266 1
Anal. Chem. 1988, 60, 2661-2665
optical fiber bundles were used to transmit both the excitation and scattered radiation, the preferred cell holder was a thermostated metal block onto which both fibers were attached with set screws. The scattering geometry is defined by the metal block, and because the fiber bundles utilized in this study have a relatively large numerical aperture, it was necessary to use a 57O scattering geometry to avoid collecting Raman spectra of the excitation fiber. However, because the two fibers are in very close proximity to the sample itself, virtually no alignment is required. This type of closed system offers maximum safety with regard to the NdYAG laser beam. Moreover, lenses and other optical components can be mounted directly in the metal block for specialized experimental requirements. That is, focusing optics can be incorporated for increasing laser power densities if ordinary liquid and solid samples are to be examined. The addition of the second fiber bundle also ensures that any polarizing effects of the beamsplitter (14)would be eliminated, since the fiber-optic serves as an analogue to the scrambler used in dispersive Raman spectroscopy. Also, the circular image of a fiber bundle is well suited to the aperture of the interferometer. Since the end of the fiber carrying the excitation radiation can be modified to match closely the physical shape of the sample cell, thereby illuminating more molecular scatterers, comparable Raman scattered signal levels can be obtained for lower excitation power densities. The combination of mid-infrared interferometers and fiber-optic assemblies should give rise to low-cost FT-Raman accessories that are readily adaptable to most standard interferometers. For low-cost interferometers not specifically designed for emission spectroscopy, a fiber accessory for carrying Raman radiation could pass through the instrument case and replace the existing infrared source. Further, with near-infrared excitation, the normal KBr/Ge beamsplitter found in all mid-infrared interferometers could be used with
only a slight sacrifice in efficiency over a quartz beamsplitter (7).Depending on the instrument's use, appropriate detectors can replace the commonly used MCT (mercury-cadmiumtelluride) or DTGS (deuteriated triglycine sulfate) units. Aside from alignment and safety considerations, the combination of near-infrared FT-Raman spectroscopy and fiber-optic assemblies has considerably extended our ability to obtain spectral measurements on biologidy relevant systems. In particular, this approach enables us to record more easily and straightforwardly Raman spectra of fragile, highly fluorescing biomolecules in dilute aqueous media. Registry No. DPPC, 2644-64-6; benzene, 71-43-2; indene, 95-13-6; amphotericin A, 1405-32-9. LITERATURE CITED Hirschfekl, T.; Chase, 0. Appl. Spectrosc. 1986, 40, 133-137. Chase. D. B. J . Am. Chem. Soc. 1986, 108, 7485-7488. Chase, D. B. Anal. Chem. 1987, 59, 081A-089A. BUijS, H. SpeCtro~cqPy1986, 7(8), 14-15. Buljs, H. Bomern Technlcal Note No. DA3-8701T: Application Note No. DA3-8702. Zimba, C. G.; Hallmark, V. M.; Swalen, J. D.; Rabdt. J. F. Appl. SpechaSC. 1987, 4 1 , 721-726. Lewis, E. N.; Kalasinsky. V. F.; Levln, I. W. Appl. Spectrosc. 1988, 42, 1188-1193. Schwab, S. D.; McCreery, R. L. Anal. Chem. IS84. 56, 2199-2204. McCreery, R. L.; Fleischmann, M.; Herdra, P. Anal. Chem. 1988, 55. 146-1 48. Trott. 0. R.; Furtak, T. E. Rev. Scl. Instrum. 1980, 57, 1493. Wilson, E. B.; Decius, J. C.; Cross, P. C. Moleculer Vlbratbns; McQraw-Hill: New York, 1955. Levin, I. W. Advances h Infrared and Raman S p e c h s q y ; Clark, R. J. H., Hester, R. E., Eds.; Wiley. Heyden: New York. l9S4: Vol. 11. . Chapter 1. Kirchhoff, W. H.; Levln, I. W. J . Res. Net/. Bur. Stand. ( U S ) 1987, 92, 113-128. Martin, D. H.; Puplett, E. Infrared Phys. 1969, 70, 105.
