Katkılı SiO2 filmlerin CHF3-O2 plazması içersinde reaktif iyon aşındırılması
Katkılı SiO2 filmlerin CHF3-O2 plazması içersinde reaktif iyon aşındırılması
Dosyalar
Tarih
1995
Yazarlar
Ertuğrul, Suat Ali
Süreli Yayın başlığı
Süreli Yayın ISSN
Cilt Başlığı
Yayınevi
Fen Bilimleri Enstitüsü
Özet
Bu çalışmada, değişen sistem parametrelerinin (basın, güç, gaz akış oranları ve elektrot malzemesi), film ve plazma üzerine etkileri incelenmiştir. Bunun için CHF3 - O2 plazması içersinde SİO2 filmler aşındırılmıştır. Bu filmler ile maskeleme malzemesinin aşınma hızlan ve kendinden kutuplama (de bias) gerilimi ölçülmüştür. Bu ölçümler sonucu plazma ve aşınmanın karakteri belirlenmiştir. Bu çalışmada reaktif iyon aşındırma tekniği kullanılmıştır. Denemeler sonucu; a) Artan güçle birlikte, aşınmanın hızı ve yönbağımlı karakterinin arttığı, buna karşın maskeleme malzemesinin bozulmaya başladığı, b) Artan basınçla, aşınma hızı ve yönbağımlı karakterinin azaldığı, c) Artan CHF3 akışı ile oksit aşınma hızının azaldığı, d) Artan O2 oram ile oksit aşınma hızının belli bir noktaya kadar arttığı ve daha sonra azalması ile maskeleme malzemesinin aşınma hızının arttığı, e) Anodize edilmiş Alüminyum elektrodun kuartz elektroda göre daha hızlı oksit aşınmasına neden olduğu gözlenmiştir.
Ionized gases (plasma, glow-discharge, discharge) are used in several tech niques that have been developed for processing (depositon, etching,...) semi conductor materials and there have been several books, articles on the prop erties of the glow discharges used under various conditions. These show that a change in the system parameters (e.g. pressure, gas flow rate, applied power,...) can drastically influence the conditions in the plasma. So understanding of physics behind the formation of plasmas and the effects of system parameters on the conditons of plasmas allow us to produce more controlled processes. The min purpose of this thesis were to investigate how the system parameter effect doped SİO2 and fotosensitive material (fotoresist) etch rate and self bias voltage in reactive ion etching. These measured quantities have been given the information about plasma conditions and etch mechanisms. In this thesis, the formation and physical properties of weakly ionized plasmas have been initially reviewed. Secondly, various types of plasma (dc and ac excited) have been described and the physical properties of these, which make one type more convenient than the other one for etching purposes, have been discussed. Thirdly, the mechanisms of the reactive ion etching have been described. Finally, experimental results of reactive ion etching of doped SİO2 and fotoresist have been given. A plasma is a gas containing charged and neutral species (electrons, pos itive ions, negative ions, atoms and molecules). A plasma may be generated when a gas is subject to an electric field. Natural radiation first ionises a small fraction of the gas species and the free electrons are accelerated by the field before they undergo ionising collisions with neutral gas species. This creates more free electrons to cause further ionisation thus initiating an avalanche of collisions producing a large and equal number of electrons and ions. The discharge can be sustained by dc or ac supplies, but it is more usual to use rf excitation because of the charging that occurs on dielectric films when etching in dc plasma. The charging of the insulator-covered electrode causes extinguishing of the discharge. This represents a significiant limitation, as there are several important aplications which call for the sputtering or etching of insulators^ or the maintenance of a continuous discharge in the face of dielectric-covered electrodes (e.g the reactive ion etching of SİO2 have been vni done on powered and a quartz covered electrot in this work). The application of an ac voltage overcomes the problem. If the low frequency ac signal is applied, the system behave like a double-ended dc discharge with similar limitations. Because charging time of the insulator is less than the ac frequency. If the ac frequency is increased to the point where the charging time is much longer than the ac period, contiuous discharge will maintain. A frequency of about 50-100 kHz is ussually sufficient to achieve this condition. But most of the commercial reactive ion etch systems are used 13.56 MHz. The other reason is the ionisation which is more efficient in radio frequency than in dc discharges. The explanation is that some electron-molecule elastic collisions are taking place at an appropriate time with respect to the phase of the electric field such that the energy of the electrons continues to increase. As a consequence it becomes easier for the electrons to reach the energy required for ionisation of the gas. This causes to decrease the minimum operation pressure which is convenient clean semiconductor processes. Because of the large difference in mass between an electron and an atom or molecule the electron can only lose