J. Phys. Chem. C 2009, 113, 17447–17454
17447
Adsorption and Thermal Reaction of Short-Chain Iodoalkanes on Ge(100) P. Y. Chuang,† W. L. Lee,† T. F. Teng,† Y. H. Lai,‡ and W. H. Hung*,† Department of Chemistry, National Taiwan Normal UniVersity, Taipei 116, Taiwan, Department of Chemistry, Tunghai UniVersity, Taichung 407, Taiwan ReceiVed: May 5, 2009; ReVised Manuscript ReceiVed: July 28, 2009
The adsorption and thermal decomposition of iodoalkanes CH3I, C2H5I, and C4H9I on Ge(100) were studied with temperature-programmed desorption (TPD) and X-ray photoelectron spectra (XPS) using synchrotron radiation. At 105 K, the iodoalkanes adsorb both molecularly and dissociatively on Ge(100); the shorterchain iodoalkane dissociates to form a surface alkyl and an I adatom to a greater extent. The chemisorbed iodoalkane gradually dissociates to form a surface alkyl and an I adatom in a temperature range 200-370 K. At 720 K, most surface CH3 desorbs directly from the surface, and other surface CH3 radicals undergo disproportionation to desorb as CH4. Surface C2H5 and C4H9 mostly undergo β-hydride elimination to desorb as C2H4 and C4H8 at ∼550 K, respectively. The temperature for C4H9 to react is slightly lower than that for C2H5 because the C4H9 chain exhibits a stronger interaction with the surface than C2H5. The I adatom can react with a H atom liberated during decomposition of a surface alkyl and subsequently desorbs as molecular HI in two temperature regimes, ∼650 and ∼720 K. Some I adatoms are removed from the surface via direct desorption in atomic form at 720 K. On annealing to 770 K, the Ge surface becomes free of I adatom but retains a deposit of residual C as adatoms. According to our data, the temperature of fabrication and operation of a Ge-based device with the alkyl monolayer is suggested to be not higher than 530 K. Introduction Germanium (Ge) is a prospective semiconductor material for high-performance integrated circuits because both electrons and holes possess great mobility. The Ge material also has a direct transition of which the energy is only slightly greater than the indirect band gap; Ge has consequently a greater absorption coefficient than silicon, making Ge desirable in many optoelectronic and photovoltaic applications. A lack of stable Ge oxides of sufficient quality is, however, a drawback in the application of Ge to electronic devices. As reported previously, to improve the electrical performance of a Ge/dielectric stack in a Ge-based device, the formation of a passivating layer before the growth of a dielectric film is essential.1,2 Much effort has thus been devoted to investigate the formation and preparation of the passivating and protective layer, as an alternative to Ge oxides, on a Ge substrate. A self-assembled monolayer (SAM) of long-chain alkyl has been considered a promising candidate for a passivating layer on a Ge surface via direct Ge-C covalent bonds as occur on a Si surface.3 It can be fabricated via a reaction of a Cl-terminated Ge surface with Grignard reagents4-6 or hydrogermylation of a H-terminated Ge surface with alkenes or alkynes.7 The resulting alkyl monolayer imparts a significant stability of the Ge surface to oxidation. Such an alkyl adlayer has been demonstrated also to provide an electrical interface of high quality on Ge without surface Fermi-level pinning.8 In addition, an alkyl SAM on Ge served as a resist for area-selective atomic-layer deposition.9 Nevertheless, the thermal reaction and stability of a surface alkyl remain critical factors in their fabrication and applications under vigorous operating conditions. For example, a polyimide * Corresponding author. Fax: +886-2-29324249. E-mail: whung@ ntnu.edu.tw. † National Taiwan Normal University. ‡ Tunghai University.
