Exploiting the P L 2 , 3 absorption edge for optics : spectroscopic and structural characterization of cubic boron phosphide thin films

The transmission of cubic boron phosphide (c-BP) thin films, prepared by chemical vapor deposition (CVD), was evaluated near the phosphorous L2,3 and boron K absorption edge. The c-BP films were analyzed with transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS) and X-ray absorption near-edge spectroscopy (XANES), to study their structural and chemical properties. The TEM analysis reveals that c-BP initially grows in islands. The merging of the P L2,3, P K and B K absorption edges culminates in a sharp absorption feature starting at 130 eV, showing that c-BP can be used in applications that require a relatively transparent material in the energy range just below that absorption feature. Due to experimental constraints the samples were grown at a temperature significantly below the temperature for optimal crystal growth. XANES analysis showed that, as a result of the reduced crystal quality, the intensities of the absorption transitions are reduced compared to those in high quality crystalline reference samples. Optimizing the quality of the BP films will increase the contrast in transmission across the absorption edge. c © 2016 Optical Society of America OCIS codes: (160.0160) Materials; (300.0300) Spectroscopy; (120.7000) Transmission. References and links 1. E. A. Spiller, Soft X-Ray Optics (Bellingham Washington USA, SPIE Publications, 1994). 2. J. Larruquert, “Optical properties of thin film materials at short wavelengths,” in “Optical Thin Films and Coatings,” A. Piegari and F. Flory, eds. (Woodhead Publishing, 2013). 3. E. A. Spiller, “Windows and filters,” in “Soft X-Ray Optics,” (Bellingham Washington USA, SPIE Publications, 1994). 4. J. F. Seely, “Extreme-ultraviolet thin-film interference in an Al-Mg-Al multiple-layer transmission filter,” Appl. Opt. 41, 5979–5983 (2002). 5. M. Hatayama, H. Takenaka, E. M. Gullikson, A. Suda, and K. Midorikawa, “High-transmittance free-standing aluminum extreme ultraviolet filter,” Jpn. J. Appl. Phys. 48, 122202–122205 (2009). 6. N. I. Chkhalo, M. N. Drozdov, E. B. Kluenkov, S. V. Kuzin, A. Y. Lopatin, V. I. Luchin, N. N. Salashchenko, N. N. Tsybin, and S. Y. Zuev, “Thin film multilayer filters for solar EUV telescopes,” Appl. Opt. 55, 4683–4690 (2016). 7. P. J. van Zwol, D. F. Vles, W. P. Voorthuijzen, M. Péter, H. Vermeulen, W. J. van der Zande, J. M. Sturm, R. W. E. van de Kruijs, and F. Bijkerk, “Emissivity of freestanding membranes with thin metal coatings,” J. Appl. Phys. 118, 213107–213111 (2015). 8. S. Yulin, “Multilayer coatings for EUV/soft x-ray mirrors,” in “Optical Interference Coatings,” N. Kaiser and H. K. Pulker, eds. (Springer Berlin Heidelberg, 2003). 9. E. Louis, A. Yakshin, T. Tsarfati, and F. Bijkerk, “Nanometer interface and materials control for multilayer EUVoptical applications,” Prog. Surf. Sci. 86, 255–294 (2011). 10. I. A. Makhotkin, E. Zoethout, E. Louis, A. M. Yakunin, S. Müllender, and F. Bijkerk, “Spectral properties of La/B based multilayer mirrors near the boron K-absorption edge,” Opt. Express 20, 11778–11786 (2012). Vol. 6, No. 12 | 1 Dec 2016 | OPTICAL MATERIALS EXPRESS 3946 #277879 http://dx.doi.org/10.1364/OME.6.003946 Journal © 2016 Received 29 Sep 2016; revised 21 Nov 2016; accepted 21 Nov 2016; published 29 Nov 2016 11. D. S. Kuznetsov, A. E. Yakshin, J. M. Sturm, R. W. E. van de Kruijs, E. Louis, and F. Bijkerk, “High-reflectance La/B-based multilayer mirror for 6.x nm wavelength,” Opt. Lett. 40, 3778–3781 (2015). 12. B. Sae-Lao and C. Montcalm, “Molybdenum–strontium multilayer mirrors for the 8–12-nm extreme-ultraviolet wavelength region,” Opt. Lett. 26, 468–470 (2001). 13. D. L. Windt and E. M. Gullikson, “Pd/B4C/Y multilayer coatings for extreme ultraviolet applications near 10 nm wavelength,” Appl. Opt. 54, 5850–5860 (2015). 14. M. Mero, F. Frassetto, P. Villoresi, L. Poletto, and K. Varjú, “Compression methods for XUV attosecond pulses,” Opt. Express 19, 23420–23428 (2011). 15. M. Hofstetter, A. Aquila, M. Schultze, A. Guggenmos, S. Yang, E. Gullikson, M. Huth, B. Nickel, J. Gagnon, V. S. Yakovlev, E. Goulielmakis, F. Krausz, and U. Kleineberg, “Lanthanum–molybdenum multilayer mirrors for attosecond pulses between 80 and 130 eV,” New J. Phys. 13, 063038–063052 (2011). 16. L. R. de Marcos, J. I. Larruquert, J. A. Aznárez, M. Vidal-Dasilva, S. García-Cortés, J. A. Méndez, L. Poletto, F. Frassetto, A. M. Malvezzi, D. Bajoni, A. Giglia, N. Mahne, and S. Nannarone, “Transmittance and optical constants of Sr films in the 6–1220 eV spectral range,” J. Appl. Phys. 111, 113533–113539 (2012). 