Neural network modelling of the inductively coupled RF plasma synthesis of silicon nanoparticles

Leparoux, Marc; Loher, Marcel; Schreuders, Cornelis; Siegmann, S.

doi:10.1016/j.powtec.2007.10.004

Cited by 17 publications

(9 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[17,18] Leparoux et al even showed that it has a more pronounced influence on the SSA than the quenching gas flow rate. [19] At low powder feed rates, the energy transfer from the plasma to the inlet particles during the evaporation stage is enhanced. Moreover, due to a low particle density in the plasma, fewer collisions are occurring between the particles leading to a reduced growth.…”

Section: Resultsmentioning

confidence: 99%

Improved Plasma Synthesis of Si‐nanopowders by Quenching

2008

View full text Add to dashboard Cite

Silicon nanopowders are investigated intensively for application in microelectronics and energy among them. [1][2][3] However, for the moment the control of the product high quality and the processing costs are limiting their breakthrough. RF-thermal plasmas could respond to both criteria as it is a continuous process achieving high production volume while ensuring a good synthesis control from the gaseous phase. The absence of electrodes for igniting the plasma and the controlled process atmosphere of RF-plasmas are beneficial for producing high purity materials. Typically a supersaturated gaseous phase containing the precursor is condensed rapidly after nucleation took place. The nanoparticles grow then subsequently by coagulation and coalescence. [4] This rapid condensation is achieved either by natural thermal gradients in the plasma or by quenching using a cold source (gas or cold surface) or by using an expansion. [5][6][7][8] This paper addresses the development of a specific quenching nozzle design combining rapid cooling with an expansion for controlling the size of silicon nanopowders processed by an inductively coupled plasma (ICP). The design of the quenching device has been supported by computational fluid dynamic (CFD) calculations aiming at modelling the plasma properties like among them the temperature and the velocity of the powderfree plasma. The modelling has been validated by in-situ plasma characterization using an enthalpy-probe coupled to a mass spectrometer. The produced nanopowders were collected either on a filter membrane, or directly in the gas phase using an on-line and in-situ sampling system made of a TEM-grid fixed on a moveable support and then ex-situ characterized by XRD, Raman, microscopy and surface specific area measurement using the BET technique. Experimental Set-up and Plasma CharacterizationThe ICP apparatus devoted to the synthesis of nanoparticles has been already described in details elsewhere. [9] Its dimensions and its geometry have been introduced in the CFD modelling code. Plasma gases typically used for processing are argon and hydrogen. This reducing atmosphere is generally used to ensure oxide-free particles. The plasma torch (PL-35, Tekna) has been also modelled with the different gas inlets; carrier gas (Ar), central gas (Ar) and the sheath gas that is a mixture of Ar and H 2 . The gas composition is referred in the following as carrier gas-central gas-sheath gas (Ar/H 2 ) in slpm. An enthalpy-probe (ENT-476, Tekna) has been used to measure in-situ the enthalpy and the velocity of the plasma in powder-free conditions at different radial positions and heights in the reactor (z=0 corresponds to the plasma torch exhaust). The plasma properties have been calculated outside the electromagnetic field, meaning below the torch exhaust where the quenching should be performed. Therefore, in a first approximation, the plasma process has been considered as a steady state process with the plasma being modelled as a heat source with a homogeneous power distribution located ...

show abstract

Section: Resultsmentioning

confidence: 99%

Improved Plasma Synthesis of Si‐nanopowders by Quenching

2008

View full text Add to dashboard Cite

show abstract

“…For particle sizes of 4 in., it is observed that 300 and 350 iterations provide best prediction accuracy and increasing iterations does not improve prediction of the TDNN model. This demonstrates that the TDNN model is affected by the network parameters [30] and the data used [31].…”

Section: Resultsmentioning

confidence: 96%

Time delay neural network modeling for particle size in SAG mills

Shang

2011

Powder Technology

View full text Add to dashboard Cite

“…The SSA of synthesized silicon nanopowder was analyzed by BET (Belsorp‐max, Bel Inc., Japan). From the BET measurement, the mean particle size of the silicon nanopowder ( D eq ) was evaluated as follows :

D_{eq} = \frac{6}{SSA \cdot ρ},

where the density of synthesized silicon, ρ , is assumed the same as that of bulk silicon at 2.33.…”

Section: Methodsmentioning

confidence: 99%

Synthesis of silicon nanopowder from silane gas by RF thermal plasma

Lee

Kim

et al. 2013

Physica Status Solidi (a)

View full text Add to dashboard Cite

Silicon nanopowder that can be applied as a selective emitter in solar cells was prepared from the decomposition of silane (SiH4) gas by a radio‐frequency (RF) thermal plasma. The RF thermal plasma offers a high‐temperature and contamination‐free environment to produce pure silicon material from SiH4. A steep temperature gradient in the thermal plasma enables the rapid quenching of decomposed material to synthesize nanosized particles. In this experiment, the input power of RF thermal plasma was controlled from 8 to 12 kW. The raw material of SiH4 gas was injected into an argon thermal plasma plume with argon carrier gas. Argon was also used as quenching gas to suppress the growth of silicon nanoparticles. In addition to the input power, the flow rates of carrier and quenching gases were employed as operating variables. The flow rates of carrier and quenching gases were controlled from 15 to 45 L min−1 and from 50 to 250 L min−1, respectively. Polycrystalline spherical silicon nanoparticles that were less than 100 nm in diameter were observed by transmission electron microscopy, Brunauer–Emmett–Teller (BET) surface area measurement, X‐ray diffractometry, and Raman spectrometry. Among the operating variables, the flow rate of carrier gas showed the most significant effect on the size of the nanopowder. The smallest mean particle size of 36 nm was obtained from the highest carrier gas flow rate of 45 L min−1, because the growth of nanopowder was limited by the enhanced axial velocity and quenching rate of in‐flight silicon nanopowder.

show abstract

Neural network modelling of the inductively coupled RF plasma synthesis of silicon nanoparticles

Cited by 17 publications

References 17 publications

Improved Plasma Synthesis of Si‐nanopowders by Quenching

Improved Plasma Synthesis of Si‐nanopowders by Quenching

Time delay neural network modeling for particle size in SAG mills

Synthesis of silicon nanopowder from silane gas by RF thermal plasma

Contact Info

Product

Resources

About