TW201623703A - Method of fabrication of an ingot of n-type single-crystal silicon with a controlled concentration of oxygen-based thermal donors - Google Patents

Abstract

The present invention relates to a method of fabrication of an ingot of n-type single-crystal silicon, with a controlled concentration of oxygen-based thermal donors, comprising at least the steps consisting of: (i) providing a bath of molten silicon comprising one or more n-type dopants, said bath being supplemented with at least germanium (Ge) and/or tin (Sn) at a content adjusted for inhibiting the formation of some or all of the thermal donors in the expected silicon ingot; and (ii) carrying out pulling of the silicon ingot from the bath in step (i) by a pulling technique of the Czochralski type, the initial pulling speed V1 being reduced to a speed V2 =V1/b, with b between 10 and 1.2, when the solidified fraction of silicon, fs, reaches a predetermined critical value. It also relates to an ingot of single-crystal silicon obtained by this method, and its use for fabrication of a photovoltaic cell by a low-temperature method.

Description

Method for producing an ingot of n-type single crystal germanium having a controlled concentration of an oxygen-based thermal donor

The present invention relates to a novel method for producing one of the n-type single crystal crucibles having a controlled concentration of an oxygen-based thermal donor.

This ingot is particularly advantageous in the context of producing photovoltaic photovoltaic (PV) cells by the so-called "low temperature" process. By "low temperature" herein is meant the different production steps of converting one wafer from a crystal rod to a PV solar cell at temperatures well below 650 °C. In the "low temperature" technology, the fabrication of PV solar cells with amorphous germanium/crystalline germanium heterojunctions (so-called "HET" cells) allows for a conversion efficiency of more than 20%.

The fabrication of HET solar cells typically begins with wafers of n-type germanium doped with phosphorus (P) obtained from ingots obtained by Czochralski pulling. An ingot produced by a Czochralski pulling technique (also referred to as a "Cz ingot") has a very high oxygen content of about 4.10 ¹⁷ cm ^-3 to 2.10 ¹⁸ cm ^-3 .

After crystallization, during the cooling of the ingot, a hot donor is formed when the temperature approaches 450 ° C (typically between 350 ° C and 500 ° C), ie, a small oxygen cluster typically formed from 3 to 20 oxygen atoms. . These hot donors behave as electron donors. At the height of a Cz ingot, the TD concentration at the end of cooling is typically about 5.10 ¹² cm ^-3 to 5.10 ¹⁵ cm ^-3 . The top of the Cz ingot (the first part of the crystal) is typically the area of the ingot containing the most hot donor. In fact, the top of the Cz ingot is typically part of the ingot with the highest oxygen concentration and the slowest cooling (which is therefore maintained at a temperature of about 450 ° C for a longer period of time), as disclosed in the publication [1] Revealed in.

Despite phosphorus doping (usually about 10 ¹⁴ cm ^-3 to 10 ¹⁶ cm ^-3 ), the high content of the hot donor in the crystallized and cooled ingot can affect the concentration of most carriers in equilibrium (marked Is n ₀ ), and thus affects the resistivity (ρ) of the material. More precisely, they can cause an increase in concentration n ₀ and a decrease in resistivity ρ in the case of a doped phosphorus ingot. The life of the charge carrier (τ) depends mainly on n ₀ . For a given density of one of crystal defects (points or extended defects), the lower n _{0 , the} longer the life of the charge carrier τ. Therefore, the formation of TD during cooling of the ingot can result in a decrease in τ and thus a decrease in the photovoltaic conversion efficiency of the fabricated battery. Furthermore, TD exhibits recombination activity (other than as a dopant) which, although slight, can aggravate this decrease in τ.

The method of manufacturing a homojunction PV cell employs a production step that occurs at a high temperature of 650 ° C or higher. The removal of the hot donor at this temperature (this is often referred to as "destruction" or "dissociation" of the hot donor) and therefore has no effect on the photovoltaic conversion efficiency of the battery. However, when using the "low temperature" method, the thermal donor can significantly alter the photovoltaic conversion efficiency of the battery, especially for cells made from wafers obtained from the top of the Cz ingot.

In order to prevent the effects of the hot donor, it has been proposed to remove the TD by an annealing operation prior to the "low temperature" manufacturing level step of the HET battery, which is commonly referred to as "TDA" (short for "hot body annealing").

These annealing operations can occur at the germanium wafer level. This is especially the case with the work described by Nakamura et al. ([1]), which uses a wafer annealing operation at 700 ° C for 30 minutes, followed by an annealing operation at about 450 ° C. Rapid cooling at a rate of °C.s ^-1 .

However, the use of this type of annealing on an industrial scale poses some problems. First of all, this is an extra step and is quite long and will therefore cause productivity problems. this In addition, it means that the manufacturing chain will add additional equipment and thus incur a significant additional cost. Finally, since this step occurs at high temperatures, it may result in undesirable contamination of the wafer volume that can change the life of the charge carriers (eg, caused by elements from the furnace composition or from the annealing atmosphere).

These annealing operations for removing the TD can also occur directly at the level of the ingot. However, again, this is also an additional annealing step with associated disadvantages in terms of productivity loss and cost. Furthermore, at the level of an ingot, considering the considerable thermal inertia, it is difficult to have sufficiently fast cooling conditions at the end of the TDA to avoid reforming the TD.

Finally, we can also mention the latest method of lifting Cz ingots, which use one of the "thermal shielding" inserted between the crucible and the bottom of the ingot when the ingot is removed from the crucible, which accelerates Cooling of the ingot. Therefore, during cooling, the length of time at 450 ° C is reduced and thus the content of TD at the end of cooling is also reduced. However, the results obtained with this technical solution are unsatisfactory because the TD concentration can exceed 10 ¹⁵ cm ^-3 despite the insertion of this shield. In addition, the photovoltaic industry tends to lift longer and/or wider rods, which therefore have greater thermal inertia and therefore slow down the cooling rate at the end of the lift. Such ingots may therefore incorporate higher levels of TD.