17. V. Medvedev, A. Yakshin, R. van de Kruijs, and F. Bijkerk, “Phosphorus-based compounds for EUV multilayer optics materials,” Opt. Mater. Express 5, 1450–1459 (2015). 18. B. Kjornrattanawanich, “Reflectance, optical properties, and stability of molybdenum/strontium and molybdenum/yttrium multilayer mirrors,” Ph.D. thesis, University of California, Davis (2002). 19. B. Sae-Lao, S. Bajt, C. Montcalm, and J. F. Seely, “Performance of normal-incidence molybdenum-yttrium multilayercoated diffraction grating at a wavelength of 9 nm,” Appl. Opt. 41, 2394–2400 (2002). 20. Z. Wang, H. Wang, J. Zhu, Y. Xu, S. Zhang, C. Li, F. Wang, Z. Zhang, Y. Wu, X. Cheng, L. Chen, A. G. Michette, S. J. Pfauntsch, A. K. Powell, F. SchaÌĹfers, A. Gaupp, and M. MacDonald, “Extreme ultraviolet broadband Mo/Y multilayer analyzers,” Appl. Phys. Lett. 89, 241120–241122 (2006). 21. N. I. Chkhalo and N. N. Salashchenko, “Next generation nanolithography based on Ru/Be and Rh/Sr multilayer optics,” AIP Adv. 3, 082130–082138 (2013). 22. F. Frassetto, P. Villoresi, and L. Poletto, “Beam separator for high-order harmonic radiation in the 3-10 nm spectral region,” J. Opt. Soc. Am. A 25, 1104–1114 (2008). 23. Silson, “http://silson.com/,” . 24. B. Padavala, C. Frye, Z. Ding, R. Chen, M. Dudley, B. Raghothamachar, N. Khan, and J. Edgar, “Preparation, properties, and characterization of boron phosphide films on 4Hand 6H-silicon carbide,” Solid State Sci. 47, 55–60 (2015). 25. B. Padavala, C. D. Frye, X. Wang, Z. Ding, R. Chen, M. Dudley, B. Raghothamachar, P. Lu, B. N. Flanders, and J. H. Edgar, “Epitaxy of boron phosphide on aluminum nitride(0001)/sapphire substrate,” Cryst. Growth Des. 16, 981–987 (2016). 26. B. Padavala, C. Frye, X. Wang, B. Raghothamachar, and J. Edgar, “CVD growth and properties of boron phosphide on 3C-SiC,” J. Cryst. Growth 449, 15–21 (2016). 27. S. P. Huber, V. V. Medvedev, E. Gullikson, B. Padavala, J. H. Edgar, R. W. E. van de Kruijs, F. Bijkerk, and D. Prendergast, “Determining crystal phase purity in c-BP through x-ray absorption spectroscopy,” Unpublished . 28. J. Underwood and E. Gullikson, “High-resolution, high-flux, user friendly VLS beamline at the ALS for the 50-1300 eV energy region,” J. Electron Spectrosc. Relat. Phenom. 92, 265 – 272 (1998). 29. E. M. Gullikson, S. Mrowka, and B. B. Kaufmann, “Recent developments in EUV reflectometry at the advanced light source,” in “Emerging Lithographic Technologies V,” , vol. 4343 E. A. Dobisz, ed. (SPIE-Intl Soc Optical Eng, 2001), vol. 4343, pp. 363–373. 30. W.-Y. Ching, S.-D. Mo, and Y. Chen, “Calculation of XANES/ELNES spectra of all edges in Si3N4 and Si2N2O,” J. Am. Ceram. Soc. 85, 11–15 (2004). 31. B. Henke, E. Gullikson, and J. Davis, “X-ray interactions: Photoabsorption, scattering, transmission, and reflection at E = 50-30,000 eV, Z = 1-92,” At. Data Nucl. Data Tables 54, 181 – 342 (1993). 32. R. Soufli, S. Bajt, and E. M. Gullikson, “Optical constants of beryllium from photoabsorption measurements for x-ray optics applications,” in “EUV, X-Ray, and Neutron Optics and Sources,” C. A. MacDonald, K. A. Goldberg, J. R. Maldonado, H. H. Chen-Mayer, and S. P. Vernon, eds. (SPIE-Intl Soc Optical Eng, 1999), pp. 251–258. 33. M. Fernández-Perea, M. Vidal-Dasilva, J. A. Aznárez, J. I. Larruquert, J. A. Méndez, L. Poletto, D. Garoli, A. M. Malvezzi, A. Giglia, and S. Nannarone, “Transmittance and optical constants of Eu films from 8.3 to 1400 eV,” J. Appl. Phys. 104, 123527–123533 (2008). 34. B. Kjornrattanawanich, D. L. Windt, J. A. Bellotti, and J. F.Seely, “Measurement of dysprosium optical constants in the 2-830 eV spectral range using a transmittance method, and compilation of the revised optical constants of lanthanum, terbium, neodymium, and gadolinium,” Appl. Opt. 48, 3084–3093 (2009). 35. B. Kjornrattanawanich, D. L. Windt, and J. F. Seely, “Optical constants determination of samarium, holmium, and erbium in the 15-850 eV spectral range using a transmittance method,” Appl. Opt. 49, 6006–6013 (2010). 36. B. Sae-Lao and R. Soufli, “Measurements of the refractive index of yttrium in the 50–1300-eV energy region,” Appl. Opt. 41, 7309–7316 (2002). 37. V. A. Mukhanov, P. S. Sokolov, Y. L. Godec, and V. L. Solozhenko, “Self-propagating high-temperature synthesis of boron phosphide,” J. Superhard Mater. 35, 415–417 (2013). 38. H. Sugimoto, M. Fujii, and K. Imakita, “Size-controlled growth of cubic boron phosphide nanocrystals,” RSC Adv. Vol. 6, No. 12 | 1 Dec 2016 | OPTICAL MATERIALS EXPRESS 3947


