Brazilian Journal of Physics, vol. 25, no. 1, March, 1995

Latent Tracks in for Future Use in Nanoelectronics: an Overview of t he Present State-of-the-Art

D. Fink and R. Klett Hahn-Meitner-.lnstitut GmbH, Glienicker Str. 100 0-14109 Berlin, Germany

Received January 10, 1994; revised manuscript received August 2, 1994

Latent ion tracks might turn out to be useful as active elements in future nanoelectronic devices, after appropriate doping. For this salíe, new work has recently been initiated for better ion track characterization, and for understanding the ion track doping mechanism. A review of the present state-of-the-art is given.

I. Introduction jor meetings in different places of tlie world every two years[ll. It is known since about half a century ago tha.t ener- Due to tlie overwhelming success of the use of etched getic , after their passage through most insulating traclis ancl the difficulties to examine latent tracks in- materials, modify a certain zone aloiig tlieir trajectory, stead, interest in latent tracks has gradually decreased the so-called latent ion track, whicli is permanently visi- in the last decades. Only recently, some new ideas have ble in e.g. the transmission microscope (TEM). emergecl which led to a renaissance of latent track stud- The possibility to document iii this way tlie existence ies. One of them is related to the possibility of using and fate of individual ions has had a big impact on moclified individual (i.e. non-overlapping) latent ion many fields of science and teclinology, sucli as geology, tracks as f~ltureelectronically active elements of nano- mineralogy, oil exploratioii, paleontology, medicine, bi- metric dimensions in large-scale electronic devicesi2]. ology, materials science, space science, planetary sci- Tlie possible future use of these latent ion tracks in ence, nuclear physics, and nuclear chemistry. electronics would meet the trend to increasing degrees In most cases however, one does not observe nowa- of miniaturization in electronic~[~I(Fig.1). Apart from days these latent tracks themselves, but instead one this idea, the use of ion to inscribe sub- makes use of a very specific property of them, whicli micrometer conducting patterns (with overlapping ion is the enhanced etchability in agressive (mostly alka- tracks) in insulating matrices is of great inter- line and oxydizing) solutions - a property whicli may est, and more applications might be found in optoelec- exceed the one of tlie corresponding bulk materials by tronics and membrane technology. orders of magnitude. Thus, by removing the chemically modified and hence sensitive part of the latent tracks, holes are created with diameters in the order of inicro- 11. Tlie coiicept of "Single Ion Track Electron- meters (so-called "etch tracks"), whicli are easily visible ics" (SITE) in optical microscopes, and thus open the way for more rapid ion tracli registration. Due to the importante of There exist severa1 approaches to realize electronic this technology, even an international society has been elements with smaller dimensions and hence higher formed tlie "Nuclear Track Societyl" l, which holds ma- storage clensities than available nowadays. One at- ' The International Nuclear Track Society (INTS), Secretary: tempt, called , is crudely spoken - noth- Dr. P.Vater, Phillips-Universitat Marburg, Germany; temporary president: Dr. V.P. Perelygin, Dubna, Russia ing but an extrapolation of today's lithographic mask D. Fink and R. Iílett 5 5

meric material along individual latent ion tracks. This material is known to be enriched in sp2 bonds (occa- sionally e.g. in the form of conjugated double bonds, which exhibit especially high stability2), if the tracks have been produced by low energy ions. With increas- ing ion energy, i.e. with increasing deposited energy density, formation of complex matter such as aromatic or heterocyclic compounds and/or fullerenes along sin- gle ion tracks is also observed. This appears to hold for practically all hitherto examined polymers and other year organic matter (e.g. for PP, PI, PC, PMMA, PET, Figure 1: Historical development of the number of atoms needed for a storage device, after Ref. 3. Added are ex- PTF, and saccharose), and therefore we dare to make pectation values for nanolithography, molecular electronics, these statements in that generalized way. and SITE. For SITE, the different values shown correspond The idea to make use of organic materials enriched to different track sizes, ranging from: 10 A track diameter 2 4 and 10 pm length R up to: 100 A diameter and 100 pm in sp bonds in polymeric electronics is not new in itself. length. For exampIe, materials rich in conjugated double bonds

- thiophenes and oligomers of thiophenes - have already techniques to smaller dimensions, by replacing today's been used for the production of FET's[~].Large blocks light sources by other ones of smaller wavelength, or of sp2 bond-rich materials are produced in industry by even by particle beams. Another one, the molecular chemical reactions, thermal treatment (pyrolysis), or electronics, can be regarded as the attempt to make by irradiation of polymers with UV, X rays, Gamma use of, and to modify God's finest technology - i.e. bi- rays or . What is new here is the use of high ology - for human requirements. This approach is char- energy heavy ion irradiation, so that these materials acterized by a number of hitherto still quite unusual form small and well-defined structures with relatively electronic peculiarities, such as the task to handle two sharp borders within an else undisturbed matrix, and types of charge transport in coexistence - electrons for thus can be used for future nanometric-size electron- short range currents, and ions for long range currents -, ics. Most of these above mentioned modified materials an aqueous environment, complicated membrane chem- with double bonds, phenyl rings or fullerene structures istry, parallel data processing, high fault-tolerance, self- are insulators in their neutra1 ground states, if consid- assembly and self-reproducibility of the whole system. ered individually. Bowever, they become semiconduct- Due to the enormous difficulties involved, it is ing upon addition or removal of electrons, i.e. upon doubtful whether this fascinating technology can be doping. realized in the near future. We have therefore pro- It is well-ltnown that after low energy high dose ion posed another simpler approach to nanoelectronics by irradiation polymers transform into highly cross-linked making use of single (i.e. non-overlapping) latent carbon-based networks. Their electronic band structure ion tracks in non-conducting polymers[2] as active might possibly be intrinsically semiconducting, so that electronic elements. There exist similar approaches, e.g. doping could just act on this basis. In fact, phases by filling up etched tracks in inorganic solids (such as of a-C and a-C:H are known to behave as conducting ) with semiconducting (e.g.si)L4] or metallic (e.g. and semiconducting material^[^^^]. It appears however Cu) matterf5], via electrolytic deposition. This tech- that this type of material modification is restricted to nique leads to somewhat larger device sizes as a result the case of multiple ion-track overlapping, and does not of the etching procedure. In this work we shall however occur in single ion iracks: We could demonstrate that concentrate on the use of latent tracks in polymers, and single ion tracks exhibit a very poor conductivity, so not discuss the application of etched tracks further.

We want to make active use of the modified poly- 'as an example, see the nodificationof PE by 60 keV O+ 161. 56 Brazilian Journal of Physics, vol. 25, no. 1, March, 1995

