Mn and As doping of β-FeSi₂ via a chemical method

Abstract

β-FeSi₂ was doped with Mn and As via the chemical reduction of the glassy phase of [{Fe₂O₃·4SiO₂} + Mn₂O₃ + Mn₃O₄] and [{Fe₂O₃·4SiO₂} + As₂O₃] respectively, using Mg-metal at 800 °C. Iron silicide doping was investigated using high resolution transmission electron microscopy (HRTEM), X-ray diffraction (XRD), Raman spectroscopy, X-ray photoelectron spectroscopy (XPS), energy dispersive X-ray analysis (EDX) and Hall measurements. The increment in the inter-planar distance, as verified by the SAED image and XRD peak shift, indicates both As and Mn doping. The Raman defect (D) peak at 300 cm⁻¹ confirms As and Mn doping. The red shift in the Raman peaks at 190 cm⁻¹, 244 cm⁻¹ and 300 cm⁻¹ due to increases in the bond lengths indicates significant doping. In addition, the upshift in the binding energy of the Si 2p XPS peak and the downshift in the binding energy of the Fe 2p³ peak indicate As and Mn doping in β-FeSi₂. Also, the low intensity As 3d and Mn 2p³ XPS peaks considerably confirm doping, which is in close agreement with the EDX result. The Hall measurements show that the As doped sample is n-type, while the Mn doped sample is p-type. The carrier concentration in the Mn doped sample is higher than that of the As doped sample, which can be justified by the higher activation energy and using two-band model theory.

---

1. Introduction

The β-phase of FeSi₂ has been attracting attention over many years due to its significant semiconducting properties. It has a direct band gap of 0.85–0.87 eV, which can absorb a wide spectral range, and it also has a high optical absorption coefficient (>10⁵ cm⁻¹ at 1 eV). Thus, it may be a promising candidate for renewable energy and photovoltaic applications.^1–8

Doped β-iron silicide can be useful for developing various semiconductor devices, particularly p–n homojunction solar cells.^2,4,7,10 In this report we are dealing with the As and Mn doping of β-FeSi₂. As and Mn doping results in n-type and p-type β-FeSi₂ respectively, as reported in the literature.^9,10 β-FeSi₂ has an orthorhombic lattice structure with a unit cell that contains 16 formula units distributed over two crystallographically inequivalent sites: 8 Fe_I, 8 Fe_II and 16 Si_I, 16 Si_II.¹¹ Comparing the computed total energies, Tani et al. reported that the preferential site for the Mn dopant is Fe_I, which is in good agreement with that reported by Kondo et al.^12–14 On the other hand, according to Tan et al., As atoms when substituting Si sites possess lower total energies than when substituting Fe sites which are energetically more favourable.¹⁵ Therefore, by comparing the total energies for both of the Si substitutional sites (Si_I and Si_II), they have concluded that As prefers the Si_I site over the Si_II site.¹⁵ According to Tani et al., the solid solution of Fe_1−xMn_xSi₂ exists within the composition range of 0% ≤ x ≤ 10% which is in good agreement with Kojima et al. also.^9,16

There have been several reports of doping iron silicide with transition metals like Co, Ni, Mn, Cr and metalloids like B and As over the years.^9,17–21 However, most of the doping techniques involved are based on physical methods and often require high-end equipment. For example, Terai et al. used an ion-beam synthesis technique to dope Mn in β-FeSi₂, Arushanov et al. have grown Mn doped β-FeSi₂ single crystals by chemical vapour transport, Liu et al. have reported As and B doping in β-FeSi₂ thin films by co-sputtering of heavily As doped Si chips and elemental B chips respectively with a Si target, and Tan et al. and Lu et al. used an ion implantation technique for the purpose of B and As doping in β-FeSi₂.^10,21–24 Therefore, there are hardly any reports on the use of chemical methods to dope iron silicide in the literature. Using a chemical method of doping has the advantage of uniformity, simplicity and it usually does not require any high-end equipment, and is therefore cost effective.

Herein, we propose a novel chemical method to dope iron silicide with Mn and As to make it p-type and n-type, respectively. This is done by using manganese acetate and arsenic tri-oxide precursors of different wt% for the Mn and As doping respectively, while synthesizing β-FeSi₂ chemically. To the best of the authors' knowledge, this is the first approach to chemically dope β-FeSi₂. The synthesis and characterization of the doped β-iron silicide will be discussed in detail in the following sections of this article.

2. Experimental section

2.1. Chemical doping procedure

2.1.1. Chemicals. Analytical grade nona-hydrate ferric nitrate [Fe(NO₃)₃·9H₂O] (≥99.99%), tetraethyl orthosilicate [Si(OC₂H₅)₄] (≥99%), ammonium acetate [CH₃COONH₄] (≥99.99%), tetra-hydrate manganese acetate (Mn(CH₃COO)₂·4H₂O) (≥99%), arsenic tri-oxide (As₂O₃) (≥99.95%), magnesium (Mg) (≥98%), glacial acetic acid (CH₃COOH) (≥99.8%) and ammonia (NH₃) (25%) were used as reagents in this synthetic procedure. All the chemicals aside from acetic acid and ammonia were purchased from Sigma-Aldrich. All the reagents are of analytical grade and were used as received from the manufacturer without further purification.