RECEIVED for review May 17,1988. Accepted September 1, 1988.
Palladium-Tin Ratios in Electroless Copper Plating Catalysts Determined by Rutherford Backscattering Spectrometry Bruce Pierson, Kenneth W. Nebesny, and Quintus Fernando* Department of Chemistry, University of Arizona, Tucson, Arizona 85721 Tetsuya Ogura
Departamento de Quimica, Universidad Autonoma de Guadalajara, Guadalajara, Jalisco, Mexico The concentratlons of tln and palladium In catalysts used In electroless copper platlng have been determined by Rutherford backscatterlngspectrometry wlth hlghemergy (2-5 MeV) 'He'. The thxpalladlum ratio In the catalyst decreases when catalyskoatsd substrates are exposed to an alkaHne solution. X-ray photoelectron spectroscopy has conflrmed thls result and has shown that the palladlum In the catalyst Is present as palladlum metal and that the tln is present, probably as an oxldlzed specles, lo a depth of about 30 A. Catalysts for the electroless platlng of copper are obtalned by the reactlon of Pd(I1) with Sn(I1). The extent of the reactlon and the concentratkns of the reaction produds depend on the sokrtkn condltlons. Confllctlng results obtalned In prevlous Investlgatlons of palladium-tln catalysts can be explalned on thls basls.
The electroless plating of copper is carried out by the re0003-2700/S8/0360-2661$01.50/0
duction of copper@) to copper metal with formaldehyde. The reduction is performed in a highly alkaline solution at a temperature of =70 "C in the presence of ethylenediaminetetraacetic acid (EDTA) and CN-. The inherent advantages of electroless (autocatalytic) plating over electrolytic plating (electrodeposition) of copper in the manufacture of printed circuit boards have been recognized for more than three decades. A critical step in the fabrication of a copper-clad board is the deposition of a palladium-tin catalyst on a nonconductive substrate for the initiation of the electroless deposition of copper metal. For many years a two-stage process was used for this purpose. The substrate was immersed in an acid solution of dilute SnClz (sensitization step) followed by immersion in an acid solution of PdClz (activation step). In the 1960s a single-step process was introduced in which the substrate was immersed in an aqueous HC1 solution containing a mixture of SnClz and PdC12. The activity of the PdC12-SnC12 catalyst in the single-step process was enhanced 0 1988 Amerlcan Chemical Society
2862
ANALYTICAL CHEMISTRY, VOL. 60, NO. 24, DECEMBER 15, 1988
by treatment of the substrate surface with a solution of HCl, H2SO4, or NaOH (acceleration step). In the single-step process, the nature of the palladium and tin species that are adsorbed on the substrate surface has been the subject of a large number of investigations. The bulk of the experimental evidence supports the following model for the single-step sensitization: colloidal particles of a Pd-Sn alloy, =20 A in diameter and stabilized by a layer of adsorbed Sn2+,are formed in the sensitizing solution and are adsorbed on the substrate when it is immersed in the sensitizing bath. When the substrate is rinsed with water, tin compounds such as tin(1V) hydroxides are precipitated on the surface of the colloidal particles. These tin(IV) hydroxides rediasolve, either in the solution that is used as an accelerator or in the copper plating solution (1-3). The chemical reactions that occur when PdClz and SnClz are mixed in an HC1 solution can be represented by
-
Pd2++ 3Sn2+
-+
Pd(II)-3Sn(II) complex PdO Sn4+ 2Sn2+ (1)
+
