Silicon has dominated semiconductor history for almost half a century. Even if the first transistor was made using germanium and other semiconductors show better electronic properties in terms of higher mobility, higher saturation velocity, and larger energy gap. It is the material used to build almost the 97% of all semiconductor-based electronic de- vices. The reason is straightforward; it is the most economical technology to make inte- grated circuits. It has been possible to fabricate integrated circuits with constantly increasing number of transistors on a single chip. The first that analyzed this trend was Gordon Moore in 1965, and he suggested that the trend was due to a constant exercise in cost reduction. The manufacturing cost for a square millimeter of Si remained constant at about 1$ for many decades, while the number of transistors and other elements has grown exponentially with time. The number of transistors for units of area that can be placed on a semiconductor wafer depends on the capacity of the wafer to dissipate such thermal energy. Materials with elevated thermal conducibility and melting temperature, like silicon, are the ideal ones. The low leakage currents that can be achieved with Si oxide and Si nitrate (its native composite with air) and the elevated thermal conducibility allowed a transistor density higher than with other semiconduc- tors. Even if Si still dominates the main branch of semiconductor technology, there are areas where the low mobility, the low saturation velocity, and the indirect bandgap per- mitted other semiconductors to develop. Being the raw material cost much higher than Si one, some techniques have been developed to reduce such costs. A way to maintain the thermal advantage of a silicon substrate and reduce the raw material cost of semi- conductors that are not Si consists of using a thin semiconductor film grown on top of a thick Si substrate (hetero-epitaxy). Hetero-epitaxy guarantees lower costs maintaining a thermally efficient substrate, but on the other side implies that the semiconductor has to be grown on a substrate with a different lattice parameter. Today’s more diffused epitaxial technologies are the epi-grow of group III-V and II-VI alloys (especially for radio frequency amplification and laser technologies) and the epitaxy of group IV semiconductors like the SiC and SiGe alloys and the pure Ge. Group IV semiconductors have the non-negligible advantage of having cheaper fabrication costs than the formers. They, of course, guarantee better performances in terms of electronic properties than Si. The main obstacle in realizing such devices stems from the fact that the lattice mismatch induces the formation of detrimental defects during epitaxial growth. Such defects hinder the possibility of an industrial, extensive appli- cation of group IV semiconductors other than Si. Typically, the generated defects are grain boundaries, misfit dislocations, stacking faults, and other extended defects. A strong academic interest exists in the comprehension of defects in epitaxial systems. In particular, two systems present at the same time exciting application perspectives and hard theoretical modeling challenges: silicon germanium and cubic silicon carbide. They indeed are attractive semiconductors that can be easily integrated into the actual silicon-based architecture. Their epitaxy on Si has been studied for years, but defect densities are still too high in those systems. In this Thesis, we will deal with the problem of modeling extended defects evolution via molecular dynamics simulations. We will tackle some of the open problems about defect evolution in both the materials under consideration. Our results provided enough information to shed light on the specific problems of the formation and evolution of multiple stacking faults in cubic SiC and the formation of ordered arrays of Lomer dislocation in Ge grown on Si.
Il numero di transistor per unità di area che possono essere posizionati su un wafer a semiconduttore dipende dalla capacità del wafer di dissipare tale energia termica. I materiali con elevata conducibilità termica e temperatura di fusione, come il silicio, sono quelli ideali. Le basse correnti di dispersione che si possono ottenere con l'ossido di Si e il nitrato di Si (i suoi composti nativi con l'aria) e l'elevata conducibilità termica hanno consentito una densità di transistor superiore rispetto ad altri semiconduttori. Anche se il silicio domina ancora il ramo principale della tecnologia dei semiconduttori, ci sono aree in cui la bassa mobilità, la bassa velocità di saturazione e il bandgap indiretto hanno permesso lo sviluppo di altri semiconduttori. Essendo il costo della materia prima molto superiore a quello del Si, sono state sviluppate alcune tecniche per ridurre tali costi. Un modo per mantenere il vantaggio termico di un substrato di silicio e ridurre il costo della materia prima dei semiconduttori che non sono Si consiste nell'utilizzare un sottile film di semiconduttore cresciuto sopra un substrato di Si spesso (etero-epitassia). L'eteroepitassia garantisce costi inferiori mantenendo un substrato termicamente efficiente, ma d'altro canto implica che il semiconduttore deve essere cresciuto su un substrato con un parametro reticolare diverso. Le tecnologie epitassiali oggi più diffuse sono l'epi-crescita delle leghe di gruppo III-V e II-VI (soprattutto per l'amplificazione a radiofrequenza e le tecnologie laser) e l'epitassia di semiconduttori di gruppo IV come le leghe SiC e SiGe e il Ge puro. I semiconduttori del gruppo IV hanno il vantaggio non trascurabile di avere costi di fabbricazione inferiori rispetto ai primi. Ovviamente garantiscono prestazioni migliori in termini di proprietà elettroniche rispetto al Si. Il principale ostacolo alla realizzazione di tali dispositivi deriva dal fatto che il disadattamento reticolare induce la formazione di difetti dannosi durante la crescita epitassiale. Tali difetti ostacolano la possibilità di un'applicazione industriale ed estensiva di semiconduttori del gruppo IV diversi dal Si. In genere, i difetti generati sono bordi di grano, dislocazioni, errori di impilamento (stacking fault) e altri difetti estesi. Esiste un forte interesse accademico nella comprensione dei difetti nei sistemi epitassiali. In particolare, due sistemi presentano allo stesso tempo interessanti prospettive applicative e difficili sfide di modellazione teorica: il silicio-germanio e carburo di silicio cubico. Sono infatti semiconduttori attraenti che possono essere facilmente integrati nell'attuale architettura basata su silicio. La loro epitassia su Si è stata studiata per anni, ma le densità dei difetti sono ancora troppo alte in quei sistemi. In questa tesi, affronteremo il problema della modellazione dell'evoluzione di difetti estesi tramite simulazioni di dinamica molecolare. Affronteremo alcuni dei problemi aperti sull'evoluzione dei difetti in entrambi i materiali in esame. I nostri risultati forniranno informazioni sufficienti per far luce sui problemi specifici della formazione e dell'evoluzione di stacking fault multipli in SiC cubico e la formazione di array ordinati di dislocazioni di Lomer in Ge cresciuto su Si.