encapsulation on electronic devices typically requires curing at temperatures up to 200 °C.10 To understand the thermal reaction of an alkyl SAM on the Ge surface is thus fundamentally important for molecular electronics and is relevant to fields as diverse as material science. Buriak and Loscutoff/Bent comprehensively reviewed the chemical reaction involved in functionalization of the Ge surface,3,11 but adsorption and thermal reaction of alkyl groups on a Ge surface have been studied much less than for Si and other metal surfaces.12,13 Little study has been devoted to the subsequent thermal reactions of SAM adlayers. An improved understanding of adsorption and the mechanism of reaction of alkyl groups might provide insight into the limitations and merits of an alkyl SAM as a passivating layer on a Ge surface. In general, an iodoalkane can dissociate to form an alkyl group and an I adatom on the surface via a selective scission of the C-I bond.14,15 The reaction paths of surface alkyls are nearly unaffected by the presence or absence of coadsorbed halogen atoms.16,17 For instance, a comparison of adsorption of CH3 radical and CH3I on Cu(111) showed little effect of coadsorbed I on the reaction of CH3 groups.16 The mechanisms of reactions of iodoalkanes on a Ge surface might thus provide suitable analogues to those of an alkyl SAM. We report here an investigation of the adsorption and thermal reaction of iodoalkanes on a Ge(100) surface using temperatureprogrammed desorption (TPD) and X-ray photoelectron spectra (XPS). The length of the carbon chain of an alkyl group is expected to have a significant influence on the reaction mechanism because of the presence of β-hydrogen and the interaction between the Ge surface and the alkyl group. To examine the thermal reaction as a function of the alkyl chain, we thus undertook a comparison of experimental results for alkyls of varied chain length (i.e., RI with R ) CH3, C2H5, and C4H9).
10.1021/jp904178a CCC: $40.75 2009 American Chemical Society Published on Web 09/11/2009
17448
J. Phys. Chem. C, Vol. 113, No. 40, 2009
Chuang et al.
Experiments The experiments were conducted in an ultrahigh vacuum (UHV) chamber with a base pressure of 2 × 10-10 Torr. This system was equipped with a quadrupole mass filter (EPIC, Hiden), a low-energy electron diffraction (LEED) apparatus, and an electron-energy analyzer (HA100, VSW). The Ge(100) samples (Sb-doped n-type, 1-10 Ω · cm) for our work had a thickness 0.3 mm. The Ge sample was mounted on a Si sample of the same dimensions. A Ta strip (thickness 0.025 mm) was uniformly pressed between Ge and Si samples with Ta foils at both ends, which were in turn mounted on a copper block. The sample could be cooled to 105 K with liquid nitrogen via conduction through the copper block and heated with resistive heating of the Ta strip. The sample temperature was monitored with a K-type thermocouple spot-welded onto a thin Ta foil inserted between the Ge and Si samples. The surface was cleaned through Ar+ ion sputtering and annealed at 870 K; according to LEED, the surface then exhibited c(4 × 2) patterns.18-20 The cleanliness of the Ge surface was verified with XPS measurements. CH3I (>99%, Acros), C2H5I (>98%, Acros), and C4H9I (>98%, Acros) were purified with several freeze-pump-thaw cycles. The alkyl iodides were filled in the glass tubes covered with aluminum foils to prevent photoinduced dissociation. During dosing, the partial pressure of iodoalkanes was controlled at 1 × 10-9 Torr. The sample was placed ∼3 cm in front of the doser to minimize the contamination of the UHV system by the dosed iodoalkanes. After a desired duration, the sample was moved away from the doser, and afterward, the doser was closed. The residual iodoalkanes between the exposures were neglected because the partial pressure of iodoalkanes was very small. XPS were measured at the HSGM beamline of National Synchrotron Radiation Research Center, Taiwan; the angle of incidence of photons was 45° from the surface normal. Emitted photoelectrons were collected with an electron analyzer at an angle of 10° from the surface normal in an angle-integrated mode. Collected spectra were numerically fitted with Voigt functions after Shirley background subtraction with a third-order polynomial to each side of an XPS peak. The onset of photoemission from a Au foil attached to the sample holder served as the Fermi level, corresponding to zero binding energy. The photon energies used to collect XPS spectra were 380 eV for C 1s and 730 eV for I 3d. The iodoalkanes were subject to dissociation induced by radiation during the XPS measurement. To diminish the undesired effect of photochemical dissociation, the XPS spectrum was recorded within 3 min, and the analyzed area of the sample surface was altered after each XPS measurement. A quadrupole mass filter served for analysis of desorption products in the TPD measurement. The mass analyzer was enclosed in a differentially pumped cylinder, at the end of which is a skimmer with an entrance aperture (diameter 2.8 mm). For TPD measurement, the sample surface was placed about 2 mm before the aperture and in the line of sight of the ionizer of the mass spectrometer; TPD scans were recorded on ramping the sample at a linear rate of ∼1.5 K/s. Results and Discussion The chemical identity of a surface species on the Ge surface was characterized with XPS measurement. Figure 1 shows XPS data of C 1s and I 3d recorded for a Ge(100) surface exposed at 105 K to CH3I for varied durations. The C 1s spectra for all exposures were deconvoluted with two components at 285.0
Figure 1. (a) XPS spectra of C 1s and I 3d for a Ge(100) surface exposed to CH3I for various durations at 105 K. Dots represent data collected after background subtraction; solid lines are fitted curves, and various components are shown with dashed lines. The photon energies used to collect these spectra are 380 eV for C 1s and 730 eV for I 3d.