that we can largely exclude the existence of tlie above of -doped ion traclt~[~]which, if connected prop- mentioned cross-linlted carbon-based networks and/or erly with some externa1 cooling facility, will act as heat extended phases of a-C or a-C:H. sinks. The relatively small track-to-track distances in Therefore, doping of latent ion tracks will be of es- such devices should enable, in principle, efficient heat sentia1 importance to activate these traclts electroni- removal. Further one might think at depositing the re- cally. Polymers can be 11- and p-doped[lO].Again, this latively thin polymer foils onto a thiclt metal substrate. is nothing new by itself - the feasibility of polymer dop- Secondly, tliere is tlie question of device stability in ing for producing electronic devices ha.s been ltnown for SITE. It is well-ltnown that many polymers undergo a long time[ll]: What is new here is tliat the dopants long-time clianges, and that dopants exhibit quite a are expected to act only in the spatially confined ion high diffusitivity in pristine polymers. If no precau- traclt regime, but not beyond. In earlier experiments tion was talten, this mobility would readily destroy any of polymer doping, the dopants have been diffused or newly prepared structural arrangement. There exist implanted at low energies into polymers. In the latter however ways to improve this situation. One is the case relatively high ion fluences hacl been used so that careful selectioii of appropriate polymeric materials. homogeneously modified materials ha.ve been formecl a.s Anotlier one is to suppress uiiwanted dopant mobility, the result of multiple track overlapping. For contrast, wliich can be clone by trapping the dopants at defects we are now interested in using ion bea.111~at very low along tlie traclts. Efficient trapping has the consequence fluences, so that a heterogeneous material results, witli that the dopants will migrate from the bulk towards the individual non-overlapping moclified and dopecl zones tracks ratlier than the otlier way round, and lience as- (latent tracks) within an else undisturbed matrix. sure device stability. Proceeding from microelectronics to nanoelectron- A favorite class of polymers with very liigh sta- ics imposes onto the device-constructing engineer sucli bility is composed by the , PI. They have weil-ltnown problems as the restriction of the overall been developed for use in microelectronics where the electric current density to prevent excessive heat deve- teniperature may reach 400" or more during process- lopment. This restriction implies that only a severely ing. The high cliemical stability and unusual heat limited number of electrons is allowecl to flow througli resistailce of these polymers are due to the presence each individual element of such a device per unit time. of aromatic heterocycles, which brings stability to tlie In order to obtain reliable switching of sucli a device, polymer slteleton[12]. Another polymer with extraor- the number of passing electrons must be sufficiently dinary stability in air is poly(paraphenylene), PPP. It high to avoid failure by fatal statistical fluctuations of can be implantation-doped with e.g. alkali-ions or halo- the electron current through tliis element. Vice versa, gen ions, to produce respectively n-type and p-type this means that there exists a lower threshold in switch- semicond~ctors[~~]. ing frequency for such an element to enable a minimum Apart from using polymers as carrier materials of number of electrons to flow through this element, before the latent traclis, it lias been proposed to use dzamond the switching status may be reversed reliably. in~tead[~>~~].Though this proposal is fascinating due From this argumentation it follows tha.t it will be to tlie many properties ~vliiclimake cliamond superior desirable to have an as high as possible overall current to any other electronic material['], it is more difficult density fiowing through the nanoinetric device. Here to realize - diamonds are still less readily available and the application of polymers is, in principle, connected more expensive than polymers, and thermal doping ex- with some disadvantage (as long known in microelec- periments with diamonds require usually high temper- tronic) due to their well-known sensitivity to heat. To atures up to 1000°C and more. Natural diamond as prevent overheating of the proposed polymeric SITE well as CVD cliamond films have been shown to exhibit devices, one could therefore thinli e.g. at overlayiiig a strong decrease of resistivity (by up to 10 orders of the electronically active ion track array with a grid magnitude) after ion implantation, a behavior similar D. Fink and R. Klett to that found in the case of polymers[~l. in the neighboring bulk material, as well as de- One of the open questions in this connection is the tailed knowledge about the dopant's modification 'riddle' of 'missing' latent ion tracks in dia.mond - hith- of the electronic bond structure of tiacks by the ert,o they have never been observed by TEM. However, dopant are necessary. Again, experiments should there have been performed experiments which point be closely accompagnied by tlieory. to the existence of an (invisible) zone of defects along c) A recipe has to be found how to position the ion an ion's traje~tor~[~~~~](realized e.g. by decoration of tracks in regularly predetermined arrangements. If the defects along the tracks of 124 MeV Xe in dia- no such provision is taken, the tracks will dis- mond with postimplanted 300 keV He, and subsequent tribute randomly across the target. with the in- dopant at 1000~~[~])which is capable to trap herent problem of how to contact a11 these indi- dopant/impurity atoms, and which we hence might cal1 vidual elements, and to maintain reproducibility a 'latent track', by using a somewhat more generous of the whole device. definition of latent ion tracks than it is done usually. d) The doped latent ion tracks have to be contacted Braunstein et al.Ig] could reconfirm already long time properly, in order to combine them to the prede- ago the formation of point defects along ion trajecto- termined logic circuits. ries in diamond by ion channeling for the low dose ion irradiation case. Hitherto, some work has been irivested in the first Hitherto performed experiments indicate that the two points, which will be reviewed below. Nothing has principal mechanisms which govern the dopant behav- yet been done to answer the questions of track posi- ior in ion tracks of diamond and polymers might partly tioning and contacting. There are some ideas - e.g. to be similar, though differing by orders of magnitude in make use of microbeams, or to use specifically modi- their absolute values. Therefore one might understand fied transmission channeling through Moiree patterns polymeric ion track research not only as a new field for precise track positioning, or of 'writing' conduct- by itself, but also as a means to enlarge the principie ing connections between the tracks by STM, or by low knowledge of ion track doping in carbonaceous materi- energy ion beams, but a11 these ideas still wait for real- als which might later be applied elsewhere. ization. In order to proceed to the proposed SITE in poly- In the following sections, we shall give an overview mers, four basic steps have to be fullfilled: of the characteristic ion track properties. Here, we shall look preferentially for general overall trends, and care a) First, we need accurate information about the for details of individual systems only if necessary. geometrical, chemical, structural, and electronic properties of latent tracks in polymers, and about 111. Latent ion track characterization their stability. Only when a latent track can be 111.1 Geometrical considerations well-characterized in a11 these âspects, reliable tai- loring of the latent tracks for specific applications The shape of a latent ion track is primarily defined will be possible. For this sake, it is highly desir- by the 3-dimensional distribution of electronic excita- able to have a comprehensive theory of ion track tion energy, transferred by the incoming projectile to formation. the polymeric target. It could be shown by study of The process of ion track doping, essential for the the gas evolution during irradiation that the damage ion tracks' electronic activation, has to be under- is localized in the tracks of the incident particles[16]. stood precisely. This involves questions of dopant High energy ions follow straight trajectories in a poly- mobility and trapping and detrapping efficiencies. mer, so that the latent tracks are in general linear (and The dopants should sharply concentrate along the for parallel ion beams hence parallel) structures with tracks and be depleted in the neighboring re- some slight lateral expansion. This is favorable for ap- gion. Diffusion coefficients along the traclrs and plication to SITE. Riaziliais Joumal of Physics, vol. 25, no. 1, March, 1995

Projectile bacltscattering (resulting in catastrophi- ties. They also become important with higher degrees cally deflected ion tracks) can occur only if the target of polymeric destruction, when many ion tracks over- atoms' masses exceed the projectile mass, so that, for lap. However, the latter case is not considered here. application to SITE, heavy ion irracliation is favorable. The lateral strucfure of an ion traclt at a certain

In this case, also small angle multiple ion scattering and deptli can 1x2 described by a number of concentric shells, electronic scattering effects are negligible. each shell corresponding to a certain threshold value The ranges of such projectiles can be dcscribed of transferred electronic energy density, Fig. 3. This with good accuracy by J.P.Biersac1c's Monte-Caclo code picture is probably oversimplified as it is not yet clear TRIM['~],or by his analytic code PRAL[~",wliere tlie whether there really exist sharp energy thresholds for Bethe-Blocli electronic stopping po~er[~~]is applied. tlie differcnt effects, or wliether these thresholds should They are in a typical order of magnitude of some 10 better be replaced hy broad distributions3 Thus, tliese micrometers for 100 MeV heavy ioiis, aiid of some 100 eiiergy thresholds - may they be sharp or smeared out

micrometers for GeV heavy ions. - primarily determine the "effective latent track radii" The distribution of energy transferred to tlie elec- for tlie various effects under consideration - e.g. for the tronic system of tlie target ( "electronic stoppzngr", Se) clopant mobility, chemical changes, visibility in TEM, is a smooth curve with a maximum shortly before the etc (for illustration see Fig. 3) - These effective radii particles' range and a l-iigh but ratlier constant due may be furtlier enlargecl by secondary effects such as in the surface-near area, Fig. 2. It is this effect which e.g. diffusion of the formed radicais. Characteristic determines most of the ioii track's properties, sucli as valucs for effcctive ion track radii are listed in Table chemical chanqes, conductivity etc. (see below) . 1 including the corresponding reference~[l~-~~].They are only a little depenclent on the specific ion and polymeric target under consideration. First attempts 2.2 GeV Au-PMM have been inade to describe the lateral traclí extension 1 tlieoretically[2".

111.2 Latent ion track chemistry

At present we are still far away from possessing a thorough overall knowledge about latent track chem- istry. Tliere liave been a number of scattered 'mosaic stones' in this field, from which one can derive already an overall picture of ion track chemistry, provided tliat one is courageous enough to fill out the still missing ltiiowledge with a good deal of speculative inter- and DEPTH [pm] extrapolations. This preliminary procedure appears to Figure 2: Depth distribution of deposited particles, nuc1ea.r be justified by the observation that many of the basic and electronic energy transfer, for tlie example: 2.2 GeV Aii radiochemical effects very closely resemble each other ions in PMMA (TRIM ~alculation[~~~~[")). for different irradiated polymeric targets, thus allowing In contrast, the collisional energy transfer ( ('nuclear for a certain ílegree of generalization of statements. The stopping", S,) is negligible along most of the ion traclt, materials upon which most of the research worlt has except for its very end. Nuclear energy transfer pro- been based hitlierto are common-type polymers such cesses usually create less but more stable defects than 3Recent TEM and STM pictures of latent ion tracks show electronic t,ransfer processes. Therefore they can occa- structures with astonishingly sharp boundaries, which might be understood as an indication that the thresholds of different ion sionally gain some importance for trapping of impuri- track properties are rather well-defined.

6 O Brazilian Journal of Physics, vol. 25, no. 1, March, 1995

as foils of PMMA, PE, PP, PI, PC, PET, PTF, PS, or subsequent chemical reactions like rearrangements and photoresist. Most of the examinations reported in the readditions can occur between the reactive species and literature have been performed with low energy ions in the modified polymer[29]. Applied to our aim of SITE, the regime of 1 1teV to 1 MeV. Hitherto there exists this means that one should use ion tracks of heavy en- only a marginal number of papers dealing with radio- ergetic ions rather tlian of light ones, to obtain best chemistry of ions with MeV to GeV energies, such as long-time device stability. e.g. Refs. 29 or 30. Electronic energy transfer processes lead primarily The latent ion track regime is characterized by two to excitation and breaking of bonds, with the con- effects: (a) destruction of existing components, and (b) sequences of chain scission, cross-linking, and/or for- formation of new materiais. Chemical bonds in any mation of new types of bonds (e.g. C=C), depend- polymer can be broken whenever the energy transfer to ing on the system under consideration. These effects the electronic system of the target exceeds the inter- have long since been studied in great detail. For ex- atomic bonding energy. The latter is in the range of ample, it became possible to distinguish experimen- a few eV only (e.g. H-CH: 4.3 eV, CH3-CH3: 3.7 eV tally between interchain and intra-chain bond cross- [31]). It has been established that upon electronic en- linlti~~~[~~].It was shown that soluble polymers trans- ergy transfer, bonds are not broken at random. Rather, form into non-soluble gel upon introducing one cross- experimental evidence shows that certain selectivity link per macromolecule~33~.Aromatic polymers irradi- rules which do not only follow bond energy consider- ated with ions at low deposited energy densities act as ations apply. Thus in linear hydrocwbons, C-H bonds efficient 'energy sinks' insofar as the aromatic rings can are broken more frequently under irradiation than C- dissipate a great deal of the excitation energy[32]. How- C bonds, in spite of the higher bond strength of the ever, for ion irradiation of these compounds at higher C-H b~nd[~'].Bond breaking will be strongly favored energy densities (i.e. heavier particles and higher flu- in the core region of a track, where most of the energy entes), the loss of aromatic conjugation is one of the is transfered, whereas it will occur only occasionally in predominant destruction processe^[^^^^^]. the track's penumbra, due to energy transfer densities Production of a vast amount of volatile irradiation which are lower by orders of magnitude. products with subsequent strong gas emission is a char- Calcagno et al.L3'] have performed optical and rhe- acteristic effect of ion track formation in polymers. As ological examinations of 100 1teV I-I and 300 1teV He the gas emission is a diffusionaliy controlled process, irradiated PS in the single ion track regime. They de- degassing will be strongest at the surface and slowest rived cross-link production yields of 12 cross linlts/ion, near the track end. Thus an inhomogeneity in the dis- respectively defect production yieids of 100 defects/ion. tribution of residual (i.e. slowly moving, large) volatile This corresponds to chemical yields of 0.07, respectively components along the ion track will emerge. 0.28 defects/100 eV deposited energy. From these ob- With increasing projectile energy the transferred en- servations they concluded that the energy distribution ergy density is further increased, until a11 bonds along inside the single ion tracks is the controlling factor for the central region of the latent ion track (the so-called the chemical yield. (ion track core 3 are broken. Thus a highly energetic ion It has been observed for PVDF modification by en- is capable to produce completely new compounds, with ergetic heavy ions (1 to 50 MeV/amu O, Kr, and Xe no connection to the originally present ones. The higher ions at fluences of some 1011 to 1012 ions/cm2 [29]) that the projectile energy - i.e. the higher the deposited en-