energy to a molecule's internal energy levels; it cannot increase its kinetic energy directly since the momentum trans fer is very inefficient. This gives rise to two leading plasma characteristics. The collisions produce a wide range of exited gas species which lose or redis tribute energy via radiation, dissociation or collision to give a further mixture of excited or ground state species. It is these species which can then give rise to etching or deposition. The second effect is the absence of thermodynamic equilibrium between the gas and the free electrons. Typically the electron energy can be charecterized by a temperature of 30000 K while the ions and molecules can be very near room temparature. In reactive ion etching system rf power is applied to the lower electrode while the larger upper electrode is grounded. Due to the rapidity of alternating voltage and to the greater mobility of the electrons the smaller driven electrode has a negative voltage for more than half cycle. The magnitude of this dc bias (self bias) depend on the cathode-anode area ratio (3:1 in our case) as the net current must be zero for each cycle. Thus for the majority of the rf cycle ions are accelerated towards the cathode. There are mainly three regions in a rf plasma. These are cathode dark space, anode dark space and glow region. The cathode dark space which is in front of the small driven electrode, has relatively few electrons which, in general, have not had the necessary ac celeration distance to gain enough energy to electronically excite gas particles. There is a large voltage drop across this dark space causing accelaretion of ions towards the cathode the resulting sputtering of substrate atoms. ix The central glow region is the true plasma described above and is under the influence of only a small electric field due to the screening effect of the cathode dark space. Because of the high mobility of the electrons the plasma potential lies slightly above the anode potential. The densitiy of ions and electrons is about 1010cm-3 giving the fractional degree of ionisation of 50 mTorr of order 6 x 10~5. The plasma is very dilute. The anode dark space sheath is thinner than the cathode region and es sentially collisionless. There is a small voltage drop across it. A material may be etched by either purely chemical means, as in wet etch ing, by purely physical bombardment, as in inert gas sputtering or combination of chemical reaction and physical sputtering as in reactive ion etching. Wet etching processes are typically isotropic. Therefore, if the thickness of the film being etched is comparable to the minimum pattern dimension, undercutting due to isotropic etching, becomes intolerable. Since many films used in semiconductor device processes are 0.5 - ljj,m thick, reproducible and controllable transfer of paterns in 1 - 5fj.m range (in our case 3[im) becomes difficult. Dry etching processes must therefore be applied to produce semicon ductor devices with such dimensions. In the physical sputtering processes, the strongly directional nature of the incident energetic ions allows substrate material to be removed in a highly anisotropic manner. Unfortunately, such material removal mechanisms are also quite non-selective against both masking material and materials under lying the layers being etched. Furthermore, since the ejected species are not inherently volatile, redeposition can occur. Another major problem of pattern transfer by physical sputtering involves the redepositon of nonvolatile species on the sidewalls of the etched feature. As a result of these drawbacks, dry etch processes for pattern transfer based on phsical removal mechanisms have not found wide use in semiconductor device production. The basic concept of reactive ion etching is rather direct. A glow discharge is utilized to produce chemically active species (atoms, radicals, and ions) from relatively inert molecular gas. The etching gas is selected so as to generate species which react chemically with the material to be etched, and whose reaction product with the etched material is volatile. The following processes take place in the system during ion enhanced etching: 1) Active species generation: In Reactive ion etch a glow discharge is used to generate from a suitable feed gas by electron-impact dissociation, ioniza tion the gas phase etching environment which consists of radical, positive and negative ions, electrons, and neutrals. 2) Formation of a dc bias for ion acceleration: The material to be etched is placed on a high-frequency-driven capacitatively caupled electrode. Since the electron mobility is much gerater than the ion mobility, after ignition of the plasma the electrode acquires a negative charge (self-bias voltage). Therefore, the electrode and material placed on the electrode will be exposed to energetic, positive ion bombardment. 