2.1.2. β-FeSi₂ synthetic procedure. The synthesis of β-FeSi₂ consisted of a chemical reaction between the nona-hydrate ferric nitrate [Fe(NO₃)₃·9H₂O] and tetraethyl orthosilicate [Si(OC₂H₅)₄] precursors. Their molar ratio was maintained at 1 : 2 for the formation of β-FeSi₂. The method involved the sol–gel polymerization of a mixed solution of tetraethyl orthosilicate and ferric nitrate in a controlled pH environment of 5–7, followed by calcinations at 600 °C for 6 h. Thereafter, the calcinated mixed oxide powder was chemically reduced at 800 °C for 6 h by Mg under an Ar atmosphere which was further followed by filtering using centrifugation at 8000 rpm. The details of the synthetic procedure of β-iron silicide is given in our previously reported work.²⁵

2.1.3. Mn doping in β-FeSi₂. Here the doping with the Mn precursor was done during the sol–gel polymerization of tetraethyl ortho-silicate while synthesizing β-FeSi₂. The precursor used was tetra-hydrate manganese acetate (Mn(CH₃COO)₂·4H₂O). It has been reported earlier that Mn atoms substitute Fe_I sites, therefore x has been considered as the atomic wt% of the Mn dopant with respect to Fe atoms as given in Table 1.^12–14 Firstly, (2 − 0.02x)/3 M of Fe(NO₃)₃·9H₂O solution was prepared with 30 ml of DI water, 0.02x M of Mn(CH₃COO)₂·4H₂O solution was prepared with 10 ml of DI water and 3.33 M ammonium acetate solution was prepared with 30 ml of DI water. Thereafter, an adequate volume of these solutions were mixed together with 4.48 M of 8.89 ml tetraethyl orthosilicate to get the desired molar ratio of Fe³⁺ : Si(OC₂H₅)₄ : Mn²⁺ : ammonium acetate = (1 − x) : 2 : x : 5 in the presence of excess ammonium acetate. Ammonium acetate was used to maintain the pH of the final solution for the homogeneous precipitation of Mn²⁺–silicate gel. After the precipitation, the precipitate was dried, powdered and calcined at 600 °C for 6 h. During the calcination, the dehydration of manganese acetate occurred within the range of 100–190 °C, followed by complete decomposition between 230 °C and 480 °C. This was further followed by the full decomposition of organics at 600 °C. Therefore, calcination resulted in the tetragonal phase of Mn₃O₄ as reported by Sahoo et al., along with the generation of Mn₂O₃ under an atmosphere of abundant oxygen.²⁶ Therefore, the product was a mixed oxide of [{Fe₂O₃·4SiO₂} + Mn₂O₃ + Mn₃O₄].

Mn(CH₃COO)₂·4H₂O + O₂ → Mn₂O₃ + Mn₃O₄ + CO₂ + H₂O

Table 1 Atomic wt% of the precursors and dopants for several doping percentages of Mn and As in doped β-FeSi₂

	Mn(CH3COO)2·4H2O	Fe(NO3)3·9H2O	Si(OC2H5)4
Molar ratio	1	1	2
Molar mass (mmol)	20 mmol = 4.90 g	20 mmol = 8.08 g	40 mmol = 8.93 ml
Mn atomic% (x)	1% = 0.05 g	99% = 7.99 g	8.93 ml
	3% = 0.15 g	97% = 7.84 g	8.93 ml
	5% = 0.25 g	95% = 7.6 g	8.93 ml
	As2O3	Si(OC2H5)4	Fe(NO3)3·9H2O
Molar ratio	1	2	1
Molar mass (mmol)	20 mmol = 3.96 g	40 mmol = 8.93 ml	20 mmol = 8.08 g
As atomic% (y)	0.025% = 0.99 mg	99.975% = 8.928 ml	8.08 g
	0.5% = 0.02 g	99.5% = 8.89 ml	8.08 g
	1% = 0.04 g	99% = 8.84 ml	8.08 g
	2% = 0.08 g	98% = 8.75 ml	8.08 g

---

During reduction by 200 mmol Mg at 800 °C under an Ar atmosphere, both Mn₂O₃ and Mn₃O₄ were reduced to Mn, whereas the {Fe₂O₃·4SiO₂} mixed oxide was reduced to β-FeSi₂.

Therefore, the reduced Mn atoms, owing to high kinetic energy at such a high reduction temperature of 800 °C, substituted the Fe_I atomic sites in β-FeSi₂. Here the filtering was done by the centrifugation of 200 ml 50% diluted acetic acid solution of the mixed powder sample [Mn doped β-FeSi₂ + MgO] at 8000 rpm. The diluted CH₃COOH was neutralized to some extent by adding 50 ml ammonia to bring its pH up to 5 to prevent reaction with β-FeSi₂ particles. Thereafter, repeated washing was performed with acetone and DI water.²⁵