the major parameter for polymer modification is the ergy density along the ion track core - the more time projectile atomic number: Destruction is greatest for it will take for dissipation of the target atoms' excita- heavy ions, and it remains stable along the ion tracks in tion energy, and hence the more complex will be the the bulk. For contrast, activated centers introduced by compounds which can be newly formed in this track lighter ions tend to migrate towards the surface, where core. D. Fink and R. Klett

This has actually been verified (e.g. for 1.5 GeV old values could not yet be derived, but they are as- Bi in Pi, 2.2 GeV Au in PMMA[~~~~~],2.1 GeV Dy in sumed to lie in a similar order of magnit~de.~To HOPG[~"~~],and 3 GeV U in sacchar~se[~~]):A large give the reader a feeling it should be mentioned that a amount of new (po1y)cyclic compounds emerges from GeV heavy ion impinging into a sugar crystal is capable

GeV irradiated organic matter - independent from the to create typically some 150 C60 and 3 molecules aromatic or aliphatic of the original materials along its trackL40]. Taking into account the relatively - and even fullerenes are seen to develop (at present, high degree of dilution of carbon atoms in sugar (only both Cso and C70 could be identifiedL4OI). Experimen- about 25% carbon atoms, among a majority of H and O tal data from GeV irradiated PMMA suggest a thresh- atoms), we should expect that the production efficiency old in linear energy transfer along the tracks of around of fullerenes by GeV ions in more carbonenriched mate- 1 ke~/Afor the production of polycyclic compounds, rial (such as graphite) should be larger by some orders corresponding to a threshold energy density in the ion of magnitude. For example, theory predicts some 105 track core of around 3 ev/A3. Above this threshold, to 106 CGOmolecules per ion in GeV-radiated graphite. the production yield of these compounds increases very However this is not yet verified experimentally - present strongly, as shown in Fig.4. For example, the pro- experimental limitations in fullerene detection from ion duction yield of the above cited polycyclic compounds tracks in e.g. graphite stem from the very low leaching formed in GeV irradiated PMMA is of the order of some efficiency of these molecules out of the tracks. 10% or rn~re[~~>~lI. There might exist an inverse correlation between the production efficiency of polycyclic compounds and of fullerenes: For GeV irradiated PMMA, where many polycyclic compounds have been found to emerge, prac- lu 1PRODUCTION EFFICIENCY I tically no fullerene was formed, whereas for irradiated OF FUNCTIONAL GROUPS 16 I saccharose we found relatively much Cso and less new aromatic compounds. It may be that the chemical pro- duction paths are influenced not only by the degree of carbon dilution in the presence of high H and O abun- dance, but also by the mobility (and hence loss) of H- and H-enriched compounds along the ion track regime. The depth distribution of the production yield of these compounds always follows closely the one of the electronic energy transfer. Only for high doses, when traclts are at least thousandfold overlapping, an in- fluente of nuclear energy transfer processes becomes visib~e[~~].

4The production of fullerenes inside the ion tracks may not be confused with the spvtter-emission of fullerenes from car- bonaceous materials, see e.g. Ref. 42: Here, it could be shown that, upon penetration of an energetic ion into polymers or into fullerite, material is ejected from the surface regime down to a depth of roughly 200 A of the emerging ion track. Upon pen- etration of this supersonic jet-like cloud of free neutra1 carbon Figure 4: Production yield of polycyclic compounds in poly- atoms through the track surface, adiabatic expansion sets in dur- mers as a function of the projectile's linear energy transfer, ing which carbon clusters are formed, among which are the above- cited fullerenes. As fullerene formation in an expanding plasma as verified by specific functional groups in the polymers' requires a lower energy density than fullerene formation inside FTIR spectra. Shown here for the example of 2.2 GeV Au highly confined solid matter (but takes longer time), sputter- ions in PMMA[~~]. emission of fullerenes could already be observed for impinging projectile energies as low as some 70 MeV [42], which is about one order of magnitude lower than the assumed threshold for For fullerene formation, the corresponding thresh- fullerene production inside ion tracks [43]. 6 2 Brazilian Journal of Physics, vol. 25, no. 1, March, 1995

The hitherto applied technique (FTIR) alone does non-overlapping) latent ion traclís could be measured not allow to distinguish between tlie different types for tlie first time. (By 'intrinsic conclucfivity' we mean of ion-created aromatic compounds. Concerning here the conductivity of as-implanted ion tracks, in con- fullerenes, only the two most abundant types, CC0and trast t,o t,he ion track conductivity after doping). Im- C7", could be identified[40]. In order to improve this mediately after the track formation, this intrinsic con- situation, a combined c11roma.tograpl-i/ mass spectrom- ductivity seems to decrease rapiclly by about at least eter should be used in future. one order of magnitude, due to the reduction of free ~heory[~~Ipredicts tliat tlie production of aromatic electrons (by annealing and eventually also due to ox- compounds and fullerenes will be restricted to tl-ie outer idation processes - observecl for irradiation of PP foils region of the ion traclí core. Only inside the core with various ions at some 30 to 100 MeV eriergie~[~']). the necessary minimum energy density thresliolcl for It turns out that the intrinsic conductivity of a sin- bond hreaking is maintained for a sufficiently long time gle ion track in e.g. PI in its stationxy state (i.e. at to enable the formation of aromatic compounds aiid least some days after the implantation) is of tlie order fullerenes. Furtliermore, another threshold conclition of 10-~' ohm-' which mems that every second only defines the maximum possible energy density transfer: 1 to 5 electrons pass a single ion traclr (as determinecl The newly formed components must be quencliecl suffi- for about 10 liV/cm electric field strengtli applied, i.e. ciently rapid after their formation, to prevent excessive still below the onset of electron avanlanches or break- decay backreactions. Tliis condition is frequently not through). Tliese data liave been derived from a wide fullfilled in the very center of traclis of energetic ions. range of measurements with different projectiles - from Thus we are restricted for tl-iis special type o€ radio- 2 MeV EIe to 3 GeV U - for low eiiergy deposition rates cliemistry in GeV ion implanted polymers to a narrow of 109 to 1013 ions/cm2 in commercial (Good- cylindrical traclí regime of typically some 30 to 200 A fellow LTD)[~~>~~~~'~. radius (as calculated by Gamaly and ~liadclerton[~~]). The overall polymeric conductivity was found to in- In the outer regions of the tracli, competitive chem- crease proportionally to the ion tracli den~ity[~~].Wlien ical reactions wliicli do not require the brealiing of a11 a11 the polymeric target is nearly homogeneously cov- bonds, with only partially destroyed polyiners miglit erecl with single (but not yet overlapping) ion tracks, take place. Thus it has been recently shown by the overall polyiner conductivity (in tlie case of poly- FTIR that aromatic esthers are crea.ted togetl~erwith imide) has risen up to typically 10-l~ohm-'cm-l. polycyclic compounds in 2.2 GeV Au ion traclís in The conductivity values for other polymers (e.g. PP, PMMA[~~].This aromatic estlier formation cai1 be ex- PC, PET, PMMA or PTF) are of a similar order of plained on the basis of classical cliemical reactions with magnitude[45]). Altliough the intrinsic conductivity of the original polymer, whicli lias been verified by a. sim- irradiated polymers exceeds the one of unirradiated ulation e~periment[~~l. polymers by some orders of magnitude, tliese values The expected radial variation of ion tracli cliemistry still lie in the range of definition of insulators (thougli has not yet been verifiecl. It inight iinply tliat, on ion bad ones): We are still far away from materials with tracl; doping, various concentric zoiies of different elec- semiconducting or even conducting quality, which only tronic conductivities emerge around the ion track core, can be obtained after doping. Tliis experience has been due to tlie different chemical compounds created. indirectly reconfirmed by many authors who studied polymeric conductivity changes after low energy low 111.3 Intrinsic electronic properties dose irradiation: ion irradiation at doses below track overlapping "does not produce an appreciable drop in A light modificatiori of tlie ~vell-ltnowntwo -point resistivity" [351. technique for resistivity measurements enabled us to measure lower conductivities than was ever possible The conductivity of an irradiated polymer becomes bef~re[~"~~].Thus, intrinsic conductivities of single (i.e. isotropic when the ion traclís start overlapping. In this D. Fink and R. Klett 63 case, the overall polymeric conductivity does no longer model of curreiit transport between localized ~tates[~'], increase witli fluence, as given in the single ioii traclc or by tlie process of three-dimensional electron hop- case, but remains constant witliin a fluence range of ping from one conducting island to an~ther[~~].Both about 3 orders of magnitude (Fig. 5). For tlie single moclels yielcl the same results, so that an experimental ion traclc case, tlie conductivity in transversal clirection decision between them appears to be rather impossi- (i.e. from one traclc to the other) equals the value of the ble. A inoclified concept of electron liopping between pristine material mitliin our measuring accuracy (shown conducting aggregates seems to be also a good model by us for traclcs in PI[~~]). for the description of the much poorer single ion track conductivity (see below).

