Velocity Flux Distributions Research Articles

Determination of deposited flux and energy of sputtered tungsten atoms on every stages of transport in HiPIMS discharge

A time-resolved tunable diode-laser (DL) induced fluorescence (TR-TDLIF) technique has been used to identify different populations of atoms (on different stages of transport) to determine their corresponding deposited energy and flux. The temporal dimension permits the splitting of the processes of sputtering during the discharge and particles transport in the post-discharge where atoms and flux velocity distribution functions (AVDF, FVDF) of each population were measured varying the discharge parameters (power, voltage, pressure, and distance from target). Tungsten (W) was chosen, being an interesting case in terms of sputtered atom transport, considering its weight which implies weak changes of directivity or energy transfer after collisions with the buffer gas. The high temporal and spectral resolutions of TR-TDLIF are the keys for the distinction of the atoms populations and the stage corresponding to the transition from the ballistic to diffusive regime of transport was observed for the first time and named quasi-diffusive regime. Thus, the ability to dissociate populations of atoms and to determine their deposited flux and energy may be of great interest to adjust film properties as desired for applications.

Characterization of energetic and thermalized sputtered tungsten atoms using tuneable diode-laser induced fluorescence in direct current magnetron discharge

In this study a tuneable diode-laser induced fluorescence (TD-LIF) technique (λ0 = 407.4358 nm) is used to determine the atoms' velocity distribution function (AVDF) of energetic and thermalized sputtered tungsten (W) atoms in direct current magnetron discharge. The AVDF is characterized by probing the plasma above the centre of the target racetrack along the magnetron cathode axis in an argon–helium (Ar–He) mixture. Quantitative absorption measurements corroborated by deposition on silicon substrates are used to calibrate the TD-LIF relative measurements. Density, flux, temperature, AVDF and the flux velocity distribution function are derived from fitting the TD-LIF signals with four Gaussians (thermalized atoms) and four (energetic atom) functions (Stepanova and Dew 2004 Nucl. Instrum. Methods Phys. Res. B 215 357) taking into account the natural abundance and resonance wavelength shifts of the four main isotopes. Measurements show transport improvement for W atoms and an increase of the ratio of Ar ions to Ar neutrals with the increase of the percentage of He. All measurements are performed at 0.4 Pa and 100 W. The mean velocity of energetic W atoms typically ranges from 1900 to 2200 m s−1. The densities of thermalized and energetic atoms are in the same order of magnitude (∼109 cm−3) and their corresponding fluxes are several tens of times higher for energetic atoms (∼1015 cm−2 s−1).

Characterization of energetic and thermalized sputtered atoms in pulsed plasma using time-resolved tunable diode-laser induced fluorescence

In this work, a time-resolved tunable diode-laser (DL) induced fluorescence (TR-TDLIF) method calibrated by absorption spectroscopy has been developed in order to determine atom and flux velocity distribution functions (AVDF and FVDF) of the energetic and the thermalized atoms in pulsed plasmas. The experimental set-up includes a low-frequency (∼3 Hz) and high spectral-resolution DL (∼0.005 pm), a fast rise-time pulse generator, and a high power impulse magnetron sputtering (HiPIMS) system. The induced TR-TDLIF signal is recorded every 0.5 μs with a digital oscilloscope of a second-long trace. The technique is illustrated with determining the AVDF and the FVDF of a metastable state of the sputtered neutral tungsten atoms in the HiPIMS post-discharge. Gaussian functions describing the population of the four W isotopes were used to fit the measured TR-TDLIF signal. These distribution functions provide insight into transition from the energetic to thermalized regimes from the discharge onset. This technique may be extended with appropriate DLs to probe any species with rapidly changing AVDF and FVDF in pulsed and strongly oscillating plasmas.

Nonequilibrium radio frequency plasma interacting with a surface

In this paper, we discuss the successive results of our series of investigations into a two-frequency capacitively coupled plasma in CF4(5%)/Ar, sustained by a very high frequency source and biased by a low frequency source in a collision-dominated region. The nonequilibrium in a collisional radiofrequency plasma is studied in terms of the velocity distribution of charged particles. In the bulk plasma, especially, we discuss the velocity distribution of electrons influenced by the quantum structure of a feed gas molecule, and in the passive ion sheath, we predict the flux velocity distribution of positive ions. We then investigate the active species interacting with a patterned SiO2 wafer surface and predict each species' contribution to the feature profile, and discuss the relationship between the pattern size and the etch rate.