(2022). Extended defects in heteroepitaxial structures on silicon by Molecular Dynamics simulations: applications to SiGe and cubic SiC. (Tesi di dottorato, Università degli Studi di Milano-Bicocca, 2022).
Extended defects in heteroepitaxial structures on silicon by Molecular Dynamics simulations: applications to SiGe and cubic SiC
BARBISAN, LUCA
2022
Abstract
Silicon has dominated semiconductor history for almost half a century. Even if the first transistor was made using germanium and other semiconductors show better electronic properties in terms of higher mobility, higher saturation velocity, and larger energy gap. It is the material used to build almost the 97% of all semiconductor-based electronic de- vices. The reason is straightforward; it is the most economical technology to make inte- grated circuits. It has been possible to fabricate integrated circuits with constantly increasing number of transistors on a single chip. The first that analyzed this trend was Gordon Moore in 1965, and he suggested that the trend was due to a constant exercise in cost reduction. The manufacturing cost for a square millimeter of Si remained constant at about 1$ for many decades, while the number of transistors and other elements has grown exponentially with time. The number of transistors for units of area that can be placed on a semiconductor wafer depends on the capacity of the wafer to dissipate such thermal energy. Materials with elevated thermal conducibility and melting temperature, like silicon, are the ideal ones. The low leakage currents that can be achieved with Si oxide and Si nitrate (its native composite with air) and the elevated thermal conducibility allowed a transistor density higher than with other semiconduc- tors. Even if Si still dominates the main branch of semiconductor technology, there are areas where the low mobility, the low saturation velocity, and the indirect bandgap per- mitted other semiconductors to develop. Being the raw material cost much higher than Si one, some techniques have been developed to reduce such costs. A way to maintain the thermal advantage of a silicon substrate and reduce the raw material cost of semi- conductors that are not Si consists of using a thin semiconductor film grown on top of a thick Si substrate (hetero-epitaxy). Hetero-epitaxy guarantees lower costs maintaining a thermally efficient substrate, but on the other side implies that the semiconductor has to be grown on a substrate with a different lattice parameter. Today’s more diffused epitaxial technologies are the epi-grow of group III-V and II-VI alloys (especially for radio frequency amplification and laser technologies) and the epitaxy of group IV semiconductors like the SiC and SiGe alloys and the pure Ge. Group IV semiconductors have the non-negligible advantage of having cheaper fabrication costs than the formers. They, of course, guarantee better performances in terms of electronic properties than Si. The main obstacle in realizing such devices stems from the fact that the lattice mismatch induces the formation of detrimental defects during epitaxial growth. Such defects hinder the possibility of an industrial, extensive appli- cation of group IV semiconductors other than Si. Typically, the generated defects are grain boundaries, misfit dislocations, stacking faults, and other extended defects. A strong academic interest exists in the comprehension of defects in epitaxial systems. In particular, two systems present at the same time exciting application perspectives and hard theoretical modeling challenges: silicon germanium and cubic silicon carbide. They indeed are attractive semiconductors that can be easily integrated into the actual silicon-based architecture. Their epitaxy on Si has been studied for years, but defect densities are still too high in those systems. In this Thesis, we will deal with the problem of modeling extended defects evolution via molecular dynamics simulations. We will tackle some of the open problems about defect evolution in both the materials under consideration. Our results provided enough information to shed light on the specific problems of the formation and evolution of multiple stacking faults in cubic SiC and the formation of ordered arrays of Lomer dislocation in Ge grown on Si.File | Dimensione | Formato | |
---|---|---|---|
phd_unimib_726289.pdf
accesso aperto
Descrizione: Tesi
Tipologia di allegato:
Doctoral thesis
Dimensione
8.87 MB
Formato
Adobe PDF
|
8.87 MB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.