Multielement Characterization of High-Purity Titanium for

Multielement Characterization of High-Purity Titanium for Microelectronics by Neutron Activation Analysis. Dieter. Wildhagen, and Viliam. Krivan. Anal...
0 downloads 0 Views 1MB Size
Anal. Chem. 1995, 67, 2842-2848

Multielement Characterization of HighmPurity Titanium for Microelectronics by Neutron Activation Analysis Dieter Wlldhagen and Viliam Krivan*

Sekiion Analytik und Hmbstreinigung, Universimt Ulm, D-89069 Ulm, Germany A radiochemical neutron activationanalysis technique for the determination of 26 elements including the a-emitting elements Th and U and Cu, Fe, K, Na, Ni, and Zn has been developed. The radiochemical separation was performed by anion exchange on a Dowex 1 x 8 column f” HF and HF/NH4Fmedium. It leads to a selective r e m d of the matrix-producedradionuclides 46Sc,47Sc,and 48sc and a nearly selective isolation of 239Npand 233Pa,the indicator radionuclides of U and Th,respectively. Counting the intensive but unspecific 51 1-keVpray of was enabled by a selective extraction of copper with dithizone from 15 M HF. For K, Na, Th,arrd U, a limit of detection of 30, 0.05, 0.03, and 0.07 ng/g, respectively, was achieved. For the other elements, the detection limits were between 0.002 ng/g for Ir and 45 ng/g for Zr. The elements As, Cr, and Mn were assayed only by instrumental neutron activation analysis. These techniques were applied to the analysis of two titanium sputter target materials of different purity grade. Results firom seven elements are compared with those of isotope dilution and glow discharge mass spectrometry. Titanium, because of its several special properties, has become an important material for advanced technology. Trace impurities have been found to influence many of these properties. Especially in microelectronic research and technology, there is an increasing demand for high-purity titanium as a basic material for the production of thin films of TIN, TiSiz,and Ti02 used for very large scale integration (VLSr) and ultra large scale integration CULsr) applications.’-1° Layers of these compounds can act as diffusion barriers, low-resistance contacts, and gate and source/drain materials. Titanium dioxide is a promising material to substitute (1) Ostling, M.;Nygren, S.; Petersson, C. S.; Norstrom, H.; Buchta, R; Blom, H.-0.; Berg, S. Thin Solid Films 1986, 145,81-88. (2)Glass, R C.;Spellman, L. M.; Tanaka, S.; Davis, R F.J. Vac. Sci. Technol. A 1992,10(4), 1625-1630. (3)Willemsen, M. F.C.; Kuiper. A. E. T.; Reader, A H.; Hokke, R; Barbour, J. C. 1.Vac. Sci. Technol. B 1988,6(1), 53-61. (4) Kolbesen, B. O.;Pamler. W. Fresenius Z. Anal. Chem. 1989,333,561-568. (5) Pramanik. D.;Jain, V. Solid State Technol. 1993,38,73-82. (6)Wicaksana, D.;Kobayashi, A: Kinbara, A J. Vac. Sci. Technol. A 1992, 10(4), 1479-1482. (7)Spitzer, A;Reisinger, H.; Willer. J.; Honlein, W.: Cerva, H.: Zom, G. In Insulating Films on Semiconductors 1991: Thin Ti02 as a film with a high dielectric c~nstant;Eccleston, W.. Uren, M., Eds.; Adam Hilger: Bristol, U K , 1991; pp 187-190. (8)Pamler, W.; Wangemann, K; Bensch. W.; Bussmann, E.; Mitwalsky, A. Fresenius 2.Anal. Chem. 1989.333,569-575. (9)Li,B.-Z.; Zhou, S.-F.:Hong, F.; Jiang, G.-B.; Liu, P.; Zhang, A.M.; Chao, M. J. Vac. Sci. Technol. B 1988,6(6), 1714-1720. (10)Shenai, IC;Piacente, P. A; Lewis,N.; Smith, G. A: McConnell, M. D.: Baliga, B. J. J. Vac. Sci. Technol. B 1988,6(6).1728-1733.