2.1.4. As doping in β-FeSi₂. Similar to the Mn doping procedure, As doping was performed during the formation of a sol–gel polymer in the synthesis of β-FeSi₂. The precursor used for As doping was arsenic tri-oxide (As₂O₃). As it has been reported previously by Tan et al. that arsenic atom substitutes Si_I site, y was considered as the atomic wt% of As with respect to Si atoms, as given in Table 1.¹⁵ At first, both 0.667 M of ferric nitrate solution and 3.334 M of ammonium acetate solution were prepared with 30 ml of water. Then 30 ml of ammonium acetate solution was added to the ferric nitrate solution to make the solution buffered. As arsenic tri-oxide is not soluble in water and is only soluble in weak acid/base, a 0.02y M As³⁺ (in the form of As₂O₃) solution was prepared with 10 ml of 53.44 M ammonia by heating at 60 °C and stirring at 500 rpm for 10 min. The As³⁺ solution was further neutralized using 1–1.5 ml acetic acid to reach a pH of 4–5, and to match that of the buffered iron nitrate solution to prevent As₂O₃ precipitation during gel formation. Thereafter, 4.48 M of (893 − 8.93y)/100 ml tetraethyl orthosilicate was added to the aforementioned volume of buffered iron nitrate and arsenic tri-oxide solutions to obtain the desired molar ratio of Fe³⁺ : Si(OC₂H₅)₄ : As³⁺ : ammonium acetate = 1 : (2 − 2y) : 2y : 5 for arsenic doping. Here, ammonium acetate was also used for the homogeneous precipitation of As³⁺–silicate gel. After precipitation, the precipitate was calcined and powdered similarly at 600 °C for 6 h. Thus, the calcination resulted in a mixed oxide of [{Fe₂O₃·4SiO₂} + As₂O₃]. Thereafter, As₂O₃ was reduced by Mg at 800 °C for 6 h under an Ar environment to obtain As. Also similar to the iron silicide synthesis, the {Fe₂O₃·4SiO₂} mixed oxide was reduced to β-FeSi₂ as reported earlier.²⁵

Thus, the reduced As atoms, due to their high kinetic energy at 800 °C, substituted the Si_I sites in β-FeSi₂. Again, the filtering was performed by the centrifugation of a neutralized and diluted acetic acid solution of the mixed powder of the As doped β-FeSi₂ + MgO at 8000 rpm. Centrifugation was followed by repeated washing with acetone and DI water.²⁵ Altogether, eight samples were prepared: one undoped, four As doped (0.025%, 0.5%, 1% and 2%) and three Mn doped (1%, 3%, 5%) iron silicide samples.

2.2. Doping characterization

The samples were extensively characterized using High resolution TEM, X-ray diffraction (XRD), Raman spectroscopy, X-ray photoelectron spectroscopy (XPS), energy dispersive X-ray (EDX) and Hall measurements to confirm the doping. HRTEM and XRD were performed using a JEM-2100 and an XPERT PANalytical X-ray diffractometer (with Cu-Kα target) respectively, to investigate the changes in the lattice inter-planar distances. Raman spectroscopy was done using a Horiba scientific T64000, with an Ar–Kr mixed ion gas laser to check the lattice defect line D, due to incorporation of Mn and As as dopants. XPS and EDX (by FESEM ZEISS SUPRA-40) were also carried out to analyze the dopants. Finally, Hall measurements were carried out with the help of an ECOPIA Hall instrument to obtain the conduction type, carrier concentration, and carrier mobility.

3. Results and discussion

3.1. Doping analysis using HRTEM

As several researchers so far have studied doping using HRTEM and electron diffraction, here we have investigated doping from the SAED profile of the Mn and As doped samples.^27–30 To the best of our knowledge, the study of doped β-FeSi₂ using selected area electron diffraction (SAED) has been hardly reported in the literature. Therefore, here we have cited some of the reports where authors have investigated the doping of Si, SnO₂ and ZnO using the electron diffraction profile, even if for lightly Li⁺ doped Si nano-wires.^27–30 As can be seen from the SAED image in Fig. 1(a), there is almost no change in the inter-planar distance (d) in the undoped β-FeSi₂ as verified from the JCPDS database. On the contrary, for the As and Mn doped cases, the inter-planar distances (d) have increased within the range of 0.02–0.06 Å as shown in Fig. 1(b)–(d). As the radius of the arsenic atom is 0.03 Å larger than that of the Si atom (R_Si = 1.11 Å and R_As = 1.14 Å) the maximum increment in the inter-planar distance is up to 0.06 Å, where As atoms are incorporated in two consecutive atomic planes. As can be seen from Fig. 1(b), the minimum increment in d is approximately 0.03 Å, where an As atom is incorporated in a single atomic plane. On the other hand, the Mn atomic radius is 0.06 Å larger than that of the Fe atom (R_Mn = 1.61 Å and R_Fe = 1.56 Å). Therefore, as a Mn atom replaces a Fe_I site, the inter-planar distance obtained for the Mn doped β-FeSi₂ is 0.05 Å–0.06 Å due to the incorporation of a Mn atom in a single atomic plane as shown in Fig. 1(c) and (d).

Fig. 1 (a) SAED image of undoped β-FeSi₂ for the (711) and (337) planes, (b) SAED image of 1% As doped β-FeSi₂ for the (023) and (711) planes, (c) SAED image of 1% Mn doped β-FeSi₂ for the (023) and (711) planes, and (d) SAED image of 3% Mn doped β-FeSi₂ for the (222) and (711) planes (β-FeSi₂ inter-planar distance verified from JCPDS data card 04-007-1080 and 01-079-5663).