111.4 Structiiral aspects

Tlie modified polymeric matter inside ion traclts is not homogeneo~islydistributed but exhibits fluctu- ations in both tlie carbon ancl electron clensity. Tliis lias been derived from small angle neutron scattering + 0.5úev I -PI (sANs)[~~],respectively small angle X ray scattering ' x ISGeVXe-PI MAXIMUM NOISE .*o 2 MeV He -PI (sAxs)[""J~~>~~],and from UV/vis. ~~ectra[~~],and this n 340MeV Xe-PI o 8OMeV 6-PI is also the basic assum~tionof the theories on intrinsic polymer ~onductivity[~~~~~]- see above. The iiidivid- ual carbon and electron enriched zones appear to have Figure 5: Dependente of intrinsic electronic conductivity in near-spherical, slightly ellipsoidal shapes, their symme- ion irradiated PI on tlie transferred electronic energy den- try axes being oriented along the ion traclí directi~n[~~I. sity, for a) tlie single ion track (SIT) regime, b) tlie miilti- ply overlapping track (MOT) regime of constant coiiductiv- The latter finding resembles earlier assumptions tliat ity, and c) the MOT regime of strong conductivity clianges. tliese inclusions miglit liave s filamentary shape elon- Straiglit lines to guide the eye, to oiitline the principle ten- gated along the traclí of the incident ions[12]. dency. For even liiglier transferred energy densities (about Typical diameters of the carbon-enriched zones are 10-' to 10' ey/A3), a tliird regime emerges with extremely liigh conductivity a a result oE carbonization processes (not reportecl to be 6 to 7 nm (as measurecl for 50 MeV shown Iiere). B traclm in PET at 1013 ions/cm2 by SANS[~~]),and tlie diameters of the electron-enriched zones are found For deposited electronic energy densities exceecling to liave sizes of 0.2-4 iim (according to UV/vis. spec- 10-I ev/A3 (i.e. at roughly thousandfold ion track trometry, obtained e.g. for some 300 1íeV He to Ar overlapping) extremely strong chemical and/or struc- ions iii PS, PE, and PET at Auences between 10i4 and tural clianges set in, wliich lead to a dramatic increase 101"ons/cm2 [54], [55]), 5 to 6 nin (as measured for in conductivity, until that, at around 102 ev,Ih3 trans- 2.2 GeV ALI ions in PMMA at 5 x 10'' ions/cm2 by ferred electronic energy density, near-graphitic concluc- SAXS[~~$~~~~]),and 10 to 50 nm according to the con- tivity lias been reached. Extensive experimental worlí ductivity tlieories (as estimatecl for some 100 1teV lieavy lias been performed for this higli-dose regime of poly- ions at fluences around 1015 ions/cm2 [49], [50]). The meric destruction in the last clecades (see, e.g. Ref. average distance between these zones is estimated to be 47). Both conductivity and optical transparency are 13 nin by SANS[~~],and 0.1 to 20 nm from intrinsic con- not only influenced hy electronic energy transfer in this ductivity res~~lts[~~~~~].As those values liave been ob- regime, but also atomic collisions begin to play ali im- tainecl for different implantation conditions (liglit and portant role[12>21>36>4"1. lieavy projectiles, low and higli energies and fluences), Theoretical models liave been derived long ago to a systeinatic depenclence on implantation parameters describe this conductivity regime, either by a linear cannot yet be derived from tliese data and they just 64 Brazilian Journal of Physics, vol. 25, no. 1, March, 1995

give a feeling for the possible orders of magnitude of 5 x 101° ions/cm2 [45] from SANS results on B irradi- this effect. Anyhow, we can conclude that the forma- ated PET[~']). tion of these aggregates is not a specific high dose effect ESR measurements on irradiated polymers (specif- resulting from ion track overlapping, as the existeiice of ically shown for 100-300 MeV Ar and Xe irradiated these aggregates has also been verified for the case of PI and PC) reveal, among other signals, some which non-overlapping single ion tra~ks[~~>~']. depend on the implantation dose. They might stem Nothing is known about the composition of these from brolten C-H bonds. For 'old' latent ion tracks (i.e. aggregates and speculations range from crystalline tracks which had been exposed to atmospheric oxygen graphite platelets[12~54~56~57],via giant fullerenes["] , or for about half a year), a density of about 500 unpaired glassy ~arbon[~~],up to benzoid ring c~usters[~~],or spins per traclr was fo~nd[~~l~.)Combining this result just simply pitch-and-tar-lilte matter[51]. Another re- on a still speculative basis with an estimated number cently developed idea[43]is to explain these agregates as of 10.000 carbon- (and electron-) enriched aggregates precipitates of the irradiation-formed fullerenes and/or per track (see above), we arrive at roughly 1 unpaired poylcyclic matter along the ion tracks. The observation spin per 20 aggregates. This might indicate that the of Davenas et a1.[12] that conductive polymeric proper- measured intrinsic polymer conductivity results from ties after low energy high dose irradiation are always a combination of electron hopping between the aggre- connected to the presence of aromatic or heterocyclic gates, with trapping/detrapping processes (at the un- structures in the polymeric precursor appears to be paired spins) for the electrons inside the aggregates, quite interesting in this connection. which would make understandable the low observed in- trinsic ion traclt conductivity. It may be mentioned in It should be pointed out at this occasion that the this connection that the density of unpaired spins in presently available experimental results derived from pure (i.e. undoped) fullerite is also very low (around UV/vis. spectrometry for the cluster diameters always 10~~spins/molecule), so that even the presence of a lead to values smaller than the ones derived from SANS major number of fullerenes along the ion tracks would or SAXS. The reason is that the underlvine:" " model is not alter the above given picture. based on the assumption that these aggregates stem from diffusional clustering of benzene rings along the 111.5 Theoretical approach for deseription of track; the observed optical energy gap is then corre- track formation lated with the number of benzoid rings by a simple formula[55]. This formula would, of course, be invalid Severa1 historic models exist for the description of if mechanisms other than benzoid ring clustering were latent ion track formation, e.g. the "Coulomb Explo- responsible for aggregate formation. We have to leave sion Modeln["] and the "Therrnal Spike ~odel"[~~].A the final decision still open. It has been shown that Monte Carlo model, based on the Binary Encounter the growth of these sp2 clusters is correlated with the Approximation, has been developed recently to study loss of hydrogen, the presence of which tends rather to details of the secondary electron emission and migra- stabilize the sp3 phase[55]. tion inside tracks of heavy charged particles[28]. Unfor- We do not yet know with certainty whether the tunately, this model is at present restricted to a very carbon-enriched aggregates (as vizualized by SANS) are special case only: the ionic energy deposition in water identical with the electron-enriched ones (as detected vapor. by SAXS), though the similarity in geometrical param- There exists a new model which puts special em- eters malres this identity highly probable. However we phasis on the formation of excited (i.e. non-bound) but do know that the depth distribution of the size of the neutra1 target atoms aiong the ion track~[~~].We pre-

aggregates follows the electronic energy transfer curve, 5This result for energetic heavy ions has to be compared with hence that the formation of these aggregates stems from those of low energy ion irradiation: around 1 spin/ion track for 60 keV Ot in PE at 5 x 1014 to 5 x 1016 ions/cm2 [6], respectively electronic processes (shown for 2.2 GeV in PMMA at for 50 keV C+ in diamond at 3 x 1013 ions/cm2 [58]. D. Fink and R. Klett fer this latter model for an analytic description of the between neighboring atoms which may lead to the pro- ion track, as it avoids the inherent difficulties of e.g. duction of new materials along the track. A precondi- the Coulomb Explosion model. The latter one assumes tion for this process is of course that the energy density that, after bond breaking and ionization of the target in the track regime exceeds the threshold energy den- atoms by the impinging projectile ion, most target elec- sity necessary for formation of these new compounds. trons are expelled from the ion track core for a very The formation of specific high temperature/high pres- long time, so that the remaining target ions are free to sure compounds such as fullerenes6 will be favored as move away from each other by Coulomb repulsion. In long as the energy density in the track core exceeds order to explain the absence of immediate electron-ion their specific formation threshold value, and as long recombination which would prevent any ion motion at as the lifetime of this hot region exceeds the time re- all, the model assumes that quired for formation of these compounds. A system of n rate equations can be established to describe the a11 electrons suffer very high energy transfer so dynamic equilibrium between formation and decay of that they are capable to move very far away (up carbon clusters C, with n atoms ea~h[~~].This pro- to a few micrometers) from the ion track core ('5- cess will come to an end when the energy density in the electrons') , track regime drops below the threshold value for forma- the attractive Coulomb field between electrons tion of the C, clusters, as a result of heat dissipation and ions is shielded by a conducting layer in case by thermal diffusion. of conducting targets, respectively, The use of these n rate equations enables one to in case of insulating targets, the electrons - once make predictions for the production yields of the C, having lost their kinetic energy - are no longer clusters, provided that "reasonable guesses" were made capable to move back to the ions in the insulating for the reaction rate constants in these equations. Ap- target, but are rather permanently trapped. plying this approach to the above mentioned case of Simple formulae, taken from plasma physics, show 1.5 GeV Dy in HOPG, one ends up in some 105 to 106 however that the average target electron gains only lit- predicted fullerene molecules per ion. (This value could tle energy and hence migrates only some 10A (as cal- not yet be reconfirmed experimentally, as our hitherto culated for the system: 1.5 GeV Dy in graphite in Ref. applied detection method is by far not yet efficient 43), so that enough to isolate a11 formed Cso molecules from the track.) It is expected that further work on this model a) there is no space for building up a shielding con- may yield more detailed informations, e.g. on the pro- ducting layer in a conducting target, respectively duction yield of fullerenes and polycyclic compounds in b) the electric field between electrons and ions still organic matter, and on their radial distribution along exceeds the breakthrough voltage of an insula- the track. tor so that the electrons cannot be permanently trapped, IV. Doping of ion tracks c) the recombination time is hence so small (ca. IV.l Experimental data base 10-l4 - 10-l5 sec) that the target ions have no chance to migrate away from their position by We have seen above that latent tracks of energetic Coulomb repulsion. (MeV... GeV) ions in a11 common polymers contain a number of promising compounds which - though being The excited but neutralized atoms will gradually re- 6According to the recently derived phase diagram of carbon turn to their ground states via emission of or allotr~~es[~~],fullerene formation is characteristic for the high temperature T / high pressure p regime. This has to be compared by electron-phonon coupling, with the consequence of with diamond (low T, high p), graphite (low T, low p), and with heating the ion track core and its immediate environ- carbon vapor (high T, low p). In other words, fullerenes and diamond are metastable allotropesunder 'normal' conditions (low ment. During this time, new chemical bonds are formed T, low p). Brazilian Journal of Physics, vol. 25, no. 1, March, 1995 bad insulators themselves iii their neutra1 grouncl states (examples: I indiffusion into PI at 60°C, preirradiated

