¹ were obtained for vacuum arcs (30–40 A) with and without external magnetic field (0.36 T). In both cases extensive statistical studies confirm the diffusion character of the motion of cathode spot fragments on 10 ns timescale. In addition to such a diffusion, the magnetic field provokes fragment steps of length ∼5 μm in the retrograde direction, approximately once in every 0.3 μs, which account for the directed spot movement in magnetic field. Thus two mechanisms are responsible for spot dynamics. Episodes of immobile spot fragments are also observed.

Temperatures have been measured in a high-pressure microwave sulfur lamp using sulfur atomic lines found in the spectrum at 867, 921 and 1045 nm. The absolute intensities were determined for 3, 5 and 7 bar lamps at several input powers, ranging from 400 to 600 W. On average, temperatures are found to be 4.1±0.15 kK and increase slightly with increasing pressure and input power. These values and trends agree well with our simulations. However, the power trend is reversed to that demonstrated by the model, which might be an indication that the skin-depth model for the electric field may be incomplete.

The applicability of `fluid' models based on analytic approximations of the electron energy distribution function (EEDF) and of kinetic models for low-pressure discharge light sources is discussed. Traditionally, `fluid' models of fluorescent lamps assume that the EEDF is Maxwellian up to the energy of the first excited state. It is shown that such an approach is sufficiently accurate in most cases of conventional as well as of `highly loaded' fluorescent lamps. However, this assumption is strongly violated for many rare gas glow discharges for mercury free light sources. As an example, a neon dc discharge is studied. The densities of the four lowest excited states and the electric field have been measured. The experimental results can be fairly well reproduced by a kinetic positive column model. This article was scheduled to appear in issue 14 of J. Phys. D: Appl. Phys. To access this special issue please follow this link: http://stacks.iop.org/0022-3727/35/i=14/

This paper presents experimental results on an atmospheric plasma jet generated by a rectangular wave-guide based microwave system operating at 896 MHz. Argon gas emerging from a copper nozzle, which is placed inside the wave-guide short circuited at one end (hereafter referred to as the cavity for convenience), jets into the atmosphere through an aperture in the cavity wall. The plasma jet can be visually divided into two parts: a very bright patch (active zone) where microwave discharge takes place and a large, weakly illuminating tail attached to the active zone. The plasma jet is characterized in this work by measuring the electron number density, temperature and an excitation temperature of the argon atoms in the active zone over a range of microwave source power from 2 to 5 kW and argon flow rate from 2 to 7 litre min⁻¹. Results show that the plasma is not in a local thermodynamic equilibrium state and the temperature of electrons is considerably higher than the excitation temperature of the heavy particles which indicates the upper limit of the translation temperature of the heavy particles.

Based on experimental observations and theoretical explanation of the streamer spatial extension for an inhomogeneous field gap in SF₆, a new model is developed, taking into account the probability of the appearance of an efficient electron. The model, which can help one to better understand the breakdown characteristics of the impulse-steepness dependence, relates the spatial extension of the streamer corona in an inhomogeneous field gap in SF₆ gas to the steepness of applied positive impulses.

Columnar structure of thin aluminium nitride (AlN) films is examined by x-ray diffractometry (XRD), scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The films were deposited on SiO₂/Si(100) substrate using radiofrequency reactive sputtering method. Strong [0001] preferred orientation is observed by XRD and confirmed by selected area diffraction pattern of TEM. Columnar grains of ∼50–100 nm inclined at an angle of ∼10° to the substrate normal are observed by SEM. As revealed by TEM, each columnar grain is composed of nano-grains of the order of 10 nm and no faceting is observed in the nano-grains and columns. The [0001] preferred orientation results as columnar grains are oriented at various azimuthal angles with their c-axes perpendicular to the substrate surface. A slight tilt of a few tenths of a degree between adjacent nano-grains within a column is also observed. The random azimuthal orientation of columnar grains and small tilt between nano-grains in the films are accommodated by the amorphous phase present in the grain boundaries.

