Oct 23, 2014 - 'M. M. Mandurah, K. C. Saraswat, C. R. Helms, and T. L Kamins, J. App!. ... R. Wilson, R. B. Gregory, W. M. Paulson, S. J. Krause, J. A. Leavitt, L.
Resistivity of borondoped polycrystalline silicon M. Y. Ghannam and R. W. Dutton Citation: Applied Physics Letters 52, 1222 (1988); doi: 10.1063/1.99164 View online: http://dx.doi.org/10.1063/1.99164 View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/52/15?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Properties of boron-doped thin films of polycrystalline silicon AIP Conf. Proc. 1569, 314 (2013); 10.1063/1.4849283 Analysis of low-frequency noise in boron-doped polycrystalline silicon–germanium resistors Appl. Phys. Lett. 81, 2578 (2002); 10.1063/1.1511815 Resistivity of boron-doped diamond microcrystals Appl. Phys. Lett. 72, 2445 (1998); 10.1063/1.121680 Solid phase epitaxial regrowth of borondoped polycrystalline silicon deposited by lowpressure chemical vapor deposition Appl. Phys. Lett. 51, 611 (1987); 10.1063/1.98363 Gallium Diffusions into Silicon and BoronDoped Silicon J. Appl. Phys. 42, 3750 (1971); 10.1063/1.1659681
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 139.141.11.40 On: Thu, 23 Oct 2014 13:44:11
Resistivity of boron ..doped po!ycrystaUine sincon M< Y. Ghannam a ) and R. W. Dutton Stanford University, AEL 201, Stanford, California 94305
(Received 19 October 1987; accepted for publication 9 February 1988) The doping dependence of the resistivity of polycrysta1line silicon deposited by low-pressure chemical vapor deposition and implanted with boron is investigated. At doping concentrations < 10 1M em -3, the resistivity is almost two orders of magnitude larger than that of crystalline silicon. At very large doping levels ( = 102fl em -3), the resistivity is comparable to that of crystalline silicon though slightly higher. Boron segregation at grain boundaries is not observed for doping levels < 1O!9 cm- 3 • At 102() cm--l, boron segregation or additional clustering at grain boundaries causes a reversible change in the resistivity upon changing the annealing temperature.
During the last decade, polycrystaHine silicon (polysilicon) deposited by low-pressure chemical vapor deposition (LPCVD) has been widely investigated. A significant amount of data concerning the structure, the physical and electrical properties of ion implanted pclysilicon layers subjected to furnace annealing l _.'i or to rapid thermal annea!ing6-l< is available. Most of the work was oriented toward phosphorus-doped poiysilicon films used as the gate materia! in metal-oxide-semiconductor (MOS) technology and to arsenic-doped polysilieon films mainly used as the emitter contact in bipoiar technology. Much less attention was given to boron-doped polysilicon. Boron-doped polysilicon is now commonly used in state of the art silicon self-aligned bipolar technology as the extrinsic base contact. A full knowledge of its properties is necessary for proper design of the device. Its resistivity is one afthe factors upon which depends strongly the speed performance of the transistor. In this work, we present experimental results concerning the doping dependence of the resistivity of boron-doped (impianted) LPCVD polysilicon. We also try to understand the phenomenon governing this dependence. Polysilicon films are deposited by LPCVD at 625°C and 500 mTorr on top oftht:rmally oxidized silicon wafers (100 nm of oxide). The film thickness is measured by means of a nanospec, which is based on optical interference analysis. A refractive index of 3 is used for the undoped polysiiicon materiaL An average thickness of 535 nm is observed with a standard deviation of 4.5 um. The samples are then implanted with boron at 60 keY which positions the implant peak nearly at the center of the polysilicon films. A low dose ( = 10 14 em- 2 ), a moderate dose ( = 10 15 cm- 2 ), and a high dose (= 10 16 cm- 2) corresponding to 1.87>< lOtH cm --'. 1.87 X 10 1'l em -3, and 1.87 X lO:w em- 3, respectively, have been investigated. Boron atoms are electrically activated and redistributed in the polysilicon films after being subjected to heat cycles at 800, 900, and 1000 °e for 12,2, and 1 fl, respectively. A final anneal at 1100·C for 20 min is also performed. The cycles are long enough to achieve maximum possible boron activation at the corresponding annealing temperature. After each heat cycle the sheet resistance is measured using a standard four point tungsten carbide probe
a)
Present address: Electronics and Communications Department, Faculty of Engineering, Cairo UniverSity, Guiza, Egypt.