Experimental evidence for the formation of the 1:3 Pd(I1)Sn(I1) complex and its slow autoreduction has been provided by lI9Sn Mossbauer spectroscopy and by Rutherford backscattering spectrometry (RBS) (2, 3). The above results however, are in disagreement with previous radiotracer work that was carried out with the y-emitters llsSn and lWPd ( 4 ) . These experiments showed that a 1:2 palladium-tin complex was formed and that palladium metal is not formed via the reaction Sn2+
-
+ Pd2+
PdO + Sn4+
An electron diffraction study of a dielectric surface treated with a SnC12-PdC12-HC1 solution followed by an accelerator solution indicated that a totally different compound, PdsSn, was formed on the surface (5). From the results that have been summarized above, it is evident that the palladium-tin ratio in the deposits on the substrate is variable and appears to be governed not only by the nature of the substrate but also by unspecified solution conditions. It is conceivable that these reported variations in the palladium-tin ratio were caused by the interaction of the catalyst on the substrate with the components of the electroless copper plating bath. Several workers have examined the variations in the palladium-tin ratios by employing Rutherford backscattering spectrometry with 2-MeV He+ ions (2,3,6). The palladium and tin peaks in the backscattering spectra show considerable overlap, especially if the energy of the incident He+ ion beam is less than 2 MeV. We have reexamined the effects on the palladium-tin ratio caused by immersion of the catalyst-coated substrates in an alkaline solution. A He+ ion beam at an optimum energy between 2 and 5 MeV was used, and from the areas under the palladium and tin peaks, which were obtained by a deconvolution technique, we have calculated the tin and palladium ratios. We have also carried out a photoelectron spectroscopic study of the surfaces of the catalyst-coated substrates before and after immersion in an alkaline solution. The variability in the Sn:Pd ratio that has been reported in previous investigations can be explained on the basis of chemical reactions that occur on the surface of the catalyst in alkaline media.
EXPERIMENTAL SECTION Pd:Sn Ratios by Rutherford Backscattering Spectrometry. The apparatus used for obtaining RBS spectra has been described in a previous paper (7). An area of about 4 mm2of the sample was irradiated with a high-energy *He+beam produced in a 6-MV Van de Graaff accelerator (High Voltage Engineering Corp.). The incident 4He+beam was normal to the sample surface, and the energies of the backscattered ions were measured with a silicon surface barrier detector (Ortec Ba-014-025-100)with a 15-keV resolution. The detector subtended a solid angle of 0.78
X sr at the sample target. The sample chamber was maintained at a pressure of =lo” Torr and was electrically isolated so that it could serve as a Faraday cup for the measurement of the number of 4He+ions incident on the sample target. A bismuth RBS standard was used to check the efficiency of charge collection
(8).
The sample targets were prepared on 1-cm2carbon or silicon substrates. In group I, 10 samples were prepared by immersing a carbon foil in a solution of a commercial catalyst (9) diluted with an equal volume of 1M NaCl solution for 5 min and rinsing the carbon foil for 1min in water. In group 11, six samples were prepared on a carbon substrate in the same manner as described above. After the 1-min rinse in water, three of the samples were immersed for 5 min in a solution of NaOH at pH 11.8 and then rinsed for 1 min in water. The remaining three samples were immersed for 5 min in a solution of NaOH at pH 11.8 that contained formaldehyde (0.027 M) and then rinsed for 1min in water. In group 111, five samples were prepared on a silicon