Advances in surface force spectroscopy (SFS) have stimulated the research on the adhesion of two solid surfaces on the microscale. The Johnson–Kendall–Roberts (JKR) adhesion theory is extensively used in investigating the adhesion behaviour of thin films by using the SFS. However, the JKR theory is based on the contact between two spheres or a sphere and an elastic half space. Its application in thin films remains an open question. This work presents a rigorous analysis of the adhesive contact between an incompressible elastic thin film and a rigid axisymmetric indenter under the condition of the contact radius much larger than the film thickness. It turns out that, for the adhesive interaction between a rigid spherical indenter and the thin film the pull-off force is proportional to the indenter diameter and independent of the film thickness with the perfectly bonded condition between the thin film and the substrate requesting the highest pull-off force to separate the indenter from the substrate. For a conical indenter, the pull-off force depends on the film thickness. For the frictionless condition, the pull-off force is proportional to the square root of the film thickness, while it is proportional to ¾ power of the film thickness for the bonded condition.

The local elastic properties and the ferroelectric domain configuration of piezoelectric ceramics have been examined by atomic force acoustic microscopy and by ultrasonic piezoelectric force microscopy. The contrast mechanisms of the two techniques are discussed. From the local contact stiffness which is obtained by evaluation of the contact resonance spectra, the elastic constants of the sample surface can be calculated. In the case of anisotropic materials these elastic constants correspond to the indentation moduli. Indentation moduli for barium titanate and for a lead zirconate-titanate ceramics were calculated theoretically and are in reasonable agreement with experiments. The non-linearity of the tip–sample interaction becomes noticeable at large vibration amplitudes or large mechanical tip loads.

Cd_1−xZn_xS nanoparticles with different compositions (0⩽x⩽1) embedded in silica matrix were prepared in thin film form by the sol–gel technique. The film texture and structural transformation with increasing x value were studied from transmission electron micrographs and electron diffraction patterns. Optical study indicated increase in optical bandgap and decrease in refractive index with increasing zinc content (x value). Photoluminescence studies showed emission from surface states with emission energy blue shifted with increasing zinc content in the films. Annealing behaviour of the films with increasing annealing temperature and time was studied.

Thin films of TiN have been deposited by reactive magnetron sputter deposition in varying partial pressures of nitrogen. Reflectivity measurements have been carried out between 1.5 and 3.5 eV and a correlation made between the film properties and optical data. Resistivity measurements carried out at room temperature are shown to exhibit the same trends as those obtained from reflectivity experiments. X-ray absorption fine structure measurements, in both electron-yield and fluorescence-yield modes, have shown the films to be identical and stoichiometric to within ±5 at.%. The use of reflectivity spectra to form the basis of a characterization tool for physical vapour deposited thin films is discussed.

The Schottky behaviour of Ni/Au contact on n-GaN was investigated under various annealing conditions by current–voltage (I–V) measurements. A non-linear fitting method was used to extract the contact parameters from the I–V characteristic curves. Experimental results indicate that high quality Schottky contact with a barrier height and ideality factor of 0.86±0.02 eV and 1.19±0.02 eV, respectively, can be obtained under 5 min annealing at 600°C in N₂ ambience.

We propounded and grew a new crystal material of ammonium nickel sulfate hexahydrate (ANSH) for an ultraviolet light filter. The crystal structure of ANSH was solved by x-ray diffraction method and the empirical formula of the title compound is (NH₄)₂Ni(SO₄)₂⋅6H₂O. The ANSH crystal belongs to the monoclinic space group P2₁/C, a = 6.2351 Å, b = 12.451(3) Å, c = 9.1798(18) Å, β = 106.88(3)°, v = 681.9(2) Å³, z = 2 and D_c = 1.924 g cm⁻³. The deep-green ANSH single crystal with a dimension of 20.5×28×15 mm³ has been grown by the cooling solution method. Thermo-gravimetry analysis showed that the dehydration temperature of the ANSH crystal is above 96.06°C, which is much higher than that of commercially available nickel sulfate hexahydrate (NSH) crystals. The optical transmission character of the ANSH crystal is discontinuous in the range from ultraviolet (UV) to near IR wavelengths, and has several transmission peaks at 250 nm, 500 nm, 875 nm and 1425 nm, respectively. The UV transmission peak (250 nm) especially has high transmission efficiency and a narrow spectrum band-width. The relationship between the structure and the optical transmission property is further discussed.