Introduction
Materials with low optical absorption are often required to develop efficient optical elements for extreme ultraviolet (EUV) radiation applications [1,2].Prospective candidates are generally selected by analyzing the characteristic absorption spectra of various elements, whose spectral features in the EUV spectral range are largely determined by the characteristic binding energies of their respective electronic orbitals [2].The absorption of photons by atoms of a certain element increases sharply for photon energies just above the binding energies of the atomic core levels.To photons with an energy just below such an absorption edge, the element is relatively transparent, which makes it ideal for the construction of a thin-film transmission filter operating near that energy [3][4][5][6][7].Alternatively, two or more elements can be combined to construct a multilayer coating [8] that has a high reflectance in the spectral range where all elements are relatively transparent.However, due to the large dispersion in the vicinity of absorption edges, the transparent spectral range is typically narrow.Therefore, the manipulation of EUV radiation in a broad spectral range often requires a combination of multiple reflective and or transmissive optical elements, composed of varying material combinations [2].
The L 2,3 -edge of silicon (Si) at approximately 100 eV and the K-edge of boron (B) around 187 eV are prime examples of elements along with their characteristic absorption edges, that are commonly employed in EUV optical applications, such as free-standing optical filters and high reflective optics [9][10][11].Methods for fabricating optical elements that are operational around the Si L 2,3 and B K-edge are well established, however, many applications require optics that are highly transparent at energies within the bounds of this spectral range [6,[12][13][14][15].Several studies have highlighted the M 4,5 and L 2,3 absorption edges of strontium (Sr) and phosphorus (P), respectively, as potential candidates for the construction of a transmission filter or reflective optic that would operate just below 130 eV [16,17].The high reactivity of Sr renders this element effectively inapplicable, as any exposure to ambient conditions would result in a rapid deterioration of the optical element's quality [12,18].Phosphorus in pure form suffers from similar problems, however some phosphides are stable.A stable phosphide that maximizes the relative phosphorus content and whose other constituent elements are relatively transparent around 130 eV, is cubic boron phosphide (c-BP).In this paper, we study the feasibility of employing c-BP to construct an optical element that would operate below the P L 2,3 -edge.
Here we report on the characterization of chemical vapor deposition (CVD) grown c-BP thin films with thicknesses below 100 nm.Structural and chemical characterization of the samples was performed using transmission electron microscopy (TEM) and x-ray photoelectron spectroscopy (XPS).The microscopy analysis revealed that in the initial stages, the film grows in islands, which resulted in discontinuous films for shorter deposition times.XPS measurements indicated that the stoichiometry of the grown films deviated slightly from the expected 1 : 1 stoichiometry of c-BP, with a small surplus of boron.XPS analysis also revealed the sample surfaces to have oxidized as a result of the exposure to atmospheric conditions.Broadband transmission measurements demonstrated that the grown BP films have a broad absorption band from 130 eV to 300 eV, as a result of the overlapping P L 2,3 , P K and B K absorption edges.This work shows that, through c-BP, the P L 2,3 -edge can in principle be employed to construct optical elements that are relatively transparent to photons with an energy below 130 eV, which can be applied in space optics [19,20], photolithography [21] and high harmonics generation [22].