substrate. Three of these samples were prepared by immersing the silicon substrate for 5 min in a 1:l mixture of the commercial catalyst and 1 M NaCl and then rinsing the silicon substrate for 1 min in water. The remaining two samples were prepared on a silicon substrate in the same manner. After the 1-min rinse in water, these two samples were immersed for 5 min in a solution of NaOH at pH 11.8 and then rinsed for 1min in water. All the samples were dried in air and analyzed by the RBS method. The samples in group I were analyzed with beam energies of 1892 and 3776 keV, the samples in group I1 with a beam energy of 3776 keV, and the samples in group I11 with a beam energy of 4700 keV. X-ray Photoelectron Spectroscopy. A Vacuum Generators ESCALAB MKII photoelectron spectrometer (Vacuum Generators, East Grinsted, U.K.) was used to obtain the photoelectron spectra of the catalysts-on the silicon substrates. An aluminum X-ray source (AlK, 1486.6 eV) was used at 300 W, and the pass energy was set at 50 eV. The full width at half-maximum (fwhm) of Ag 3d5j2varied between 1.5 and 1.8 eV for our system. Data were collected at 0.2-eV intervals, and the samples were scanned repetitively to obtain a suitable signal:noise ratio. The spectra were corrected for the transmission function of the spectrometer, (KE)-1/2. The iterative method that was used for background subtraction was based on the assumption that the background is proportional to the scaled integrated intensitiesof the corrected peaks (10). The binding energies and the areas of the Sn 3d and Pd 3d peaks were then determined. The binding energies were corrected on the basis that the carbon 1s peak occurred at 284.6 eV. Relative ratios of Sn:Pd were calculated from (3) Where I represents the peak area, N the number of atoms, u the cross sections ( I I ) , and X the inelastic mean free path (12). The calculated Sn:Pd ratios are shown in Table 11. Reaction of PdClP with SnC12in Acidic Media,The extent to which reaction 2 occurs in acidic solutions was determined in the following manner: a crystal of solid SnC12.2H20(1.633 mmol) was dissolved in 10 mL of 0.35 M HCl, and 0.500 mL of an aqueous solution containing 0.496 mmol of NazPdC14was added with vigorous stirring. The reaction was carried out in a nitrogen atmosphere to prevent the oxidation of Sn2+.The solution turned brown immediately, and after 20 min a dark brown solution was obtained. Solid Na2EDTA (2.270 mmol) was added to the dark brown solution, followed by the dropwise addition of 15 mL of 2 M NaOH with continued vigorous stirring for 10 min. The reaction mixture was heated to boiling for about 15 min, cooled, and filtered under nitrogen after the addition of Celite or filter aid. The concentration of Sn2+in the filtrate was determined by a back-titration technique; a known excess of 1, was added and the unreacted 13-titrated with a standard solution of Naa203 to the starch endpoint. It was found that 1.249 mmol of Sn2+ remained unreacted. Hence, in reaction 2, 0.112 mmol of Pd2+ remained unreacted; i.e. the reaction was only 77% complete under the above conditions, which are much more extreme than normally employed in the preparation of a mixed catalyst. The unreacted Sn2+in solution was determined by an alternative method in the following experiment. A crystal of solid
ANALYTICAL CHEMISTRY, VOL. 60, NO. 24, DECEMBER 15, 1988
2663
800 :
Z 3
0
0 400:
CHANNEL NUMBER
Flgure 1. RBS spectrum of the Sn-Pd catalyst on a carbon substrate after a 5-min immersion in the catalyst mixture and a 1-min rinse in water. The ?ie+ beam energy was 1892 keV (total dose = 30.02 pC). The energy/channei is 1.812 f 0.003 keV.