- cai1 be readily turned to semiconductors by appropri- with some 100 1teV noble gas i~ns[~~],or diffusion of ate doping. Therefore ion traclí doping is a crucial point LiC1 soliition into PI at R.T. and at 100°C, preirradi- i11 tlie tailoring of SITE. To this aim, we have to linotv ated with some 100 1ieV Ne, respectively 340 MeV Xe io~is[~").Tliis ineans that the dopant always enters tlie how to dope tlie ion traclís, track via its surface. how the dopants clistribute along tlie traclís and the neigliboring pristine material, In the case of solicl phase doping, tlie dopant can be what are the specific changes in the band struc- either clepositecl as a solid onto tl-ie surface of the tracli ture resulting from doping, i.e. wliat is tlie most contaiuing polymer (e.g. by evaporation), so that it can suitable dopant. cliffuse subsequently into the polymer through its sur- face, iii tlie same wa.y as gaseous or liquid dopants do It is further important to línow liow the stability of (example: cliffusion of A1 from evaporated contact lay- tlie ion traclis influences the doping procedure, to lmve ers into p~l~acet~lene[~~).For contrast,, easily mobile information about tlie stahility of dopants i11 the ioii clopants can a.lso be incorporated homogeneously in tlie traclis themselves. polymer during its production phase (example: Li in Li- At the moment, there does not yet exist aiiy detailecl graf'hted PE[~']). Subsequently introduced ion tracks worli about tlie consecluences of ion tracli doping on tlie will act as getteriiig centers so that the mobile dopant electronic beliavior of ion traclís. EIowever we have al- a.toms will finally concentrate along the ion track (ex- ready some first ideas how the ion t,ra.clr doping process ample: 135 MeV Ar or 100 1íeV He in Li graffited PE itself proceeds, how the dopants clistribute tliereupon, at 10'' ions/cm2 [20, 701). ancl how the ion traclí stability influences tlie cloping process. Tlie transieiit distribution of dopants along the ion In principle, doping of a solid can be done by ther- tracli cluriiig the inital cloping proceclure is cletermined mal or ballistic processes. The htter ones can be largely by tlie cliffrision coefficient of tlie dopant atoms and the excluded in our case, as tlie impinging dopant ions trapping/clet,rapping behwior of tlie (intrinsic and ion would usually introduce additional uncontrolled dam- induced) defects, as well as by their lifetimes; the fi- age to the target with eventually fatal coilsequences. nal dopa.nt distrihution is essentially determined by the The only exception is tlie so-called 'C;elf-ion track distribution of trapping centers along (and transverse dopiny"[611,which means tlie occasional decoration of a to) a track. Iii this coniiection, tlie energy of bonding track by the saine ion wliich lias formed the tracli itself to tlie traps plays an important role, as clopants may be before (examples: implantation of 120 l

tive. Therefore this type of defect is denoted as "sat- (as determined by RBS for Pb in the first case, and urable trap" from the point of view of diffusion theory. for I in the latter case) arranged according to the The other possible source of defects in a polymer predicted (TRIM)[~~]electronic energy transfer. are collisionally (so-called "nuclear") projectile-target b) After doping of 100 keV Ne tracks in PI with interactions. Here, atoms can be ltnoclted away from 5 mole/l LiC1 solution at 100~~[~"],or of tracks any point in the polymeric chains, thus introducing stemming from various noble gases in PI with bond breaking with some empty space in the target. gaseous iodine at ~o~c[~~](as determined by NDP This additionally available space can be used to ac- for Li in the first case, and by RBS for I in the lat- comodate dopants, independently from their chemical ter case), the dopants arranged according to the reactivity and the chemical nature of the site - i.e. also predicted nuclear energy transfer distribution. noble gas atoms can be trapped here. The capability As immediately after the ions' passage, electronic of these defects for dopant uptake is in principle re- as well as nuclear defects have been formed (with elec- stricted by the available volume, but can be easily en- tronic defects ltnown to be in the vast majority), one larged upon widening of this cavity by strain, induced ha.s to conclude from these experiments that in the case by the dopant atoms present in this cavity. (This case had been discussed extensively, e.g. for He bubble for- of irradiated PI these electronic defects have vanished, 'but not (yet) in the case of irradiated PE. In a11 cases mation in metal^[^^]). These tra~sare considered to examined, the ion irradiation took place long time (days be large-size potential depressions. These defects are denoted as ('unsaturable traps" in diffusion theory. Of to years) before doping; hence we can estimate the life- course, this distinction between saturable and unsat- time of electronic defects in PI to be considerably less urable traps appears to be still somewhat schematic, than one year, but in case of PE to be much larger than one year. This is, in principle, nothing new. It is but it is a reasonable simplification to enable first the- oretical modelling. As, for swift ions, electronic energy known long since that, whereas in liquids or polymers Tg transfer exceeds by far the nuclear energy transfer, nu- above the glassy transition temperatures the life- clear (unsaturable) traps are normally less abundant times of free radicals are extremely short, correspond- than electronic (saturable) ones, and hence frequently ing lifetimes in polymers in vitreous state may become extremely long (weeks, months,.. .) and are essentially masked by the latter ones. The overall amount of trapped dopants is deter- determined by the temperature: the farther away is mined by the diffusion coefficient of the impurities and the system from its T,, the longer the lifetime of the their trapping efficiencies at polymeric defects. If more created defects, due to strongly reduced mobilities and detrapping probabilities of trapped free radical^[^^]. than one dopant/impurity is present, a competition Trapped ions and traiped radicals absorb light between the species for chemical bonding will occur. and may be responsible for more or less transient Dopants can be trapped not only at electronically or collisionally induced defects, but also undergo bond- disc~loration[~~].Thus one can study their long-time ing at other dopant or impurity atoms present along behavior easily. This has in fact already been done ear- the track, or even compensate each other. This type lier, e.g. for hydrogen irradiated polyimide, where at of chemical compensation has been verified e.g. for least two different activation energies for the detrapping Kf and I ions in 15-30 keV K, Na, or I irradiated process of trapped radicals were f~und[~"].Regarded polyacetylene[lOI. from the point of view of polymeric device stability for The hitherto performed ion track doping experi- SITE applications, this means that one should on one Tg ments yielded two different types of results: side care for polymers with an as high as possible, and on the other hand for as low device operation tem- a) After doping of 150 keV F or As tracks in PE peratures as possible. Furthermore one should try to with 0.5 mole/l lead acetate s~lution[~~]or with 5 introduce as 'deep' trapping centers as possible, which mole/l KI soluti~n[~~]~[~~Iat 100 'C, the dopants implies ion track formation with heavy ions rather than 68 Braziijan Journal of Physics, vol. 25, no. 1, March, 1995 with light ones. the other side, to sketch the possible recipes for tailor- termines the annihilation of electronic defects. lt is a) Aim: Shape of dopant depth distribution = shape definitely not the indiffusion of oxygen, a.s was antici- of electronic energy transfer distribution: Enable pated until recently: Hnatowicz and c~worlters[~~]could rapid dopant diffusion, use any polymer. show that oxygen readily had diff~~sedinto the above b) Aim: Shape of dopant distribution = shape of nu- mentioned F or As-irradiated PE, even heyond the ir- clear energy transfer distribution (near the end of radiated zone, but that the electronic defects for dopant the íon tracks): Use appropriate polymeric mate- trapping were not affected by this oxidation process[7". rial (e.g. PI), then wait until metastable electronic Therefore we tend now to think that the lifetime of elec- traps have vanished, finally perform dopant diffu- tronic defects depends on the material's capability for sion. self-repair, i.e. on the capability for radical recombina- c) For high energy keavy ions, the nuclear trap distri- tion, respectively for undergoing cross-linlcing or form- bution is near-zero in the surface-near region and. ing double bonds. non-negligible only near the end of the particles' Trapping centers resulting from electronic energy ranges. Therefore, one will gain regular thermal transfer ("electronic traps", usually free radicals) are dopant mobility in the surface-near track regime, occasionally (apparently depending on the poiymer; if applying recipe (h) for surface-near polymer re- e.g. in PI) metastable due to recombination processes, gions: use of appropriate polymer - e.g. PI -, and or due to impurity uptalte. In these cases, by recombi- wait for annealing of electronic damage centers). nation or capture of an impurity or dopant atom, the electronic traps might eventually annihilate, so that a The latter approach is the basis for a11 permeation subsequently passing dopant atom cannot be trapped experiments where gases or liquids are transported any longer by them ("saturable traps"). The much less through thin foils containing latent ion track~[~~I.From abundant defects of collisional origin ("nuclear traps") this type of experiment, information about the magni- can, in principie, accomodate many dopant atoms ("un- tude of the dopant diffusion can be gained via the speed saturable traps"), so that they might become nucleation of penetration. For liquid phase penetration, values are centers for dopant precipitation[76], in case that elec- found to be in the typical order of magnitude of 10-'L to tronic defects have been annihilated already, and a suf- 10-%m2s-l, depending on the polymer, the ion track ficiently large supply of dopant atoms is available. specification and the ion traclt density (as derived for Thus, tailoring the shapes of dopant distributions 135 MeV Ar and 340 MeV Xe ions in PI ands PC at appears to be possible: By using appropriate polymers doses from 109 to 1012 ions/cm2 [68]). (e.g. PI) and waiting for sufficiently long time, elec- In this worlc, we discuss only the fate of chemically tronic traps have largely vanished, so that subsequently moderately active dopants in traclts, which can bond indiffusing dopant atoms will be trapped only at tlie re- to defects in these traclts, but do not destroy them fur- maining permanent (nuclear darnage) centers. On the ther. Enhanced chemical reactivity of dopants leads to other side, "rapidly" introduced dopants (i.e. dopants the well-known effect of ion track etching, where sensi- which arrive at the traps before these ones anneal) will tive molecules in the ion track are either destroyed or be bound to both electronic and nuclear traps. As the dissolved in the etchant, and finally rinsed out from the number of nuclear traps is in general much lower than track with the etchailt solution. This etching process the one of the electronic trapping centers, the dopant thus readily creates an excess of free volume along the depth distributions will then primarily follow the depth latent traclt, which enhances further dopantletchant distribution of electronic energy transfer. Thus one can diffusion into the track. Depending on the special cir- use the competition of electronic damage annealing in cumstances, this diffusion enhancement might be useful suitable polymers on one side, and polymer doping on for the practical application of deep ion track doping. D.Fink and R. Klett