Sample deposition
Boron phosphide thin films were deposited on silicon nitride membrane substrates by chemical vapor deposition (CVD).The substrates were purchased from Silson Ltd [23] and comprise a crystalline silicon wafer with a surface area of 10×10 mm 2 and a thickness of 381 µm, on top of which a 100 nm thin layer of Si 3 N 4 is grown.Part of the underlying silicon is removed by chemical etching, exposing in the center of the silicon wafer, a square surface area of 0.5×0.5 mm 2 of the silicon nitride layer on both sides.Boron phosphide was deposited on both the substrates with silicon nitride window and a plain silicon substrate, which served as reference samples for the x-ray absorption spectroscopy measurements.The CVD setup and experimental methods used in this work have been described in greater detail elsewhere [24][25][26].The precursor gases were ultra-high purity phosphine (99.999 %) and diborane (1 % in H 2 ) in an ultra-high purity hydrogen carrier gas.Detailed deposition conditions employed are listed in Table 1.The ideal deposition temperature for high quality crystalline c-BP films is around 1200 • C, however, the silicon nitride membranes delaminated from the silicon membranes under these deposition conditions and therefore, the deposition temperature had to be reduced to 700 • C.

Sample characterization
After deposition, the grown boron phosphide films were characterized by x-ray photoelectron spectroscopy (XPS), transmission electron microscopy (TEM) and x-ray absorption near-edge spectroscopy (XANES), the results of which will be presented in the following sections.

X-ray photoelectron spectroscopy
The binding energy of the boron 1s core level overlaps with the phosphorus 2s core level and therefore special care has to be taken when fitting XPS data to extract the elemental stoichiometry of the analyzed samples.The intensity ratio of the P2s signal with respect to the P2p 3/2 signal was determined from a MoP reference sample that ostensibly contains phosphorus but no boron.This ratio was 1.15 and was kept constant during the subsequent XPS analysis of the BP samples.First, the P2p signal was fit to determine the total intensity of the elemental P2p core-level electrons, which in combination with the earlier determined P2s/P2p 3/2 intensity ratio, fixes the total expected intensity of the P2s signal.While keeping the peak area of the P2s component fixed, the combined P2s and B1s signal was fit into its individual components.
The results of the XPS analysis are shown in Fig. 1.The left column displays the signal collected from electrons excited from the P2p core-level with an approximate binding energy of around 131 eV.The experimental data, indicated by the black dots, is fit by four individual contributions, whose sum is indicated by the solid gray line labeled "Envelope", taking into account a Shirley background.Due to spin-orbit coupling, the P2p core-level is split energetically Binding energy E (eV) Data Envelope P2s B1s Fig. 1.In rows, from top to bottom, the measured and analyzed XPS data for samples S1, S2 and S3, respectively.The columns, from left to right, represent the collected signal from excitations of electrons from P2p and B1s/P2s core-levels, respectively.The experimental data is represented by black dots and the dashed colored lines represent the partial contributions to fit the experimental data.The solid gray line, labeled "Envelope", is the sum of these partial fit contributions.
into the two distinct core states P2p 3/2 and P2p 1/2 .These two core-level states have a relative electron occupancy ratio of 2 : 1, which is reflected in the ratio of the areas of the corresponding distributions.Two additional contributions, with significantly lower intensities, can be found at a higher energies, which correspond to the same two P2p core states, but in this case associated with phosphorus atoms that have reacted with oxygen.
The right column of Fig. 1 shows the measured signal centered around an energy of 189 eV, which corresponds to the overlapping populations of excited electrons from the B1s and P2p core levels, as described previously.Due to the convolution of these two signals, the uncertainty in the fit of the experimental data is slightly increased.For this reason, no attempt was made to fit a peak corresponding to boron oxide, even though a minimal contribution could arguably be observed on the high energy side of the main peak.The almost negligible signal of the oxide (estimated at the order of 1 %), combined with the error due to the B1s/P2p overlap, would render the resulting fit effectively meaningless.
From the fit to the experimental data, atomic content percentages can be extracted from the areas of the various peaks, the results of which are displayed in Table 2.All samples show a significant presence of oxygen on the sample surface of 7 % to 9 %, which usually takes the form of various adsorbed carbohydrates and hydroxyl groups, but the presence of the phosphorus oxide peak shows that part of the oxygen has reacted with the phosphorus.As discussed in the previous paragraph, a small fraction of the oxygen also has reacted with the boron, as evidenced by the small shoulder on the high energy side in the B1s peaks of Fig. 1.Due to the surface sensitivity of XPS, the adsorbed carbohydrates on the sample surface yield a significant contribution of C1s excitations.The non-oxidized boron and phosphorus contents were almost stoichiometric, as one would expect for c-BP, however in all three samples, there seems to be a small surplus of boron atoms of 1.7 % to 4.0 % absolute.Recent research has shown that the cubic boron phosphide and elemental boron phases, have comparable formation enthalpies, and therefore during deposition, the formation of these two phases can compete in parallel [27].The excess phosphorus may be offgassed or deposited in the form of oxides as evidenced by the XPS measurements.Even though the excess of boron is small, these XPS measurements indicate that the grown c-BP films may contain (amorphous) elemental boron phases.