3.2. Doping analysis by X-ray diffraction

Fig. 2(a) and (b) show that the XRD peaks of doped iron silicide shift towards a lower diffraction angle (2θ) with respect to undoped β-FeSi₂, due to an increase in the inter-planar distance. As both As and Mn dopants have larger atomic radii compared to Si and Fe atoms respectively, the inter-planar distance in doped β-FeSi₂ increases for both As and Mn dopants. This has been shown in Table 1 for the (422) and (515) planes. The increase in the inter-planar distance also depends on the orientation of planes, as well as on the doping percentage. Thus, the peak shifts more towards lower 2θ values, indicating an increase in the inter-planar distance (d) with an increase in the doping percentage. The increase in the inter-planar distance is almost 0.001 Å on average for the (422) plane, whereas Δd is almost 0.0006 Å on average for the (515) plane. The increases in the inter-planar distance with respect to the dopant atomic percentages and planes are summarized in Table 2.

Fig. 2 XRD profile of (a) the (422) plane and (b) the (515) plane for undoped, As (0.025%, 0.5% and 1%) and Mn (1%, 3% and 5%) doped β-FeSi₂ (verified from JCPDS data card 04-007-1080 and 01-079-5663).

Table 2 Increase in the inter-planar spacing of β-FeSi₂ with an increase in dopant atomic%

	Undoped β-FeSi2	As doped β-FeSi2			Mn doped β-FeSi2
		0.025%	0.5%	1%	1%	3%	5%
d422	1.8329 Å	1.8337 Å	1.8339 Å	1.8343 Å	1.8340 Å	1.8333 Å	1.8340 Å
Δd422	—	0.0008 Å	0.001 Å	0.0014 Å	0.0011 Å	0.0004 Å	0.0011 Å
d515	1.1994 Å	1.1998 Å	1.1999 Å	1.2002 Å	1.2001 Å	1.1999 Å	1.2002 Å
Δd515	—	0.0004 Å	0.0005 Å	0.0008 Å	0.0007 Å	0.0005 Å	0.0008 Å

---

Moreover, the crystallite sizes were calculated from the diffraction angle (2θ) and FWHM (β) of the XRD peaks at 49.7° and 79.9°, as shown in Fig. 2(a) and (b) respectively with the help of Debye–Scherrer formula (1),

L = Kλ/β cos θ (1)

where K = Scherrer constant = 0.94, λ = 1.5406 Å is the Cu target Kα₁ X-ray wavelength, β is the FWHM of the XRD peak at the 2θ diffraction angle and L is the crystallite size. The XRD peaks shown in Fig. 2 are quite convoluted (complex) in nature. Therefore, at first these complex peaks were deconvoluted/fitted with the help of Origin software. Thereafter, the crystallite sizes at diffraction angles (2θ) of 49.7° and 79.9° have been determined by Scherrer's formulae as summarized in Table 3. The trend shows that as the doping percentage increases, the crystallite size decreases quite significantly for both dopants. This can be justified by the increase in local strain in the crystallite planes (422) and (515), due to the incorporation of more and more dopant atoms within the Fe and Si lattice sites. Aside from this, the average crystallite size for the As doped sample is around 45–50 nm and 55–60 nm for the (422) and (515) planes respectively. Likewise, the average crystallite size for the Mn doped samples is 53–55 nm and 55–60 nm for the (422) and (515) planes respectively. On the other hand, the average crystallite size of the undoped samples is 42.55 nm and 54.4 nm for the (422) and (515) planes respectively. Therefore, it can be observed that the crystallite size is larger for the As and Mn doped samples compared to the undoped samples for both the lattice planes. This may be due to lattice strains because of the incorporation of As and Mn dopants whose atomic sizes (R_As = 1.14 Å and R_Mn = 1.61 Å) are larger than the Si and Fe atomic sizes (R_Si = 1.11 Å and R_Fe = 1.56 Å) respectively, in β-FeSi₂.

Table 3 Summarized crystallite size and lattice strain of undoped and Mn and As doped β-FeSi₂ for both the (422) and (511) planes

		Undoped β-FeSi2	As doped β-FeSi2			Mn doped β-FeSi2
			0.025%	0.5%	1%	1%	3%	5%
Crystallite size (L) (nm)	L422	42.55	56.81	52.87	42.15	59.78	59.0	42.54
	L515	54.4	65.21	51.79	56.07	66.4	66.0	46.24
Lattice Strain (ε) (%)	ε422	0.2	0.15	0.16	0.2	0.14	0.15	0.2
	ε515	0.1	0.09	0.11	0.1	0.08	0.09	0.12

---

The strain has also been calculated for both peak positions for both the dopants using the following relation between peak broadening and strain:

β = 4ε tan θ (2)

where ε is the doped particle micro-strain. Therefore, the resolved crystallite strain of the As and Mn doped sample from eqn (2) is summarized in Table 3. The trending features of the strain show that the lattice strain increases as the doping percentage increases. This indicates that as more and more dopant substitutes the Fe/Si atomic lattice sites, the lattice strain continues to increase. The trend is identical for both the (422) and (515) planes for the cases of both As and Mn dopants. However, the average strain for both the Mn and As dopants and also the undoped sample for both planes are almost trivial, at around 0.1–0.2%.

