jellison_1992#
G. E. Jellison Jr., “Optical functions of silicon determined by two-channel polarization modulation ellipsometry”, Opt. Mater. 1, 41-47 (1992)
Source record · Catalogue · Review register
DOI: 10.1016/0925-3467(92)90015-F
Source kind: measurement
Recorded technique: Two-channel polarization modulation ellipsometry (2-C PME; pp. 1-2): 75 W Xe arc lamp, 0.25 m monochromator, polarizing prism and photoelastic modulator (50.2 kHz), Wollaston prism and two phototubes, measuring N = cos 2psi, S = sin 2psi sin Delta and C = sin 2psi cos Delta (Eq. 1), normalised by beta = (N^2 + S^2 + C^2)^(1/2) (Eq. 3). Angle of incidence 67.24 +- 0.02 deg, from measurements of polished fused silica fitted to Malitson’s index. 187 wavelengths from 234 to 840 nm, at room temperature. n and k are calculated from the data by a two-dimensional Newton’s method after mathematically removing the native oxide (a Sellmeier layer, Eq. 4: n^2 - 1 = A lambda^2/(lambda^2 - lambda_0^2), A = 2.0, lambda_0 = 92.27 nm), whose thickness, 0.77 +- 0.02 nm for (100), 0.58 +- 0.02 nm for (111) and 0.62 +- 0.02 nm for (110), was set by normalising the data to Dash and Newman’s absorption coefficient at 765-805 nm (p. 2).
Materials in structured entries: Si
Dataset quantities: extinction_coefficient, refractive_index
Contents: 6 datasets; 0 formulas.
Curation notes#
Pdf pages 1-7. Identity: the library’s record is VALIDATED_WITH_WARNINGS; title, author, journal, volume and pages agree with the first page. The tables are unnumbered in the library (the first page of Table 1 lost its label); the paper calls them Table 1 and Table 2.
SAMPLES (p. 2). (100), n-type, 0.2-0.3 Ohm cm; (111), n-type, 0.2-0.5 Ohm cm; (110), p-type, ~11 Ohm cm; all factory polished; the oxide removed with an HF etch and the sample left in air ~1 h before measurement (so a native oxide regrew, which the analysis removes). ‘At room temperature’ (abstract, Figs. 1-3, p. 4); no number.
TABLE 1 (pp. 3-4; datasets n_100, k_100, primary). ‘The refractive index n and extinction coefficient k of (100) silicon from 234 nm to 840 nm, including error limits’: 187 rows, wavelength (nm, 2 nm steps to 450 nm and 5 nm above), energy (eV), n +- error, k +- error, three decimals. The error limits ‘reflect the experimental uncertainties in the monochromator wavelength, in N, S, and C, and in errors introduced from the beta correction. Additional uncertainties in the angle of incidence and in the oxide layer thickness were also incorporated’ (p. 5): propagated errors, level not stated; 0.4-1.2 % of n (0.4-0.6 % from 400 nm), 0.001-0.003 in k above 450 nm (0.004 +- 0.003 at 840 nm). The k values at 765-805 nm are forced to Dash and Newman’s (W. C. Dash and R. Newman, Phys. Rev. 99, 1151 (1955), the paper’s Ref. 10, not in the library) by the oxide-thickness normalisation, so they are excluded rows; elsewhere k below ~3 eV still depends on that normalisation. WAVELENGTHS: the energy column is 1239.5/lambda (nm) to three decimals in all 186 other rows (1239.842/lambda differs by 1 to 2 in the last digit in 141 rows), which is also 1239.842/(n_air lambda) for lambda in standard air: the medium of the wavelengths is not stated; vacuum is recorded. The difference (2.8e-4 in wavelength) moves n by up to 0.012 (288 nm, error 0.026) and k by up to 0.015 (372 nm, error 0.021), and n by at most 0.0013 above 450 nm: under the error limits.