Transmission electron microscopy
Before cross-sectional transmission electron micrographs of the deposited boron phosphide thin films could be recorded, the samples had to undergo further preparation.First, a platinum layer of 100 nm was deposited on both sides of the samples.Subsequently, an FEI Nova 600 Nanolab Dualbeam focused ion beam (FIB) was employed to cut a cross-section of the sample with a thickness of approximately 300 nm, using a gallium ion beam at an acceleration voltage of 30 kV.At this point, the acceleration voltage was reduced to 5 kV, in order to minimize structural damage of the eventual preparate, while further reducing its cross-sectional thickness to the minimally required 100 nm.After sample preparation, cross-sectional transmission electron micrographs were recorded with a Philips CM300ST-FEG TEM at an operating acceleration voltage of 300 kV.Calibrated by the MAG*I*CAL standard, errors in the magnification are on the order of at most 1 % to 2 %.Thicknesses were measured with energy-filtered transmission electron microscopy (EFTEM) employing a GATAN Tridiem energy filter.
Figure 2 shows a close up view of a TEM recording of the cross-section for each one of the three studied samples.The TEM images show a layered structure, where the bottom layer can be identified as residue of the FIB milling process that was used to extract a cross-section from the samples.The layer directly on top corresponds to the platinum layer that was deposited on the silicon nitride membrane, prior to the FIB milling process.The Si 3 N 4 membrane is seen directly on top of the Pt-layer and clearly has a uniform thickness and low surface roughness.On top of the smooth nitride membrane, the deposited c-BP layer has a significantly higher surface roughness, followed by another Pt-layer and FIB residue layer.
For samples S1 and S2, shown in Fig. 2(a) and Fig. 2(b), respectively, the phosphide layer is discontinuous and exposes the underlying the Si 3 N 4 , as indicated by the red squares.Sample S3 is the only sample that has a continuous boron phosphide layer that fully covers the underlying nitride membrane.The discontinous c-BP layers of samples S1 and S2 seem to comprise multiple regions, hemispherical in shape, which suggests that in the initial stages of the c-BP crystal growth, the film grows in isolated islands, which eventually agglomerate and merge into a continous film, as in the case of sample S3.This indicates, that in order to grow a continous c-BP film under these growth conditions, a deposition time of at least 60 s is required.Due to the discontinuous character of the c-BP films of samples S1 and S2, an effective thickness is meaningless.However, the TEM analysis assigns an effective thickness of 90 nm to the boron phosphide layer of sample S3.
The grown c-BP films typically have a polycrystalline character with crystal grain size on the order of 10 nm.An example of such a grain is shown in Fig. 3(a), with a further magnification in Fig. 3(b).
Regions with amorphous character are also found, two examples of which are shown in Fig. 4.These regions are found throughout the layer with a propensity of occurring near the c-BP sample surface.The presence of amorphous regions is not wholly surprising, given that the deposition temperature had to be lowered from the ideal temperature of 1200 • C to 700 • C in order not to destroy the silicon nitride membrane upon which the c-BP film was grown.The amorphous regions could partly also be associated with amorphous elemental boron regions that are suggested by the excess of boron as found in the XPS analysis.