3.3. Doping characterization using Raman spectroscopy

Fig. 3 shows the comparative Raman spectrum of undoped and doped β-FeSi₂. The characteristic Raman lines (β) for the undoped sample at 244 cm⁻¹ and 190 cm⁻¹ are contributed by the Raman active (A_g) mode due to first order Raman scattering by the orthorombhic β-FeSi₂ lattice.³¹ On the other hand, for doped β-FeSi₂, the Raman peak is red shifted towards lower frequencies and also FWHM becomes broader with increasing dopant atomic percentage. This red shift is due to the increase in the bond length as the bigger As and Mn dopant atoms introduce defect density into the β-FeSi₂ lattice, further weakening the force constant due to the longer bond lengths.¹² On the contrary, the additional D peak at 300 cm⁻¹ for the doped sample is caused by defect-induced Raman scattering due to lattice imperfection and stress introduced by dopants. Fig. 3 reveals that the red shifted β peak at 244 cm⁻¹ is merged together with the defect peak D at 300 cm⁻¹ for the doped sample, resulting in a broad shoulder at around 230 cm⁻¹. As the dopant (As and Mn) concentration increases, the Raman line at 300 cm⁻¹ is intensified and shifts towards red wavelengths due to an increase in the bond length introduced by the enhanced lattice defects.¹² This is because, as the bigger As atom (R = 205 Å) substitutes the smaller Si_I (R = 117.6 Å) atom, the Si_I–As (2.53 Å) and Si_II–As (2.51 Å) bond lengths and also the adjacent Si_I–Si_I and Si_I–Si_II bonds lengths become larger than the relaxed Si_I–Si_I and Si_I–Si_II bonds lengths, due to tensile strain. Similarly, the As–Fe_I (2.34 Å) and As–Fe_II (2.33 Å) bonds lengths become longer than the normal Si_I–Fe_I and Si_I–Fe_II bonds lengths.¹⁵ Thus an increase in the Si–Si and Si–Fe bonds lengths will result in the red shift of the Raman peaks. In addition, an exceptional fact is that as the As dopant increases from 1% to 2%, the Raman line becomes slightly blue shifted towards higher frequencies with respect to the 1% curve. This may be due to fewer defects in the 2% doped sample compared to the 1% As doped sample, because as the doping percentage increases towards the solubility limit, segregation of the As dopant out of the iron silicide is very probable, which further results in slightly relaxed β-FeSi₂. Thus, the Raman peak at 300 cm⁻¹ which is absent in undoped β-FeSi₂ indicates As doping in FeSi₂.

Fig. 3 Raman spectroscopy of the undoped, As (0.5%, 1% and 2%) and Mn (1% and 3%) doped samples.

Similarly, as the size of the Mn atom (r_a = 1.61 Å) is slightly larger than that of the Fe atom (r_a = 1.56 Å), the Fe_I–Si (2.36 Å) and Fe_II–Si(2.38 Å) bond lengths in the Mn doped sample become larger than the normal Fe_I–Si and Fe_II–Si bond lengths in relaxed β-FeSi₂ due to tensile strain.¹² Therefore, the red shifting of the β and D Raman lines towards lower frequencies proves that there is significant Mn doping in our synthesized samples.

3.4. Doping analysis using X-ray photoelectron spectroscopy

Fig. 4 shows the XPS spectra for undoped and doped β-FeSi₂. For undoped β-FeSi₂, the binding energy of the Si 2p peak and Fe 2p³ peak are observed at 100.14 eV and 707.7 eV respectively. A trivial upshift in the Si 2p binding energy has been observed in Fig. 4(a) for both the As and Mn doped samples with respect to the undoped sample. This upshift of the binding energy for the As doped samples can be explained by the rising of the Fermi energy level towards the conduction band due to the formation of a covalent bond between Si and As. In addition, for the As doped sample, as doping increases from 0.025% to 1% the Si 2p peak becomes broader due to an increase in structural disorder, because the Si–As bond length (almost 2.51 Å) is longer than the Si–Si bond length.¹⁵ From the XPS spectrum (Fig. 4(b)) it can be observed that there is also a significant bottom shift in the Fe 2p³ peak for both the As and Mn doped samples compared to that of the undoped sample. Thus the shift in the Si 2p and Fe 2p³ peaks in the case of doped sample indicates the effective chemical doping of β-FeSi₂. Fig. 4(c) shows a low intensity peak of substitutional As 3d at 49.77 eV which confirms As doping. Fig. 4(d) exhibits a weak peak of Mn 2p³ at 636.77 eV for the Mn doped sample, which indicates the substitutional Mn dopant. However, the Si : Fe atomic ratio has been determined to be around 2 for both the doped and undoped samples from the Si 2p and Fe 2p³ peak intensity as given in Table 4, which confirms the formation of β-FeSi₂.³²

Fig. 4 XPS peak of (a) Si 2p for undoped, As (0.025% and 1%) and 1% Mn doped β-FeSi₂, (b) Fe 2p³ for undoped, As and Mn doped β-FeSi₂, (c) As 3d for 0.025% and 1% As doped β-FeSi₂ and (d) Mn 2p³ for 1% Mn doped β-FeSi₂.

