Distributions of Energy Losses by Fast Ions along Their Propagation Paths in Solids

The integro-differential equations for cumulants of the distribution function that describes energy losses by fast ions during their propagation in solids have been obtained. The equations differ from those obtained by other authors by one new term. The term describes accurately the process of slowi...

Full description

Saved in:
Bibliographic Details
Date:2010
Main Authors: Ilyina, V.V., Makarets, M.V.
Format: Article
Language:English
Published: Відділення фізики і астрономії НАН України 2010
Subjects:
Online Access:https://nasplib.isofts.kiev.ua/handle/123456789/13389
Tags: Add Tag
No Tags, Be the first to tag this record!
Journal Title:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Cite this:Distributions of Energy Losses by Fast Ions along Their Propagation Paths in Solids / V.V. Ilyina, M.V. Makarets // Укр. фіз. журн. — 2010. — Т. 55, № 2. — С. 236-243 . — Бібліогр.: 14 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id nasplib_isofts_kiev_ua-123456789-13389
record_format dspace
spelling Ilyina, V.V.
Makarets, M.V.
2010-11-05T14:28:59Z
2010-11-05T14:28:59Z
2010
Distributions of Energy Losses by Fast Ions along Their Propagation Paths in Solids / V.V. Ilyina, M.V. Makarets // Укр. фіз. журн. — 2010. — Т. 55, № 2. — С. 236-243 . — Бібліогр.: 14 назв. — англ.
2071-0194
PACS 61.72.uf; 61.72.uj
https://nasplib.isofts.kiev.ua/handle/123456789/13389
539.534.9
The integro-differential equations for cumulants of the distribution function that describes energy losses by fast ions during their propagation in solids have been obtained. The equations differ from those obtained by other authors by one new term. The term describes accurately the process of slowing down of an ion at the start of its path. The equations have been numerically solved for the first seven cumulants of the distribution function for both elastic and inelastic energy losses, and the results have been compared with the results for ion ranges. It has been found that: 1) for energies in the interval 1 keV-1 GeV, the average ranges with energy losses are approximately 30-90% of the ion ranges; 2) for low energies, the straggling of the distribution of energy losses are slightly larger than or equal to the straggling of the distribution of ion ranges, while, for high energies, the former can be 10 times as large as the latter; 3) for low energies, the skewnesses and excesses of the distributions of energy losses and ion ranges are approximately the same, while their changes for the former at higher energies are several orders smaller than those for the latter. This implies that the distribution of energy losses are wider and closer to the normal distribution than the distribution of ion ranges. We show that these properties of energy loss distributions are a result of the inclusion of the new terms in the equations which dominate at high energies.
Отримано iнтегродиференцiальнi рiвняння для кумулянтiв функцiї розподiлу втрат енергiї швидких iонiв вздовж їх шляху у твердому тiлi, у яких враховано втрати енергiї починаючи iз точки старту. Рiвняння для перших семи кумулянтiв розподiлiв втрат енергiї у пружних, непружних та обох типах зiткнень розв’язанi чисельно за допомогою методу, розвинутого авторами ранiше, на iнтервалi енергiй iонiв 1 кеВ–1 ГеВ. Їх порiвняння з параметрами розподiлу пробiгiв iонiв показало, що: 1) середнiй шлях за розподiлом втрат енергiї становить 30–90% повного пробiгу iонiв; 2) при низьких енергiях страгглiнг за розподiлом втрат енергiї дещо бiльший або однаковий iз страгглiнгом за розподiлом пробiгiв iонiв, а при високих енергiях перший перевищує останнiй у десятки разiв; 3) скiсностi i ексцеси розподiлiв при низьких енергiях вiдповiдно близькi, у той час як при високих енергiях їх змiна для розподiлу втрат енергiї на кiлька порядкiв менша, нiж для розподiлу пробiгiв iонiв. Звiдси випливає, що розподiл втрат енергiї ширший i значно ближчий до нормального, нiж розподiл пробiгiв iонiв при всiх енергiях iмплантацiї. Показано, що цi властивостi розподiлу втрат енергiї зумовленi новими членами у рiвняннях, якi не враховувалися ранiше, але домiнують при високих енергiях.
Получены интегродифференциальные уравнения для кумулянтов функции распределения потерь энергии быстрых ионов вдоль их пути в твердом теле, в которых учтены потери энергии начиная с точки старта. Уравнения для первых семи кумулянтов распределений потерь энергии в упругих, неупругих и обеих типах столкновений решены численно с помощью метода, развитого авторами ранее, на интервале энергий ионов 1 кэВ–1 ГэВ. Их сравнение с параметрами распределения пробегов ионов показало, что: 1) средний путь с потерей энергии составляет 30–90% полного пробега иона; 2) при низких энергиях страгглинг распределения потерь энергии немного больше или одинаков со страгглингом распределения пробегов ионов, а при высоких энергиях первый превышает последний в десятки раз; 3) скошенности и эксцессы распределений при низких энергиях соответственно близки, в то время как при высоких энергиях их изменение для распределения потерь энергии на несколько порядков меньше, чем для распределения пробегов ионов. Отсюда следует, что распределение потерь энергии более широкое и ближе к нормальному, чем распределение пробегов ионов для всех энергий имплантации. Показано, что эти свойства распределения потерь энергии обусловлены новыми членами в уравнениях, которые не учитывались ранее, но доминируют при высоких энергиях.
en
Відділення фізики і астрономії НАН України
Тверде тіло
Distributions of Energy Losses by Fast Ions along Their Propagation Paths in Solids
Розподіл втрат енергії швидких іонів вздовж їх шляху у твердому тілі
Распределение потерь энергии быстрых ионов вдоль их пути в твердом теле
Article
published earlier
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
title Distributions of Energy Losses by Fast Ions along Their Propagation Paths in Solids
spellingShingle Distributions of Energy Losses by Fast Ions along Their Propagation Paths in Solids
Ilyina, V.V.