Transmission spectroscopy
The transmittance of the grown c-BP samples in the soft x-ray range was measured at room temperature at beam line 6.3.2 of the Advanced Light Source of Lawrence Berkeley National Laboratory.A detailed description and characterization of the beamline and measurement chamber can be found elsewhere [28,29].The synchrotron radiation is passed through an array of optical elements, including a grating to select the required wavelength and an order suppressor to reduce intensity from higher orders, resulting in a monochromatic and linearly polarized beam with a spectral purity of approximately 99.98 %.Before each measurement the photon energy of the beam was calibrated with respect to the absorption edges of calibrated silicon and boron filters, installed at the beam line.With a wavelength precision of approximately 0.01 %, the beam line gives an energy resolution of approximately 0.02 eV.
Prior to each measurement, the samples was aligned with respect to the beam such that the beam was fully centered over the square aperture of the silicon nitride membrane and was not clipped by the silicon substrate.The measured transmittance is normalized with respect to the intensity of the direct beam as a function of photon energy, in order to correct for the fluctuations in the intensity of the photon beam.The normalized transmittance is then further normalized with respect to the transmittance of the bare Si 3 N 4 membrane substrate.The final normalized transmittance for photon energies from 50 eV to 1250 eV is shown in Fig. 5(a).
From 130 eV to 215 eV the combined absorption features due to electronic excitations from P2p and B1s core levels appear.A close-up of this phosphorus L 2,3 and boron K absorption edge feature is shown in Fig. 5(c).At an energy of 130 eV, the transmittance decreases sharply as incident photons have sufficient energy to excite electrons from the P2p core-level.Around 188 eV transitions of electrons from boron 1s core states are activated resulting in another sudden increase in absorption, and therefore a decrease in transmittance.For energies increasing beyond the two absorption edges, the transmittance monotonically increases towards unity, where the film almost becomes transparent.Over the entire photon energy range, the difference between the transmittance of the three studied samples was generally linearly proportional to the difference in the effective thicknesses of the c-BP film.This is what one would expect if the only difference between the samples were the c-BP film thickness, however there are exceptions, for example just below the phosphorus L 2,3edge, which is shown in Fig. 5(b).The thickest sample S3 shows an approximately monotonically increasing transmittance until the P L 2,3 absorption edge is encountered, which is the expected behaviour for a continuous homogeneous c-BP layer.In contrast, the transmittance of samples S1 and S2 shows a dip around 106 eV, which can be related to electron excitations from Si L 2,3 core-level states [30].Since these spectra are normalized with respect to the blank Si 3 N 4 reference membrane, these contributions from Si core states should have been normalized out, as is the case for sample S3.The reason that a signal is still visible in the transmission spectra of S1 and S2 indicates that the Si 3 N 4 of these samples was affected significantly by the deposition of the BP film, however since this would therefore also have to apply to S3, there has to be an additional cause.The most plausible explanation is that the deposition of the BP film on the Si 3 N 4 membrane perturbs its surface structure, which renders the usually stable Si 3 N 4 susceptible to oxidation in the case of S1 and S2, which, due to the discontinuous character of the BP film, leave certain areas of the Si 3 N 4 membrane exposed to atmospheric conditions.
Optical properties of materials in the soft x-ray and extreme ultraviolet energy range, such as transmission, are often simulated by approximating the absorption coefficient from tabulated scatter factors for the elemental constituents of the material.Using a single layer model for the normalized transmittance data of sample S3, we have fit the layer thickness, where we approximated the optical constants for stoichiometric c-BP through the tabulated scatter factors from Henke [31].The density of the BP film was set to 2.9 g cm −3 , the density of the pristine crystal structure near room temperature.The results are shown in Fig. 6.
The fit to the experimental data yielded a BP film thickness of 90.1 nm, which agrees well with the value estimated from the cross-section TEM studies presented in the previous section.Far away from the phosphorus and boron absorption edges, both at lower and higher energies, the simulated transmission using tabulated scatter factors agrees well with the experiment.However, the influence of local chemistry and structure on the absorption coefficient near an absorption edge is not taken into account in this simplistic model, which explains the significant discrepancy between the experimental data and the fit for photon energies close to the P L 2,3 and B K-edge.The close up image of this energy range shown in Fig. 6(b), shows that there is a lot of fine structure in the experimental spectroscopy that is not captured by the approximated absorption coefficient.The observed onset of the P L 2,3 absorption edge is also red-shifted by approximately 5 eV with respect to the value reported by Henke.These results highlight that optical constants in the soft x-ray and extreme ultraviolet can in principle be approximated from tabulated scatter factors, except near absorption edges.The true optical constants of c-BP could in principle be extracted from transmission measurements like those presented in this work, as has been done for a wide selection of materials in this energy range [16,[32][33][34][35][36].However, this approach requires the transmission spectroscopy of multiple homogeneous thin films of varying thickness with a negligible surface roughness on the order of a few nm, criteria that were obviously not satisfied by the thin films presented in this work.