Table 4 Si 2p, Fe 2p³ XPS peak intensities and corresponding Si : Fe atomic ratios for both doped and undoped β-FeSi₂

β-FeSi2	Si 2p peak intensity	Fe 2p^3 peak intensity	Si : Fe atomic ratio
Undoped	1517.88	862.12	1.7
0.025% As doped	9971.61	5424.18	1.8
1% As doped	6682.87	3403.49	1.9
1% Mn doped	8011.01	3412.57	2.3

---

3.5. Doping characterization using EDX

EDX results for various percentages of As and Mn doped β-FeSi₂ and the corresponding elemental wt% analyses are shown in Fig. 5. These experimental results of the atomic wt% of dopants matches quite well with the calculated values as shown in Fig. 5 and Table 5. A measured sample EDX profile for the 0.5% As doped β-FeSi₂ is given in Fig. 6, as it is not possible to include the EDX profiles of all Mn and As doped samples. Also, Table 5 shows that the Si : Fe atomic ratio for all the As and Mn doped samples ranges between 1.6 and 2.0 as determined from Fig. 5. Thus, the Si : Fe atomic ratio ranging from 1.6–2.0 indicates the formation of β-FeSi₂ as reported by Takakura et al.³²

Fig. 5 EDX analysis of Si, Fe and As and Mn dopants for the doped samples.

Table 5 Atomic wt% for As and Mn dopant in doped β-FeSi₂ and Si : Fe atomic percentage ratio measured by EDX

	Undoped β-FeSi2	As doped β-FeSi2				Mn doped β-FeSi2
Cal. atomic wt%	—	0.025	0.5	1	2	1
EDX atomic wt%	—	0.015	0.34	0.54	0.77	0.64
Si : Fe	1.9	1.6	2.0	2.1	1.8	1.9

---

Fig. 6 Sample EDX profile of 0.5% As doped β-FeSi₂.

3.6. Doping analysis using Hall measurements

Hall measurements were carried out on the Mn and As doped β-iron silicide samples. The acceptor and donor atom concentration for the Mn and As dopants were theoretically calculated from the dopant atomic wt%. The Hall mobility was detected for both the Mn and As dopants. Apart from this, the carrier mobility calculations were done using the Caughey–Thomas mobility model as follows.³³where μ_n,p is the majority carrier free electron/hole mobility in n/p-type β-FeSi₂ respectively, μ_nmin,pmin is the minimum electron/hole mobility at higher carrier concentration, μ_nmax,pmax is the maximum electron/hole mobility at lower carrier concentration, N_D,A is the donor/acceptor concentration, N_refn,refp is the reference concentration for donor/acceptor and α/β is the fitting parameter for free electrons/holes respectively. The mobility model parameter values were taken from the reported work of Yuan et al., as shown in Table 6.³⁴ The activation energy at room temperature was determined theoretically from the model eqn (5) as reported by Tani et al. and Arushanov et al., respectively.^9,35where N_A,D is the acceptor/donor density, E_a is the activation energy, N_v,c is the valence/conduction band density and is the effective mass of holes/electrons. Considering = = m₀ as reported by Udonu et al., the valence band and conduction band density of states are calculated to be N_v = N_c = 2.508 × 10¹⁹ cm⁻³.³⁶ Finally, using eqn (5) we have determined the activation energy E_a at room temperature for different acceptor/donor concentrations N_A/D and corresponding Hall carrier concentrations. The Hall measurement results, along with the calculated theoretical values, are summarized in Table 7.

Table 6 Caughey–Thomas mobility model parameters value for β-FeSi₂ [ref. 34]

Materials	Carriers	μmax (cm^2 V^−1 s^−1)	μmin (cm^2 V^−1 s^−1)	Nref (×10^17 cm^−3)	α/β
β-FeSi2	Electron	515	2	0.8	0.93
	Hole	256	0.5	1.2	0.82

---

Table 7 Measured Hall parameters and calculated activation energy and carrier mobility for Mn and As doped β-FeSi₂

Atomic wt%	Hall co-efficient (RH​) (m^3 C^−1)	Acceptor conc. (NA​) (cm^−3)	Hall conc. (p) (cm^−3)	Activation energy (Ea​) (meV)	Cal. Mob. (cm^2 V^−1 s^−1)	Hall Mob. (cm^2 V^−1 s^−1)	Resistivity (ρ) (Ω cm)
Mn					Hole (μpcal​)	Hole (μp)
1%	9.35 × 10^−1	2.55 × 10^20	9.65 × 10^18	91.4	7.31	8.69	1.07 × 10^−1
3%	7.31 × 10^−3	7.65 × 10^20	8.54 × 10^20	112	0.68	0.6	5.93 × 10^−2
5%	2.1 × 10^−2	1.28 × 10^21	2.98 × 10^21	44.2	0.92	0.23	8.97 × 10^−2