and 285.6 eV, corresponding to two adsorption features.21 The latter component is assigned to chemisorbed CH3I molecules, in which the C atom is bound to the electronegative I atom; the former component is attributed to surface CH3 that is bound to the Ge surface via the C atom. The surface CH3 is produced on dissociation of the CH3-I bond, CH3-I(g) f CH3(ad) + I(ad). Like the C 1s spectrum, the I 3d spectrum contains two d5/2 features at 620.1 and 621.9 eV. The latter component of I 3d5/2 is attributed to chemisorbed CH3I molecules, whereas the former component of I 3d5/2 is assigned to a surface I adatom bonded to the surface Ge atom. The relative intensities of the I 3d5/2 features at 620.1 and 621.9 eV change with the exposure as observed for the C 1s features at 285.0 and 285.6 eV. This observation is consistent with the assignments of C 1s and I 3d5/2 peaks. Accordingly, the XPS data show that CH3I can adsorb both molecularly and dissociatively on the Ge surface at 105 K. The spectra of C 1s and I 3d show little change after the exposure duration exceeds ∼70 s, indicating that the surface sites for chemisorption become saturated and the sticking coefficient for physisorption is small. To detect possible products evolved during thermal decomposition of CH3I on Ge(100), we observed several possible fragments. Figure 2 shows the composite TPD scans taken from a Ge(100) surface at 105 K exposed to CH3I for 150 s. The desorption of H2 (m/e ) 2) is below the detection limit, which is commonly observed for decomposition of alkyl-containing species on other surfaces; for instance, Si(100).22 CH3 (m/e ) 15), CH4 (m/e ) 16), HI (m/e ) 128), and CH3I (m/e ) 142) are the desorption products. The signal of CH3I is observed with maxima at 245 and 725 K. The low-temperature feature corresponds to molecular desorption from a fraction of chemisorbed CH3I. The high-temperature feature is attributed to originate from the recombination of surface CH3 and I adatom. The TPD scan of m/e ) 142 shows no presence of physisorbed CH3I at low temperature for an exposure duration up to 150 s, consistent with the XPS data. No signal for C2H6 (m/e ) 30) is observed in the decomposition of CH3I, although that signal was obtained from the coupling of two surface CH3 on other surfaces.23-25
Short-Chain Iodoalkanes on Ge(100)
J. Phys. Chem. C, Vol. 113, No. 40, 2009 17449
Figure 2. Composite TPD spectra collected from Ge(100) exposed to CH3I for 150 s at 105 K. Figure 4. XPS spectra of C 1s and I 3d for a Ge(100) surface exposed to CH3I for 150 s at 105 K and subsequently heated to indicated temperatures.
Figure 3. TPD spectra of CH3 (m/e ) 15) and HI (m/e ) 128) collected from Ge(100) exposed to CH3I at 105 K for various durations as indicated.