2842 Analytical Chemistry, Vol. 67, No. 17, September 1, 1995

for silicon dioxide in dynamic random access memory (DRAM) above 16 MbiL7 In these applications, the purity requirements are extraordinarily high for the natural radioactive elements thorium and uranium which, due to emission of a-particles, can cause changes in the potential of the data storage cells.11J2 Further trace impurities of interest include the “mobile ions” of alkali and alkali earth elements and some other metallic impurities such as Co, Cr, Cu, Fe, Mn, Ni, and Zn.11J2 As a consequence of these purity requirements, the develop ment of adequate analytical methods is necessary. Among the mass spectrometry methods using solid samples, glow discharge mass spectrometry (GDMS) has become the most important one for bulk trace characterization of high-purity materials and it has also been applied to analysis of titar1ium.~~3’~ However, for accurate quantification, these methods require matrix-containiig standards, and inhomogeneities can lead to serious errors due to the extremely low sample consumption. Atomic and mass spectrometry solution techniques such as graphite furnace atomic absorption spectrometry (GFAAS), inductively coupled plasma atomic emission spectrometry (ICPAES), inductively coupled plasma mass spectrometry (ICPMS) and isotope dilution mass spectrometry (IDMS) require sample digestion and, in the most instances, matrix/analyte separation. Recently, an essential improvement of the detection limits of the solution atomic and mass spectrometry methods as applied to analysis of titanium by matrix/analyte separations prior to determination was reported.15-18However, the blank can be a considerable limitation and, for example, in the work by Beckmann and Wunsch,I6 this limitation was not considered at all. Beer and Heumann18 determined eight elements including Th and U by IDMS based on matrix/analyte separation using several procedures including cation exchange and electrodeposition for Cd, Cr, Cu, Ni, and Pb, anion exchange and extraction for Fe and U, and coprecipitation with LaF3 for Th. Detection limits of 0.07 ng/g for Th and U and of 1-35 ng/g for the other elements were achieved. However, the separation procedures described seem to be rather time-consuming, and several detrimental impurities were not assayed. (11) Ortner, H.M.;Blodom, W.; Friedbacher, G.; Grasserbauer. M.; Krivan, V.; Virag, A; Wilhartitz, P.; Wunsch, G. Kontakte 1988,3,39-52. (12)Krivan, V.Nachr. Chem. Tech. Lab. 1991,39(5). 536-540. (13)Guidoboni, R.J.; Leipziger, F. D.1.Cyst. Growth 1988,89,16-20. (14)Venzago. C.; Weigert. M. Fresenius J. Anal. Chem. 1994,350,303-309. (15)Yoshikawa. H.;Ishibashi. Y.; Gunji, N.; Misumi, T. Bunseki Kagaku 1990, 39(12), 829-833. (16)Beckmann, K; Wunsch, G. Fresenius 2.Anal. Chem. 1992,342,469-472. (17)Kozuka. S.;Kon, M.; Hayashi, M.: Matsunaga, M. Bunseki Kagaku 1993, 42(1),T19-T22. (18)Beer, B.; Heumann. K G. Anal. Chem. 1993,65, 3199-3203. 0 1995 American Chemical Society 0003-2700/95/0367-2842$9.00/0

Only a few applications of instrumental neutron activation analysis (INAA) or radiochemical neutron activation analysis (RNAA) to the analysis of titanium have been reported dealing with the determination of a small group of elements in samples of lower p ~ r i t y . ~Some ~ - ~radiochemical ~ multielement separation procedures were published earlier by Neirinckx et al.23-25However, they are not well suited for analysis of the high-purity titanium used in microelectronics as they were not primarily developed for the determination of those trace elements relevant in this application field. In the present work, two radiochemical separation procedures for the determination of 27 elements, including Th and U, by RNAA were developed, which, in addition to IN& were applied to a comprehensive characterizationof two titanium sputter target materials. EXPERIMENTAL SECTION