3) Transport of plasma-generated reactive species from the bulk of the plasma to the surface of the material being etched: This occurs by dif fusion which, for particular structures such as narrow deep trenchs, can limit the etch rate. 4) Adsorption step: Reactive radicals adsorb on the surface of the mate rial to be etched. This step can be strongly enhanced by concurent ion bombardment which serves to produce "active sites" 5) Reaction step: A reaction between the adsorbed species and the material to be etched must take place, because of the plasma-induced formation of reactive radicals, the reaction rate is very large relative to reaction rates in non-plasma enviroments. The reaction step can be greatly enhanced by ion bombardment. 6) Desorption of volatile reaction product: The desorption of the reaction product into the gas phase is one of the most critical steps in the over all etching reaction. This requires that the reaction product has a high vapor pressure at the substrate temperature. The removal of reaction product from the surface can be greatly accelerated by ion bombardment via sputtering. 7) Pumpout of volatile reaction product: This requires that the desorbed species diffuse from the etching surface into the bulk of the plasma and are pumped out. Otherwise plasma induced dissociation of product molecules will occur and redeposition can take place. Reactive ion etching rely on the conversion of the material into some volatile species which can therefore be removed from the surface. In the case of SİO2 this involves surface adsorption of some HF containing plasma species and chemical recation to give SİF4 followed by desorption of this product. If this were the only mechanisms occuring the etching would be totaly isotropic and would have few advantages over the use of liquids. However, there is also ion bombardment perpendicular to the surface which causes not only directionality but also higher etch rates. Two principal mechanisms have been postulated to be operative in directional etching processes. Firstly, ion bombardment may influence certain surface chemical reactions. Effect which is believed to be important in reactive ion etching systems is the lattice damage caused by the high energy (> 50 eV) ions. This can give rise to active sites extending several monolayers below the surface at which reaction can proceed at an accelerated rate. The second model is the formation etch inhibiting layer (non-volatile polymer layers) which is removed from the horizontal surfaces by sputtering but remains on the sidewalls as a protecting film. This adsorbed layer prevents lateral etching either by reacting with the etchant species or by simply denying etchant species access to the substrate surface. In this work, CHF3 - O2 gases have been used for reactive ion etching of XI doped SİO2 (BPSG). Three sets of experiment have been set up to investigate the effects of pressure, power, gas flow rate and electrode material on plasma conditions. The main criteria taken into account were etch rates of SİO2 and fotoresist, and dc bias to understand the effects of the physical paremeters of the system. These experiments were: 1) The effects of varying pressure and power on etch rates and dc bias under constant gas flow rate, 2) The effects of varying gas flow rates on etch rates and dc bias under constant pressure and power, 3) The effects of electrode materials on etch rates and dc bias. First observation was the etch rates and dc bias have been increased with power at constant pressure. The first reason of that is, the increased number of electrons increases the number of reactive species avalibale for etching. The second one is, the increased dc bias voltage raises the ion bombardment energy thus enhancing the physical component of the etching. This increased ion energy has two other important effects: It usually raises the uniformity in etch rate and it may also have the disadvantageous effect of reducing selectivity to the masking photoresist and to underlying material. Second observation was the etch rates and dc bias have been decreased with increasing pressure at constant power. A rise in process pressure causes a rise in the number of reactive species giving an increased chemical attact. This enhance the chemical component of the etching process. So, the isotropic chracter of etching is increased. And, a decrease in the bias voltage reducing the sputtering mechanisms. Third observation was the etch rates have been decreased after a maximum (which was not observed) with increasing CHF3 gas flow rate at constant power and pressure. The increasing CHF3 gas flow rate causes to form more thinner polymer film which is limited the etch rates. But these polymer film enhaced the unisotropic character of etching. Fourth observation was the etch rate of SİO2 have been decreased after a maximum with increasing O2 gas flow rate at constant power and pressure. But photoresist etch rate is increased with increasing O2 gas flow rate. Fifth observation was the etch rates have been increased when anodized Al used as an electrode material instead of quartz electrode.