Figure 3 depicts TPD scans of CH3 and HI as a function of duration of exposure to CH3I. Their desorption intensities attain maxima at durations greater than 70 s because the chemisorption of CH3I becomes saturated, consistent with XPS data. Two desorption features of CH3 are observed at ∼150 and 725 K. The first broad feature exhibits a similar intensity for all exposures and likely originates from desorption from locations other than the Ge surface; for example, the sample holder. Desorption of CH3, CH4, HI, and CH3I concur in exhibiting signal maxima at ∼725 K. These desorption features are associated with surface CH3 that results from the dissociation of CH3I. The desorption of CH3 is due to direct desorption of surface CH3 on breaking its bond to the surface Ge atom. The desorption of CH3I in this temperature range is attributed to the recombination of surface CH3 and I, as described above. The desorption of CH4 and HI are proposed to be initiated with the decomposition of CH3. A fraction of the surface CH3 decomposes to liberate the H atom that subsequently reacts with a neighboring CH3 and an I adatom to yield CH4 and HI, respectively, but no desorption of H2 is observed
in the decomposition of CH3I. Overall, the surface CH3 can participate in several reaction paths at 725 K, which results in various desorption products. Among the possible reaction paths, the direct desorption of surface CH3 is the major reaction on Ge(100). Recorded in our TPD measurements, Figure 2 shows a signal at m/e ) 127 that is attributed to the I+ ion resulting from an electron-induced fragmentation of HI and CH3I in the mass spectrometer. The ratio of signals at m/e ) 127 and 128 obtained from the decomposition of CH3I is ∼0.7, but the ratio of signals at m/e ) 127 and 128 for the HI molecule is about 0.5 when the HI molecule is measured in our mass spectrometer. In contrast, the amount of desorbed CH3I is so small that the fragment of its I+ ion is an insignificant factor for the intensity at m/e ) 127. Hence the signal at m/e ) 127 is partially contributed from the I atom desorbing from the surface. A fraction of surface I adatoms thus undergo direct desorption, in addition to the reactions with CH3 and H to desorb CH3I and HI. We applied the thermal evolution of XPS spectra to characterize the variation of surface composition during thermal decomposition of CH3I and to correlate with TPD results to elucidate the reaction intermediates. Figure 4 shows the C 1s and I 3d5/2 spectra for the Ge(100) surface exposed to CH3I at 105 K for 150 s and subsequently annealed to various temperatures. All XPS spectra were recorded for samples at 105 K after being heated to a desired temperature at a rate less than 1 K/s and cooled immediately on terminating the heating abruptly. The intensities of C 1s at 285.6 eV and I 3d5/2 at 621.9 eV due to chemisorbed CH3I began to attenuate with a similar rate on annealing the sample to 200 K and completely disappeared at 320 K. In contrast, the intensities of features of C 1s at 285.0 eV and I 3d5/2 at 620.1 eV increase by ∼15%. These spectral changes reveal that chemisorbed CH3I partially desorbs intact from the surface, corresponding to the desorption maximum of CH3I at 245 K. The remaining CH3I dissociates to form additional surface CH3 and I. The XPS data show only CH3 and I to be present on the surface at temperatures above 320 K. Upon further annealing to 700 K, the intensities of C 1s and I 3d gradually decrease.
17450
J. Phys. Chem. C, Vol. 113, No. 40, 2009
Chuang et al.
The decrease in C 1s intensity corresponds to the desorption of carbon-containing species; that is, CH3, CH4, and CH3I as shown in TPD data. The decrease in I 3d5/2 intensity is due to the desorption of iodine-containing species; that is, I, HI, and CH3I. The binding energy of C 1s shifts slightly toward the small energy, indicating that the CH3 dissociates to liberate H and subsequently forms CH2 or CH, as described above. At 770 K, the intensity of C 1s is significantly decreased, whereas the intensity of I 3d5/2 disappears. The eventual intensity of C 1s is due to the residual C that remains on the surface after the complete decomposition of surface CH3. The decomposition of surface CH3 liberates H atoms that can react with other surface CH3 or I and so lead to desorption of CH4 or HI, as described above. The disappearance of the I 3d5/2 intensity indicates all I adatoms are removed via desorption of I, HI, and CH3I. On the basis of the TPD and XPS data, we summarize the adsorption and decomposition of CH3I according to the following reactions.
CH3I(g) f CH3I(ad)
or
CH3(ad) + I(ad)
105 K
CH3I(ad) f CH3I(g)
or
CH3(ad) + I(ad)
200-320 K
(1)
(2) CH3(ad) f CH3(g) 670-770 K CH3(ad) + CHx(ad) f CH4(g) + CHx-1(ad) (x ) 1, 2, or 3) I(ad) + CHx(ad) f HI(g) + CHx-1(ad) I(ad) f I(g) (3) Although Ge(100) and Si(100) exhibit similarities of surface structure and reconstruction,26,27 the thermal reaction of CH3I molecules on these surfaces shows a subtle distinction. Electronenergy-loss spectra showed that all CH3I molecules adsorb dissociatively on Si(100).23 The resulting CH3 group and I adatom were proposed to bond to separate Si atoms of the Si dimer on Si(100), respectively.22 Our XPS data show that CH3I molecules dissociate mainly to form surface CH3 and I on a Ge(100) surface with the presence of molecular adsorption at 105 K. For dissociative adsorption of CH3I, the Ge(100) surface is thus less reactive than Si(100). Chemisorbed CH3I molecules can partially dissociate to form additional surface CH3 and I at 320 K. At 720 K, most CH3 desorbs directly from the surface, and a fraction of CH3 (