CHECKS (2026-10-14). Both pages compared with their images. Three misprints, printed as the library reads them (errata): 595.0 nm n 2.948 for 3.948 (neighbours 3.960, 3.937); 705.0 nm n 2.766 for 3.766 (neighbours 3.772, 3.760); 495.0 nm error of n 0.002 for 0.020 (neighbours 0.021, 0.020; every other error of n is 0.4-1.2 % of n). The energy at 386.0 nm is printed 3.221 for 3.211 (between 3.228 at 384 nm and 3.195 at 388 nm), not extracted. 292.0 nm n is printed ‘4.688 +- 0 027’ (a lost point; the library reads 0.027). The (100) rows of Table 2 equal Table 1’s at all seven wavelengths (checked by the script).
TABLE 2 (p. 6; datasets n_111, k_111, n_110, k_110, check). ‘A comparison of the literature values of the optical functions of Si determined by several workers at selected wavelengths’: 248, 292, 362, 400, 450, 515 and 635 nm, faces (100), (111) and (110) of this work (a), with Ref. [4] (b: D. F. Edwards in Palik’s Handbook, 1985, p. 547: ‘most of the data in ref. [4] in this wavelength region has been taken from ref. [3], while the values of k were taken from ref. [11]’, i.e. Aspnes and Studna and Hulthen) and Ref. [1] (c: Jellison and Modine, J. Appl. Phys. 53, 3745 (1982)); b and c are not extracted. The faces differ by 1-2 % in epsilon at the critical points (292 and 362 nm) and agree below the direct edge (p. 5). The b value at 248 nm, n = 1.589, k = 3.354, is Aspnes and Studna’s 5.2 eV row (238.4 nm), not their 5.0 eV row (248 nm: 1.570, 3.565), so the ‘~0.1 (~6 %) discrepancy in n at 248 nm’ the text reports (p. 7) compares different wavelengths (against the 5.0 eV row it is 0.113, 7 %); the b value at 400 nm (5.570, 0.387) is Aspnes and Studna’s 3.1 eV row. Ref. [3] is printed as ‘Phys. Rev. B 27 (1983) 7466’.
AGAINST aspnes_1983 (2026-10-14; this table interpolated to Aspnes and Studna’s energies with E = 1239.842/lambda): Aspnes’s n differs by at most 0.015 at 1.5-3.0 eV (higher below 2.8 eV), is lower by 0.02-0.10 at 3.1-3.3 eV (the E1 edge), differs by -0.08 to +0.04 at 3.4-4.2 eV and is lower by 0.10-0.135 at 4.3-5.3 eV (5.0 eV: 1.570 against 1.683 +- 0.013); Aspnes’s k is higher by 27-63 % at 2.0-3.0 eV (3.0 eV: 0.269 against 0.196 +- 0.003; below 2.84 eV Aspnes’s k is Dash and Newman’s or blended with it), 20 % at 3.2 eV, and within 1-4 % at 3.3-5.3 eV. Peak epsilon_2 = 2nk here 46.5 at 4.246 eV; Aspnes’s 46.80 at 4.25 eV.
FINDINGS (p. 7). Above the direct edge (3.4 eV) this work and Ref. [4] agree within the error limits; below it n agrees and k is 25-50 % lower here, attributed to the RAE’s inaccuracy in Delta and the residual oxide of Ref. [4]’s data.
COMPARISON (2026-10-14). The refractiveindex.info page Si/Jellison (186 rows, wavelength in um, n and k without errors) is Table 1 with the two n misprints corrected (0.595 um 3.948, 0.705 um 3.766) and the 640.0 nm row left out; the other 185 rows are equal.
Recorded bibliographic identity status: VALIDATED_WITH_WARNINGS. This concerns identification of the source; it does not establish the scientific correctness of its contents.