Makarets, M.V.
Тверде тіло
title_short Distributions of Energy Losses by Fast Ions along Their Propagation Paths in Solids
title_full Distributions of Energy Losses by Fast Ions along Their Propagation Paths in Solids
title_fullStr Distributions of Energy Losses by Fast Ions along Their Propagation Paths in Solids
title_full_unstemmed Distributions of Energy Losses by Fast Ions along Their Propagation Paths in Solids
title_sort distributions of energy losses by fast ions along their propagation paths in solids
author Ilyina, V.V.
Makarets, M.V.
author_facet Ilyina, V.V.
Makarets, M.V.
topic Тверде тіло
topic_facet Тверде тіло
publishDate 2010
language English
publisher Відділення фізики і астрономії НАН України
format Article
title_alt Розподіл втрат енергії швидких іонів вздовж їх шляху у твердому тілі
Распределение потерь энергии быстрых ионов вдоль их пути в твердом теле
description The integro-differential equations for cumulants of the distribution function that describes energy losses by fast ions during their propagation in solids have been obtained. The equations differ from those obtained by other authors by one new term. The term describes accurately the process of slowing down of an ion at the start of its path. The equations have been numerically solved for the first seven cumulants of the distribution function for both elastic and inelastic energy losses, and the results have been compared with the results for ion ranges. It has been found that: 1) for energies in the interval 1 keV-1 GeV, the average ranges with energy losses are approximately 30-90% of the ion ranges; 2) for low energies, the straggling of the distribution of energy losses are slightly larger than or equal to the straggling of the distribution of ion ranges, while, for high energies, the former can be 10 times as large as the latter; 3) for low energies, the skewnesses and excesses of the distributions of energy losses and ion ranges are approximately the same, while their changes for the former at higher energies are several orders smaller than those for the latter. This implies that the distribution of energy losses are wider and closer to the normal distribution than the distribution of ion ranges. We show that these properties of energy loss distributions are a result of the inclusion of the new terms in the equations which dominate at high energies. Отримано iнтегродиференцiальнi рiвняння для кумулянтiв функцiї розподiлу втрат енергiї швидких iонiв вздовж їх шляху у твердому тiлi, у яких враховано втрати енергiї починаючи iз точки старту. Рiвняння для перших семи кумулянтiв розподiлiв втрат енергiї у пружних, непружних та обох типах зiткнень розв’язанi чисельно за допомогою методу, розвинутого авторами ранiше, на iнтервалi енергiй iонiв 1 кеВ–1 ГеВ. Їх порiвняння з параметрами розподiлу пробiгiв iонiв показало, що: 1) середнiй шлях за розподiлом втрат енергiї становить 30–90% повного пробiгу iонiв; 2) при низьких енергiях страгглiнг за розподiлом втрат енергiї дещо бiльший або однаковий iз страгглiнгом за розподiлом пробiгiв iонiв, а при високих енергiях перший перевищує останнiй у десятки разiв; 3) скiсностi i ексцеси розподiлiв при низьких енергiях вiдповiдно близькi, у той час як при високих енергiях їх змiна для розподiлу втрат енергiї на кiлька порядкiв менша, нiж для розподiлу пробiгiв iонiв. Звiдси випливає, що розподiл втрат енергiї ширший i значно ближчий до нормального, нiж розподiл пробiгiв iонiв при всiх енергiях iмплантацiї. Показано, що цi властивостi розподiлу втрат енергiї зумовленi новими членами у рiвняннях, якi не враховувалися ранiше, але домiнують при високих енергiях. Получены интегродифференциальные уравнения для кумулянтов функции распределения потерь энергии быстрых ионов вдоль их пути в твердом теле, в которых учтены потери энергии начиная с точки старта. Уравнения для первых семи кумулянтов распределений потерь энергии в упругих, неупругих и обеих типах столкновений решены численно с помощью метода, развитого авторами ранее, на интервале энергий ионов 1 кэВ–1 ГэВ. Их сравнение с параметрами распределения пробегов ионов показало, что: 1) средний путь с потерей энергии составляет 30–90% полного пробега иона; 2) при низких энергиях страгглинг распределения потерь энергии немного больше или одинаков со страгглингом распределения пробегов ионов, а при высоких энергиях первый превышает последний в десятки раз; 3) скошенности и эксцессы распределений при низких энергиях соответственно близки, в то время как при высоких энергиях их изменение для распределения потерь энергии на несколько порядков меньше, чем для распределения пробегов ионов. Отсюда следует, что распределение потерь энергии более широкое и ближе к нормальному, чем распределение пробегов ионов для всех энергий имплантации. Показано, что эти свойства распределения потерь энергии обусловлены новыми членами в уравнениях, которые не учитывались ранее, но доминируют при высоких энергиях.
