Nuclear Instruments and Methods in Physics Research B80/81 (1993) 721-725 North-Holland
Boam Ia4eraetioaa with IlaterintsiAtoms
Low temperature recrystallization of ion implanted InP P. Müller a, W. Wesch 3, v.S . Sr",~owe1 b, P.I . Gaidizk i,, E. Wendler ', F.F . Komarov b and G. Götz a '
''Friedrich-Schiller-Universität Jena, lnstitut für Festkörperphysik, Max-Wien-Platz 1, 0-6900 Je ia, Germany "Institute of Applied Physics Problems, Minsk, Belarus
The low temperature recrystallization behaviour of completely amorphized InP layers is investigated by means of RBS channeling and TEM measurements in the temperature region 150 to 400°C as a funct on of the annealing time . The resulting lajrers consist of a well annealed and a defective near surface layer containing a high density of microtwins. The thickness of the weal annealed region increases with the annealing time and reaches a saturation value which depends on the temperature and is independent of the thickness of the initial amorphous layer. With increasing temperature the time to reach the saturation value de,:reases . 1 . Introduciian For several technologically relevani AnjBv semiconductors ion implantation is the most promising technology for doping of thin layers to produce optoelectronic as well as electronic devices. However, ion implantation is connected with the displacement of lattice atoms, and the degree of disorder produced depends on the implantation parameters . For sufficiently high ion fluences under certain conditions amorphous layers can be created. Generally, after the implantation a heat treatment has to be performed to reorder the lattice and to obtain e.g . electrical activation of the implanted dopants . Contrary to Si, in the AttjBv compounds the recrystallization of amorphous layers occurs at relatively low temperatures (150 to 400°C), but the annealing of the residual defects requires temperatures higher than 800°C (see e .g . ref . ). It has been previously shown for GaAs and 1nP that it is advantageous either to prevent aniorphization during the implantation or to recrystallize amorphous layers at relatively low temperatures and then to activate the dopants by a second high temperature annealing step . In the past in several works the annealing behaviour of weakly damaged InP layers at different temperatures (room temperature and higher) [2-5] and of amorphous InP layers at higher temperatures (>_ 300°C) has been reported [6-9] . Similar as in GaAs, amorphous layers recrystallize at temperatures around 300°C, and the resulting layers consist of a well annealed and a defective near surface layer [10,11] . However, the influence of the annealing temperature and the time is not yet investigated in detail in this temper-
ature region . Therefore, in the present paper the time and temperature dependence of the annealing behaviour of ion implanted amorphous InP layers, produced by Si' and Si'/I" coimplantation at room temperature in the temperature range between 150 and 400°C is discussed. 2. Experimental N-type (n e = 10''' cm 3 ) (WO) orienicd hrF crystals were implanted at room temperature under the conditions given in table 1 . The ion current density was kept constant lower than 50 nA/cm2 to prevent sample heating. The multiple implantations in samples #2 and #3 were carried out to ensure that continuous amorphous layers up to the surface were produced and that the recrystallization during annealing could only start from the amorphous-crystalline interface. The coimplantation of phosphorus gives the opportunity to check the influence of stoichiometric changes in the samples on annealing behaviour. The samples were isochronally (20 min) annealed in a quartz tube under flowing N2 atmosphere at different temperatures (200, 235, 300 and 400°C). Some of the samples were isothermally annealed at 300 and 400°C for 8, 20, 40, and 90 min . Annealing at 150° was carried out in high vacuum in the scattering chamber during the Rutherford backscattering (RBS) experiments. The disorder in the implanted and annealed InP crystals was investigated by RBS-channeling measurements along the (100) direction using 1 .4 MeV He` ions and backscattering angles ® =170 and 130°, respecthnly . The energy resolution of the detector was
0168-583X/93/$06.00 0 1993 - Elsevier Science Publishers B.V . All rights reserved
P. Müller et at JLaw temperature recryltallization of MP
Table 1 Implantation conditions and thickness of the amorphous layers _ .-._ _ .__._ ..., .. . . . . ._ Sample In - Energy Dose Thickness of amor[keV] [cm - `] phous layer amorph [nm]
x 1(" ,-!
î80 130 45 25
1 .5 4.4 4.9 1 .7
x 10 14 x 10" x 10 13 x 10 13
180 130 45 25 180 130 45 25
7.5 XI 013 2.2 x 1013 2.45 x 1013 `:.5 x 10 13 7.5 x 10 13 2.2 XI0' 3 2.45X 1013 8.5 x 10 12
. .... .. ......... .. ... . . .....