X-ray absorption near-edge spectroscopy
The transmittance spectrum provides a picture of the loss channels for photons traveling through the c-BP layer, which includes absorption but also specular reflection and scattering.To isolate the absorption process, we looked at core level electron excitations through x-ray absorption near-edge spectroscopy (XANES).XANES measurements were carried out in the total electron yield mode at beamline 6.3.2 of the ALS.As the samples on the Si 3 N 4 membrane substrates would suffer from charging effects due to the insulating nitride layer, reference samples on crystalline Si(100) wafers were used for this measurement.The measured absorption spectra at the P L 2,3 and B K-edge, normalized with respect to the direct beam spectrum, are shown for all three samples in Fig. 7(a) and Fig. 7(b), respectively.All spectra are normalized such that the area under the curve, for the energy range that is plotted, is unity, which allows for meaningful comparison of relative feature intensity differences.In addition to the three thin c-BP films deposited for this work, the absorption spectrum of a reference sample is included, which was grown at comparable experimental conditions but at the optimal temperature for crystal growth of 1200 • C. The growth of these samples and the analysis of the x-ray absorption spectra have been described in detail elsewhere [27].At both the P L 2,3 and B K-edge, the three c-BP samples grown for this work, have comparable spectra and therefore exhibit similar differences with respect to the highly crystalline reference samples.By comparison, the features of the reference sample are "sharper" (i.e. the changes in absorption are larger within the same or a smaller energy window), which can be attributed to the higher degree of crystallinity of this c-BP layer [27].The effectiveness of both reflective and transmissive optics, such as multilayer mirrors and transmission filters, respectively, will be determined among others by the sharpness of the edge.This shows that if the c-BP thin films could be grown with a higher crystal quality, for example by alleviating the constraint of the lowered deposition temperature, the absorption features would be more strongly defined, leading to more effective optics.In this work, the maximally allowable deposition temperature was limited by the use of substrates containing a thin silicon nitride membrane, which was required for the recording of the transmission spectrum of the c-BP film.However, for the fabrication of real devices, such as transmission filters, other substrates exist that do not impose a restriction on the deposition temperature.

Discussion
In this work we investigate the applicability of c-BP for the manufacture of EUV optics, which can roughly be divided into reflective and transmissive optics.Both types of optics employ the sharp change in optical properties of a material that occurs near an absorption edge, in order to optimize the reflectivity or minize the transmissivity for photons with an energy within a specific range.Reflective optics such as multilayer mirrors require, in addition to this optical property, smooth thin films with a thickness on the order of nanometers and minimal surface roughness.The thin films discussed in this work, which are not thin nor smooth, do not meet either of these requirements, which is due to the employed deposition method of CVD, which is not suitable for growing thin films with these qualities.Techniques that áre commonly employed for this purpose, are physical vapor deposition methods such as magnetron sputtering, however, these techniques require high purity sputter targets of the desired material, which at the time of this research were not available for c-BP.However, several recent studies [37][38][39][40] have published promising results towards producing sintered sputtering targets, which would make magnetron sputtered thin films of c-BP, and therewith c-BP based reflective optics, a possibility.Compared to reflective optics, transmissive optics such as transmission filters do not impose as stringent requirements on film thickness and roughness.In the absence of a concrete application, there are no performance specifications available for a c-BP transmission filter, however, the measured transmittance shown in Fig. 6 clearly proves that the films produced in this work could be used as a transmission filter operating near the P L 2,3 -edge.An actual transmission filter would probably be thicker compared to the films presented in this work, to ensure that the target wavelength is almost fully suppressed.The large surface roughness of the films is irrelevant in the case of a transmission filter, as the transmittance of a film near the P L 2,3 -edge is reduced typically only by a factor of 10 −3 approximately, even for a surface roughness on the order of tens of nanometers.The oxidation observed by XPS will likely also not pose a problem, as c-BP is a highly chemically inert and stable compound [41] and the reported amount of oxide is not expected to increase over time.
The XANES analysis showed that the sharpness of the absorption edge is determined by the crystallinity of the material, where a higher degree of crystallinity will yield a sharper edge.As mentioned before, both reflective and transmissive optics rely on the sudden change in absorption across the edge and as such their performance will benefit from a sharper transition.For both types of applications the goal therefore should be to maximize the sharpness of the absorption edge.From the transmittance measurements, we observe that the actual transmission contrast across the P L 2,3 -edge is smaller than what is predicted by employing optical constants that are approximated from tabulated atomic scatter factors [31], as shown in Fig. 6.The reduced transmission contrast indicates that the actual extinction coefficient of the grown c-BP is smaller than predicted by tabulated scatter factors, which would mean that the predicted reflectivity values of c-BP based multilayer mirrors by Medvedev et al. [17], who employed the same approximated optical constants for their calculations, may be overestimated.Nevertheless, the demonstrated sharpness of the P L 2,3 -edge shows that it may still be exploited for the construction of a reflective multilayer mirror, provided that the c-BP thin films can be grown sufficiently thin and smooth.