Datasets#
jellison_1992/n_100#
{
"id": "n_100",
"material": "Si",
"file": "Si/jellison_1992__n_100.csv",
"table": "tbl_290c9b5e3da6",
"quantity": "refractive_index",
"abscissa_unit": "nm",
"use": "primary",
"uncertainty": {
"column": true,
"level": "not stated"
},
"parameters": {
"form": "bulk",
"technique": "two-channel polarization modulation ellipsometry, native oxide removed in the analysis"
}
}
Table 1 (pp. 3-4, tbl_290c9b5e3da6 and tbl_6b86dd05f9d8): n of (100) silicon (n-type, 0.2-0.3 Ohm cm, factory polished, HF etched, native oxide of 0.77 nm removed in the analysis), 234-840 nm, 187 rows, with the paper’s error limits in sigma (propagated errors, level not stated). Room temperature. Errata: 595.0 nm (2.948 for 3.948), 705.0 nm (2.766 for 3.766), 495.0 nm error (0.002 for 0.020). Wavelength medium not stated (see notes).
jellison_1992/k_100#
{
"id": "k_100",
"material": "Si",
"file": "Si/jellison_1992__k_100.csv",
"table": "tbl_290c9b5e3da6",
"quantity": "extinction_coefficient",
"abscissa_unit": "nm",
"use": "primary",
"excluded_rows": {
"765.0": "Forced to Dash and Newman's absorption coefficient by the oxide-thickness normalisation at 765-805 nm (p. 2).",
"770.0": "Forced to Dash and Newman's absorption coefficient by the oxide-thickness normalisation at 765-805 nm (p. 2).",
"775.0": "Forced to Dash and Newman's absorption coefficient by the oxide-thickness normalisation at 765-805 nm (p. 2).",
"780.0": "Forced to Dash and Newman's absorption coefficient by the oxide-thickness normalisation at 765-805 nm (p. 2).",
"785.0": "Forced to Dash and Newman's absorption coefficient by the oxide-thickness normalisation at 765-805 nm (p. 2).",
"790.0": "Forced to Dash and Newman's absorption coefficient by the oxide-thickness normalisation at 765-805 nm (p. 2).",
"795.0": "Forced to Dash and Newman's absorption coefficient by the oxide-thickness normalisation at 765-805 nm (p. 2).",
"800.0": "Forced to Dash and Newman's absorption coefficient by the oxide-thickness normalisation at 765-805 nm (p. 2).",
"805.0": "Forced to Dash and Newman's absorption coefficient by the oxide-thickness normalisation at 765-805 nm (p. 2)."
},
"uncertainty": {
"column": true,
"level": "not stated"
},
"parameters": {
"form": "bulk",
"technique": "two-channel polarization modulation ellipsometry, native oxide removed in the analysis"
}
}
Table 1 (pp. 3-4, tbl_290c9b5e3da6 and tbl_6b86dd05f9d8): k of (100) silicon (n-type, 0.2-0.3 Ohm cm, factory polished, HF etched, native oxide of 0.77 nm removed in the analysis), 234-840 nm, 187 rows, with the paper’s error limits in sigma (propagated errors, level not stated). Room temperature. The oxide thickness was set so that k matches Dash and Newman’s at 765-805 nm: those nine rows are theirs by construction and are excluded. Wavelength medium not stated (see notes).
jellison_1992/n_111#
{
"id": "n_111",
"material": "Si",
"file": "Si/jellison_1992__n_111.csv",
"table": "tbl_364290431ef8",
"quantity": "refractive_index",
"abscissa_unit": "nm",
"use": "check",
"uncertainty": {
"column": true,
"level": "not stated"
},
"parameters": {
"form": "bulk",
"technique": "two-channel polarization modulation ellipsometry, native oxide removed in the analysis"
}
}
Table 2 (p. 6, tbl_364290431ef8; image checked): n of (111) silicon (n-type, 0.2-0.5 Ohm cm; native oxide of 0.58 nm removed in the analysis) at 248, 292, 362, 400, 450, 515 and 635 nm, with error limits; the paper’s comparison of faces. A check on the face dependence (1-2 % in epsilon at 292 and 362 nm), not data to fit beside the (100) table.