issn 2071-0194
url https://nasplib.isofts.kiev.ua/handle/123456789/13389
citation_txt Distributions of Energy Losses by Fast Ions along Their Propagation Paths in Solids / V.V. Ilyina, M.V. Makarets // Укр. фіз. журн. — 2010. — Т. 55, № 2. — С. 236-243 . — Бібліогр.: 14 назв. — англ.
work_keys_str_mv AT ilyinavv distributionsofenergylossesbyfastionsalongtheirpropagationpathsinsolids
AT makaretsmv distributionsofenergylossesbyfastionsalongtheirpropagationpathsinsolids
AT ilyinavv rozpodílvtratenergííšvidkihíonívvzdovžíhšlâhuutverdomutílí
AT makaretsmv rozpodílvtratenergííšvidkihíonívvzdovžíhšlâhuutverdomutílí
AT ilyinavv raspredeleniepoterʹénergiibystryhionovvdolʹihputivtverdomtele
AT makaretsmv raspredeleniepoterʹénergiibystryhionovvdolʹihputivtverdomtele
first_indexed 2025-11-26T10:00:51Z
last_indexed 2025-11-26T10:00:51Z
_version_ 1850618361354911744
fulltext V.V. ILYINA, M.V. MAKARETS DISTRIBUTIONS OF ENERGY LOSSES BY FAST IONS ALONG THEIR PROPAGATION PATHS IN SOLIDS V.V. ILYINA, M.V. MAKARETS Taras Shevchenko National University of Kyiv (64, Volodymyrs’ka Str.,01033 Kyiv, Ukraine; e-mail: mmv@ univ. kiev. ua ) UDC 539.534.9 c©2010 The integro-differential equations for cumulants of the distribu- tion function that describes energy losses by fast ions during their propagation in solids have been obtained. The equations differ from those obtained by other authors by one new term. The term describes accurately the process of slowing down of an ion at the start of its path. The equations have been numerically solved for the first seven cumulants of the distribution function for both elastic and inelastic energy losses, and the results have been com- pared with the results for ion ranges. It has been found that: 1) for energies in the interval 1 keV-1 GeV, the average ranges with energy losses are approximately 30-90% of the ion ranges; 2) for low energies, the straggling of the distribution of energy losses are slightly larger than or equal to the straggling of the distribution of ion ranges, while, for high energies, the former can be 10 times as large as the latter; 3) for low energies, the skewnesses and excesses of the distributions of energy losses and ion ranges are approx- imately the same, while their changes for the former at higher energies are several orders smaller than those for the latter. This implies that the distribution of energy losses are wider and closer to the normal distribution than the distribution of ion ranges. We show that these properties of energy loss distributions are a result of the inclusion of the new terms in the equations which dominate at high energies. 1. Introduction The first system of equations for the distribution func- tion of ions implanted into a solid was obtained over half a century ago [1]. The methods to study these equations have been developed in several stages [2,3]. In the recent years, the authors of [4–6] have proposed the cumulant approach and developed numerical methods that allow cumulants up to the 6th order to be calculated and ana- lyzed for the whole energy range, where the assumptions of the classical implantation theory [1] hold. The main advantage of this approach is that cumu- lants of the distribution are smoother functions of the ion energy than its moments – their amplitudes are less by several orders. It has allowed one to find an exact enough analytical approximation and construct a stable numerical method. In general, the hardly soluble equa- tions with the obvious physical meaning were reduced to a more easily soluble equation but with a more fuzzy physical meaning. Now we use the cumulant approach for the investiga- tion of energy losses of fast ions in a solid. When we have applied this approach for the first time to equations from [7] which describe the energy loss dis- tributions of ions and knocked-out particles, we obtained unexpected results. Namely, the norm and other cumu- lants of the nuclear energy losses distribution were equal to zero. On the contrary, the norm of the electronic en- ergy losses distribution was equal to the ion energy E, and its other cumulants were also non-zero. These re- sults contradict the physical meaning, but the sum of norms is in complete agreement with the energy conser- vation law. We felt obliged to investigate the reasons for having the solution to the well-established equations that con- tradicts the basic physics laws. This investigation [8] has shown that the equations for the spatial energy losses [2, 7] contain a hidden assumption that the Energy Loss Distribution (ELD) function for an ion at the start of its path has a zero value. This assumption is a con- sequence of the even earlier assumption that the losses inside an infinitely small volume δV are proportional to δV . The same assumption was made while obtaining the equations for the ranges of implanted ions [1]. In this case, however, the assumption is completely valid, be- cause an ion cannot stop more than once inside δV . Therefore, the number of random stops is δn = 1 if an ion stopped inside δV and δn = 0 if an ion stopped outside δV . Then, using the first-order