1222
AppL Phys. Lett 52 (15), 11 April 1988
system. The resistivity of each film is calculated by multiplying its sheet resistance by the average thickness. Due to a continuous increase in grain size, a reduction in the resistivity with increasing annealing temperature is observed for all implant doses. A greater dependence of the resistivity on the grain size is, however, observed for lower doping levels. The grain size reaches its largest value and is stabilized after the 1100 °C anneal. Figure I displays the dependence of the resistivity of boron-doped polysiHcon on the implant dose (and on the estimated doping level) after a final anneal at 1100 de in argon. The value of the resistivity of the low dose films is scattered over a relatively wide range and averaged at 2.67 n em (with a standard deviation of 0.56 n em). This value is almost two orders of magnitude larger than that of crystalline silicon doped with boron to the same level (1.87X 10 18 cm---'). Since boron segregation at the grain boundaries was found negligible, 3 this high resistivity is attributed to carrier trapping at the grain boundaries 9 which is detrimental for electrical conduction for two reasons; First, this phenomenon results in a relatively important depletion of free carriers inside the grains. Second, car-
Doping concentration (cm-3) 18
E
10
10 20
10
21
I
:;-
(.)
E
:-:-
..r:::.
-2-
:E'
19
10
10
10 2
10- 1 l:"
> ~ -tij 10- 2 m
I-
a:
10-3 i'-
10- 4 10 13
\.,
,~
\\,'
~.
~ 10 14
10 15
lO i6
10 17
Impiant dose (cm-2) FIG. L Resistivity of boron-doped LPCVD polysilicol1 vs boron implant dose (and boron doping concentration). These values are determined experimentally after a final anneal at l100"C for 20 min. Also, the resistivity of crystalline silicon IS plotted for comparison.
0003-6951 188/15i 222-03$01.00
@ 1988 American Institute of Physics
1222
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 139.141.11.40 On: Thu, 23 Oct 2014 13:44:11
rier trapping at the grain boundaries creates a potential bar~ rier which impedes the free transfer of carriers from one crystallite to another (referred to as carrier reflection at the grain boundaries). The result is a strong reduction in the average carrier mobility in the material. It should be mentioned that defect scattering at the grain boundaries also re~ duces the carrier mobility in polysilicon and can be considered as a second component of carrier reflection at the grain boundaries. Because most of the traps at the grain boundaries become saturated at high doping concentrations, the conductivity should be strongly enhanced. In this case, oniy a smaH fraction of the carriers is trapped and the rest are free to move inside the grains and to contribute to the electrical conduction, In addition, the probability of tunneling through the potential barrier from one crystallite to another increases strongly as the doping level approaches degeneracy which renders the carriers more mobile throughout the material. Our experimental results are consistent with these expectations. Indeed, the film doped with a moderate dose shows an average resi.stivity of 15 mn cm (with a standard deviation of 0.57 mn cm), which is about 2,3 times larger than that of crystalline silicon doped with beron up to the same doping level, The resistivity of the high dose film is estimated to L4 mn em (with a standard deviation of 0.047 mn em), which is still higher than but very dose to that of boron-doped crystalline silicon (0.8 mn em). Any heat cycle performed at T < 1000 °C after the 1000 °C anneal does not affect the established value of the resistivity of the low dose (10 14 em - 2) film or ofthe moder~ ate dose (lO!S em -2) film [Fig. 2 plot (a)]. On the other hand, when the annealing temperature is decreased and increased repeatedly between 800 and 1000 ce, the resistivity of the high dose film alternates between a high value (for 800"C anneal) and a low value (for 1000 °C anneal) as shown in Fig. 2 (plot (b)]. If this behavior is caused by dopant segregation at grain boundaries, as is the case for arsenic- and phosphorus~doped polysilicon.2.~ it should have been observed at aU doping ranges. 3 However, we cannot really rule out that boron segregation at grain boundaries