Conclusions
We presented the growth of c-BP thin films on Si 3 N 4 membrane substrates, as a proof of principle of the exploitation of the P L 2,3 absorption edge for applications requiring a high transparency below that energy.Transmission electron microscopy analysis showed that the BP layers initially grow in islands to form discontinuous films that, given enough deposition time, will eventually form a continuous film.Regions with both polycrystalline and amorphous character were found.X-ray photoelectron spectroscopy measurements revealed that surface oxidation due to the exposure to atmospheric conditions causes oxidation primarily of the phosphorus species.Partly as a result of this effect, the B:P ratio was slightly non-stoichiometric, with a surplus of boron.
Transmittance measurements of the composite c-BP/Si 3 N 4 were conducted, which showed a strong absorption feature from 130 eV to 300 eV resulting from the merging of the P L 2,3 , P K and B K absorption edges.Comparisons of x-ray absorption near-edge spectroscopy measurements on highly crystalline c-BP reference samples showed that an increase in crystal quality sharpens absorption features.This result implies that if the c-BP films could be grown at the ideal deposition temperature, instead of the lowered temperature used in this work, the resulting quality in crystal structure would be higher, yielding a larger transmission contrast across the absorption edge.We have shown by proof of principle that, through c-BP, the P L 2,3 absorption edge can be exploited to construct optical elements, such as transmission filters or reflective optics.Despite the fundamental applicability of c-BP however, the morphology of the grown films resulting from the CVD method, characterized by a high surface roughness, precludes them from being applied in x-ray reflective optics, such as multilayer mirrors, which require extremely thin and smooth layers.We discussed how physical vapor deposition based techniques may be employed to overcome this technical issue.Since the transmittance of a thin film is less susceptible to its surface roughness, CVD grown c-BP could certainly be used for the construction of transmission filters.

Fig. 2 .
Fig. 2. From left to right, transmission electron microscopy recordings of the cross-section of samples (a) S1, (b) S2 and (c) S3.The images show a layered structure, where the bottom layer corresponds to residue created by the focused ion beam process in creating the cross-section of the sample.The layer directly on top of that is the platinum layer (black) that was deposited on the Si 3 N 4 membrane, which can be seen directly above the Pt layer.On top of the relatively smooth Si 3 N 4 layer, one can see the deposited c-BP layer, which was much rougher compared to the previous layers.The penultimate visible layer is yet another Pt-layer (black) that was deposited prior to the FIB milling process, on top of which another layer, of the residue this process created, can be seen.There are visible differences between the c-BP layer of the three different samples.Where sample S3 has a c-BP layer that fully covers the silicon nitride membrane underneath, the phosphide layer for both S1 and S2 contains holes that expose the underlying nitride, as marked by the red squares.

Fig. 3 .
Fig. 3. (a) Crystalline grain in sample S2 of approximately 10 nm by 30 nm.Panel (b) shows a further magnification of the crystalline grain.The dashed white line indicates the interface between the Si 3 N 4 membrane and the grown c-BP film.

Fig. 4 .
Fig. 4. Two cross-section transmission electron micrograph recordings of sample S3.The dashed white lines indicate the interfaces between the Pt, Si 3 N 4 and c-BP layers.The dashed red circles approximately highlight the location of the regions in the c-BP layer with an amorphous character.The apparent color difference between (a) and (b) is due to the usage of a relative grayscale and the fact that the Pt layer is only visible in (a).

Fig. 5 .
Fig. 5. (a) Normalized transmittance of the three samples with different c-BP film thickness for photon energies from 50 eV to 1250 eV.(b) A close up view of the absorption feature around 106 eV that appears in the transmittance spectrum for samples S1 and S2, that can be attributed to Si2p excitations from the exposed Si 3 N 4 membrane under the discontinous c-BP film.(c) Close up view of the feature due to the P L 2,3 and B K absorption edge fine structure.

Fig. 6 .
Fig. 6.(a) Transmission spectroscopy data for sample S3 (solid red line) and the simulated transmission for a single layer BP model where the absorption coefficient is approximated from tabulated scatter factors.(b) Close up of the same data shown in (a) near the absorption edges.

Fig. 7 .
Fig.7.Normalized x-ray absorption spectra for the (a) phosphorus L 2,3 -edge and the (b) boron K-edge for the three c-BP samples on silicon wafer substrates, as obtained in the total electron yield mode.The dashed gray line represents the spectrum for a reference sample that was deposited under similar experimental conditions but at the ideal temperature of 1200 • C.

Table 1 .
Experimental conditions during deposition of the three samples studied in this work, with flowrates of precursor gases in cm 3 /min, total pressure p in torr, temperature T in • C and deposition time t in s.

Table 2 .
Elemental abundances for each of the three samples as determined from the experimental XPS analysis.