Üretim parametrelerinin, silisyum tek kristalinin özellikleri üzerine etkisi

Abstract

ÖZET Elektronik endüstrisinde, entegre devre ve güneş pili üretiminde, yarı iletken taban malzemesi olarak kullanılan silisyum tek kristallerinin üretim parametreleri elde edi len kristal özelliklerini değiştirmekte ve bu özellikler ise direk olarak entegre devrelerin kullanıldığı cihazları ve güneş pili verimlerini önemli ölçüde etkilemektedir. Ci haz üretiminde kullanılacak olan silisyum tek kristalinin dislokasyonsuz olması, içerisindeki doping elementinin homo jen olarak dağılması ve değişken olmayan özdirenç değerleri ne sahip olması arzu edilmektedir. Yapılan çalışmada, sıvıdan kristal büyütme metodlarından özellikle yarı iletken teknolojisinde büyük çaplarda kristal elde etmek için kullanılan Czochra'lski [1] yöntemi seçilmiş ve yapılan kristal büyütme deneyleri sonucu elde edilen kris taller üzerinde yapılan incelemelerde, silisyum tek kristali nin büyüme şartları, tek kristallikten çok kristalli hale ge çiş şartları bulunmuştur. üretim parametrelerinin kristal özeliklerine olan tesirlerini incelemek üzere, kesilen kris taller üzerinde radyal ve eksenel doğrultuda dislokasyon yo ğunluğu ve özdirenç ölçümleri yapılmıştır. Genel olarak tek kristal elde edebilmek için çekme hızı ve empürite konsantrasyonu belirli değerlerin üzerine çıka mamakta ve düşük empüriteye sahip sıvıdan daha yüksek hızda kristal büyütülebilmektedir. Dislokasyon yoğunluklarının incelenmesinde, kristal merkezinde daha az dislokasyona rast lanırken cidarlara doğru gidildikçe dislokasyon yoğunlukları nın arttığı görülmüştür. Büyütülen kristallerde homojen öz direnç değişimini sağlayan büyütme parametreleri bulunmuş ve günümüzde entegre devre üretiminde kullanılan dislokasyonsuz ve homojen özdirenç değişimine sahip silisyum tek kristalle rinin üretimi için gerekli koşullar ortaya konulmuştur. ıx