SnClZ.2Hz0(1.590 mmol) was dissolved in 10 mL of 0.20 M HCl, and 0.500 mL of an aqueous solution containing Na2PdCI, (0.493 mmol) was added slowly with vigorous stirring in a nitrogen atmosphere. The reaction mixture was maintained at room temperature, under nitrogen, for 3 h. The unreacted Sn2+in solution was oxidized by the addition of 1.686 mmol of 13-. After 10 min, when the oxidation was complete, 1.063 mmol of NazSz03 was added to react with the excess I,. The reaction mixture was filtered after the addition of Celite, and the unreacted NazS2Os, when backtitrated, was found to require 0.252 mmol of Is-. Hence, 0.825 mmol of Sn2+was left unreacted in the reaction mixture. This alternative procedure, in which an excess of NazSz03was added to the reaction mixture containing unreacted 13-,minimized the loss of Iz during the filtration step. The reaction of Is- with Snz+is extremely fast, but the reaction with Pdo is relatively slow. If the oxidation and filtration steps are carried out rapidly, there is little likelihood that the PdO will be oxidized by the Is-. If the reduction reaction, Pd2++ Sn2+ PdO + Sn4+,goes to completion, 1.097 mmol of Sn2+should be found in the reaction mixture. Instead, only 0.825 mmol of Sn2+was found; hence, 0.272 mmol of Sn2+was lost from solution either by coprecipitation with Sn02 or by the formation of Snovia the disproportionation of Sn(I1). Reaction of PdC142-with SnC12in Alkaline Media. The extent to which reaction 2 occurs in an alkaline medium was determined as follows: a crystal of solid SnC1.2H20 (1.227 mmol) was dissolved in 10 mL of water, and 1.0 mL of a solution containing 0.993 mmol of Na2PdCI,was added dropwise with vigorous stirring, under nitrogen. Thirty milliliters of a solution containing 40 mmol of NaOH was added dropwise with stirring. A bulky black precipitate was formed. The reaction mixture was heated to boiling for 'Iz h, and the dense black precipitate that remained was filtered, dried in a stream of nitrogen, and found to weigh 0.1202 g. No evidence for the presence of Sn2+was found in the filtrate. The concentration of the NaOH in the solution is sufficiently high to dissolve any hydroxide or hydroxy complex of Sn2+that may have been precipitated in the course of this reaction. If the reduction reaction, Pd2++ Sn2+ Pdo + Sn4+,goes to completion, 0.1057 g of PdO should be obtained. If this reduction reaction together with the disproportionation of Sn2+to SnO and Sn4+occurs, the precipitate should weigh 0.1196 g. This value is in good agreement with the experimental value, and it may be concluded, therefore, that the reduction reaction proceeded to completion and was catalyzed either by Pdo or Pd2+. The disproportionation reaction 2Sn2+ Sno + Sn4+occurs when the Pd2+:Sn2+ratio is greater than l / p Even when this ratio is less than 1/2, the disproportionation reaction occurred in the majority of the experiments that were carried out.
-
-
-
RESULTS AND DISCUSSION The RBS spectrum of the tin-palladium catalyst on a carbon substrate obtained with a 1892-keV He+ ion beam is shown in Figure 1. The locations of the Sn, Pd, C1, and 0 peaks are indicated by arrows. It is apparent that there is considerable overlap between the Sn and P d peaks. The peak shapes are, however, not Gaussian, probably because the coverage of the carbon substrate is not uniform. The areas
200 1
l 800
0
0
.'
o
o
0
s
400
200
3
0 800
(b) I
850
I
I
1
1
'
1
900
-'
9
channel number Figure 2. RBS spectra of the Sn-Pd catalyst on a carbon substrate after a 5-min immersion in the catalyst mixture and a 1-min rinse in water: (a) 'He+ beam energy = 1892 keV, total dose = 30.02 pC, energy/channei = 1.812 f 0.003 keV; (b) 'He+ beam energy = 3776 keV, total dose = 100.02 pC, energy/channel = 3.742 f 0.003 keV.
under these two skewed overlapping peaks were obtained by a curve-fitting method that has been described for skewed Gaussian distributions (13). The concentrations of Sn and Pd, (Nt)snand ( N t ) p d , were calculated from the peak areas and eq 4 where A = peak area (number of counts), DTR =
Nt = A(DTR)(CF,i)/QCh
(4)
dead time ratio for the detector, CFBi= correction factor from bismuth standard, Q = number of 4He+particles incident on the sample, s2 = solid angle subtended by the detector, D = Rutherford scattering cross section for Sn or Pd, N = number of atoms in 1 cm3, and t = thickness of catalyst layer in centimeters. The samples were analyzed by using three different beam energies (1892,3776, and 4700 keV). The higher beam energies were used in order to improve resolution of the Sn and P d peaks. These two peaks were much better resolved when the energy of the incident He+ beam was increased from 1892 to 3776 keV (Figure 2). The fwhm of these peaks (20-25 keV) was governed primarily by the detector rebolution (15 keV) since the catalyst coatings were very thin. With an increase
2664
ANALYTICAL CHEMISTRY, VOL. 60, NO. 24, DECEMBER 15, 1988
Table I. Concentrations of Sn and Pd by Rutherford Backscattering Spectrometry' concentrations of Sn and Pd (atoms cm-2) sample treatmentb
Sn:Pd
(Nt)s,
A
Group I (Carbon Substrate) 1.5 i 0.2 13.8
A and B A and C
Group I1 (Carbon Substrate) 0.83 i 0.2 1.9 1.0 f 0.1 2.1
A A and B
Group I11 (Silicon Substrate) 1.5 i 0.2 8.7 0.7gC 3.0
X
(Nt)pd
9.1
VI 600:
h
-i
2.3 2.0
z
3
0
.. ' .