approximation, it can be assumed that an ion takes up a volume of ΔV ≈ 1/N0, where N0 is the atom concentration of a solid. Then the probability of the ion stopping in a volume δV ≤ ΔV is defined as δP ≈ δnδV/ΔV ∼ δV and has a norm equal to 1. Then we can introduce a density of implanted ions as a limit Π (~r,E) = lim δV→0 δP/δV . In the same way, one can introduce the density of the ions distribution along their paths by considering an infinitely small length δl. 236 ISSN 2071-0194. Укр. фiз. журн. 2010. Т. 55, №2 DISTRIBUTION OF ENERGY LOSSES Let us now consider the energy losses of an ion. While an ion only stops once, it looses energy in many col- lisions during its trajectory. Energy losses can occur many times in the same volume δV if the characteristic length of a collision-free path of an ion λ is small or if ion’s trajectory is very complex. Therefore, an ion passes through the same volume δV many times before leaving it or stopping in it, and thus the energy lost in δV will be equal to 0 if an ion did not cross δV or proportional to a number of collisions δn if an ion crossed δV . Now, the number of random collisions inside δV is δn ≈ δl/λ, where δl ∼ (δV )1/3 is the ion path, and we can assume in the first-order approximation that the energy is transferred from an ion to a solid through a cylinder with a volume ΔV ≈ πR2λ. The axis of the cylinder connects the locations of two neighboring col- lisions. For elastic collisions, the radius of the cylinder Rn is equal to the half-distance between atoms of a solid Rn ≈ a/2. For inelastic collisions, the radius Re depends on the ion energy and the properties of the electronic subsystem of a solid [9]. It can be defined generally as Re ≡ R(E,N0, . . .). If we denote the energy lost in one collision of the α- type (where α = n, e for nuclear and electronic collisions) as 〈Tα〉, then the probable loss of energy inside δV ≤ ΔV can be determined from δPα ≈ 〈Tα〉δnδV/ΔV . This probability has a norm of κα0 (E), which is the en- ergy lost in one collision of the α-type. It follows from here that δPα ∼ (δV )4/3, and therefore we cannot deter- mine the volume density for energy losses, at least within the limits of this approach. However, we can introduce the density of energy losses along an ion’s path δl if we consider the probable loss of energy δPα ≈ 〈Tα〉δn ∼ δl. 2. Main Equations and Their Analysis In this paper, we consider the distribution functions for energy losses per unit length of an ion along its trajectory in the solid. To be more precise, we study the cumulants of these distributions. Let Πα(E,R)dR be the most probable energy lost by a fast ion with initial energy E in the α-type collision in the vicinity of a point dR along ion’s path R, where α = t, n, e correspond to the total, nuclear, and electronic losses, respectively. We follow [1] to obtain the equations for the density of the distribution: ∂Πα(E,R) ∂R = −N0 ∫ dσ(E, T ) {Πα(E,R) − −Πα(E − T,R)} , (1) where dσ(E, T ) is the differential cross-section of the elastic and inelastic scattering process involving the ion and a target atom, and T is the energy lost in this scat- tering process. This equation is identical in form to the equation for the density distribution of ion ranges obtained in [1]. However, the meaning of the function Πα(E,R) is dif- ferent from the one used in [7]. This leads to an initial condition (at R = 0) for this function being different from the one used in [2, 7]. Namely, the function Πe(E, 0)dR for electronic losses is the most probable energy lost in inelas- tic collisions during the path dR near the start point of an ion. Then it follows that Πe(E, 0) = Se(E), where Se(E) ≡ ∫ Tedσe(E, Te) is the inelas- tic energy loss per unit length of an ion with en- ergy E or the electronic stopping, and dσe(E, Te) is the differential cross-section of electronic scatter- ing. For nuclear losses, Πn(E, 0) = Sn(E), where Sn(E) ≡∫ Tndσn(E, Te) is the elastic energy loss per unit length of an ion with energy E or the nuclear stopping, and dσn(E, Tn) is the differential cross-section of nuclear scattering. It is obvious that, for the total losses, Πt(E, 0) = Se(E) + Sn(E). Therefore, for the α-type losses, we can write down Πα(E, 0) = Sα(E), where Sα(E) is the α-type stopping for an ion with energy E. These initial conditions mean that a fast ion will start to loose energy immediately after the start. We can also include the ion ranges distribution in the common scheme of energy losses calculations, if we taken its initial condition Π(E, 0) = 0 into ac- count and then set the index α = i and formally put Si(E) = 0. In this case, the most probable num- ber of ions which are stopped at the start is zero, but there is no reason to believe that this initial condition holds for energy losses. Since this zero initial condi- tion was incorrectly used in [2, 7] for the elastic energy loss distribution of fast particles stopped into a unit vol- ume, the corresponding equations have the above solu- tions. Let us now change to the Lindhard’s dimension- less units by replacing R ⇒ ρ and E ⇒ ε following [1, 2] in Eq. (1). We will also replace