SUMMARY THE EFFECT OF GROWTH PARAMETERS ON THE PROPERTIES OF SILICON SINGLE CRYSTAL Integrated circuits and solar cells have been developed almost exclusively around silicon technology. The growth parameters affecting the silicon crystals also affect the quality of integrated circuits and efficiency of solar cells. The silicon sigle crystals used in electronic industry have to be dislocation free and have to have a homogenous resis tivity variation. In order to achive those properties, the silicon crystals have to be grown in certain growth conditions and parameters. In this study, the silicon crystals are grown by Czochralski 1 method and the growth parameters, kinetics of crystalization, and crystal perfection are investigated. Based on the preliminary research, the growth parameters affecting the crystal properties can be listed as follows; 1. The pulling speed of the crystal. 2. The rotation rate of crystal and crucible. 3. The temperature gradiants. 4. The impurity concentration of the melt. 5. The cleanless of the crystal growth equipment. 6. The diameter of the crystal to be grown. The crystals are grown with various growth parameters and whole data is recorded on an Apple II computer. After the crystal growth experiments the crystals are cut from its certain parts to be examined under the following subjects. 1. Silicon single crystal growth boundry conditions, In this part the maximum growth rates of crystals are found and they are compared with theoretical growth rates. In same crystals the crystal growth has begun as single crystals and during the growth by differing the growth parameters the single crystal growth breakdowns and poly - crystal growth is continued. The growth condi tions and parameters of this critical point are investigatedby searching the impurity concentration is achieved. 2. The radial and axial dislocation spread of silicon crystals. Dislocations can be introduced at nearly every stage of processing. In initial crystal growth, dislocations in the seed will propagate into the new growth as each succeding atomic layer is added. When sudden growth - rate changes occur, extra dislocations will generally occur. Initial growth on to unclean surfaces will lead at best to a higher dislocation density but more often will also produce poly- crystallinity. When precipitation of a second phase occur during the initial crystal cooldown, dislocations are likely to be formed becuse of thermal contraction. In silicon SiO- precipitation can also cause dislocation networks and loops. The dislocation density of the crystals are found out by etch - pit formation. The region near a dislocation line usually etches more rapidly than the rest of the crystal and thus etch pits develop. The growth parameters of silicon crystals like growth rate, rotation rate, temperature gradiant and crystal diame ter chosen are affecting the dislocation density of the crys tals. In order to examine the axial dislocation density the crystals are cut paralel to the axis, and for the radial dis location density the crystals are cut perpendicular to the axis of the crystals. One special crystal is also grown as with a smaller and larger diameter parts on, in order to examine the difference of the dislocation density variation. 3. The radial and axial resistivity spread of the silicon crystals. The growth parameters also affect the resistivity spread of the crystals grown. In order to find out the effect of the growth rate, rotation rate of the crystal and crucible and diameter of the crystals on to the radial resistivity variation, crystals are grown with various growth parameters. Then they are cut perpendicular to the vertical axis and the resistivity measurements are made by four point probe method. During the crystal growth from melt the impurities have a trend to stay in the melt proportional with their segregation coefficients. For that reason as the crystal grows the impurity amount introducing to the crystal be comes more and more. Subsequently the resistivity measure ments made along the crystal started from the seed differs xirelated to the impurity amount and type. In order to find out this proces, crystals are grown with various doping elements and then cut paralel to the axis. The measurements are again done by four point probe method. CONCLUSIONS In this study, it is aimed to improve the growth con ditions of silicon single crystals by means of growth para meters affecting the properties of crystals, so that the better semiconductor material with a low cost and accurate measurement conditions are achieved. The below mentioned conclusions are arrived at; 1. Silicon single crystal growth and conditions. The two important growth parameters in crystal growth of silicon are impurity concentration of melt due to the segregation coefficient and the pulling rate of crystal from the melt. 1.1 It has been proved that the crystal growth of silicon with bigger diameter is necessary when the economic growth is concerned. 1.2 As it is proved experimentaly in order to grow single crystal silicon, maximum %50 of the theoretical growth rate could be applied. 1.3 At the crystals with bigger diameter, the impurity concentration in the melt is not limiting the growth rate. In contrast, on the crystals with smaller diameters, when the maximum growth rate is applied, the impurity concentration of the melt has a great effect on the single crystal structure. 1.4 In order to achieve single crystal silicon, the impurity concentration is low for the higher growth rates and in contrast high for the lower growth rates. In other words it is possible to grow single crystal silicon with a high impurity concentration when the low pulling rate is applied. 1.5. At the last stage of crystal growth process, because of the chemical reaction of the crucible some small parts of Si02 begins to flow near the interface of crystal- melt, and results the single crystal breakdown. xix2. Dislocation density distributions. 2.1. The radial dislocation density distribution. By comparing the dislocation density distributions resulting by the different growth and rotation speeds it is found out that the optimum growth parameters providing the crystal perfection and the minimum dislocation density are in a system of 20 rpm. crucible rotation speed, 25 rpm. crystal rotation on the opposite direction of crucible and 75 mm/hr. pulling speed. Because of the different cooling rates, less dislocation density around the center of the wafers and an increase of dislocation density towards the periphery is observed. 2.2. Axial dislocation density distributions, 1. A crystal which is grown with two different diame ters (Fig. 7. 11) is cut axially and it is observed that be cause of the different cooling rates, the dislocation den sity existing along the axis is always less than the dis location density along the periphery. 2. When the grown crystal diameters are more than 20 mm. the above difference becomes greater, and the crys tals with big diameters have a very low dislocation density at the center, and it reaches its maximum along the perip hery. In the crystals grown under 19-20 mm. diameter this difference becomes very low. 3. Resistivity distributions and measurement. 3.1. Radial resistivity variation. Following results are obtained by the investigation of the growth parameters to the resistivity distributions in the crystals grown by the CZ method. 1. In the crystals grown only by the rotation of the crucible the radial resistivity variation increases until the 25 rpm. crucible rotation rates for all pulling speeds. Starting from 25-30 rpm. crucible rotation, a decrease occurs at the resistivity distribution of the crystals and continues to decrease inverselly proportional to the in crease of the pulling speed of the crystals. In addition to that as the pulling speed increases the radial distri bution increase. The result.of this experiments; to obtain a minimum resistivity change, pulling speed must be under 50 mm/hr. and the rotation rate of the crucible must be over 50 mm/hr. xiii2. In the crystal growth operations with only a crystal rotation and constant crucible, the radial resis tivity variation increases until 10 rpm. crystal rotation rate and decreases over this value. When the crystal rota tion rate is over 50 rpm. the resistivity distribution is independent of the the pulling speed and this value is about 10-15%. In the crystals grown only by crystal rota tion in order to keep the resistivity variation in a mini mum level the crystal rotation speed must be above 50 rpm. 3. When crystal is grown by the rotation in the oppo site direction of the crucible, the radial resistivity vari ation reaches to a maximum level on 10 rpm. crystal rotation rate for all pulling speeds. For the crystal rotation speed of 25 rpm. and above, radial resistivity variation becomes independent of pulling speed of the crystal. If rotation of the crucible or the crystal is under 25 rpm. the radial resistivity variation becomes around 10-20%: 4. As the diameter of the crystal increases, the ra dial resistivity variation decreases. In the crystals with diameters 30 mm. and over, the radial resistivity variation stays under 10%. 3.2. Resistivity variation along the crystal 1. The most important parameters affecting the axial resistivity variation in the crystals grown by CZ method are the existence of doping elements in silicon melt and their segregation coefficients. If 25% resistivity varia tion is wanted in crystals grown by doping with a n-type dopant, only less than 50% of the crystal.starting from the seed can get this rate. 2. In order to grow crystal having 25% resistivity vaiation, in more than half portion of a crystal, at the beginning stage of the growth the pulling speed is chosen high and it is decreased gradually or a compansating doping element is added during the whole proces. 3. As a result of the resistivity measurements on the grown crystals which are cut axially, it is found out that the unique factor determining the conductivity type and the resistivity variation is the doping element in the melt. In case of existing one or more doping elements, the one with the biggest segregation coefficient determines the resisti vity value and conductivity type. As the crystal growth continues the other doping element becomes active when the effect of the first one decreases. xiv