Another recipe for dopant enhancement is to treat the in the polymer's chemical structure. This is indeed ion track with special organic solvents before doping, quite an important phenomenon which can decrease whereby soluble fragments will leave the tracl~[~"]. the effective track cross section (and hence the track's Further, the amount of dopant which can permeate dopant uptalting efficiency, its conductivity etc.) by through the tracks in a given time gives us informa- orders of magnitude. Observable swelling of e.g. irra- tion about the effective open track radius, respectively diated PI or PC[~~]around ion tracks sets in typically cross section, see Table 1. Depending on system and some 30 sec ... 30 min after the onset of the water per- measuring technique, typical effective open. radii of 1 meation, and continues up to 112... 5 hours at room to 10 A, corresponding to open track cross sectipns of temperature - depending on the system under consider- some 10-l4 to 10-l3 cm2 are found (values derived for ation - before it has come to saturation. The amount up 100'MeV to 3 GeV heavy ion (Ar to U) tracks in PI, to which swelling influences the track's properties dif- PC, PP, PET, and PMMA for penetration by water fers greatly, depending essentially on the material and and LiC1 salt s~lutions[~~]).These "effective" values do the ion track density: reduction of track cross section however not necessarily give us an information about and conductivity by factors between 2 and 1000 have the real dimensions: an effective open track radius of, been found, see e.g. Fig. 6a. Swelling can be even so say 10 A, can either mean that there exists a real hole dramatic that the effective open track radius vanishes. with 100% transmission of 10 A radius in the track core In this case, the only path remaining for the dopant region, or it might mean that a region of 100 A radius is bulk diffusion. If a strong electric field applied is in has an average transmission of only 1%. In general, the this case, one observes repeated current breakthrough latter interpretation of the latent ion track as a porous pulses along the partly closed tracks, Fig. 6b. zone of reduced density with enhanced permeability will be more realistic than the pinhole model, as the poly- mer destruction by ions usually leaves solid residuals. A remarkable exception is cellulose nitrate which, upon accurate irradiation conditions, decomposes into exclu- sively volatile reaction products (as demonstrated for -a 30 keV He to 300 keV Xe at fluences between some C o O 5 1,O 15 min 10' to 10'' cm-2 [79]), and hence should form real pin- holes along tracks of energetic ions. This has not yet been studied further. These permeation experiments are usually per- formed for the stationary flow of gases or liquids Figure 6: Influence of swelling on the water permeability through thin films with thickness << track length, long of ion-irradiated polymers (shown here for 2 pm thick PC, time after the electronic traps have been annihilated. It irradiated with 1 x 10" Ar ions of 135 MeV in this case), as determined by conductivity changes: (a) In the first few sec- is recommendable to examine also the track permeation onds after exposure to water, rapid water uptake of a11 the immediately after the track creation. By measuring irradiated depth regime occurs through the latent ion tracks the exact shapes of the dopant depth profiles along the which act as 'irrigation pipes'. This is acompagnied by an increase in conductivity of severa1 orders of magnitude. This track as a function of time, one has got a to01 to follow effect comes to saturation after less than typically a minute. the fate of the traps along the considered track regime. Subsequently, reduction in conductivity sets in, which is as- cribed to a reduction in the effective latent track radius for By following the time dgendence of a permeation permeation, as a consequence of swelling due to polymeric process of water or aqueous salt solution through tracks, water uptake. (b) Occasionally the swelling is so strong that one can check for the influence of polymer swelling by the effective latent track radiiis for permeation nearly van- ishes. A sufficently high electric field applied in this case water uptake of the polymer through the track. Poly- (here: 15 V per 2 pm thickness = 75 kV/cm), periodic mer swelling is the result of hydratation, i.e. of a change breakthrough pulses can be observed. 70 Brazilian Journal of Physics, vol. 25, no. 1, March, 1995

Whenever an energetic ion impinges into a solid, it dures from the gaseous or liquid phases. In the second will emit Mach's supersonic pressure shock wave in the case, the polymer PAis irradiated whilst in contact with shape of a cone with opening angle tg a = vsound/vlon. the monomer MB. This is the counterpart of ion track It is this shoclr wave which will lead to plastic deforma- doping from the solid phase. A potential drawback of tion of the polymer in the vicinity of the new ion traclí, this method is the eventually dominant - formation of in the way that the open polymeric structuie itself as homopolymers PBOOH. well as also older tracks are compressed. As a result tlie In the last case, the polymer is irradiated in the average effective track diameter decreases with increas- presence of air to produce peroxides PAOOP and hy- ing fluence. This has been repeatedly verified by both droperoxides PAOOH, which are thermally or photo- liquid and gaseous doping experiments (examples: LiCI decomposed in contact with the monomer to form the solution and I vapor permeation into PI and PC[~'J~~]). graft copolymer PAOPB. The counterpart of this From the dose dependence of the effective track diame- method in ion track doping wouid be ion track oxi- ter for permeation, one can estimate the compressabil- dation during or soon after formation, with subsequent ity of ion tracks. It turns out that their compressability indiffusion and bounding of the dopant. is lower by some orders of magnitude than the one of Grafting has hitherto nearly exclusively been per- the pristine material[22]- in other words, ion traclts are formed after Gamma-, X ray-, or electron beam irradi- harder than their environment. Due to the reduction ation. To our knowledge, it has been tried only once of polymeric free interna1 volume by the ion-induced after swift heavy ion irradiationf8']. In this case, liq- target compression, the dopant mobility will in general uid styrene monomer was grafted by the peroxide tech- be reduced as well in tracks as in the pristine polymer nique onto 132 MeV 0, 40.7 MeV Xe, and 472 MeV Xe itself. ion tracks in PVDF. Interestingly, it turned out that This target compression will also give rise to an en- the overall grafting yield was rather independent from hancement of the overall polymeric density. Accord- tlie irradiation dose, i.e. the higher the track density ing to the above model, we should expect the den- in the polymer, the less the grafting efficiency per ion sity change to be correlated with the electronic energy track. This - for the authors highly puzzling - effect transfer. It has been shown h~wever['~]for low energy can be easily understood by remembering the strong high dose irradiation (25-250 keV Ar, Kr, and Xe at flu- compaction of both the pristine polymer and neighbor- entes of 1014 to 1016 ions/cm2) of polystyrene that the ing ion traclrs during formation of a new ion track: the densification is mainly due to energy transfer by atomic higher the track density, the smaller will become the collisions with the target atoms. It has still to be found available free volume along a track; hence, the lower out by future experiments in which way the transition will be the graft yield. from the low energy liigh dose case to the high energy It would be interesting not only to dope ion tracks single ion track case will affect the target density. with small electrically active ions such as LiS, BS,As f,