the function Πα(ε, ρ) by its cumulants καk , where k = 0, 1, 2, . . ., following [4]. Taking the norm of the energy distri- bution function as the energy lost in an α-type col- lision κα0 (ε), we can write down the following ex- pression for the energy loss distribution density func- ISSN 2071-0194. Укр. фiз. журн. 2010. Т. 55, №2 237 V.V. ILYINA, M.V. MAKARETS tion [4]: Πα(ε, ρ)= κα0 π Δρα ∞∫ 0 exp [ − 1 2! s2+ Exα1 4! s4−Ex α 2 6! s6+. . . ] × × cos [ −rαs− Skα1 3! s3 + Skα2 5! s5 − . . . ] ds, (2) where Δρα = √ κα2 (ε)/κα0 (ε) is the straggling of the mean path ρα(ε) = κα1 (ε)/κα0 (ε), rα ≡ rα(ρ, ε) is the centered dimensionless mean ion path, and Skαj and Exαj are the distribution’s skewness and excess, re- spectively, which depend on the energy ε only. These quantities can be expressed through the cumulants by rα = ρ− ρα(ε) Δρα(ε) , Skαj = κα2j+1(ε) κα0 (ε) [Δρα(ε)]2j+1 , Exαj = κα2j+2(ε) κα0 (ε) [Δρα(ε)]2j+2 , j = 1, 2, . . . (3) At a fixed ion energy ε, rα(ρ) is the dimensionless distribution’s variable, and Skαj and Exαj are dimen- sionless distribution’s parameters. The cumulant of the jth order has the units of (length)j multiplied by the units of κα0 . It follows from Eq. (2) that the cu- mulant κα0 defines the amplitude of the distribution, while its form depends on the parameters χαj (ε) = καj (ε)/κα0 (ε), where j = 1, 2, . . . Equations (2) and (3) remain true for the ion ranges distribution, since κi0(ε) ≡ 1. Following [4] and using Eq. (1), we obtain the equa- tions for the cumulants using the initial condition Πα(E, 0) = Sα(ε). For the norm of the distribution function κα0 and the mean path distribution κα1 , these equations become L̂ {κα0 } = Sα(ε), (4) L̂ {κα1 } = κα0 (ε), (5) where the left-hand operator is defined as L̂ {καk} ≡ N̂ {καk (ε− γτ)}+ Se(ε) ∂καk (ε) ∂ε , (6) N̂ {. . .} ≡ − 1 γ ε∫ 0 σ(ε, τ) ∂∂τ {. . .} dτ, (7) where N̂ is the nuclear scattering operator, σ(ε, τ) is defined in [4] as the so-called summarized cross-section of ion’s nuclear scattering on the target atom, γ = 4M1M2/ (M1 +M2) 2, and M1,2 are the masses of an ion and the target atom, respectively. The equations for the cumulants of the k ≥ 2 order are: L̂ {καk} = N̂ { κα0 (ε− γτ)Fk ( χα1 , . . . , χ α k−1 )} + +Sα(ε)fk ( χα1 , . . . , χ α k−1 ) , (8) where the functions Fk(χα1 , . . . , χ α k−1) and fk(χα1 , . . . , κ α k−1) do not depend on the type of a collision explicitly. In particular, for the cumulants of the second-sixth orders which define the straggling, two skewnesses, and two excesses, the functions Fk do not depend on the norm κα0 explicitly: F2 = [Δα 1 (ε, τ)]2 , (9) F3 = Δα 1 (ε, τ) { [Δα 1 (ε, τ)]2 + 3Δα 2 (ε, τ) } , (10) F4 = Δα 1 (ε, τ) { [Δα 1 (ε, τ)]3 + 6Δα 2 (ε, τ)Δα 1 (ε, τ) + +4Δα 3 (ε, τ) } + 3 [Δα 2 (ε, τ)]2 , (11) F5 = Δα 1 (ε, τ) { [Δα 1 (ε, τ)]4 + 10Δα 2 (ε, τ) [Δα 1 (ε, τ)]2 + +10Δα 3 (ε, τ)Δα 1 (ε, τ) + 15 [Δα 2 (ε, τ)]2 + +5Δα 4 (ε, τ) } + 10Δα 2 (ε, τ)Δα 3 (ε, τ), (12) F6 = Δα 1 (ε, τ) { [Δα 1 (ε, τ)]5 + 15Δα 2 (ε, τ) [Δα 1 (ε, τ)]3 + +20Δα 3 (ε, τ) [Δα 1 (ε, τ)]2 + 15Δα 4 (ε, τ)× ×Δα 1 (ε, τ) + 45 [Δα 2 (ε, τ)]2 Δα 1 (ε, τ) + +60Δα 2 (ε, τ)Δα 3 (ε, τ) + 6Δα 5 (ε, τ) } + +15 [Δα 2 (ε, τ)]3 + 10 [Δα 3 (ε, τ)]2 + +15Δα 2 (ε, τ)Δα 4 (ε, τ), (13) where Δα i (ε, τ) = χαi (ε− γτ)− χαi (ε), and Δα i (ε, 0) = 0 for all i. The functions fk are obtained from Fk by the substitution χαj (ε− γτ) = 0 for j = 1, 2, . . . , k − 1. Note that the equations for the cumulants contain the terms which are proportional to Sα(ε) on the right-hand side and take the energy losses of an ion at the start of its path into account. Equations (4), (5), and (8) are inhomogeneous second- order Volterra-type integral equations with integrable singularity core σ(ε, τ → 0)→∞ [4] which describe the first derivative of the cumulants. They have non-trivial solutions only if the right-hand sides are non-zeros [10]. The equations should be solved one by one, since Eq. (4) has the explicitly given right-hand side. The ion ranges distribution is an example where the contribution of Sα (ε) is equal to zero. In this case, Eq. 238 ISSN 2071-0194. Укр. фiз. журн. 2010. Т. 55, №2 DISTRIBUTION OF ENERGY LOSSES (4) for the norm becomes homogeneous with respect to its derivative, ε∫ 0 σ(ε, τ) { ∂κi0(ε− γτ) ∂ε } dτ + Se(ε) ∂κi0(ε) ∂ε = 0, (14) and has the trivial solution ∂κi 0(ε) ∂ε = 0. Using the norm conditioning it follows that κi0(ε) ≡ 1, and Eq. (5) re- duces to the equations for ranges obtained in [1]. For the cumulants of order k ≥ 2, the right-hand sides of Eqs. (8) do not contain the second term, and the inte- grals in Eqs. (9)–(11) coincide with the results obtained in [4]. Therefore, we have shown that while the equations for the ion ranges distribution functions and the ELD functions are the same, the equations for their cumu- lants, moments, and central moments differ. The reason for this is that while it is very unlikely that an ion will stop right at the beginning of its path, it is almost cer- tain that it will loose some energy at the beginning of its path. The consequence of this difference is the differ- ent initial conditions for the distribution functions which define the right-hand sides of the equations for the cu- mulants of these distributions. 