Thus, there remains a need for an inexpensive method for obtaining a Cz(R) wafer that can be used for "low temperature" fabrication of photovoltaic cells with a reduced thermal donor content or even no thermal donor.

The object of the present invention is to strictly meet this need.

Therefore, according to one of the first aspects of the aspect, it relates to a method for producing one of the n-type single crystal crucibles having a controlled concentration of an oxygen-based thermal donor, the method comprising at least A step consisting of: (i) providing a molten bath comprising one or more n-type dopants to adjust to inhibit the formation of some or all of the hot donor in the desired twine in the bath Adding at least bismuth (Ge) and/or tin (Sn); and (ii) lifting the strontium rod from the bath in step (i) by one of the Czochralski type pulling techniques, when the curing fraction of bismuth When f _s reaches the minimum value of the values f _s1 and f _s2 , the initial pulling speed V _{1 is} reduced to a speed V ₂ =V ₁ /b, where b is between 10 and 1.2, and .f _s1 is for The concentration of Ge in the solid ruthenium [Ge] reaches the value: [Ge] _fs1 = a ₁ × [Ge] The cure fraction of _crit after _crit , and the concentration of .f _s2 is the concentration of Sn incorporated into the solid ruthenium [Sn ] _Achieved value: [Sn] _fs2 = a ₂ × [Sn] The cure fraction of _crit , where: -a ₁ represents a constant between 0.3 and 1; -a ₂ represents a constant between 0.2 and 1 ; - [Ge] _crit expressed in step (ii) of Conditions of the pulling speed V ₁ may be incorporated in the solid silicon germanium predetermined maximum concentration of, if the concentration exceeds a predetermined maximum, the crystal growth of silicon will undergo a morphological instability; and - [Sn] _crit shown in step ( Ii) a predetermined maximum concentration of tin that can be incorporated into the solid niobium at a rate V ₁ pulling condition, and if the predetermined maximum concentration is exceeded, the crystal growth of the crucible undergoes a morphological instability.

In the following, the oxygen-based heat donor will be labeled as "hot donor" or more simply by the abbreviation "TD".

The name "Cz Ingot" will be used for the ingot obtained by one of the Czochralski type lifting techniques, and the "Cz wafer" will be used for the wafer obtained from this Cz ingot.

The inventors have found that an ingot having an excellent crystal quality with a curing fraction exceeding at least 90% can be obtained, and the ingot contains little, if any, thermal donor during continuous pulling.

Advantageously, the method of the present invention thus eliminates the above-mentioned additional annealing steps for removing the hot donor, which cause numerous problems on an industrial scale.

As illustrated in the examples below, the method of the present invention can, for example, divide the amount of hot donor by 3, or even divide by, for a cure fraction of 10% relative to a "standard" Cz ingot. 5 and more preferably divided by 10.

The ingot produced by the method of the present invention may have, for example, less than or equal to 2.10 ¹⁵ cm ^-3 , in particular less than or equal to 5.10 ¹⁴ at the top of the start of the lift (e.g., for one of 10% cure fraction). A thermal donor concentration of cm ^-3 and more specifically less than or equal to 2.10 ¹⁴ cm ^-3 .

According to another aspect of the aspect of the invention, the invention relates to a crystallized rod obtainable by the method defined above.

In the "low temperature" manufacturing method for photovoltaic cells, particularly "HET" batteries, it is advantageous to use Cz wafers obtained from ingots obtained at the end of the pulling technique according to the present invention.

Therefore, according to still another aspect of the present invention, the present invention relates to the manufacture of a photovoltaic cell by a low temperature method using one of the single crystal germanium obtained by the method of the present invention, in particular, a HET battery. .

Other features, advantages and embodiments of the method according to the present invention, as well as features and advantages of the ingot thus obtained, will be apparent from the following description, examples and drawings of the invention. And examples.

In the following, statements such as "between ... and ...", "range between ... and ..." and "between ... and ..." are equivalent. It is intended to indicate that the limits of the scope are included unless otherwise specified.

Unless otherwise specified, a sentence of "having/including one/one" will be understood as "having/including at least one".

Manufacture of twin rods

Step (i): Melting bath

As stated above, step (i) of the method of the present invention consists of providing a molten bath comprising one of at least one n-type dopant (also designated as a "liquid" bath).

According to an important feature of the method of the invention, the molten bath is doped with at least germanium (Ge), tin (Sn) or both antimony and tin.

The or n-type dopants may be selected from the group consisting of phosphorus (P), arsenic (As), antimony (Sb), bismuth (Bi), and the like.

Preferably, the n-type dopant is phosphorus.

More specific words, the (equal) n type dopant may be present in one of the content of the molten bath between each cm ³ 3.10 ¹³ atoms and each cm ³ 8.6 × 10 ¹⁷ atoms.

The content of Ge and/or Sn in the initial molten bath in step (i) is adjusted such that formation of some or all of the hot donor in the desired twine can be inhibited.

The effect of bismuth and tin on the activation power of hot donors has been mentioned in the literature ([2]).

More specifically, the content of Ge and/or Sn is selected relative to the desired resistivity (ρ) of the twin rod and thus to the "acceptable" TD concentration in the twin rod.

In fact, as mentioned above, the resistivity (ρ) of tantalum is closely related to the content of the hot donor body having an electron donor action in the twin rod, as in, for example, application WO 2014/064347 Detailed description.

The target value for selecting the resistivity (ρ) for the application envisaged for the twin rod is, for example, between 0.5 Ω·cm and 10 Ω·cm for the manufacture of photovoltaic cells.

Therefore, prior to the practice of the method of the present invention, it may be based on an acceptable TD concentration in the desired crystallizing rod (in other words, the desired electrical resistivity ρ) and in accordance with one of the melting baths formed in the self-deficient Ge and Sn. At the end of the pull test, the maximum TD concentration in one of the Cz ingots is obtained to determine the amount of Ge and/or Sn to be used.