---

Atomic wt%	Hall co-efficient (RH​) (m^3 C^−1)	Donor conc. (ND​) (cm^−3)	Hall conc. (n) (cm^−3)	Activation energy (Ea​) (meV)	Cal. Mob. (cm^2 V^−1 s^−1)	Hall Mob. (cm^2 V^−1 s^−1)	Resistivity (ρ)(Ω cm)
As					Electron (μncal​)	Electron (μp)
0.025%	−6.51 × 10^−1	2.55 × 10^20	1.59 × 10^19	11.8	5.06	13.2	4.95 × 10^−2
0.5%	−4.01 × 10^−1	2.55 × 10^20	1.9 × 10^18	175	24.54	23.1	1.74
1%	−3.16 × 10^1	5.1 × 10^20	1.98 × 10^17	310.4	102.51	0.235	1.34 × 10^2
2%	−6.99	1.02 × 10^21	8.92 × 10^17	250	41.84	0.66	1.07 × 10^1

---

The average Hall co-efficient R_H of the Mn doped sample is positive, indicating the sample to be p-type and that of the As doped sample is negative, indicating it to be n-type, both of which agrees quite well with the literature.^9,21 The trend in activation energy shows that as the Mn wt% increases from 1% to 5%, E_a decreases from 91.4 eV to 44.2 eV, which is quite obvious. Therefore, the hole concentration becomes higher from 9.65 × 10¹⁸ cm⁻³ for 1% doping to 2.98 × 10²⁰ cm⁻³ for 5% doping. As a result, the measured hole mobility gradually decreases from 8.69 cm² V⁻¹ s⁻¹ to 0.23 cm² V⁻¹ s⁻¹ with an increase in Mn atomic wt%.

On the other hand, Table 7 shows that the activation energy increases from 11.8 meV to 310.4 meV with an increase in the As dopant concentration from 0.025% to 1%.³¹ Consequently, this decreases the electron concentration from 1.59 × 10¹⁹ cm⁻³ to 1.98 × 10¹⁷ cm⁻³ and increases the corresponding Hall mobility. Moreover, E_a decreases quite significantly from 310.4 meV to 250 meV with an increase in As doping up to 2%, resulting in a Hall concentration increase of up to 8.92 × 10¹⁷ cm⁻³.

The Hall carrier concentration in the As doped sample is significantly less than that of the Mn doped sample, as shown in Table 7. As a consequence, the Hall mobility in the As doped sample is at least 10–20 times higher than that of Mn doped sample. This may be due to larger lattice strain and lower lattice defects in the As doped sample. This can be justified using the two band model with the existence of a deep acceptor/donor level and an additional defect/unknown impurity level as suggested by Arushanov et al. as well as Tani et al.^9,35 They both have explained the temperature dependence of R_H for Al and Mn doped β-FeSi₂ with this two band model.^9,35 Therefore, the As doped β-Fe(Si_1−xAs_x)₂ sample can also be considered to have two donor levels: a deep donor level of the As dopant and another shallow donor level of structural defects. Above all, the lower carrier concentration in the As doped sample can be well explained by (i) the higher activation energy of the deep donor level of the As doped sample compared to that of the deep acceptor level of the Mn doped sample as shown in Table 7 and (ii) the lower defect densities in the shallow defect level of the As doped sample compared to those of Mn doped sample.

4. Conclusions

In conclusion, β-FeSi₂ was doped with Mn and As dopants through a simple, cost effective chemical reduction technique. This was done by using manganese acetate and arsenic-tri-oxide as precursors for Mn doping and As doping respectively, while chemically synthesizing iron silicide. The increase in the inter-planar distance observed from the SAED and XRD profiles proves the fact of doping. On the other hand, the red shift and broadening of the β-FeSi₂ Raman peak due to an increase in the bond length shows significant proof of doping. The D peak (Raman) at 300 cm⁻¹ for the doped sample due to the presence of lattice defects further confirms the effective doping. The shift in the binding energy of the Si 2p and Fe 2p³ peaks even further establishes the reality of the As and Mn doping. The weak As 3d and Mn 2p³ XPS peaks and also the EDX analyses infer the presence of the dopants. Hall measurements show that Mn doped β-FeSi₂ is p-type and As doped β-FeSi₂ is n-type, both of which agrees quite well that with the literature. The average Hall concentration is appreciably higher in the case of the Mn doped sample, due to lower activation energy in the Mn doped sample compared to that of the As doped sample. Another significant reason is the higher defect densities in the Mn doped sample compared to the As doped sample. Therefore, the above characterizations validate the effective doping of β-FeSi₂ by our chemical doping approach. So we strongly believe that our chemical technique will be useful for fabricating future iron silicide based semiconductor devices.