3. Results and Discussion The first-order series expansions of the integral func- tions, Eqs. (4) and (5), for the energy loss γτ allows us to obtain the analytical approximations for the norm of the distribution function and for the mean path: κα0 (ε) ≈ ε∫ 0 Sα(ε′) Sn(ε′) + Se(ε′) dε′, (15) ρα(ε) ≈ 1 κα0 (ε) ε∫ 0 κα0 (ε′) Sn(ε′) + Se(ε′) dε′. (16) Approximation (15) satisfies the energy conservation law, because κt0(ε) = ε, and the integrands in this equa- tion can be interpreted as the probability of the energy losses of the α-type per unit path of an ion with the en- ergy ε′. To our knowledge, the equations have not been presented elsewhere. Expression (16) is a generalization of a well-known result for the ion mean range [1]. We have solved Eqs. (4), (5), and (8) numerically, by using the method developed in [4, 5]. Within this method, the left-hand sides of the equations are trans- formed into a system of linear equations for the expan- sion coefficients καi (ε). The expansion is carried out for the energy intervals, whose lengths grow geomet- rically until they cover the range of energies, where the assumptions of binary elastic ion-atom collisions hold. For the lower energy limit, we assume the energy, for which the distance between an ion and an atom during the head-on collision is 1/4 of the half-distance of atoms in the target. For the upper energy limit, we assume the energy, for which the distance between an ion and an atom during the head-on collision is 1/4 of the sum of their radii and the radius of the nuclear force action. For inelastic and elastic collisions, we use, respectively, the approximation of [11] and the potential proposed by Ziegler–Biersack–Littmark [12]. In Fig. 1, we show the dependences κt,n,e0 (ε) and κt,n,e0 (ε)/ε calculated for the implantation of phospho- rus ions into a silicon target. The lower and upper en- ergy limits are chosen to represent energies, where the assumptions of [1] hold. They are equal to 2.8 keV and 281 MeV, respectively. We have chosen this ion-target combination because of its practical applications, but also because γ ≈ 1 for this combination, and the expan- sion in γ in integrands has the worst asymptotic behav- ior. One can see from Fig. 1,a that, for large energies, nearly all losses are caused by the electronic scattering, however the nuclear losses slowly increase as well. We can see from Fig. 1,b that the energy lost in elastic, inelastic, and all collisions satisfies the energy conserva- tion law, though it has not been used in obtaining the equations. We believe that the accuracy, to which the law is satisfied (more than 0.2%), can be interpreted as the indirect evaluation of the numerical method accu- racy. To approximate the equations for cumulants of the order k ≥ 2 analytically, we have to take into account the expansion of the integrands Fk in Eq. (8) in γτ up to the second-order terms. Therefore, the approximation of the integral begins with the straggling of the nuclear scattering W (ε) = ∫ τ2dσ(ε, τ) = 2 ∫ τσ(ε, τ)dτ . The right-hand sides have the term proportional to Sα(ε), and we can rewrite the equations as {Sn(ε) + Se(ε)} ∂καk (ε) ∂ε ≈ Sα(ε)fk ( χα1 , . . . , χ α k−1 ) + + 1 2 W (ε)κα0 (ε)F (2) k ( χα1 , . . . , χ α k−1 ) , (17) where F (2) k is the second derivative of Fk with respect to τ at τ = 0, which is a polynomial of the second order in the first derivatives of the cumulants of the (k − 1)th order inclusive. In case of the distribution ISSN 2071-0194. Укр. фiз. журн. 2010. Т. 55, №2 239 V.V. ILYINA, M.V. MAKARETS Fig. 1. Dependence of the norm of the energy losses function – а and the relative norm – b on the implantation energy of phosphorus ions into a silicon target. Letters n, e, and t next to the curves denote the distributions of nuclear, electronic, and total losses, respectively of ranges, when κα0 (ε) ≡ 1 and Si(ε) ≡ 0, the right- hand sides of Eq. (17) can be simplified, and the first derivative of the cumulant of the kth order can be ex- pressed through the moments of nuclear scattering ex- plicitly [4]. In the case of energy losses, the second term on the right-hand side of Eq. (17) is proportional to the cumulants of the (k − 1)th order, and this can change the dependence of cumulants on energy signifi- cantly. Figure 2 presents the graphs of dependences of pa- rameters from Eqs. (2) and (3) on the ion energy for the distributions α = n, e, t and i for the combina- tion phosphorus-silicon. The arrow and the number in Fig. 2,с–f near the curve for the parameter of the ions distribution show the direction of its monotonic change and its value for an ion energy of 1 GeV, re- spectively. One can see from Fig. 2,a that the mean of the total ELD is smaller than the mean of the ions dis- tribution corresponding to the picture of the energy losses of an ion just before it stops. One can also see that the mean of the elastic losses distribution and the ranges distribution have similar values for high ener- gies suggesting that this mechanism dominates at the