The concentration of Ge and/or Sn to be introduced into the ingot can be estimated, for example, using an empirical equation.

For example, the inventors have established the following empirical mathematical expressions (obtained for annealing operations performed at 450 ° C and convertible to any temperature that allows for the formation of TD) for expressing the effect of the presence of germanium on TD formation:

Where: [TD](T, t ) represents the TD concentration (in cm ^-3 ) at the end of annealing at a temperature T for one sample containing one of the contents of [Ge]; TD] _{[Ge] = 0} indicates the TD concentration (in cm ^-3 ) after annealing at a temperature T for a time t without containing ruthenium.

In this sentence, [Ge] is expressed in units of cm ^-3 .

Regarding tin doping, tin is one atom larger than yttrium. It has been found that yttrium-doped tin and doped yttrium have a similar effect on oxygen-related defects, because the content of tin is ten times lower than that of yttrium.

The formation of the self-deficient Ge and Sn can be inferred by comparing the resistivity measured for the Cz ingot obtained at the end of the test of the lift before and after applying one of the annealing heat treatments for facilitating the elimination of the hot donor. The concentration of the hot donor in the Cz ingot obtained at the end of the pull test of the bath.

According to another alternative, the experimental value of the resistivity measured for the ingot obtained at the end of the pull test can be compared to the lack of a thermal donor (only the presence of the (n) n-type dopant is considered) The theoretical value of the desired resistivity is compared to infer the TD concentration formed in the Cz ingot obtained at the end of the test from the absence of one of Ge and Sn.

Since the change in resistivity can be attributed to the formation of the hot donor, the concentration of the hot donor formed in the ingot can be easily inferred therefrom, for example, using the relationship specified in the application WO 2014/064347.

For example, FIG. 1 shows the resistivity of the ingot obtained in the pull test in the case where no Ge is added, which is described in detail in Example 1, and the value expected when the TD is absent, considering only phosphorus as a dopant.

More specifically, it can be applied to the "top" portion of the ingot (ie, the area corresponding to the start of curing) The "acceptable" TD concentration is determined (and thus the concentration of Ge and/or Sn to be introduced into the twin rod) is determined.

In fact, as mentioned above, this is part of the ingot that incorporates the highest level of thermal donor and thus has the greatest effect on resistivity for the appearance of its TD, as shown schematically in Figure 1.

Specifically, about 10% of the curing fraction will be used as a reference for determining the maximum content of TD formed in the ingot at the end of the pulling test, and thus determining the Ge and/or Sn to be introduced into the crucible. content. In practice, the end of the ingot is usually removed by standard cut-off, especially corresponding to the tip of a cure fraction that is strictly below 10%.

The content of Ge and/or Sn to be introduced into the top of the ingot as determined as described above (especially for a curing fraction of about 10%) can be easily inferred using Scheil's law to be introduced into the initial melting enthalpy between the Sn and / or concentration of Ge, Scheil's law within the bath for a given impurity concentration C _fs and k show the curing fraction of solid formed in the impurity _eff of the effective distribution coefficients of the impurities through the f _s relationship:

The content of Ge and/or Sn in the bath in step (i) is usually between 10 ¹⁹ cm ^-3 and 3.10 ²² cm ^-3 .

According to a first variant embodiment, the bath in step (i) is added with hydrazine.

Preferably, the Ge content of the liquid bath is adjusted so that the Ge content in the twine bar is between 3.10 ¹⁹ cm ^-3 and 3.10 ²¹ cm ^{-3 for a} curing fraction of about 10%, specifically 10 ²⁰ Between cm ^-3 and 6.10 ²⁰ cm ^-3 and more specifically about 3.10 ²⁰ cm ^-3 .

According to another variant embodiment, the bath in step (i) is supplemented with tin.

Preferably, the Sn content of the liquid bath is adjusted such that for a curing fraction of about 10%, the Sn content in the twin rod is between 3.10 ¹⁸ cm ^-3 and 3.10 ²⁰ cm ^-3 , specifically 10 ¹⁹ Between cm ^-3 and 6.10 ¹⁹ cm ^-3 and more specifically about 3.10 ¹⁹ cm ^-3 .

According to still another variant embodiment of the invention, the bath in step (i) is supplemented with tin and antimony.

The preparation of the molten bath used in step (i) of the method of the present invention is part of the common knowledge of those skilled in the art.

The enthalpy charge used to form the molten bath can be composed of electronic grade blocks obtained from a chemical purification process.

A molten bath can be prepared in a vermiculite or graphite crucible (as appropriate coated with a layer of SiC). These crucibles are known to be heat resistant at elevated temperatures suitable for obtaining molten baths.

The incorporation of bismuth and/or tin into the ruthenium charge is within the abilities of those skilled in the art.

Helium and/or tin may be added to the ruthenium charge before, during or after the formation of the molten bath.

According to a variant embodiment, niobium and/or tin are added to the niobium charge in the form of a powder or beads.

Preferably, the introduced element has a high purity (typically greater than 4 N) to avoid any unintentional contamination of the crucible.

Furthermore, it is especially possible to avoid the problem of low melting point of tin by using technical solutions known to those skilled in the art (for example, encapsulating Sn in "beads").

Alternatively, niobium and/or tin may be introduced into an inner coating of a crucible intended to contain a molten bath which can, for example, be dissolved into the molten bath by dissolution.

Of course, those skilled in the art will be able to adapt the volume of the molten bath used in step (i) of the method of the present invention to the size of the desired crystallizing rod.

According to a particular embodiment, the twin rod formed by the method of the invention may be additionally doped with carbon (C) and/or nitrogen (N).

The effect of carbon on the power to form a hot donor during an annealing operation at about 450 ° C has been investigated in the literature ([3], [4]). Carbon slows down the formation of TD.

Yang et al. ([5]) also pointed out that nitrogen may slow the formation of TD.