The heterogeneous mixture properties depend on its constituents' characteristics. We examine the effective permittivity of a two-phase composite material made of epoxy resin host matrix and barium titanate (BaTiO₃) filler for different volume fractions in the matrix. The task we undertake consists in finding a model of BaTiO₃ particles through the computer simulations executed in PHI3D-electric field calculating package, based on the resolution of the Laplace equation using boundary integral equation method. With this aim in view we compare the measured results of the effective permittivity of the BaTiO₃–epoxy resin composite samples with the simulation results for different BaTiO₃ particle geometric models and for the same experimental conditions, with regard to the given volume fraction of the powder in the matrix. The experimental results are obtained through the measurements with an impedance meter in the range of frequencies from 50 Hz to 1 MHz.

A novel method to detect reversibly high concentrations of organic vapours in air has been developed by combining an intrinsically low conductive membrane of the calix[4]resorcinarene [C₇H₁₅] derivative with a charge flow transistor. The modulation of the turn-on response for the transistor upon exposure to acetone, chloroform, methanol, hexane and water is presented. The increase in the membrane conductivity is partially attributed to condensation of the vapours in the highly microporous membrane even below the saturation vapour pressure and partially to the effect of the polar analyte molecules complexing inside and between the OH groups of the cavities. The observed sensitivity is in the order chloroform >> acetone >> methanol >> hexane >> water.

The concept of finite time thermodynamics is used to determine the ecological function of irreversible Stirling and Ericsson heat engine cycles. The ecological function is defined as the power output minus power loss (irreversibility), which is the ambient temperature times, the entropy generation rate. The ecological function is maximized with respect to cycle temperature ratio and the expressions for the corresponding power output and thermal efficiency are derived at the optimal operating conditions. The effect of different operating parameters, the effectiveness on the hot, cold and the regenerative side heat exchangers, the cycle temperature ratio, heat capacitance ratio and the internal irreversibility parameter on the maximum ecological function are studied. It is found that the effect of regenerator effectiveness is more than the hot and cold side heat exchangers and the effect of the effectiveness on cold side heat exchanger is more than the effectiveness on the hot side heat exchanger on the maximum ecological function. It is also found that the effect of internal irreversibility parameter is more than the other parameters not only on the maximum ecological function but also on the corresponding power output and the thermal efficiency.

Ricochet of a tungsten heavy alloy long-rod projectile from oblique steel plates with a finite thickness was investigated numerically using a full three-dimensional explicit finite element method. Three distinctive regimes resulting from oblique impact depending on the obliquity, namely simple ricochet, critical ricochet and target perforation, were investigated in detail. Critical ricochet angles were calculated for various impact velocities and strengths of the target plates. It was predicted that critical ricochet angle increases with decreasing impact velocities and that higher ricochet angles were expected if higher strength target materials are employed. Numerical predictions were compared with existing two-dimensional analytical models. Experiments were also carried out and the results supported the predictions of the numerical analysis.

A new type of microwave discharge—near-surface non-self-sustained discharge (NSND)—has been realized and investigated. A physical model of this discharge is presented.

For the first time NSND application for microwave air jet engines has been proposed. Measurements under laboratory conditions modelling the microwave air jet engine operation shows the qualitative agreement between the model of NSND and actual processes near the target irradiated by a powerful microwave beam. Characteristic dependences of recoil momentum of target on the background pressure and microwave pulse duration obtained in experiments are presented. Measured cost of thrust produced by the NSND is no more than 3.0 kW N⁻¹, which is close to the predicted values.