In situ measurement of plasma charging on SiO 2 hole bottoms and reduction by negative charge injection during etching

Charging damage in the fabrication of a micro- and nanoelectronic device is one of the electrical damages during plasma etching and caused basically by a huge difference of the flux velocity distribution between positive ions and electrons toward the wafer to be processed. Beam-like positive ions are accumulated on the bottom of a miniaturized structure during etching. With the evolution of the technology node, charging damage will increase due to several factors, increase of plasma exposure time, decrease of annealing temperature, and narrow process window, etc., caused by the increase of the number of metal layers and the introduction of new materials such as low-k and high-k instead of SiO 2. The progress of a top-down nanotechnology depends on the development of in situ diagnostics regarding plasma damage to lower-level elements and on the development of charging-free plasma process. In this paper, in situ charging measurements by using a test chip and negative charge injection to the wafer by optical computerized tomography are first demonstrated. Second, we discuss the characteristics of the charging potential on the bottom of SiO 2holes during etching in a two-frequency capacitively coupled plasma (2f-CCP), and refer to the procedure to reduce the positive potential by utilizing the negative charge acceleration to the hole bottom under the artificial formation of a double-layer close to the wafer. In addition, the charging’s effect on the aspect ratio of the hole and the antenna ratio are discussed.

Numerical Modeling of RF Magnetron Sputtering with Metallic or Dielectric Target

A self-consistent modeling of two-dimensional and temporal (2D-t) structures of RF magnetron plasma with a metallic (copper) and a dielectric (SiO2) target is performed at 5 mTorr in argon by using plasma hybrid model consisting of a particle-in-cell/Monte Carlo (PIC/MC) simulation for electrons and the relaxation continuum (RCT) model for ions. The erosion profiles of both targets are numerically predicted through the flux velocity distribution of ions incident on the surface. The mechanism of plasma maintenance in a dielectric target is quite different from that in a metallic one. Two major differences exist in an erosion profile between both targets. One is the presence of dual peaks in the erosion profile of the dielectric target. The other is the discrepancy in the position of the maxima between incident ion flux and erosion depth on the dielectric target. These are direct influences of the radially localized structure in an interfacial sheath based on the charge accumulation on a dielectric surface.

Effects of Plasma Molding on Feature Profile of Silicon Micro-Electro-Mechanical Systems through Flux Ion Velocity Distributions in Two-Frequency Capacitively Coupled Plasma in SF6/O2

We numerically investigated large-scale Si etching of several hundreds of micrometers such as that used in micro-electro-mechanical systems (MEMS) fabrication. This was carried out in SF6 (83%)/O2 at 300 mTorr in two-frequency capacitively coupled plasma using an extended VicAddress. We estimated the plasma molding's local characteristics including potential distribution and flux ion velocity distribution adjacent to an artificial microscale hole pattern. The sheath thickness is comparable to or even smaller than the size of the hole, and the sheath tends to wrap around the hole on a Si wafer. The distorted sheath field caused by the plasma molding directly affects the incident ion flux velocity distribution. The incident angle of SF5+ ions with a low energy strongly deviates from the normal surface at the hole's edge. The ion flux becomes radially nonuniform in the vicinity of the hole pattern, suppressing anisotropy of the etch profile. The feature profile of the Si hole indicates that the etching is enhanced particularly at the bottom corner and sidewall.

Modeling of dc magnetron plasma for sputtering: Transport of sputtered copper atoms

A self-consistent modeling of two-dimensional structures of a dc magnetron plasma with a copper target is performed at 5mTorr in Ar by using a hybrid model consisting of a particle-in-cell/Monte Carlo simulation for electrons and relaxation continuum model for ions. The erosion profile of the copper target is estimated by the flux velocity distribution of ions incident on the target. In the case of a dc magnetron, the discharge is mainly sustained by an E×B drift motion of energetic electrons near the position Bz∼0 with a static doughnutlike magnetic field. Then, a strongly localized profile appears in the plasma structures. The position of a maximum erosion on the copper target exactly coincides with the position where the incident Ar+ ion flux is maximum. Under the system between the light ion and the heavy target atom, the energy of the sputtered Cu atom is low (<10eV) despite the high-energy ion injection (∼200eV) to the target surface. In particular, we have developed a technique to predict the transport of sputtered particles in the gas phase within a reasonable computational time. The spatial distribution of the sputtered particles is divided into two components: “directional” fast-moving particles that do not collide with Ar feed gas in the gas phase and “random” slow-moving particles whose energy is relaxed by collision. The sputtered Cu atoms are widely dispersed from the doughnutlike region on the target. At 5mTorr, the flux of sputtered Cu atoms at the substrate is mainly affected by the random particles with relaxed energy by collision.

Laser induced desorption of NO from NiO(100): Characterization of potential energy surfaces of excited states

In order to interpret experimental results such as velocity flux distributions and rotational/vibrational populations of the state resolved UV-laser induced desorption of NO from NiO(100) ab initio calculations at the configuration interaction (CI) and complete active space self consistent field (CASSCF) levels have been performed for the electronic ground state and those excited states which are important for the desorption process. The NO NiO(100) system was described by a NiO 8− 5-cluster embedded in a Madelung field of point charges with NO adsorbed in the on-top position on the central Ni 2+ ion. Two-dimensional potential energy surfaces for several electronic states have been calculated as a function of the NNi distance and the tilt angle of NO towards the surface normal. The excited states involved in the desorption process are charge transfer states in which one electron is transferred from the oxygen 2p-shell into the NO 2π-orbitals. The dependence of the potential energy surfaces on the NNi distance is dominated by a strong Coulomb attraction between the NO − ion formed as an intermediate and the hole created within the cluster. The angular dependence of the potentials favours an upright adsorption geometry if NO − is approaching the surface. This offers an explanation of the strong coupling between translation and rotation, which has been observed experimentally for the system NO NiO(100) , as well as the absence of such a coupling in the system NO NiO(111) .