References

1. A. B. Filonov, D. B. Migas, V. L. Shaposhnikov, N. N. Dorozhkin, G. V. Petrov, V. E. Borisenko, W. Henrion and H. Lange, J. Appl. Phys., 1996, 79, 7708.
2. G. K. Dalapati, S. L. Liew, A. S. W. Wong, Y. Chai and S. Y. Chiam, Appl. Phys. Lett., 2011, 98, 013507.
3. G. K. Dalapati, S. M. Panah, A. Kumar, C. C. Tan, H. R. Tan and D. Chi, Sci. Rep., 2015, 5, 17810.
4. A. Bag, S. Mallik and C. K. Maiti, J. Renewable Sustainable Energy, 2014, 6, 023110.
5. Z. Liu, S. Wang, N. Otogawa, Y. Suzuki, M. Osamura, Y. Fukuzawa, T. Ootsuka, Y. Nakayama, H. Tanoue and Y. Makita, Sol. Energy Mater. Sol. Cells, 2006, 90, 276.
6. A. Kumar, G. K. Dalapati, H. Hidayat, F. Law, H. R. Tan, P. I. Widenborg, B. Hoex, C. C. Tan, D. Z. Chi and A. G. Aberle, RSC Adv., 2013, 3, 7733.
7. G. K. Dalapati, A. Kumar, C. C. Tan, S. L. Liew, P. Sonar, H. L. Seng, H. K. Hui, S. Tripathy and D. Chi, ACS Appl. Mater. Interfaces, 2013, 5, 5455.
8. M. Yoshimizu, R. Kobayashi, M. Saegusa, T. Takashima, H. Funakubo, K. Akiyama, Y. Matsumoto and H. Irie, Chem. Commun., 2015, 51, 2818.
9. J. Tani and H. Kido, J. Appl. Phys., 1999, 86, 464.
10. Z. Liu, M. Osamura, T. Ootsuka, S. Wang, Y. Fukuzawa, Y. Suzuki, R. Kuroda, T. Mise, N. Otogawa, Y. Nakayama, H. Tanoue and Y. Makita, Opt. Mater., 2005, 27, 942.
11. Y. Dusansoy, J. Protas, R. Wandij and B. Roques, Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem., 1971, 27, 1209.
12. J. Tani and H. Kido, J. Solid State Chem., 2002, 163, 248.
13. S. Kondo, M. Hasaka, T. Morimura and Y. Miyajima, Nucl. Instrum. Methods Phys. Res., Sect. B, 1993, 76, 383.
14. S. Kondo, M. Hasaka, T. Morimura and Y. Miyajima, Phys. B, 1994, 198, 332.
15. K. H. Tan, K. L. Pey and D. Z. Chi, J. Appl. Phys., 2009, 106, 023712.
16. T. Kojima, Phys. Status Solidi A, 1989, 111, 233.
17. J. Tani and H. Kido, J. Appl. Phys., 1998, 84, 1408.
18. S. Brehme, G. Behr and A. Heinrich, J. Appl. Phys., 2001, 89, 3798.
19. S. Kondo, M. Hasaka, T. Morimura and Y. Miyajima, Nucl. Instrum. Methods Phys. Res., Sect. B, 1993, 76, 383.
20. Z. J. Pan, L. T. Zhang and J. S. Wu, Mater. Sci. Eng., B, 2006, 131, 121.
21. K. H. Tan, K. L. Pey and D. Z. Chi, J. Appl. Phys., 2009, 106, 023712.
22. Y. Terai and Y. Maeda, Appl. Phys. Lett., 2004, 84, 903.
23. E. Arushanov, L. Ivanenko, H. Vinzelberg, D. Eckert, G. Behr, U. K. Rößler, K. H. Müller, C. M. Schneider and J. Schumann, J. Appl. Phys., 2002, 92, 5413.
24. H. T. Lu, L. J. Chen, Y. L. Chueh and L. J. Chou, J. Appl. Phys., 2003, 93, 1468.
25. S. Sen, N. Gogurla, P. Banerji, P. K. Guha and P. Pramanik, Mater. Sci. Eng., B, 2015, 200, 28.
26. B. Sahoo and P. K. Panda, J. Adv. Ceram., 2013, 2, 26.
27. Q. Zhao, D. Ju, X. Deng, J. Huang, B. Cao and X. Xu, Sci. Rep., 2014, 5, 7874.
28. G. W. Zhou, H. Li, H. P. Sun, D. P. Yu, Y. Q. Wang, X. J. Huang, L. Q. Chen and Z. Zhang, Appl. Phys. Lett., 1999, 75, 2448.
29. J. St John, J. L. Coffer, Y. Chen and R. F. Pinizzotto, J. Am. Chem. Soc., 1999, 121, 1888.
30. O. D. Jayakumar, I. K. Gopalakrishnan, K. Shashikala, S. K. Kulshreshtha and C. Sudakar, Magnetic properties of hydrogenated Li and Co doped ZnO nanoparticles, Appl. Phys. Lett., 2006, 89, 202507.
31. Y. Maeda, K. Umezawa, Y. Hayashi and K. Miyake, Thin Solid Films, 2001, 381, 219.
32. K. Takakura, H. Ohyama, K. Takarabe, T. Suemasu and F. Hasegawa, J. Appl. Phys., 2005, 97, 093716.
33. D. M. Caughey and R. E. Thomas, Proc. IEEE, 1967, 55, 2192.
34. J. Yuan, H. Shen, L. Lu, H. Huang and X. He, Phys. B, 2010, 405, 4565.
35. E. Arushanov, C. Kloc and E. Bucher, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 2653.
36. H. Udono and I. Kikuma, Thin Solid Films, 2004, 461, 188.

---