Ionized gases (plasma, glow-discharge, discharge) are used in several tech niques that have been developed for processing (depositon, etching,...) semi conductor materials and there have been several books, articles on the prop erties of the glow discharges used under various conditions. These show that a change in the system parameters (e.g. pressure, gas flow rate, applied power,...) can drastically influence the conditions in the plasma. So understanding of physics behind the formation of plasmas and the effects of system parameters on the conditons of plasmas allow us to produce more controlled processes. The min purpose of this thesis were to investigate how the system parameter effect doped SİO2 and fotosensitive material (fotoresist) etch rate and self bias voltage in reactive ion etching. These measured quantities have been given the information about plasma conditions and etch mechanisms. In this thesis, the formation and physical properties of weakly ionized plasmas have been initially reviewed. Secondly, various types of plasma (dc and ac excited) have been described and the physical properties of these, which make one type more convenient than the other one for etching purposes, have been discussed. Thirdly, the mechanisms of the reactive ion etching have been described. Finally, experimental results of reactive ion etching of doped SİO2 and fotoresist have been given. A plasma is a gas containing charged and neutral species (electrons, pos itive ions, negative ions, atoms and molecules). A plasma may be generated when a gas is subject to an electric field. Natural radiation first ionises a small fraction of the gas species and the free electrons are accelerated by the field before they undergo ionising collisions with neutral gas species. This creates more free electrons to cause further ionisation thus initiating an avalanche of collisions producing a large and equal number of electrons and ions. The discharge can be sustained by dc or ac supplies, but it is more usual to use rf excitation because of the charging that occurs on dielectric films when etching in dc plasma. The charging of the insulator-covered electrode causes extinguishing of the discharge. This represents a significiant limitation, as there are several important aplications which call for the sputtering or etching of insulators^ or the maintenance of a continuous discharge in the face of dielectric-covered electrodes (e.g the reactive ion etching of SİO2 have been vni done on powered and a quartz covered electrot in this work). The application of an ac voltage overcomes the problem. If the low frequency ac signal is applied, the system behave like a double-ended dc discharge with similar limitations. Because charging time of the insulator is less than the ac frequency. If the ac frequency is increased to the point where the charging time is much longer than the ac period, contiuous discharge will maintain. A frequency of about 50-100 kHz is ussually sufficient to achieve this condition. But most of the commercial reactive ion etch systems are used 13.56 MHz. The other reason is the ionisation which is more efficient in radio frequency than in dc discharges. The explanation is that some electron-molecule elastic collisions are taking place at an appropriate time with respect to the phase of the electric field such that the energy of the electrons continues to increase. As a consequence it becomes easier for the electrons to reach the energy required for ionisation of the gas. This causes to decrease the minimum operation pressure which is convenient clean semiconductor processes. Because of the large difference in mass between an electron and an atom or molecule the electron can only lose energy to a molecule's internal energy levels; it cannot increase its kinetic energy directly since the momentum trans fer is very inefficient. This gives rise to two leading plasma characteristics. The collisions produce a wide range of exited gas species which lose or redis tribute energy via radiation, dissociation or collision to give a further mixture of excited or ground state species. It is these species which can then give rise to etching or deposition. The second effect is the absence of thermodynamic equilibrium between the gas and the free electrons. Typically the electron energy can be charecterized by a temperature of 30000 K while the ions and molecules can be very near room temparature. In reactive ion etching system rf power is applied to the lower electrode while the larger upper electrode is grounded. Due to the rapidity of alternating voltage and to the greater mobility of the