There is a close similarity of what we cal1 here "ion P+, etc., but also with larger entities such as C60 (in track doping" to what chemists use to cal1 "grafting" order to increase the fullerene content above the natu- of ion tracks in polymers. Three standard methods of ral production rate), and to see in how far they can be grafting are available, (a) the consecutive, (b) the si- rendered immobile by trapping along the tracks - or in multaneous, and (c) the peroxide methods["]. In the chemist's terminology, by grafting them onto the host first case, the polymer PA is irradiated in the absence polymer along the track. The first step in realizing such of oxygen prior to exposure of the liquid or g;aseous experiments is to establish a technique for unambigu- monomer MB which then bonds chemically to form the ous depth profiling of e.g. fullerenes in carbonaceous graft polymer PAMB. This corresponds to our hith- matter. Such as technique has just been developed by erto most frequently applied ion track doping proce- US[~~],and we could already show that fullerenes really D. Fink and R. Klett

accomodate along latent GeV heavy ion traclts in PI between creation and annihilation (doping) of saturable up to depths of some 3 ,um, after doping them with traps lias been established, insaturable traps grow by C60/toluene solution at R.T. for 1 day. capturing more dopant atoms, thus slowly becoming Ion track oxidation - applied above for a first ion nucleation centers of future dopant precipitations. track grafting experiment - seems to be one of the cru- This model is valid both for the single-ion-track cial factors which limits the stability of polymeric im- (i.e. non-overlapping) as for the multiple-ion-track- planted devices. Recent ~ork[~~l~~1suggests that indif- overlapping case. It has however not yet been expanded fusing oxygen preferentially bonds to surface-near poly- to the more realistic case of an irradiated polymer hav- mer zones and to nuclear damage centers, and eventu- ing additional properties such as swelling and/or com- ally also to the track-creating projectile ions, provided paction. As ion traclts are essentially linear struc- that they are chemically reactive. After prolonged time, tures, dopant mobility in them can be described well the oxygen may even diffuse towards depths exceeding by one-dimensional diffusion codeswhich include source the ion track length. Not only oxygen, but also nitrogen (or sink) terms at the surfaces, and trapping and de- has found to be eventually incorporated in irradiated trapping at saturable and unsaturable traps. Such a polymers (as shown for the case of 2 keV He in PET code has been developed recently by us from an older at 1017 ions/cm2 [83]). It is still unknown whether this 0ne[~~-"1,which was based on the mathematical tech- finding can be generalized to tracks of higher energetic nique of finite differences["l. ions. It was mentioned above that the diffusional behavior There exists another important question in the con- of the dopants will be of crucial importance for the long- text of polymeric ion track doping for SITE applica- term stability of the polymeric SITE devices. In fact, tions, namely the question of dopant homogeneity. Do dopant mobility in irradiated polymers differs consider- the dopants distribute homogeneously along the ion ably from tlie well-known high mobility in unirradiated track, or do they eventually cluster at special sites along materials: on one side, radiation-induced polymer com- the track? At the moment, this question cannot yet paction reduces dopant bulk mobility considerably, due be answered. Surely, the overall dopant distribution to the reduction of open volume for diffusion. (This exhibits homogeneity along the ion tracks within the can even lead to the extreme of case of bulk diffusion depth resolution of corresponding depth profiling tech- inhibition, as could be shown by us recently for the niques, i.e. typically within a few 100 A. But this does example of 135 MeV Ar irradiation of PI at 10'' to not exclude possible dopant clustering in the sub-100 1012 trac1ts/cm2, with subsequent subjection to aque- A range. In order to answer this question, it will be ous LiCl dopant: it was found in this extreme case that necessary to perform small angle neutron (or X ray) no Li dopant could penetrate into the polymer at all, scattering measurements on the dopant distribution in whereas we could accomodate in unirracliated PI nearly tracks, which has not yet been done hitherto. up to 10'' Li atoms/cm3 after just 1 h doping at room temperature.) IV.2 Theory for ion track doping "Fresh" tracks (where electronic defects still exist) A first attempt has been made to describe the ion appear to act as quite efficient trapping centers for mo- track doping theoretica~l~[~~].For this sake, a set of rate bile dopants, thus depleting the surrounding polymer equations has been established to follow the fate of the from eventually present dopants. This property effi- mobile dopant atoms, the saturable traps and tlie in- ciently counteracts the tendency of regular thermal dif- saturable traps. The solution of these equations nicely fusion to spread radially from the track a11 over the shows that, with increasing fluence, first an excess of matrix. So one should expect that by bonding of the saturable traps is produced, which then are filled by dopants to the electronic defects along the tracks the the mobile dopant atoms. Once a dynamic equilibrium dopant distribution will gain the required long-time sta- 72 Brazilian Journal of Physics, vol. 25, no. 1, March, 1995 bility. Of course, this has still to be verified experimen- 7) understanding the changes in electronic properties tally. upon doping, and 8) handling better the factors which influence degra- V. Summary and conclusion dation and hence ion track stability.

This is a review about our present understanding Furthermore, the above proposed ways for precise of latent ion tracks in polymers, and their possible fu- positioning and contacting of individual ion tracks have ture application to nanometric electronic technology. to be tested. Only then it will be possible to tailor and In this overview we have tried to underline the basic arrange ion tracks so that they can fullfill specific re- general trends, and to avoid going into systemspecific quirements and be incorporated in future hyperdense details too much. The latter should be done later, when electronic devices as elements of nanometric lateral di- more detailed knowledge has been accumulated. Some mensions. progress has been made concerning the knowledge of ex- terna1 and interna1 geometric track structures, includ- Acknowledgrneiits ing the influence of compaction, the basic chemical re- We gratefully acknowledge financia1 supports from actions, and the intrinsic conductivity of latent tracks. the BMFT, CNPq, CONACYT, CSAV, CSIRO, The principal mechanisms which govern the doping pro- DAAD, GSI, HMI, ILL, JAERI, and the UFRGS, re- cedure - diffusion, trapping, detrapping, and swelling - ceived in tlie last decade at various occasions. We also are known. Theories have been established for latent thank Profs./Drs. M. Behar, J. P. Biersack, V. Cer- ion track formation and for ion track doping. venna, L. T. Chadderton, S. A. Cruz, P. F. P. Fichtner, Based on this fundament, a deeper underdanding Y. Gamaly, P. Grande, R. B. Guimaraes, V. Hnatow- of latent ion tracks has to be developed now by icz, F. Hosoi, X. Hu, J. Kaschny, G. Kvitek, M. Mueller,

1) improving the theoretical models by including a11 H. Omichi, T. Sasuga, G. Schiwietz, P. Szimkoviak, J. Vacik, L. Wang, F. C. Zawislak, and last not least to important parameters and less restrictive assump- the accelerator and reactor staffs of the HMI Berlin, ILL tions, 2) getting more information on the lateral ion track Grenoble, and GSI Darmstadt for their steady help and for numerous discussions. Finally, thanks to the referee structure - e.g. concerning conductivity and chem- istry, eventually by application of tomographic for valuable critics and helpful hints, on how to improve this paper. techniques[">"18", 3) learning how the composition, size and density of References the current-carrying clusters are influenced by de- cisive parameters such as the deposited electronic energy density along tlie track, and on the mate- 1. For the proceedings of the last meeting see Nucl. rial, Tracks Radiat. Meas. 22 (1993). 4) getting absolute production yields of polycyclic 2. D. Fink, L. T. Chadderton, S. A. Cruz, W. R. compounds and fullerenes along Lhe tracks, and Fahrner, V. Hnatowicz, E. H. te Kaat, A. A. Mel- knowing which of the many possible polycyclic niltov, V. S. Varichenko and A. M. Zaitsev, Radiat. compounds are specifically created, Eff. Def. Solids 126, 247 (1993). 5) examining whether dopants in tracks cluster or 3. H. W. Vollmann, J. Information Recording Mate- distribute homoqeneously along them, rials 20, 1 (1992). 6) defining precisely the electronic energy leve1 4. C. Martin (Colorado State University), personal scheme of ion tracks as a function of their param- communication (1993). eteres, 5. J. Vetter, GSI-Nachrichten 5, 1 (1992). D. Fink and R. Klett