According to a particular embodiment, the formed twin rod is additionally doped with carbon. In this case, the initial bath in step (i) may be added with carbon, especially by techniques for incorporating carbon into the ruthenium charge, such as mentioned above for Sn and/or Ge.

Preferably, the carbon content of the molten bath in step (i) is adjusted so that the carbon content in the twin rod is between 1.10 ¹⁷ and 8.10 ¹⁷ cm ^{-3 for a} curing fraction of about 10%, in particular It is between 2.10 ¹⁷ and 6.10 ¹⁷ cm ^-3 and more specifically about 4.10 ¹⁷ cm ^-3 .

According to another particular embodiment, the formed twin rod is doped with nitrogen in addition to Ge and/or Sn and optionally C. In this case, doping can be carried out by circulating a stream of nitrogen on the surface of the molten bath during the pulling of the ingot.

Preferably, the nitrogen-doped stream is adjusted to achieve between 10 ¹⁵ and 10 ¹⁷ cm ^{-3 for a} cure fraction of about 10%, in particular between 3.10 ¹⁵ and 3.10 ¹⁶ cm ^-3 and more specific It is said that one of the crystal rods of about 10 ¹⁶ cm ^-3 has a nitrogen content.

Step (ii): Lifting the crystal rod

In a second step of one of the methods of the present invention, directional solidification from a crucible bath is performed by one of the Czochralski type pulling techniques.

The "Czochralski type lifting technique" indicates one of the original Czochralski lifting techniques or one derived from the Czochralski technique.

In general, the Czochralski type of pulling technique consists of recrystallization of a crucible starting from a seed crystal and a molten bath. The seed crystal oriented relative to the crystal axis of the solid helium is first infiltrated in the molten bath. The seed is then slowly pulled up. Therefore, the solid twin rod gradually grows from the liquid bath.

The use of equipment suitable for Czochralski type technology is within the common knowledge of those skilled in the art.

When the curing fraction of the crucible (labeled "f _s ") reaches a defined value as described in detail below, by reducing the initial pulling speed V ₁ to a pulling speed V ₂ =V ₁ /b The twin rod lifting is carried out according to the method of the invention, wherein b is between 10 and 1.2.

The initial pulling speed V ₁ can be, for example, between 2.10 ^-6 ms ^-1 and 2.10 ^-4 ms ^-1 , specifically between 4.10 ^-6 and 1.10 ^-4 , and more specifically at 8.10 ^-6 Between 5.10 and ⁵ ms ^-1 .

Usually, it can be between 0.1 rad.s ^-1 and 15 rad.s ^-1 , specifically between 0.5 rad.s ^-1 and 10 rad.s ^-1 and more specifically ¹ rad.s ^-1 and 4 rad. The crystal between the .s ^-1 is lifted with respect to the rotational speed ω of the bath. The rotational speed ω of the crystal can be kept constant throughout the lift of the ingot.

When the solidification fraction f _{s of the} crucible reaches the minimum of the values f _s1 and f _s2 , the initial pulling speed V _{1 is} lowered to a speed V ₂ .

.f _s1 system for the solid silicon which is incorporated in a concentration of Ge, a [Ge] reaches a value _{of: [Ge] fs1 = a 1} × [Ge] _crit curing of the silicone fraction, .f _s2 incorporated into the system for which The concentration of Sn in the solid [ [Sn] reaches the value: [Sn] _fs2 = a ₂ × [Sn] The cure fraction of _crit , where: -a ₁ represents a constant between 0.3 and 1, in particular Between 0.4 and 1 and more specifically a value of 0.7; -a ₂ represents a constant between 0.2 and 1, in particular between 0.3 and 1 and more specifically with a value of 0.5; -[ Ge] _crit represents the predetermined maximum concentration of ruthenium which can be incorporated into the solid ruthenium in the condition of pulling at a speed V ₁ in step (ii), and if it exceeds the predetermined maximum concentration, the crystal growth of the ruthenium undergoes a morphological instability And -[Sn] _crit represents the predetermined maximum concentration of tin that can be incorporated into the solid cerium in the condition of pulling at a speed V ₁ in step (ii), and if it exceeds the predetermined maximum concentration, the crystal growth of bismuth Experience a form of instability.

f _S1 And / or f _S2 Judgment

Since tantalum and tin have a low partition coefficient (Ge is about 0.38 and Sn is about 0.023, such as As shown in the examples given below, the concentration of Ge or Sn incorporated into the solid thus increases during the curing of the crucible. This increase is also described in Schcil's Law (2) mentioned above, except that a transient initial transient state of one of the solute boundary layers is allowed to form before the growth front.

The concentration [Ge] _crit or [Sn] _crit (hereinafter referred to as "critical concentration") means the perturbation phenomenon of single crystal growth (for example, polycrystalline or equiaxed growth phenomenon) from which Ge or metal which appears in the ingot The maximum concentration of Sn.

According to a first variant, the method may be implemented prior to analysis by the present invention since the addition of germanium at a constant pulling speed V ₁ for pulling one end of the test (or added with tin, added separately) melted silicon The bath obtains the crystal quality of one of the ingots and determines the critical concentration [Ge] _crit (or [Sn] _crit ).

The crystal quality of the ingot can be analyzed by observing the wafer obtained by cutting the ingot at different heights. If it is known that the crystal begins to grow from the height at which it is disturbed (and thus the cure fraction is known), the corresponding concentration of Ge (or Sn) can be determined by applying Scheil's law (2).

According to another variant embodiment, the inventors have determined that the following equation can be used to determine the critical concentration ([Ge] _crit or [Sn] _crit ).

Therefore, the concentration [Ge] _crit can be determined using Equation (3) below before the step (ii) of pulling the ingot according to the method of the present invention:

Where: -V ₁ represents the initial pull rate (in ms ^-1 ); -G represents the temperature gradient at the solid/liquid interface, G typically has a value of approximately 2.10 ³ Km ^-1 ; -C ₁ at 0.3 and Between 3, specifically between 0.5 and 2 and more specifically has a value of 1.