electrons the smaller driven electrode has a negative voltage for more than half cycle. The magnitude of this dc bias (self bias) depend on the cathode-anode area ratio (3:1 in our case) as the net current must be zero for each cycle. Thus for the majority of the rf cycle ions are accelerated towards the cathode. There are mainly three regions in a rf plasma. These are cathode dark space, anode dark space and glow region. The cathode dark space which is in front of the small driven electrode, has relatively few electrons which, in general, have not had the necessary ac celeration distance to gain enough energy to electronically excite gas particles. There is a large voltage drop across this dark space causing accelaretion of ions towards the cathode the resulting sputtering of substrate atoms. ix The central glow region is the true plasma described above and is under the influence of only a small electric field due to the screening effect of the cathode dark space. Because of the high mobility of the electrons the plasma potential lies slightly above the anode potential. The densitiy of ions and electrons is about 1010cm-3 giving the fractional degree of ionisation of 50 mTorr of order 6 x 10~5. The plasma is very dilute. The anode dark space sheath is thinner than the cathode region and es sentially collisionless. There is a small voltage drop across it. A material may be etched by either purely chemical means, as in wet etch ing, by purely physical bombardment, as in inert gas sputtering or combination of chemical reaction and physical sputtering as in reactive ion etching. Wet etching processes are typically isotropic. Therefore, if the thickness of the film being etched is comparable to the minimum pattern dimension, undercutting due to isotropic etching, becomes intolerable. Since many films used in semiconductor device processes are 0.5 - ljj,m thick, reproducible and controllable transfer of paterns in 1 - 5fj.m range (in our case 3[im) becomes difficult. Dry etching processes must therefore be applied to produce semicon ductor devices with such dimensions. In the physical sputtering processes, the strongly directional nature of the incident energetic ions allows substrate material to be removed in a highly anisotropic manner. Unfortunately, such material removal mechanisms are also quite non-selective against both masking material and materials under lying the layers being etched. Furthermore, since the ejected species are not inherently volatile, redeposition can occur. Another major problem of pattern transfer by physical sputtering involves the redepositon of nonvolatile species on the sidewalls of the etched feature. As a result of these drawbacks, dry etch processes for pattern transfer based on phsical removal mechanisms have not found wide use in semiconductor device production. The basic concept of reactive ion etching is rather direct. A glow discharge is utilized to produce chemically active species (atoms, radicals, and ions) from relatively inert molecular gas. The etching gas is selected so as to generate species which react chemically with the material to be etched, and whose reaction product with the etched material is volatile. The following processes take place in the system during ion enhanced etching: 1) Active species generation: In Reactive ion etch a glow discharge is used to generate from a suitable feed gas by electron-impact dissociation, ioniza tion the gas phase etching environment which consists of radical, positive and negative ions, electrons, and neutrals. 2) Formation of a dc bias for ion acceleration: The material to be etched is placed on a high-frequency-driven capacitatively caupled electrode. Since the electron mobility is much gerater than the ion mobility, after ignition of the plasma the electrode acquires a negative charge (self-bias voltage). Therefore, the electrode and material placed on the electrode will be exposed to energetic, positive ion bombardment. 3) Transport of plasma-generated reactive species from the bulk of the plasma to the surface of the material being etched: This occurs by dif fusion which, for particular structures such as narrow deep trenchs, can limit the etch rate. 4) Adsorption step: Reactive radicals adsorb on the surface of the mate rial to be etched. This step can be strongly enhanced by concurent ion bombardment which serves to produce "active sites" 5) Reaction step: A reaction between the adsorbed species and the material to be etched must take place, because of the plasma-induced formation of reactive radicals, the reaction rate is very large relative to reaction rates in non-plasma enviroments. The reaction step can be greatly enhanced by ion bombardment. 