end of the ion path, when its energy is low. Strag- glings of losses of all types, Fig. 2,b, exceed the strag- gling of the ranges distribution, by suggesting that the energy losses are distributed more evenly during the ion path. Therefore, the ELDs will have maxima to- ward the beginning of its path, and the distributions will be wider than the ranges distribution. In addi- tion, the ratio of the straggling to the mean Δρα/ρα has a maximum for the total losses distribution and has a minimum for the ranges distribution. Therefore, these distributions will be narrower and wider, respec- tively. Figure 2,с–f demonstrates the common property for all parameters of the order k ≥ 3 for the ELDs, which is their relatively small change of amplitudes. Namely, the changes in amplitudes for the ELDs are several times smaller than those for the ranges distribution. Then it follows that these distributions are closer to the normal distribution, as all cumulants of the or- der k ≥ 3 are equal to zero [13]. We can also see that the parameters for the total loss distribution for low energies are similar to parameters for the elas- tic loss distribution, while, for high energies, they be- come similar to the parameters for the inelastic ener- gies distribution. This corresponds to the fact that different loss mechanisms dominate at different ener- gies. We can now conclude that, for the ELDs for different types of collisions, we can use the normal distribution as the zero-order approximation. An indirect criterion for its applicability will be the initial condition Πβ(E, 0) = Sβ(E) for β = n, e, t which can be written explicitly as κβ0 (ε)√ 2π Δρβ(ε) exp { −1 2 [ ρβ(ε) Δρβ(ε) ]2} ≈ Sβ(E). (18) This condition links three first cumulants. We have also calculated the ratio of the left-hand side of Eq. (18) to its right-hand side for the phosphorus/silicon combination. The value of this ratio shows that the normal distribu- tion at |Sk|, |Ex| ≤ 1 gives values of the ELDs at the coordinate origin that are approximately 1/3 of the true values. If |Sk| and |Ex| are large, then the difference increases further, and it becomes necessary to use one of 240 ISSN 2071-0194. Укр. фiз. журн. 2010. Т. 55, №2 DISTRIBUTION OF ENERGY LOSSES Fig. 2. Dependence on the implantation energy of phosphorus ions into a silicon target for: а, b – mean path distribution and its straggling; c, d – the first and second skewnesses; e, f) – the first and second excesses. The letter i near the curves denotes the ion distribution the Pearson distributions [2] or the Johnson distribu- tion [14]. We have also carried out the numerical analysis of the contributions of both members on the right-hand side of Eq. (8). The analysis has shown that the stopping mem- ber with Sα(ε) dominates, while the nuclear scattering only contributes significantly at low energies. This jus- tifies the use of the approximation χαk (ε) ≈ 1 κα0 (ε) ε∫ 0 Sα(ε′)fk ( χα1 (ε′), . . . , χαk−1(ε ′) ) Sn(ε′) + Se(ε′) dε′, (19) for k ≥ 2. The analysis has shown that, for all ion- target combinations, the error of approximations (15), (16), and (17) is the largest for k = 2, when M1 ≈ M2. For all types of losses, the error of the norm can be negative or positive and generally does not exceed 5%. For the mean and the straggling, the results of using this approximation gives values which are smaller than true ones by no more than 15% and 25%, respectively. The skewnesses and excesses change the sign, and the errors can be large. In Fig. 3, we show the comparison of the numerical results obtained by solving Eqs. (4), (5), and (8) and the results obtained using approximations for the implanta- tion of boron ions in a germanium target. One can see from Fig. 3 that the approximate values of the parameters of the order k ≥ 3 are very close to the calculated values for high energies. For low energies, the approximate values have the correct order of mag- nitude and the general dependence structure. The max- imum errors for the mean and the straggling occur for the same energies. Therefore, we can use the approxima- tions in Eqs. (15), (16), and (17) or (19) for estimating the cumulants for the whole energy range used in the implantation. The results above show that, for the ELDs of vari- ous types, we can use the Gaussian distribution for the zero-order approximation. This approximation will be ISSN 2071-0194. Укр. фiз. журн. 2010. Т. 55, №2 241 V.V. ILYINA, M.V. MAKARETS Fig. 3. Parameters for the ELDs for the implantation of boron ions into the germanium target. Solid lines denote the numerical solutions, and the dotted lines denote the approximations true for the whole energy range used in the implanta- tion. If the parameters of higher order are large enough to distort the Gaussian distribution [2], we can use the Pearson-4 distribution. 