Furthermore, the concentration [Sn] _crit can be determined using Equation (3') below before the step (ii) of pulling the ingot according to the method of the present invention:

Where: -V ₁ represents the initial pull rate (in ms ^-1 ); -G represents the temperature gradient at the solid/liquid interface, G typically has a value of approximately 2.10 ³ Km ^-1 ; -C ₂ at 0.3 and Between 3, specifically between 0.5 and 2 and more specifically has a value of 1.

The curing fraction f _s1 corresponding to one of the Ge concentrations [Ge] _fs1 (equal to a ₁ × [Ge] _crit ) incorporated in the solid ruthenium can be inferred from Scheil's law (2) stated above.

The same inference applies to the cure fraction f _s2 .

In order to determine f _s1 and f _s2 , the use of Scheil's law requires knowledge of the effective partition coefficient k _eff of the elements considered.

The coefficient k _{eff of 锗} can be obtained using the following relation (4):

Wherein V ₁ is the initial pulling speed (in ms ^-1 ) and ω is the rotational speed (in rad.s ^-1 ) of the crystal used for pulling in step (ii).

The coefficient of tin k _eff can be obtained using the following relation (5):

Wherein V ₁ is the initial pulling speed (in ms ^-1 ) and ω is the rotational speed (in rad.s ^-1 ) of the crystal used for pulling in step (ii).

During the pulling of the ingot, the self-tanning curing fraction starts to count when the minimum value of the determined values f _s1 and f _s2 is reached, and the pulling speed is reduced to a value V ₁ /b, where b is 10 and 1.2 between.

Preferably, the initial pull rate is reduced to a velocity V ₁ /b, where b is between 5 and 1.7, and specifically b has a value of 2.

It should be understood that in the case where the molten bath in the step (i) is only added with a crucible, the pulling speed is decreased when the curing fraction of the crucible reaches the value f _s1 as determined as described above.

Further, particularly in the case where the molten bath in the step (i) is only added with tin, the pulling speed is decreased when the curing fraction of the crucible reaches the value f _s2 as determined as described above.

According to a variant embodiment of the method of the invention, hydrazine (in solid or liquid form, preferably in liquid form) is added to the molten bath during the pulling in step (ii).

This addition reduces the concentration of impurities by the dilution effect during the pulling process.

Twin rod

As seen above, in accordance with another aspect of the aspects, the present invention is directed to an ingot of single crystal germanium obtainable by the method described above.

One of the ingots obtained in accordance with the present invention advantageously has a controlled concentration of thermal donor which is substantially reduced in contrast to the twinked rod system obtained by standard pulling techniques.

The ingot is advantageously at least 70% above its height, in particular at least 80% of its height, having the desired resistivity (ρ) without being affected by the presence of the thermal donor.

The twin rods may have a cylindrical shape.

The twin rods may, for example, have a height between 10 cm and 3.5 m, in particular between 20 cm and 2 m.

According to a particular embodiment, the crystallized rod obtained by the method of the invention has a cure fraction of about 10% (specifically at a height corresponding to a cure fraction of greater than or equal to 2%) having less than or Equivalent to 2.10 ¹⁵ cm ^-3 , specifically less than or equal to 5.10 ¹⁴ cm ^-3 , and more specifically less than or equal to a thermal donor concentration of 2.10 ¹⁴ cm ^-3 .

According to another variant embodiment, the twin rod may be doped with ruthenium. Specifically, the twin rod may have a curing fraction of about 10.10 ¹⁹ cm ^-3 and 3.10 ²¹ cm ^-3 for about 10%, specifically 10 ²⁰ cm ^-3 and 6.10 ²⁰ cm ^-3 . More specifically, it is about 3.10 ²⁰ cm ^-3 Ge concentration.

According to another variant embodiment, the twin rods may be doped with tin. Specifically, the twin rod may have a curing fraction of about 10.10 ¹⁸ cm ^-3 and 3.10 ²⁰ cm ^-3 for about 10%, specifically 10 ¹⁹ cm ^-3 and 6.10 ¹⁹ cm ^-3 . More specifically, it is about one concentration of 3.10 ¹⁹ cm ^-3 of Sn.

According to another particular embodiment, in addition to being doped with Ge and/or Sn, the twin rod is doped with carbon. In particular, the twin stick may have a cure fraction of between about 1.10 ¹⁷ and 8.10 ¹⁷ cm ^-3 for about 10%, specifically between 2.10 ¹⁷ and 6.10 ¹⁷ cm ^-3 and more specifically It is approximately one carbon concentration of 4.10 ¹⁷ cm ^-3 .

According to yet another particular embodiment, the germanium ingot is doped with nitrogen in addition to being doped with Ge and/or Sn. In particular, the twin stick may have a cure fraction of between about 10 ¹⁵ and 10 ¹⁷ cm ^-3 for about 10%, specifically between 3.10 ¹⁵ and 3.10 ¹⁶ cm ^-3 and more specifically A nitrogen concentration of approximately 10 ¹⁶ cm ^-3 .

It will be appreciated that the twin rods according to the invention may be doped with antimony and/or tin, in addition to being doped with carbon and/or nitrogen.

Advantageously, the introduction of the previously mentioned levels can also limit the formation of crystal defects, such as "voids" (gap clusters of octahedral structures) that are particularly detrimental to the performance of high efficiency PV cells.

Furthermore, the contamination of the nitrogen with the above-mentioned content of nitrogen can advantageously limit the formation of "vortex defects".

Furthermore, the twin rods obtained according to the invention advantageously have an excellent crystal quality at a height corresponding to at least 80%, in particular at least 90% or even at least 95% of the solidification fraction. In particular, the content of niobium (or tin) may exceed the critical concentration of the curing fraction corresponding to greater than 90%, specifically greater than 95%, as defined above, at the bottom end (the last solidified portion) of the ingot. [Ge] _crit (or [Sn] _crit ) (associated with the pull speed V ₁ ).