.. .r ..,..~.. .~ ..}
20v " °
backscattering at displaced P atoms (channel numbers < 120) is superimposed to the intense yield of He' ions backscattered at deeper lying In atoms. Because of the small mass of P compared to In, the increase of the backscattering yield is small and, therefore, no additional information can be obtained from the phosphorus damage signal . As a consequence only the behaviour of the In damage ;,ignal will be discussed in the following. After the first annealing step (15 min at 150°C) the interface between the amorphous and crystalline bulk region moved towards the surface . In addition no changes of the damage in the near surface region are observed. That means, that a good recrystallization o^!y from the underlying crystalline bulk material occurred. For longer annealing periods (35 min and 150 min) a further moving of the interface to the surface is observed, but the speed of the interface motion is reduced and the thickness of the well annealed layer dperf approaches a saturation value. It is clearly to be seen that for a 150°C anneal only 30 nm annealed well .
Fig . 1 . Energy spectra of 1 .4 MeV He` ions backscattered at si + implanted InP (sample #1) and annealed at 150°C for different times.
~ + P+ [sample #31
P 1 ~""°
MeV He' ions backscattered at #2) and annealed at different temperatures for 20 min .
+random .-mpml,,] ^ISO°C, , cmml " I5a °C, lo nun ) amx".akd " c.lmm .ulaalhnc
random .a -planted 2-:, an nr .lrd . 2o m,n? ,ryaallmn
_- . 4al°c,
Fig. 1 shows the RBS spectra of sample # 1 annealed at 150°C and different times. As can be seen, the damage peak for the as-implanted sample connected with displaced In atoms (channel numbers 205 to 240) reaches the level of the random spectrum up to the surface of the sample indicating the formation of a continuous amorphous layer . The signal due to lsample #II
Fig. 2. Energy spectra of 1 .4 Si' implanted InP (sample
12 keV . For the energy-to-depth conversion we have used the stopping power cross section data of Ziegler [1'2] and assumed the density of the bulk material for the amorphous layers . To get more detailed information about the structure of the disordered near surface layers electron diffraction rn_rasurements at the surface were perfonacd for selected sampl".
InV Si "
..a tmpWlcd -2W°C,2e m,n
~nm"alr l -4.C, 211 W,a) , . .rvNall,.r Inml
Fig . 3 . Energy spectra of 1 .4 MeV rie ' ions backscattered at (Si + +P + ) implanted InP (sample #3) and annealed at different temperatures for 20 min .
P. Müller et al. j Low temperature recrystallization of biP
Figs . 2 and 3 show the backscattering spectra of samples #2 and #3 annealed at a constant time of 20 min at 2(10 and 400°C. In both cases the moving of the solid phase gritaxy front is clearly noticeable. It can be sees-, that the thickness of the well annealed layer uncreases with the annealing temperature . A cur .rpa: :son of the spectra in figs. 2 and 3 shows, that no significant difference between the annealing behaviour of the Si implanted and the sample coimplanted with P exists : the thickness of the layer, which is well annealed, is the same for the Si + + P + implantation as in the case of the Si implantation alone. Obviously a possible mean s(oichiomettic imbalance as assumed to be responsible for the faster increasing roughness of the interface in InP compared to GaAs  at least seems not to influence the thickness of the well annealed layer . The results of the annealing experiments at 150, 300 and 400°C are summarized in table 2 which gives the thickness of the well annealed layer dpe,r as a function of the anneaüug time . From the backscattering spectra it can be seen that for high annealing temperatures (see the spectra for 400°C in figs. 2 and 3) or for longer annealing times at lower temperatures (see the spectrum tot 1,U ^,in in fig. i) the residual damage peak does not reach the random level. The backscattering yield in this part of the corresponding spectra is only 90% of that of the random spectrum. This indicates the existence of nonamorphous defective residual layers . To get information about the structure of these defective layers TEM investigations were performed . As an example, fig. 4 shows electron diffraction patterns of an InP singe crystal (fig . 4a) and of %ample #1 annealed at 235°C for 40 min (fig. 4b) . From the pattern one can see that the surface layer is or. principle single crystalline, but the splitting of the diffraction spots indicates the existence of a high density of microtwins. This microtwin struc-
Table 2 Thickness of the well annealed layers ing conditions Sample #1 OA'- 150°C
Samples #2,3 OA = 300°C
[min] 7 15 25 35 55 75 115
[nm] 10 18 23 25 27 30 30 30
[min] 3(270°C) 8 20 40 80 -
for various anneal-
[nm] 45 50 64 68 70 -
[min] 8 20 40 80 -
[nm] 92 100 100 1() -
Fig. 4. Electron diffraction pattern of (a) InP single crystal, (b) sample #1 annealed at 235°C for 40 min, (c) micrograph of sample # 1 annealed at 235°C for 40 min. ture is also clearly visible in the electron micrograph shown in fig. 4c. From the results it can be concluded that in the temperature region discussed here the amorphous layers fully recrystallize and that two different regions are produced: a well annealed region and a near surface disordered recrystallized layer. As already shown for annealing at 150'C (see fig. 1), also for annealing at 300°C and 400°C a maximum thickness of the well annealed region is found . This is illustrated in fig. 5 giving the dependence of the thickness dp,,f of the well annealed layers versus the annealing time to for the Ille . SEMICONDUCTOR MODIFICATION (c)