Mechanism of the UV-laser-induced desorption of CO from Cr 2O 3(0001)

The UV-laser-induced desorption of CO from an epitaxially grown film of Cr 2O 3(0001) is reported using a (1 + 1′) REMPI (resonance enhanced multiphoton ionisation) detection technique via the B 1 Σ + electronic state for the desorbing molecules to obtain information on single rovibrational states. The final state distributions at three different wavelengths (6.4, 5.0 and 3.5 eV) were almost identical showing bimodal velocity flux distributions for v″ = 0 and monomodal distributions for v″ = 1. The desorption cross-sections varied between (5.1 ± 1) × 10 −18 cm 2 at 3.5 eV and (3.5 ± 1) × 10 −17 cm 2 at 6.4 eV. The overall behaviour indicates that the excitation mechanism is due to a primary electronic excitation step within the substrate. Possible mechanisms are discussed.

Two-dimensional state resolved imaging after UV-laser induced desorption: NO/NiO(111)

We present an experimental set-up which can image state selectively the two-dimensional angular distribution of neutrals or ions desorbing from a surface after UV-laser excitation together with the velocity flux distribution of a single rovibronic state in a single experiment. The method combines a state selective detection method like resonance enhanced multiphoton ionization (REMPI) with an imaging technique. This method is particularly fast in comparison to other one-dimensional methods and ideal for systems where the substrate is modified via irradiation with laser light. The normally used set-up in state resolved laser desorption experiments for two-dimensional imaging [1] is limited to look at the azimuthal distribution. We report on results of the determination of angular distributions of NO desorbing from NiO(111) after excitation with 193 nm (6.4 eV) laser light. The velocity flux distributions also obtained are compared to former one-dimensional experiments performed in our laboratory.

Dynamical studies of UV-laser-induced NO-desorption from the polar NiO(111) versus the nonpolar NiO(100) surfaces

We have studied the UV-laser-induced desorption of NO adsorbed on an epitaxial film of NiO(111) grown on Ni(111). The desorbing molecules were detected state selectively via a resonance enhanced ionization technique [REMPI(1+1)] using the A 2Σ(v′=0,1,2)←X 2Π(v″=0,1,2) transition as intermediate state. Our results are compared with our experiments on NO desorption from NiO(100). The similarities and differences of the results due to the different surface structure of the polar NiO(111) and the non polar NiO(100) are discussed. For both surfaces we observe bimodal velocity flux distributions independent of the rovibrational state. Due to a rotational temperature of about 400 K and a vibrational temperature of 1800 K thermal processes can be ruled out. The wavelength dependence of the desorption cross section strongly correlates with the electronic structure of the NiO indicating a surface mediated excitation process. The spin orientation in the NO molecules influences the life time of the excited state depending on the magnetic property of the NiO surface.

UV laser-induced desorption of CO from Cr 2O 3(111). A fully quantum state resolved mapping of the desorbing species

The laser-induced desorption of CO adsorbed on an epitaxially grown film of Cr 2O 3(111) has been studied using 193 nm (6.4 eV) laser light. The velocity flux distributions of the desorbing neutral CO in its ground electronic state were recorded quantum state specifically by making use of a (1 + 1′) REMPI (resonance enhanced multiphoton ionization) technique. Single rovibrational states of the desorbing CO could be selected by using the (B 1Σ + ← X 1Σ +) transition in the VUV at ≈ 115 nm for the first excitation step of the REMPI process. Excellent resolution was obtained in contrast to the most commonly used (2 + 1) REMPI process for the same intermediate state. The velocity flux distributions were bimodal for ν″ = 0 and monomodal for ν″ = 1. The high desorption cross section of (3.5 ± 1) × 10 −17 cm 2 is typical for the desorption from oxide surfaces.

Time-of-flight characterization of pulsed supersonic helium free-jet expansions

A synchronous time-of-flight (TOF) technique is employed to make velocity distribution measurements in pulsed supersonic free-jet expansions. For helium expansions, the flow attains a steady–state condition with a terminal Mach number of 80, on a 50-μs time scale. The transient behavior is primarily due to the mechanical action of the valve. For a 650-μs-wide pulse, greater than 95% of the atoms issuing from the pulsed nozzle can be described by the steady-state parameters. However, the velocity dispersion increases at both the leading and trailing edges of the pulse. This dispersion can be quantitatively described by incorporating a time-dependent Mach number into the standard steady-state flux velocity distribution function.