&
'"
"\ -Mi M2?'R-~__r-~~B
II! (J
800°C 900 C iOOO C 11 CO C
(,,)
Hi
2
3
4
5
6
7
8
9
10
0
11
12
Annealing sequence FIG. 2. Resistivity of boron-doped LPCVD polysHicon vs the annealing seqnence. Two different temperature-sequence schemes arc investigated: (a) M 1 andM2 are performed for the moderate dose tUms (10 15 em -2) and (b) H 1 and H2 for the high dose films (lo'n ern- 2). 1223
Appl. Phys. Lett., Vol. 52, No. 15, i i April 1968
might be much more significant at degenerate doping levels. Another explanation for the reversible behavior of the resistivity involves clustering, It has been reported that clustering (formation of dopant-point defect complex) in boron-doped crystalline silicon,1O and in arsenic-doped crystalline silicon II is significant only at very high doping levels (1020 em - 3), and totally negligible for lower doping levels. Therefore, based on the fact that grain boundaries are regions with very high defect density, we suggest that the reversible change in resistivity u.pon changing annealing temperature might be the result of successive clustering and declustering at the grain boundaries. It is clear that a high duster density is associated with an increased. resistivity for two reasons. First, dopant atoms that are included in the duster do not contri.bute to electrical conduction at room temperature, and second, the carrier mobility is reduced due to scattering at cluster sites. At higher annealing temperatures less clusters are formed which results in a smaner resistivity. Since the resistivity p = l/qp IL p ' where q is the electronic charge, p is the free-hole concentration, and f.lp is the average hole mobility in the polysilicon film, this latter can be extracted from resistivity measurements. We failed to extract any value for the mobility in the low dose film because, as explai.ned earlier, the fraction of carriers trapped at the grain boundaries is significant and therefore the free-carrier concentration cannot be determined accurately. On the oth~ er hand, one can assume that the free-carrier concentration in the moderate dose film is equal to the doping concentration. The minimum resistivity obtained after the 1100 ·C anneal results in a hole mobility equal to 21 cm 2 ;V s which represents 40% that of crystalline silicon having the same doping concentration (1,87 X lOll) cm -}), When carrier trapping is considered, the free-carrier concentration is smaHer than the doping concentration and therefore we expect the mobility to be slightly higher than 21 cm 2 jV s. Nevertheless, we may conclude that at this doping level, carrier reflection at the grain bou.ndaries is limiting the electrical conductivity in the polysi.licon film. In the high dose film., the free-carrier concentration is taken as 1.75x 1020 cm- 1 based on an extrapolation of Schwettmann's data 1O to in~ elude 1100 °C annealing temperature. Neglecting additional clustering (or segregation) at grain boundaries (which is a reasonable assumption for the 1100 "C anneal), the hole mobility in this high dose film is found equai to 25 cm 2/V s which represents 60% of that of crystaHine silicon doped with boron to the same level (L87x 1020 cm- 3 ). In summary, we have studied the resistivity of boronimplanted LPCVD polysiHcon films. Resistivities comparable to single~crystal silicon resistivities can be obtained in polysilicol1 only at very high dopi.ng levels ( 1020 cm _.) and only after high~temperature annealing (1100 ee) The resistivity of this very heavily doped material is, however, very sensitive to any post processing at lower temperatures due to dopant clustering (or segregation) at the grain boundaries. For boron concentrations> 1O l9 cm- 3 we found that the hole mobility i.n polysiHcon is 40--60% that of crystalline silicono IT. l. Kamins, I. ElectroGhem. Soc. 126,833 (1979). 'T. L Kamins, J. Electrochem. Soc. 127. 686 (1980). M. Y. Ghannam and R. W. Duttor
1223
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 139.141.11.40 On: Thu, 23 Oct 2014 13:44:11 .......• •.• •.•.•;•••..•.•••.• •.•;-..•.•••.'..... •.•. ·,'.'.·.'.;.".:-:·;·;·7.·.·.·.'.·.·.·.·.··-······· ..•.. ;.. ~.:.;
~
:-.~.;
-n'.:.:.:.;.;.~