P. Miller et al. / Lowtemperature recrystallization of InP
temperatures 0. investigated. The graphs show that (i) da, t for .ach temperature increases with the time to and reaches a certain saturation value (cf. table 2), and (ü) with increasing annealing temperature vA the time to for reaching the saturation value decreases, i.e . the recrystallization velocity increases with 7YA . The latter result indicates that the low temperature annealing occurs by the motion of a solid phase epitaxy front, which is in agreement with the result of Licoppe et al . 1111 . It should be mentioned that in fig. 5 the thickness of the well annealed layer, as determined by RBS, is not consistent with the position of the phase front. For annealing temperatures of 300 and 400°C the solid phase epitaxy front reaches the surface for annealing C es of about 1 min which can be concluded from the lowering of the backscaitcring yield in the damage region (not shown). From results of other authors [10,11] it is known that low temperature epitaxial recrystallization of completely amorphous layers results in the formation of a defective near surface layer and a thin, underlying, well annealed region . Our results show that such a structure is already produced at 150°C for sufficiently high annealing times, and that the thickness of the well annealed layer dce,r depends on the annealing temperature (see fig . 5). This is in contrast to the behaviour of ion implanted GaAs, in which for low temperature recrystallization independent of the annealing temperature, a good regrowth of an amorphous layer within - 40 nm is found . In fig. 6 the thickness of the well annealed InP layers is depicted versus the reciprocal annealing temperature. The figure summarizes results of samples with different amorphous layer thicknesses . However, the results published in ref.  show that the thickness of the well annealed region is independent of the thickness of arc initial amorphous layers . This
Inp 9 -400"C
Fig. 5. Thickness of the well recrystallized layer annealing time .
*5A 1^C 1
tQ -20m , n
Auvray et a1.1982
Thickness of the well recrystallized layer d-r versus the reciprocal annealing temperature .
allows to plot the experimental results for all temperatures versus the annealing temperature. Surprisingly one finds the measurement points on a straight line with an Arrhenius-like dependence do, .t (TA ) with an activation energy EA = 0.16 eV. This value remarkably deviates from that reported by Licoppe et al .  for epitaxial recrystallization (EA = 1 .55 eV) . The reason for this deviation is that with our experimental method only the well annealed region and not the movement of the epitaxial phase from is considered. 4. Summary The low temperature recrystallization of InP layers amorphized by Si' and S + + P+ implantation was investigated by means of Rutherford backscatteringchanneling measurements (RBS) and by electron microscopic (TEM) methods. From the spectra of backscattered He' ions the thickness of the well annealed regions d,,t was determined . RBS and TEM measurements show that already at an annealing temperature of 150°C at sufficiently long annealing times the whole amorphous layer has been recrystallized resulting in a well annealed region and a defective near surface layer which contains a high density of microtwins. With increasing annealing temperature dpa,1 increases. This is in contrast to the results obtained for GaAs, where a conscant value of d,, r was found . In our investigations no influence of P+ coimplantation or. the recrystallization behaviour could be found . The results show that for sufficiently thin amorphous InP layers a good annealing should be possible already at low annealing temperatures .
P. Midler et al. / Low temperature recrystallization oflnP Acknowledgement The authors wish to thank T. Bachmann for helpful discussions . References [I] B.J . Sealy, Int. Mater. Rev. 33 (1988) 38 .  D.E. Davies,J.P . Lorenzo, T.G . Ryan andJ.J . Fitzgerald, Appl. Phys. Lett. 35 (1979) 631. [31 D.E. Davies, J.J . Comer, J.P. Lorenzo and T.G. Ryan, ibid ., p. 192.  D.E . Davies, Mater. Res. Sec. Symp. Proc . 45(1985) 261. [51 U.G. Akano, I.V. Mitchell and F ^. Shepherd, Appl . Phys. Lett . 59 (1991) 2.*) 10.
 P. Kringhpj, V.V. Gribkovsklt fin' A.N . Larsen, Appl. Phys . Lett . 57 (1990) 1514. [7j P. Auvray, A. Guivarc'h, Maxidon, G. Pelous . M. Salvi and P. Hen-, J. Appl . Phys. 53 (1982) 6202. [81 S.S. Gill and B.J. Sealy, J. Appl. Phys . 56 (1984) 1189.  G. Bahir, J.L. Merz, J.R . Abelson and T.W . Sigmon, Mater. Res. Sec. Symp. Proc. 45 (1985) 297.  C. Licoppe, Y.I. Nissim, P. Kranz and P. Henoc, Appl. Phys. Lett . 49 (1986) 316.  C. Licoppe, Y.I. Nissim, C. Wriadec and P. Hen-, Appl. Phys . Lett . 50 (1987) 1648 .  J.F. Ziegler, J.P. Biersack and U. Littmark, The Stopping and Range of Ions in Solids (Pergamon, New York, 1985).  J. Herold, H. Bartsch, W. Wesch and G. Götz, Phys. Status Solidi AIII (1989) 59.
Illc. SEMICONDUCTOR MODIFICATION (c)