0 4001 5.8 3.8
? .*
'Uncertainties shown in the table are standard deviations. b A = 5 rnin in catalyst, 1-min rinse in water; B = 5 min in NaOH (pH 11.8), 1-min rinse in water; C = 5 rnin in NaOH (pH 11.8) and HCHO (0.027 M), 1-min rinse in water. CThenumber of samples was insufficient for calculation of uncertainty.
in beam energy from 1892 to 3776 keV, the separation between the peaks was increased from approximately 25 to 50 keV, which reduced the overlap of the peaks (14). The average values of the Sn:Pd ratios are summarized in Table I. For the carbon substrates that were immersed for 5 min in the catalyst mixture and rinsed for: 1 min in water, o the average o Sn:Pd ratio was 1.5 f 0.2. The uncertainty in the ratio is largely due to the uncertainties in the measured peak areas (10-20%). The concentrations of Sn and Pd were found to be of the same order of magnitude that has been reported previously (2). The concentrations of Sn and Pd are significantly lower on the silicon substrates than on the carbon substrates, but the Sn:Pd ratios are the same on both substrates. After the catalyst-coated substrates were exposed to the action of a solution of NaOH at pH 11.8 in the presence or absence of formaldehyde, for 5 min, the Sn:Pd ratio decreased significantly (Figure 3 and Table I). The concentrations of both Sn and P d are lowered, but the Sn concentration is reduced to a greater extent as a result of the formation of soluble hydroxy complexes of Sn(I1) and Sn(1V). The results from the X-ray photoelectron experiments confirmed the conclusions that were drawn from the RBS experiments. The Sn:Pd ratio that was obtained for the catalyst on a silicon substrate was about 5:l. This ratio decreased to about 1:l after the catalyst was exposed to the action of a solution of NaOH at pH 11.8 in the presence or absence of formaldehyde. Similar trends were observed with carbon substrates (Table 11). These values of Sn:Pd ratios are not directly comparable with the Sn:Pd ratios that were obtained from the RBS spectra, because the sampling depth is about 30 8,with the X e s technique in contrast to a depth of =lOOOO A that is attainable with the RBS method. The binding energies (Table 11) indicate that the Sn is present in an oxidized form, probably as an oxide, and that the Pd is present as palladium metal, Pdo. It may be inferred from the results that the catalyst is a mixture of tin and palladium metal and that the outer layer consists primarily of oxidized tin to a depth of about 30 A. In alkaline solutions, the oxidized tin species are solubilized and the tin:palladium ratio decreases. When PdC12- and SnC12 solutions are mixed in an acid medium, the brown Pd(I1)-Sn(I1) complex that is formed decomposes slowly into PdO and Sn(IV). These experimental results are in agreement with previous findings from the Miissbauer spectroscopy of frozen solutions (I). The reaction conditions used in this work. the reactant concentrations used for the Mossbauer spectroscopy, and those used in the for-
**
200 :
* f
2150
io l m
0 0
channel number Figure 3. RBS spectra of the Sn-Pd catalyst on a carbon substrate: (e) after a 5-min immersion in the catalyst mixture and a 1-min rinse in water; (b) after a 5min immersan in the catalyst mixture and a l-min rinse in water, followed by a 5-min immersion in NaOH (pH 1 1.8) and a 1-min rinse in water. The %e+ beam energy = 3776 keV, total dose = 100.0 KC, and energyhhannei = 3.742 f 0.003 keV.