After standard removal of the peripheral regions of the ingot, the ingot can be cut into pieces by techniques known to those skilled in the art.

Next, wafers can be prepared from such blocks starting from such blocks as is known to those skilled in the art, particularly by cutting the blocks, grinding the faces to adjust the dimensions of the wafer, and the like.

Advantageously, the wafer produced by cutting a twin rod in accordance with the present invention can be used directly to fabricate a photovoltaic cell by "low temperature" technology without the need for an additional annealing step for removing the thermal donor.

Advantageously, the wafer can be obtained from the top of the ingot (starting at a cure fraction of 2%) or from the middle or bottom of the ingot (from 50% to 90% cure fraction).

"Cryogenic technology" means the different steps of converting a wafer into a PV solar cell at temperatures strictly below 650 °C.

In particular, the wafers can be used to fabricate a photovoltaic cell having an amorphous germanium/crystalline germanium heterojunction, a so-called "HET" cell.

For example, one or more of the following steps are performed at the end of wafer fabrication: - depositing a first layer of essentially amorphous germanium (typically having a thickness of about 5 nm) on each side of the wafer and overdoping p+ and/or n+ well or zone; - depositing a conductive transparent oxide layer based on indium oxide (ITO) on the surface of the amorphous germanium layer; - printing at a low temperature, especially by silver paste screen printing in front of the device and / One or more metallizations (also referred to as "conductive contacts") are formed on the back side.

The invention will now be described by way of the following examples and drawings, which are, of course, given and not limited to the invention.

Figure 1 : Variation in resistivity of a phosphorus-doped Cz ingot as a function of cure fraction. It shows the value expected without TD, the value obtained without adding Ge and without annealing for removing TD, and the addition of Ge (concentration in the molten bath is 7.9×10 ²⁰ cm ^-3 ) and no The expected value in the case of annealing to remove TD in the pull-up case of Example 1.

Instance

The examples given below are for Cz pulling of an n-type single crystal twin rod which is mainly doped with phosphorus, starting from a high purity cerium charge (electron level).

The phosphorus content added to the ruthenium charge is 1.4 × 10 ¹⁵ cm ^-3 .

When no hot donor is formed, the resistivity ρ should decrease monotonically from the top to the bottom of the ingot (and then the phosphorus content increases with the cure fraction), as depicted in Figure 1.

Starting from an n-type doped bath, a pull-up test of one of the crystal rods was carried out at a pulling speed of V ₁ and a crystal rotation speed ω of 1.9 rad.s ^-1 .

Figure 1 shows the resistivity measured at different heights of the ingot (corresponding to different cure fractions) at the end of the lift.

The resistivity can be measured by known techniques, for example by a four-point method or by a contactless method, for example by inductive coupling.

The value of the resistivity obtained indicates a "bell-shaped" variation of the resistivity, and the formation of a hot donor during the cooling of the ingot will significantly affect the resistivity of the top of the ingot (the portion that is first cured).

The concentration of the hot donor formed during the cooling of the ingot can be estimated by comparing the experimental value of the resistivity obtained for the ingot with the theoretical value expected in the absence of the hot donor (assuming that the hot donor will be double ionized) ).

For a curing fraction of about 10%, it is estimated that the concentration of the hot donor is 3.2 × 10 ¹⁴ cm ^-3 .

Example 1

Indole doping

Will be introduced into the molten bath

Since n-type doped silicon bath starts to one pulling speed ^-1 1.05 × 10 ^-5 ms V _1, one embodiment of a silicon rod pulled test. Since it is intended to remove the top end of the ingot by cutting off the formed ingot (curing fraction from 0 to 2%), the acceptable TD concentration in the desired twine (depending on the desired resistivity ρ) And the ruthenium content to be introduced into the solid ruthenium was evaluated for the TD concentration of one percent of the Cz ingot obtained at the end of the pull test described above.

Therefore, the desired niobium content in the twin rods for a curing fraction of about 10% is estimated to be 3.10 ²⁰ cm ^-3 .

It can be determined from Scheil's law (relationship (2) given in the text) that it will be introduced into the molten bath to provide a Ge concentration of this content in the formed ingot: [Ge] _fs = k _eff × [Ge] _{l ,i} ×(1-f _s ) ^(keff-1)

Where: f _s : cure fraction of bismuth, [Ge] _fs : 锗 content in solid 矽 under curing fraction f _s , [Ge] _{l, i} : initial concentration of enthalpy in the molten bath, and k _eff : use The effective partition coefficient (equal to 0.38) of the estimated equation (4) given in the paper.

Therefore, the initial molten bath was added with a yttrium content [Ge] _{l,i of} 7.9 × 10 ²⁰ cm ^-3 .

Determine the critical concentration of bismuth

The "critical" concentration [Ge] _{crit of the crucible} incorporated into the solid crucible can be determined by the following relation (3), and when the critical concentration is exceeded, the crystal growth of the crucible undergoes a form instability:

Assume that V ₁ is 1.05 × 10 ^-5 ms ^-1 ; G is the temperature gradient at the solid/liquid interface, which is 2.10 ³ Km ^-1 ; C ₁ has a value of 1.

Therefore, we obtain [Ge] _crit equal to about 1.1 × 10 ²¹ cm ^-3 .

Lifting the bismuth rod from a molten bath with a enthalpy of enthalpy at an initial pulling speed V _{1 of} 1.05 × 10 ^-5 ms ^-1 , and when the concentration of germanium in the solid reaches a value [Ge] _fs1 = a ₁ × [Ge ] _crit (where a ₁ is chosen to be equal to 0.7) (this value corresponds to one of the 80% cure fraction f _s estimated from Scheil's law (2)), this speed is reduced to one of V ₁ /2 Lifting speed V ₂ .