Reagents and Radiotracers. All reagents used for the separation procedures were of "pro analysi" quality and supplied by Merck, Darmstadt, Germany. The original concentration of nitric acid and hydrofluoric acid was about 65 and 40%, respectively. The exact concentration of the hydrofluoric acid was determined by titrimetry. The anion-exchange resin used was Dowex 1 x 8 (100-200 mesh) in C1- form (Fluka, Buchs, Switzerland). The radiotracers of the elements investigated were prepared by irradiation of pure metals or suitable compounds in the nuclear reactors described below, and they were monitored for radiochemical purity by pray spectrometry. Samples and Standards. The titanium sputter target sample Ti-1 was obtained from Degussa, Hanau, Germany and the sample Ti-2 from Demetron, Hanau, Germany. Sample portions of 60120 mg were cut with a diamond saw. Before irradiation, the samples were etched three times with a mixture of 2%suprapure HF and 5% subboiled HN03 for 1 min and washed with water purified using a Milli-Q system (Millipore, Neu-Isenburg, Germany). For irradiation times of >2 h, the samples were sealed into Suprasil quartz ampules (Heraeus, Hanau, Germany), and for irradiation times of 2 d (mode C), sample portions between 60 and 120 mg were irradiated for 5 d in the FRG1 reactor of the GKSS Research Center (Geesthacht, Germany) at neutron fluxes of 0 t h = 5.2 x l O I 3 cm-2 s-l, Qepi = 2.0 x 10'2 cm-' s-l and @fast = 1.3 x 1013cm-2 s-l. Before counting the prays, decay times of 26 h (mode A), 220 h (mode BANAA), 22 d (mode B/RNAA), 2 3 d (mode C/INAA), and 2 4 d (mode C/RNAA), respectively, have been applied. Radiochemical Separation Procedures. The procedures described below were developed by using the radiotracer technique for checking yields. The flow charts are shown in Figures 1 and 2. For surface decontamination, the irradiated titanium samples were etched two times with a mixture of 2% HF/5% HN03 for 1min, washed with demineralized water, dried, and weighed. After addition of 5 pg of carrier for each element, 0.5 mL of 65% HN03, and 0.5 mL of 10 M HF, the samples (120 mg) were decomposed in 50-mLTeflon beakers (VIT-LAB, Seeheim-Jugenheim, Germany) under the dropwise addition of -0.5 mL of 20 M HF. The solution was evaporated to dryness under an infrared lamp. After addition of 0.2 mL of 20 M HF and evaporation to dryness, the residue was redissolved in 0.2 mL of 20 M HF and diluted to 4 mL with demineralized water. The Teflon beaker was washed twice with 1 mL of 1 M HF. The resulting solution was transferred into 25mL Teflon pressure liners (Berghof, Eningen, Germany) and heated at 180 "C for 4 h. The separation columns were made of polystyrene syringes of 120 mm x 8 mm i.d. (irradiation mode B) and 160 mm x 8 mm i.d. (irradiation mode C) having an active bed of 6 and 8 mL, respectively. They were filled with Dowex 1 x 8 (100-200 mesh), C1- form. Before use, the resin was converted to the F- form with 20 mL of 40%HF and pretreated with 40 mL of 1 M HF at a flow speed of 0.8 mL/min. Radiochemical Separation of Short- and Medium-Lived Indicator Radionuclides(see Figure 1). The sample solution was passed through the column at a flow rate of 0.4 mL/min, and the column was then eluted with 1 M HF until 25 mL of eluate Analytical Chemistry, Vol. 67, No. 17, September 7, 1995

2843

+

I

I

ntrlrr etching

aditimofwricnnndd-pition

d i s o l u h in 0.2 mL 20 M

dissalutianin 0.2mL 20 M HF

HF

I hating 01 I8VC for 1h

elution 1: with 1 M HF until 25mLclu!cdsolution

elution 2 with 40 mL 0.5 M HF/0.3 M NH,F

cluiion 3 with 30 mL 4M+10mLmns.HN03

elution 1: with IM HF d Y l i o n 2 with 5OmL3 M HF

d Y t i o n 3 wilh 60 mL

0.5 M HF10.3 M NHIF

elution 4 wiih 35 mL IM HF10.5 M NHIF clution 5 with 40mL 4 M + IOmLcone. HNO,

a,co.(Q).cr.

81.

Fc. In. Ir. Rb. Ru. Sc.

Hf Mo. MO Nb. Nb Re. Re (Sb). Sb Hf. cxva~iionof Cu w h

Figure 1. Flow chart of the radiochemical separation procedure for RNAA of titanium via short- and medium-lived indicator radionuclides (mode E).