6) Desorption of volatile reaction product: The desorption of the reaction product into the gas phase is one of the most critical steps in the over all etching reaction. This requires that the reaction product has a high vapor pressure at the substrate temperature. The removal of reaction product from the surface can be greatly accelerated by ion bombardment via sputtering. 7) Pumpout of volatile reaction product: This requires that the desorbed species diffuse from the etching surface into the bulk of the plasma and are pumped out. Otherwise plasma induced dissociation of product molecules will occur and redeposition can take place. Reactive ion etching rely on the conversion of the material into some volatile species which can therefore be removed from the surface. In the case of SİO2 this involves surface adsorption of some HF containing plasma species and chemical recation to give SİF4 followed by desorption of this product. If this were the only mechanisms occuring the etching would be totaly isotropic and would have few advantages over the use of liquids. However, there is also ion bombardment perpendicular to the surface which causes not only directionality but also higher etch rates. Two principal mechanisms have been postulated to be operative in directional etching processes. Firstly, ion bombardment may influence certain surface chemical reactions. Effect which is believed to be important in reactive ion etching systems is the lattice damage caused by the high energy (> 50 eV) ions. This can give rise to active sites extending several monolayers below the surface at which reaction can proceed at an accelerated rate. The second model is the formation etch inhibiting layer (non-volatile polymer layers) which is removed from the horizontal surfaces by sputtering but remains on the sidewalls as a protecting film. This adsorbed layer prevents lateral etching either by reacting with the etchant species or by simply denying etchant species access to the substrate surface. In this work, CHF3 - O2 gases have been used for reactive ion etching of XI doped SİO2 (BPSG). Three sets of experiment have been set up to investigate the effects of pressure, power, gas flow rate and electrode material on plasma conditions. The main criteria taken into account were etch rates of SİO2 and fotoresist, and dc bias to understand the effects of the physical paremeters of the system. These experiments were: 1) The effects of varying pressure and power on etch rates and dc bias under constant gas flow rate, 2) The effects of varying gas flow rates on etch rates and dc bias under constant pressure and power, 3) The effects of electrode materials on etch rates and dc bias. First observation was the etch rates and dc bias have been increased with power at constant pressure. The first reason of that is, the increased number of electrons increases the number of reactive species avalibale for etching. The second one is, the increased dc bias voltage raises the ion bombardment energy thus enhancing the physical component of the etching. This increased ion energy has two other important effects: It usually raises the uniformity in etch rate and it may also have the disadvantageous effect of reducing selectivity to the masking photoresist and to underlying material. Second observation was the etch rates and dc bias have been decreased with increasing pressure at constant power. A rise in process pressure causes a rise in the number of reactive species giving an increased chemical attact. This enhance the chemical component of the etching process. So, the isotropic chracter of etching is increased. And, a decrease in the bias voltage reducing the sputtering mechanisms. Third observation was the etch rates have been decreased after a maximum (which was not observed) with increasing CHF3 gas flow rate at constant power and pressure. The increasing CHF3 gas flow rate causes to form more thinner polymer film which is limited the etch rates. But these polymer film enhaced the unisotropic character of etching. Fourth observation was the etch rate of SİO2 have been decreased after a maximum with increasing O2 gas flow rate at constant power and pressure. But photoresist etch rate is increased with increasing O2 gas flow rate. Fifth observation was the etch rates have been increased when anodized Al used as an electrode material instead of quartz electrode.
Açıklama
Tez (Yüksek Lisans) -- İstanbul Teknik Üniversitesi, Fen Bilimleri Enstitüsü, 1995
Anahtar kelimeler
Aşınma,
Film,
Plazma,
İyon,
Wear,
Film,
Plasma,
Ion