Table 11. Binding Energies and Surface Concentrationsof Sn and Pd by X-ray Photoelectron Spectroscopy sample treatment'
Sn:Pd
A A A and B A and B A and C
5.6 5.4 0.84 1.1 0.73
Silicon Substrate 486.7 335.5 486.6 335.4 486.0 335.3 485.8 335.2 486.5 335.2
A A and B A and C
6.9 1.8 0.57
Carbon Substrate 487.1 336.1 487.7 337.5, 336.2b 487.7 336.1
binding energies, eV Sn(3dep) Pd(3d5/,)
difference in binding energies, eV 151.2 151.2 150.7 150.6 151.3 151.0 150.2, 151.5 151.6
A = 5 min in catalyst mixture, 1-min rinse in water; B = 5 min in NaOH (pH 11.8), 1-min rinse in water; C = 5 rnin in NaOH (pH 11.8) and HCHO (0.027 M), 1-min rinse in water. bThe shoulder at 337.5 eV is attributed to the Dresence of Pd(I1). . .
ANALYTICAL CHEMISTRY, VOL. 60, NO. 24, DECEMBER 15, 1988 ~
~
~~
Table 111. Molar Concentrations of Reactants Used for the Preparation of Palladium-Tin Catalysts [Pd(II)] (a) this work (b) frozen solution
[Sn(II)] 0.06
2.5 X
1.1 X lo-* (2-7) X
[H+][Cl-] ref 0.2
0.3
4
4
1
0.3
0.5
9
used for Mossbauer
spectroscopy (c) single solution
10-3
0.1
sensitizer-activator mulation of a typical single solution sensitizer-activator are shown in Table 111. We have found that, in general, the rate of formation of Pdo decreases when the initial concentration of Pd(I1) is low and when the initial concentrations of Sn(II), H+, and C1- are high. Our reaction conditions (a) will, therefore, result in a much faster rate of formation of Pdo than either of the reaction conditions (b) or (c) shown in Table I. We have found that about 20% of the Pd(I1) remained unreacted 20 min after mixing the Pd(I1) and Sn(I1) solutions, whereas Sn(I1) was detected, 5 days after mixing, in the frozen solutions prepared for the Mossbauer spectroscopic studies. Commercial Pd-Sn catalysts, consisting of a combined sensitizer and activator, therefore, will contain negligible atnounta of PdO, especially in the presence of high concentrations of Sn(II), H+, and C1-. In alkaline solutions, Sn(1I) disproportionates readily into Sn(1V) and Sno in the presence of either Pd(I1) or PdO, but in the absence of Pd(I1) or PdO, Sn(I1) is stable in boiling solutions. In highly acidic solutions, however, the disproportionation of Sn(I1) is not favored (15). When PdC142-reacts with SnC12in an acidic medium, the color changes that occur indicate that a Pd(II)Sn(II) complex is formed. Our experimental results show that this complex decomposes slowly into Pdo and Sn(1V). This is in agreement with the conclusion obtained from previous Mossbauer and RBS experiments (1-3). The rate of formation of Pdois slow when the initial concentration of PdC14%is low and the initial concentrations of Sn(II), H+, and C1- are high. Even under our reaction conditions (Table 1111, about 20% of the Pd(I1) remained unreacted 20 min after mixing the solutions containing Sn(I1) and Pd(I1) and boiling the reaction mixture. It is not surprising therefore, that when the reaction of Sn(1I) with Pd(I1) is carried out under less extreme conditions, Sn(II) is found in solution several days after the reactants were mixed (1). The failure to detect PdO in a commercial catalyst ( 4 ) consisting of a combined sensitizer and activator can also be rationalized because the formation of Pdo is very slow when the initial concentrations of Sn(II), H+, and C1- are high. When a nonconductive substrate is treated with a sensitizer and activator and rinsed with water, a decrease in the hydrogen ion concentration on the catalyst surface occurs and the formation of PdO is accelerated. Hence the species present in the catalyst are Sn(II), Pd(II), and PdO, which probably is present as a stabilized colloid. When the substrate is subsequently immersed in a hot alkaline plating bath, the Sn(I1)