At the end of the pulling, one of the crystallized rods which is perfectly crystallized up to about one of the curing fractions of about 96% is obtained, which has a resistivity which is mainly determined by the phosphorus content.

Example 2

Doping the ingot with tin

The amount of tin that will be introduced into the molten bath

Starting from the n-type doping bath, a pull-up test of one of the twin rods was carried out at a pulling speed V _{1 of} 1.8 × 10 ^-5 ms ^-1 . Since it is intended to remove the top end of the ingot by cutting off the formed ingot (curing fraction from 0 to 2%), the acceptable TD concentration in the desired crystallizing rod (depending on the desired resistivity ρ) And the tin content to be introduced into the solid crucible was evaluated at the TD concentration of 10% of the Cz ingot obtained at the end of the pull test described above.

Therefore, the desired tin content in the twin rods for a curing fraction of about 10% is estimated to be 3.10 ¹⁹ cm ^-3 .

The concentration of Sn to be introduced into the molten bath to provide the content of the formed ingot can be determined from Scheil's law (relationship (2) given herein): [Sn] _fs = k _eff × [Sn] _l,i ×(1-f _s ) ^(keff-1)

Where: f _s : curing fraction of bismuth, [Sn] _fs : tin content in solid bismuth at curing fraction f _s , [Sn] _{l, i} : initial concentration of tin in liquid, and k _eff : used in the text Give an estimate of the effective partition coefficient of tin (equal to 0.023) in equation (5).

Therefore, the initial molten bath was added with a tin content [Sn] _{l,i of} 1.3 × 10 ²¹ cm ^-3 .

Determine the critical concentration of tin

The "critical" concentration [Sn] _crit of the tin incorporated into the solid ruthenium can be determined by the following empirical relationship (3'), when the critical concentration is exceeded, the crystal growth of the ruthenium undergoes a morphological instability:

Assume that V ₁ is 1.8 × 10 ^-5 ms ^-1 ; G is the temperature gradient at the solid/liquid interface, which is 2.10 ³ Km ^-1 ; C ₂ has a value of 1.

Therefore, we obtain [Sn] _crit equal to about 2.75 × 10 ²⁰ cm ^-3 .

The crystal rod is lifted from a molten bath with tin added at an initial pulling speed V _{1 of} 1.8 × 10 ^-5 ms ^-1 , and the concentration of tin in the solid reaches a value [Sn] _fs1 = a ₂ × [ Sn] _crit (where a ₂ is selected to be equal to 0.5) (this value corresponds to a solidification fraction f _s estimated to be close to 81% from Scheil's law (2)), and this velocity is reduced to be equal to V ₁ /2 A pulling speed V ₂ .

At the end of the pulling, one of the crystallized rods which is perfectly crystallized up to about 95% of one of the curing fractions is obtained, which has a resistivity which is mainly determined by the phosphorus content.

reference

[1] Nakamura et al., "Dependence of properties for silicon heterojunction solar cells on wafer position in ingot", Proceedings of the 27th European Photovoltaic Solar Energy Conference and Exhibition.

[2] Claeys et al., "Tin Doping of silicon for Controlling Oxygen Precipitation and Radiation Hardness", J. Electrochem. Soc. 148, G738-745 (2001).

[3] Wijaranukula, "Formation Kinetics of oxygen thermal donors in silicon", Appl. Phys. Lett. 1991, 59: 1608-10.

[4] Babitskii et al., "Kinetics of generation of low-temperature Oxygen donors in silicon containing isovalent impurities", Sov. Phys. Semicond. 22, 187, 1988.

[5] Yang et al., "Nitrogen effects on thermal donor and shallow thermal donor in silicon", Appl. Phys. 77, 943 (1995).

Figure 1 shows schematically the resistivity of an ingot.

Claims (23)