Figure 2. Flow chart of the radiochemical separation procedure for RNAA of titanium via medium- and long-lived indicator radionuclides (mode C).

was gathered (eluate 1). After addition of 25 mL of 50% HF, the resulting hydrofluoric acid solution was extracted twice with 10 mL of 0.005 M dithiione in chloroform followed by washing with 10 mL of chloroform to achieve a selective separation of ~ 4 Cu. Subsequently, the column was eluted with 40 mL of 0.5 M HF/0.3 M N&F (eluate 2) and finally with 30 mL of 4 M and 10 mL of concentrated HNO3 (eluate 3), both at a flow rate of 0.8 mL/min. Eluates 1and 3 and the organic phase of the extraction were counted. Radiochemical Separation of Medium- and Long-Iived Indicator Radionuclides (see F i 2). The sample solution was passed through the column at a flow rate of 0.4 mL/min, and the column was then eluted with 1 M HF until 50 mL of eluate was gathered (eluate 1). The column was subsequently eluted first with 50 mL of 3 M HF (eluate 2). then with 60 mL of 0.5 M HF/0.3 M NHQ (eluate 3) followed by elution with 35 mL of 1M HF/0.5 M N&F (eluate 4) and, finally, with 40 mL of 4 M and 10 mL of concentrated HN03 (eluate 5), all at a flow rate of 0.8 mW min. Eluates 1,2,4, and 5 were counted. Extraction of Anthony. For extraction of %b and Iz5Sb with diethylammonium diethyldithiocarbamate (DDTC), Wmg aliquots of the irradiated samples were decomposed in the above described manner, evaporated to dryness, and taken up with 120 pL of 20 M HF; the Teflon beaker was washed twice with 0.5 mL of 1 M HF. The resulting solution was transferred into 2 5 ” Teflon pressure liners and heated at 180“C for 4 h. After cooling, 2 mL of 37%HCI, 5 mL of a saturated solution of boric acid (to

mask fluoride), 3 mL of 20% tartaric acid (to mask titanium), and 0.5 mL of a solution containing 15%potassium iodide and 2.5% ascorbic acid (to reduce S b O to SbUlI)) were added. The FTFE liners were closed and then heated for 1.5 h at 80 “C. After cooling, 0.1 mL of 15%ascorbic acid was added and the sample solution was extracted twice with 5 mL of 0.025 M DDTC in chloroform and washed with 5 mL of chloroform. The organic phase was counted.

2844 Analyiical Chemisfry, Vol. 67, No. 17, September 1, 1995

RESULTS AND DISCUSSION

INAA. The nuclear reactions that can be induced by reactor neutrons on titanium are listed in Table 1. Figure 3 shows the dependence of the produced activities of the most important mahix-formed radionuclides on the irradiation time, assuming a sample weight of 100 mg. From this figure, it is evident that an instrumental performance of the neutron activation analysis is principally possible, especially when use is made of short- and medium-lived indicator radionuclides applying the irradiation modes A and B. The high mahix activity of 3% (t,12 = 5.8 min) decays to a sufiicient degree after a cooling time of 2 h. The radionuclides %c, 47Sc,and “RSC generated via the fast-neut” induced (n,p) reactions significantly increase the Compton background and, consequently, the limits of detection, too. Serious limitations in INAA via short- and medium-lived indicator radionuclides (irradiation mode B) and via long-lived indicator radionuclides (irradiation mode C) are caused hy the medium-lived and the long-lived %c. giving rise to a high Compton

Table I. Relevant Nuclear Reactions Induced on Titanium by Reactor Neutronsm**7

nuclear reaction

abundance, %

4T1 (n,2n)4yTi

8.0

4Tl(n,p) %c 47Ti(n,p) 47Sc 4Qi (n,p) 4sSc 4Qi 1"

cross section, barn

istopic

7.8 x 1.1 x 1.6 x 2.7 3.4 1.4 x 0.18 1.3 8.1 x

7.5 73.7

(n,a)45Ca 5.5 5.3

(n,p)49Sc

T i(n,y)"Ti

j(Ti(n,p)50Sc

j0Ti (n,a)47Ca Im

10m

I h

I d

10-2 10-4 10-5 10-3 10-5

half-life

main y-lines, keV (intensity, %)

3.08 h 83.8 d 3.41 d 43.7 h 163 d 57.3 min 5.76 min 1.71 min 4.54 d

511.0 (172) 889.3 (loo), 1120.5 (100) 159.4 (68) 983.4 (loo), 1311.8 (100) no y-lines 1780.0 (0.03) 320.0 (95), 928.5 (5.0) 1121.0 (loo), 1553.7 (100) 1296.8 (75)

10d

The extent of the primary interference of Mg via the reaction in the determination of Z4Mg ( n , ~ ) ~ ~and N aof Al via 27Al(n,a)24Na Na was estimated by using the applied fast neutron flux and the contents of these elements given by the producers. The interference of Mg in both samples and of Al in the sample Ti-1 was negligible (