jellison_1992/k_111#
{
"id": "k_111",
"material": "Si",
"file": "Si/jellison_1992__k_111.csv",
"table": "tbl_364290431ef8",
"quantity": "extinction_coefficient",
"abscissa_unit": "nm",
"use": "check",
"uncertainty": {
"column": true,
"level": "not stated"
},
"parameters": {
"form": "bulk",
"technique": "two-channel polarization modulation ellipsometry, native oxide removed in the analysis"
}
}
Table 2 (p. 6, tbl_364290431ef8; image checked): k of (111) silicon (n-type, 0.2-0.5 Ohm cm; native oxide of 0.58 nm removed in the analysis) at 248, 292, 362, 400, 450, 515 and 635 nm, with error limits; the paper’s comparison of faces. A check on the face dependence, not data to fit beside the (100) table.
jellison_1992/n_110#
{
"id": "n_110",
"material": "Si",
"file": "Si/jellison_1992__n_110.csv",
"table": "tbl_364290431ef8",
"quantity": "refractive_index",
"abscissa_unit": "nm",
"use": "check",
"uncertainty": {
"column": true,
"level": "not stated"
},
"parameters": {
"form": "bulk",
"technique": "two-channel polarization modulation ellipsometry, native oxide removed in the analysis"
}
}
Table 2 (p. 6, tbl_364290431ef8; image checked): n of (110) silicon (p-type, ~11 Ohm cm; native oxide of 0.62 nm removed in the analysis) at 248, 292, 362, 400, 450, 515 and 635 nm, with error limits; the paper’s comparison of faces. A check on the face dependence, not data to fit beside the (100) table.
jellison_1992/k_110#
{
"id": "k_110",
"material": "Si",
"file": "Si/jellison_1992__k_110.csv",
"table": "tbl_364290431ef8",
"quantity": "extinction_coefficient",
"abscissa_unit": "nm",
"use": "check",
"uncertainty": {
"column": true,
"level": "not stated"
},
"parameters": {
"form": "bulk",
"technique": "two-channel polarization modulation ellipsometry, native oxide removed in the analysis"
}
}
Table 2 (p. 6, tbl_364290431ef8; image checked): k of (110) silicon (p-type, ~11 Ohm cm; native oxide of 0.62 nm removed in the analysis) at 248, 292, 362, 400, 450, 515 and 635 nm, with error limits; the paper’s comparison of faces. A check on the face dependence, not data to fit beside the (100) table.
Formulas#
No formulas are recorded.
Corrections#
[
{
"dataset": "n_100",
"row": "595.0",
"column": "value",
"printed": "2.948",
"corrected": "3.948",
"reason": "A misprinted leading digit: n of silicon falls smoothly from 3.960 (590 nm) to 3.937 (600 nm); 2.948 would be a drop of 1.0 in 5 nm, and the row's error limit (0.018) is that of n ~ 3.9 (0.45 %). The refractiveindex.info page Si/Jellison gives 3.948.",
"evidence": "Table 1 (tbl_6b86dd05f9d8, p. 4) and its image, which prints 2.948."
},
{
"dataset": "n_100",
"row": "705.0",
"column": "value",
"printed": "2.766",
"corrected": "3.766",
"reason": "A misprinted leading digit: n falls smoothly from 3.772 (700 nm) to 3.760 (710 nm); the row's error limit (0.016) is that of n ~ 3.8 (0.42 %). The refractiveindex.info page Si/Jellison gives 3.766.",
"evidence": "Table 1 (tbl_6b86dd05f9d8, p. 4) and its image, which prints 2.766."
},
{
"dataset": "n_100",
"row": "495.0",
"column": "sigma",
"printed": "0.002",
"corrected": "0.020",
"reason": "A dropped digit in the error limit of n: the neighbours print 0.021 (490 nm) and 0.020 (500 nm), and every other error limit of n in Table 1 is 0.4-1.2 % of n (0.4-0.6 % at 400-840 nm) (4.317 x 0.0046 = 0.020); 0.002 would be 0.05 %.",
"evidence": "Table 1 (tbl_6b86dd05f9d8, p. 4) and its image, which prints 4.317 +- 0.002."
}
]
Formula corrections, where recorded, appear with their formulas above. Curation notes may also discuss corrections or unresolved interpretation. Consult the review register for the checks and decisions.