2865
and Pd(I1) species on the surface are converted into SnO, Sn(IV), and PdO. In addition, the Sn(I1) and Sn(1V) species surrounding the Pdo dissolve in the hot alkaline solution. The active catalyst surface is therefore either Pdo or a mixture or alloy of PdO and SnO. CONCLUSIONS The chemical composition of Sn-Pd catalysts used in the electroless plating of copper depends on the solution conditions that are employed in the preparation of the catalyst and on the alkalinity of the solution in contact with the catalyst during the electroless plating process. The important variables in the preparation of the Sn-Pd catalysts are the initial concentrations of Sn(II), Pd(II), H+, and C1-. Commercial catalysts are prepared by the addition of dilute PdClz to a solution that has a high concentration of Sn(II), H+, and C1-. Under these conditions, negligible amounts of Pdo are formed. When a substrate such as carbon or silicon is coated with the catalyst mixture containing Pd(II), Sn(II), H+, and C1-, the rate of formation of Pdo is governed by the Concentration of H+ in the catalyst layer. When the H+ ions in the catalyst layer are decreased by rinsing with water, the formation of Pdo is accelerated. Exposure of the catalyst layer to an alkaline solution results in the formation of Snoand Pdo; the oxidized tin species that are also formed subsequently dissolve in the alkaline solution. ACKNOWLEDGMENT We gratefully acknowledge the assistance of John A. Leavitt and Lawrence McIntyre, Jr., Department of Physics, University of Arizona. LITERATURE CITED Cohen, R. L.; West, K. W. J . Electrochem. Soc. 1973, 120, 502-508. Meek, R. L. J . Electrochem. Soc. 1975, 122. 1177-1185. Meek. R. L. J . Electrochem. Soc. 1975, 122, 1478-1481. de Minjer. C. H.; u.d. Boom, P. F. J. J . Electrochem. SOC. 1973, 120, 1644- 1650. Feklstein, M.; Schlesinger, M. P.; Hedgecock, N. E.; Chow, S. L. J . Electrochem. Soc. 1974, 121, 738-744. Svendsen, L. 0.; Osaka, T.; Sawai, H. J . Electrochem. Soc. 1883, 130, 2252-2255. Leavltt, J. A.; Rollins, D. K.; Fernando, 0.Anal. Chem. 1986, 58, 90-93. Eshback, H. L. Central Bureau of Nuclear Measurements, Steenweg Op Retie, 2240 Geel, Belgium. private communication. Zeblisky, R. J. U.S. Patent 3672671, 1972, and 3672938, 1972. Shirley. D. A. phvs. Rev. 6: SolM State 1972, 5 , 4709-4714. Scofield, J. H. J . Electron Spectrosc. Relet. Phenom. 1976, 8 , 129-137. Seah, M. P.; Dench, W. A. S I A , Surf. Interface Anal. 1979, 1 , 2-11. Gladney, H. M.; Dowden, B. F.; Swalen, J. D. Anal. Chem. 1969, 41. 883-888. Chu, W.; Mayer, J. W.; Nicolet, M.-A. Backscattering Spectromeby; Academic: New York, 1978; Chapter 7. Barry, B. T. K.; Thwaltes, C. J. Tln and Its Alloys and Compounds; Ellis Horwood: Chichester, U.K., 1983; p 198.
RECEIVED for review January 21,1988. Resubmitted June 10, 1988. Accepted September 7, 1988. Financial support was provided under a SUR contract, No. 44 (1984), by the IBM Corp., Endicott, NY.