4. Conclusions In the present work, it is shown that it is necessary to take into account in equations for the cumulants of the energy loss distribution (ELD) function that the energy loss is a continuous process, while the stop of an ion is an event. Therefore, the energy losses happen continuously along the whole path of an ion with high probability, while the probability of the ion stopping at the beginning of its path is very low. Accounting for these differences leads to different ini- tial conditions that must be used in the equations for the range distributions and ELDs. As a consequence, the equations for the cumulants of the loss distribution functions have additional terms on the right-hand side which are proportional to the stopping per unit length. This term gives the main contribution into the values of the cumulants and the corresponding parameters. On the other hand, the equations for the cumulants of the range distribution function do not contain this term and account only for nuclear scattering, which makes a con- tribution to the values of the cumulants of the ELD func- tion to be insignificant. 1. J. Lindhard, M. Scharf, and H.E. Schiott, Kgl. Danske Videnskab. Selskab, Mat.-Fys. Medd. 33, No. 14, 1 (1963). 2. A.F. Burenkov, F.F. Komarov, M.A. Kumakhov, and M.M. Temkin, Tables of Parameters of the Spatial Dis- tribution of an Implanted Admixture (Bel. State Univ., Minsk, 1980) (in Russian). 3. D.G. Ashworth, M.D.J. Bowyer, and R.J. Oven, J. Phys. D: Appl. Phys. A 24, 1376 (1991). 4. M.V. Makarets and S.N. Storchaka, Ukr. Fiz. Zh. 46, 486 (2001). 5. V.V. Ilyina and M.V. Makarets, Ukr. Fiz. Zh. 49, 815 (2004). 6. M.V. Makarets, V.V. Ilyina, and V.V. Moskalenko, Vac- uum. 78, 381 (2005). 7. J. Lindhard, V. Nielsen, M. Scharff, and P.V. Thomsen, Kgl. Danske Videnskab. Selskab, Mat.-fys. Medd. 33, No. 10, 1 (1963). 8. V.V. Ilyina and M.V. Makarets, Abstr. of the 2nd Con- ference FMMN’2008 (SCPT, Kharkiv, 2008), 2, p. 534. 9. E.G. Gamaly and L.T. Chaddernon, Proc. Roy. Soc. Lond. 449, 381 (1995). 10. A.D. Polyanin and A.V. Manzhirov, Handbook of Integral Equations (CRC Press, Boca Raton, 1998). 11. www.srim.org 12. J.F. Ziegler, J.P. Biersack, and U. Littmark, The Stop- ping and Ranges of Ions in Solids (Pergamon Press, New York, 1985), V.1. 13. W. Feller, An Introduction to Probability Theory and Its Applications (Wiley, New York, 1970). 14. M.D.J. Bowyer, D.G. Ashworth, and R. Oven, J. Phys. D: Appl. Phys. 29, 1274 (1996). Received 07.07.09 242 ISSN 2071-0194. Укр. фiз. журн. 2010. Т. 55, №2 DISTRIBUTION OF ENERGY LOSSES РОЗПОДIЛ ВТРАТ ЕНЕРГIЇ ШВИДКИХ IОНIВ ВЗДОВЖ ЇХ ШЛЯХУ У ТВЕРДОМУ ТIЛI В.В. Iльїна, М.В. Макарець Р е з ю м е Отримано iнтегродиференцiальнi рiвняння для кумулянтiв функцiї розподiлу втрат енергiї швидких iонiв вздовж їх шля- ху у твердому тiлi, у яких враховано втрати енергiї починаючи iз точки старту. Рiвняння для перших семи кумулянтiв розпо- дiлiв втрат енергiї у пружних, непружних та обох типах зi- ткнень розв’язанi чисельно за допомогою методу, розвинутого авторами ранiше, на iнтервалi енергiй iонiв 1 кеВ–1 ГеВ. Їх по- рiвняння з параметрами розподiлу пробiгiв iонiв показало, що: 1) середнiй шлях за розподiлом втрат енергiї становить 30–90% повного пробiгу iонiв; 2) при низьких енергiях страгглiнг за розподiлом втрат енергiї дещо бiльший або однаковий iз страг- глiнгом за розподiлом пробiгiв iонiв, а при високих енергiях перший перевищує останнiй у десятки разiв; 3) скiсностi i екс- цеси розподiлiв при низьких енергiях вiдповiдно близькi, у той час як при високих енергiях їх змiна для розподiлу втрат енер- гiї на кiлька порядкiв менша, нiж для розподiлу пробiгiв iонiв. Звiдси випливає, що розподiл втрат енергiї ширший i значно ближчий до нормального, нiж розподiл пробiгiв iонiв при всiх енергiях iмплантацiї. Показано, що цi властивостi розподiлу втрат енергiї зумовленi новими членами у рiвняннях, якi не враховувалися ранiше, але домiнують при високих енергiях. РАСПРЕДЕЛЕНИЕ ПОТЕРЬ ЭНЕРГИИ БЫСТРЫХ ИОНОВ ВДОЛЬ ИХ ПУТИ В ТВЕРДОМ ТЕЛЕ В.В. Ильина, Н.В. Макарец Р е з ю м е Получены интегродифференциальные уравнения для куму- лянтов функции распределения потерь энергии быстрых ио- нов вдоль их пути в твердом теле, в которых учтены потери энергии начиная с точки старта. Уравнения для первых семи кумулянтов распределений потерь энергии в упругих, неупру- гих и обеих типах столкновений решены численно с помощью метода, развитого авторами ранее, на интервале энергий ионов 1 кэВ–1 ГэВ. Их сравнение с параметрами распределения про- бегов ионов показало, что: 1) средний путь с потерей энергии составляет 30–90% полного пробега иона; 2) при низких энерги- ях страгглинг распределения потерь энергии немного больше или одинаков со страгглингом распределения пробегов ионов, а при высоких энергиях первый превышает последний в деся- тки раз; 3) скошенности и эксцессы распределений при низких энергиях соответственно близки, в то время как при высоких энергиях их изменение для распределения потерь энергии на несколько порядков меньше, чем для распределения пробегов ионов. Отсюда следует, что распределение потерь энергии бо- лее широкое и ближе к нормальному, чем распределение про- бегов ионов для всех энергий имплантации. Показано, что эти свойства распределения потерь энергии обусловлены новыми членами в уравнениях, которые не учитывались ранее, но до- минируют при высоких энергиях. ISSN 2071-0194. Укр. фiз. журн. 2010. Т. 55, №2 243