A method of making an ingot of an n-type single crystal crucible having a controlled concentration of an oxygen-based thermal donor comprising at least the steps of: (i) providing one or more n-type dopants a molten bath to adjust to inhibit the formation of some or all of the hot donors in the desired twine in the bath by adding at least germanium (Ge) and/or tin (Sn); and (ii) The pulling of the twin rod by the bath in step (i) by one of the Czochralski type pulling techniques, when the solidification fraction f _s of the _crucible reaches the minimum value of the values f _s1 and f _s2 , the initial The pulling speed V _{1 is} reduced to a velocity V ₂ = V ₁ /b, where b is between 10 and 1.2, and f _s1 is at a concentration [Ge] for the Ge incorporated into the solid enthalpy: [Ge] _Fs1 = a ₁ × [Ge] the curing fraction of _crit ; f _s2 is the concentration [Sn] for the Sn incorporated into the solid enthalpy: [Sn] _fs2 = a ₂ × [Sn] _crit And the curing fraction; wherein: a ₁ represents a constant between 0.3 and 1; a ₂ represents a constant between 0.2 and 1; [Ge] _crit represents a velocity V ₁ in step (ii) The conditions of the pulling can be incorporated into the solid raft Thereafter, the predetermined maximum concentration, if the predetermined maximum concentration is exceeded, the crystal growth of the crucible will undergo a morphological instability; and [Sn] _crit indicates that the step (ii) is at a speed V ₁ The predetermined maximum concentration of tin incorporated into the solid crucible, if the predetermined maximum concentration is exceeded, the crystal growth of the crucible undergoes a morphological instability. The method of the requested item 1, wherein the molten silicon bath comprises one of an n-type dopant content of phosphorus in particular lines between each cm ³ 3.10 ¹³ atoms and each cm ³ 8.6 × 10 ¹⁷ atoms (s). The method of claim 1 or 2, wherein the ruthenium and/or tin is added to the ruthenium charge before, during or after the formation of the fused bath. The method of claim 1 or 2, wherein the ruthenium and/or tin is added to the ruthenium charge in the form of a powder or a bead, or is introduced into an internal coating of the ruthenium intended to contain the fused bath, the coating being capable of The bismuth and/or tin is allowed to enter the molten bath. The method of claim 1 or 2, wherein the heat treatment in one of the Cz ingots is obtained before the method of the present invention is performed with respect to the formation of one of the Cz ingots at the end of one of the melting baths of one of Ge and Sn. The maximum concentration of the body, specifically for a curing fraction of about 10% and the desired resistivity (p) of the ingot, to determine the content of Ge and/or Sn of the molten bath in step (i) . The method of claim 1 or 2, wherein the comparison is made to the ingot obtained at the end of the pulling test by applying an annealing heat treatment to facilitate completion of the annealing of the thermal donors The resistivity is determined to be formed at the end of the pull test by comparing the resistivity measured for the ingot obtained at the end of the pull test with the theoretical resistivity desired in the absence of a hot donor. The concentration of the hot donor in the Cz ingot is obtained. The method of claim 1 or 2, wherein the bath in step (i) comprises a Ge and/or Sn content between 10 ¹⁹ cm ^-3 and 3.10 ²² cm ^-3 . The method of claim 1 or 2, wherein the Ge content in the twin rod is adjusted to be between 3.10 ¹⁹ cm ^-3 and 3.10 ²¹ cm ^{-3 for a} curing fraction of about 10%, preferably A content of between 10 ²⁰ cm ^-3 and 6.10 ²⁰ cm ^-3 and more preferably about 3.10 ²⁰ cm ^-3 is added to the bath in step (i). The method of claim 1 or 2, wherein the Sn content in the twin rod is adjusted to be between 3.10 ¹⁸ cm ^-3 and 3.10 ²⁰ cm ^{-3 for a} curing fraction of about 10%, in particular The tin is added to the bath in step (i) at a content between 10 ¹⁹ cm ^-3 and 6.10 ¹⁹ cm ^-3 and more specifically about 3.10 ¹⁹ cm ^-3 . The method of claim 1 or 2, wherein the specific content is adjusted such that the carbon content in the twin rod is between 1.10 ¹⁷ and 8.10 ¹⁷ cm ^{-3 for a} curing fraction of about 10%, in particular 2.10 ¹⁷ and 6.10 ¹⁷ cm ^-3 and more specifically about 4.10 ¹⁷ cm ^-3 of one of the additional addition of carbon in the bath in step (i). The method of claim 1 or 2, wherein the twin rod formed by the pulling in the step (ii) is additionally doped with nitrogen, in particular such that the solidification fraction is in the solidification fraction for about 10% The nitrogen content is between 10 ¹⁵ and 10 ¹⁷ cm ^-3 , in particular between 3.10 ¹⁵ and 3.10 ¹⁶ cm ^-3 and more specifically about 10 ¹⁶ cm ^-3 . The method of claim 1 or 2, wherein between 2.10 ^-6 ms ^-1 and 2.10 ^-4 ms ^-1 , specifically between 4.10 ^-6 and 1.10 ^-4 , and more specifically at 8.10 ^{- The} pulling in step (ii) is carried out at an initial pulling speed V ₁ between ⁶ and 5.10 ^-5 ms ^-1 . The method of claim 1 or 2, wherein between 0.1 rad.s ^-1 and 15 rad.s ^-1 , specifically between 0.5 rad.s ^-1 and 10 rad.s ^-1 and more specifically The pulling of the crystal between ¹ rad.s ^-1 and 4 rad.s ^-1 is carried out in step (ii) with respect to one of the rotational speeds ω of the bath. The method of claim 1 or 2, wherein the end of the pull test is performed by one of the melting baths of Ge or Sn separately added at a constant pulling speed V ₁ prior to performing the method of the present invention The crystal quality of one of the ingots is obtained to determine the critical concentration [Ge] _crit or [Sn] _crit . The method of claim 1 or 2, wherein the concentration [Ge] _{crit is} determined using the following equation (3) before pulling the ingot according to step (ii): Where: V ₁ represents the initial pulling speed (in ms ^-1 ); G represents the temperature gradient at the solid/liquid interface, in particular G has a value of approximately 2.10 ³ Km ^-1 ; C _{1 is} at 0.3 Between 3, specifically between 0.5 and 2 and more specifically with a value of 1; and/or determining the concentration [Sn] _crit using equation (3') before the pulling of the ingot: Wherein: V ₁ represents the initial pulling speed (in ms ^-1 ); G represents the temperature gradient at the solid/liquid interface, in particular G has a value of about 2.10 ³ Km ^-1 ; C ₂ is Between 0.3 and 3, specifically between 0.5 and 2 and more specifically a value of 1. The method of claim 1 or 2, wherein a ₁ is a constant between 0.4 and 1, and preferably has a value of 0.7. The method of claim 1 or 2, wherein a ₂ is a constant between 0.3 and 1, and preferably has a value of 0.5. The method of claim 1 or 2, wherein the initial pulling speed V _{1 is} reduced to a speed V ₂ = V ₁ /b during the pulling in step (ii), wherein b is between 5 and 1.7, And b preferably has a value of 2. The method of claim 1 or 2, wherein the crucible, preferably liquid helium, is added to the molten bath in a solid or liquid form during the pulling in step (ii). An ingot of a single crystal germanium obtainable by the method as defined in any one of claims 1 to 19. An ingot as claimed in claim 20 having a height between 10 cm and 3.5 m, in particular between 20 cm and 2 m. A twinked rod as claimed in claim 20 or 21 which has a cure fraction of about 10%, in particular a height of less than or equal to 2.10 ¹⁵ cm at a height corresponding to a cure fraction of greater than or equal to 2% ^{- 3} , specifically less than or equal to 5.10 ¹⁴ cm ^-3 , and more specifically less than or equal to a thermal donor concentration of 2.10 ¹⁴ cm ^-3 . A use of an ingot of a single crystal germanium as defined in claim 20 or 21 for producing a photovoltaic cell by a low temperature method, in particular for fabricating a photovoltaic having an amorphous germanium/crystalline germanium heterojunction Hit the battery.