A. Ishitani, K. Shoda, H. Ishida, T. Wanatabe, I<. A. Licciardello, O. Puglisi, L. Calcagno and G. Yoshida and M. Iwaki, Nucl. Instr. Meth. B39, Foti, Nucl. Instr. Meth. B32, 131 (1988). 783 (1989). X. L. Xu, J. Davenas and G. Boiteux, Proc. of the See, e.g. G. Horowitz, X. Peng, D. Fichou and F. Annual Meeting of the Chinese Materials Research Garnier, J.Appl. Phys 67, 528 (1990), and refs. Society, Beijing (1990). cited therein. D. Fink, X. Hu, R. Klett, M. Muller, L. T. Chad- S. Prawer, A. Hoffmann and R. Kalish, Appl. derton and L. Wang, submitted to Nucl. Instr. Phys. Lett. 57, 2187 (1990). Meth., (1994). G. Braunstein, A. Talmi, R. Kalish, T. Bernstein T. Sasuga, personal communication (1993). and R. Besserman, Radiat. Eff. 48, 139 (1980). F. Petersen and W. Enge, GSI Annual Report S. Lin, K. Sheng, J. Bao, T. Rong, Z. Zhou, L. (1992) p.286. Zhang, Z. Shen and M. Yan, Nucl. Instr. Meth. M. Kramer and G. Kraft, Report GSI-94-08, B39, 778 (1989). "Gesellschaft fur Schwerionenforschung", Darm- For an early work in this field, see: e.g. N. Koshida stadt, Germany (1994). and Y. Wachi, Appl. Phys. Lett. 45, 436 (1984). A. LeMoel, J. P. Durand, C. Lecomte, M. T. Valin, J. Davenas, G. Boiteux, X. L. Xu and E. Adem, M. Henriot, C. LeGressus, C. Darnez, E. Balanzat Nucl. Instr. Meth. B32, 136 (1988). and C. M. Demanet, Nucl. Instr. Meth. B32, 115 J. L. Duroux, A. Moliton and G. Froyer, Nucl. (1988). Instr. Meth. B34, 450 (1988). A. LeMoel, J. P. Durand and E. Balanzat, Nucl. D. Fink, L. T. Chadderton, S. A. Cruz, W. R. Instr. Meth. B18, 59 (1986). Fahrner, V. Hnatowicz, E. H.te Kaat, A. A. Mel- A. Chapiro, Nucl. Instr. Meth. B32, 111 (1988). nikov, V. S. Varichenko and A. M. Zaitsev, Radiat. L. Calcagno, G. Compagnini and G. Foti, Phys. Eff. Def. Solids 126, 247 (1993). Rev. B46, 10573 (1992). A. M. Zaitsev, A. A. Melnikov, V. S. Varichenko, A. Licciardello, O. Puglisi, L. Calcagno and G. V. F. Stelmakh, W. R. Fahrner, B. Burchard, D. Foti, Nucl. Instr. Meth. B39, 769 (1989). Fink and E. H.te Kaat, Proc. of the "Worbshop on D. Xu, X. L. Xii, G. D. Du, Z. X. Lin, R. Wang Ion Beam Applications", Hahn-Meitner-Institute and S. C. Zou, phys. stat. sol. (a) 136, 433 Berlin (1992). (1993). J. Davenas, X. L. Xu and G. Boiteux, Proc. G. Marletta, S. Pignataro and C. Oliveri, Nucl. 2nd French-Polish Symp., Nancy, France (1987) Instr. Meth. B39, 773 (1989). [Synth Met. 24, 81 (1988)l. D. Fink, F. Hosoi, H. Omichi, J. Cardoso, M. H. J. P. Biersack and L. G. Haggmark, Nucl. Instr. Vazquez, R. Montiel, L. T. Chadderton and L. Meth. 174, 257 (1980). Wang, Proc. UMRS-ICAM, Tokyo (1993). J. P. Biersack, Z. Phys. A315, 95 (1982). D. Fink, F. Hosoi, H. Omichi, T. Sasuga and L. H. Bethe, Ann. Phys. 5, 325 (1930). Amaral, Radiat. Eff. Def. Solids, in print (1994). D. Fink, H. Omichi, F. Hosoi, M. Tamada, V. L. T. Chadderton, D. Fink, H. Mockel, I<. K. Hnatowicz, J. Vacik, L. T. Chadderton and R. Dwivedi and A.Hammoudi, Radiat. Eff. Def. Klett, Proc. UMRS-ICAM Tokyo (1993). Solids 127, 163 (1993). D. Fink, L. T. Chadderton, F. Hosoi, H. Omichi, L. S. Chadderton, D. Fink, Y. Gamaly, H. Mockel, T. Sasuga, A. Schmoldt, L. Wang, R. Klett and J. L. Wang, H. Omichi and F. Hosoi, Nucl. Instr. Hillenbrand, Nucl. Instr. Meth. B91, 146 (1994). Meth. B91, 71 (1994). R. Klett and D. Fink, to be published in Nucl. D. Fink, H. J. Mockel, H. Melzer, R. Klett, F. Instr. Meth.. (1995). Hosoi, H. Omichi, L. T. Chadderton and L. Wang, 74 Brazilian Journal of Physics, vol. 25, no. 1, March, 1995

to be published in Nucl. Instr. Meth., (1994). 59. For a recent rep- 41. D. Fink, F. Hosoi, H. Omichi, L. T. Chadderton resentative discussion of the Coulomb explosion and L. Wang, to be published in Radiat. Eff. Def. moclel, see e.g. S. Klaumunzer, M-D. Hou and G. Solids, (1994195). Schumacher, Phys. Rev. Lett. 57, 850 (1986). 42. G. Brinlrmalm, P. Demirev, J. Eriksson, D. Fenyo, 60. For a recent representative on the thermal spike P. ~Akanssonand B .U.R.Sundquist, Nucl. lnstr. inodel, see e.g. C. Dufour, E. Paumier and M. Meth. B84, 37 (1994). Toulemond, Radiat. Eff. Def. Solids 126, 119 43. J. Gamaly and L. T. Chadderton, personal com- (1993). munication (1993). To be published (1994195). 61. See, e.$. R. B.Guimaraes, L. Amaral, M. Behar, 44. J. Cardoso, R. Montiel, M. H. Vazquez, personal D. Fink and F. C. Zawislak, J. Mat. Res. 3, 1422 communication (1993). (1988). 45. D. Finlr, R. Klett, M. Muller, X. Hu, G. Xiao, 62. J. P. Biersack and D. Fink, in "Ion Implantation G. Schiwietz, A. Schmoldt, L. T. Chadderton, A. in Semiconductors", edited by S. Namba (Plenum, A. Karanovich, R. Montiel, S. Cruz-Jimenez and New York, 1975), p.211. L.Wang, to be published in Proc. of the 17th 63. I(. Tjan, Ph.D. Thesis, Free University Berlin Intl Conf. on "Nuclear Traclrs in Solids", Dub- (1985). nar (1994). 64. L. Wang, Ph.D. Thesis, Free University Berlin 46. G. Schiwietz and G. Xiao, personal communica- (1990). tion, (1993). 65. As a representative example for numerous appli- 47. R. Klett, Ph. D. Thesis, Humboldt-University cations of this diffusion program to dopant mo- Berlin, to be published in 1996. bility in polymers, see: M. Behar, P. L. Grande, 48. D. Finlr, M. Muller, L. T. Chadderton, P. H. L. Amaral, J. R. Kaschny, F. C. Zawislak, R. B. Cannington, R. G. Elliman and D. C. McDonald, Guimaraes, J. P. Biersack and D.Fink, Phys. Rev. Nucl. Instr. Meth. B32, 125 (1988). B41, 6145 (1990). 49. N. F. Mott, Philos. Mag. 19, 835 (1969). 66. D. Fink and L. Wang, Radiat. Eff. Def. Solids 50. S. R. Forrest, M. L. Kaplan, P. H. Schmiclt, T. 114, 343 (1990). Venlratesan and A. J. Lovinger, Appl. Phys. Lett. 67. J. Davenas and X. L. Xu, Nucl. Instr. Meth. 41, 708 (1982). B71, 33 (1992). 51. D. Fink, I<, Ibel, P. Goppelt, J. P. Biersack, L. 68. D. Fink, V. Hnatowicz, J. Vacik and L. T. Chad- Wang and M. Behar, Nucl. Instr. Meth. B46, derton, Radiat. Eff. Def. Solids, 132, 1 (1994). 342 (1990). 69. J. Davenas, X. L. Xu, M. Maitrot, C. Mathis and 52. Y. Eyal, personal communication (1993). B. Francois, Nucl. Instr. Meth. B32, 166 (1988). 53. P. Fratzl and S. Schreiber, personal communica- 70. D. Finlr, R. Klett, H. Omichi, F. Hosoi, to be pub- tion (1994). lished in Appl. Phys., (1995). 54. R. M. Papaleo, M. A. de Araujo and R. P. Livi, 71. D. Fink, J. P. Biersack, J. T. Chen, M. Stadele, K. Nucl. Instr. Meth. B65, 442 (1992). Tjan, M. Behar, C. A. Olivieri and F. C. Zawislak, 55. L. Calcagno, G. Compagnini and G. Foti, Nucl. J. Appl. Phys. 58, 668 (1985). Instr. Meth. B65, 413 (1992), and refs. therein. 72. See numerous contributions in "Fundamental As- 56. See, e.g. W. L. Brown, Radiation Effects 98, 115 pects of Inert Gases in Solids", edited by S. E. (1986). Donelli and J. H. Evans (Plenum, New York, 57. J. Davenas, personal communication (1990). 1991). 58. M. E. Adel, R. Kalish and S. Prawer, J. Appl. 73. V. Hnatowicz, V. Havranek, J. Kvitek, V. Perina, Phys. 62, 4096 (1987). V. Svorcilr, and V. Rybka, Jpn. J. Appl. Phys. D. Fink and R. Klett

32, 1810 (1993). 81. N. Betz, E. Balanzat, A. Le Moel and J. P. Du- 74. V. Hnatowicz and J. Vacik, personal communica- raud, Radiat. Eff. Def. Solids 126, 221 (1993). tion (1994). 82. D. Fink, C. Mathis, R. Klett, J. Vacik, V. Hantow- 75. V. Hnatowicz, J. Kvitek, V. Svorcik and V. Ry- icz, L. T. Chadderton, submitted to Nucl. Instr. bka, Eur. Polym. J. 29, 1255 (1993). Meth., (1994). 76. S. Cruz, L. T. Chadderton, D. Fink, Proc. Intl. 83. P. Bertrand, Y. dePuydt, J.-M. Beuken, P. Lutgen Conf. on "Nuclear Tracks in Solids", Beijing, and G. Feyder, Nucl. Instr. Meth. B19/20, 887 1992. (1987). 77. As a representative example, see: I<. Schaupert, 84. For a representative overview for this method see, D. Albrecht, P. Armbruster and R. Spohr, Appl. e.g. G.D.Smith, Numerical Solutions of Partia1 Phys. A44, 347 (1987). Differential Equations: Finite Difference Methods, 78. B. Heinrich, H. B. Luck, M. Tamada, B. E. Fis- 2nd ed., (Clarendon Press, Oxford, 1978). cher, R. Spohr, C. Trautmann, J. Vetter and N. 85. M. Muller and D. Fink, Nucl. Instr. Meth. B34, Angert, Report GSI 92-1, "Gesellschaft fur Schwe- 162 (1988). rionenforschung", Darmstadt, Germany (1991) p. 86. D. Fink and M. Muller, J. Appl. Phys. 68, 958 265. (1990). 79. J. P. Moliton, T. Trigaud and A. Moliton, Nucl. 87. D. Fink, M. Muller, A. Schmoldt, J. K. Zhou, L. Instr. Meth. B65, 428 (1992). T. Chadderton and X. L. Xu, Nucl. Instr. Meth. 80. L. Calcagno and G. Foti, Nucl. Instr. Meth. B65, 432 (1992). B19/20, 895 (1987).