In this study, we demonstrate a tolerant and durable Cr/Ni bilayer metal etch mask that allows us to realize approximately 150:1 etch selectivity to diamond. This result is achieved through the use of a very thin initial Cr layer of <10 nm thickness as part of the bilayer metal mask, which results in five to ten times improved selectivity than thick single metal layer masks or bilayer masks with thicker combinations. A finite element analysis was employed to design and understand the physics and working mechanism of the bilayer metal masks with different thicknesses. Raman spectroscopy and energy-dispersive x-ray spectroscopy on the diamond surface were also performed to investigate the changes in diamond quality before and after the deep diamond etching and found that no noticeable etch damage or defects were formed. Overall, this mask strategy offers a viable way to realize deep diamond etching using a high heat and chemistry tolerant and durable bilayer metal etching mask. It also offers several technological benefits and advantages, including various deposition method options, such as sputtering and physical vapor deposition, that can be used and the total thinness of the bilayer metal mask required given the higher selectivity allows us to realize fine diamond etching or high-aspect ratio etching, which is a critical fabrication process for future power, RF, MEMS, and quantum device applications.

Diamond as a semiconductor for electronic, optoelectronic, and optical applications has gained attention due to its exceptional material properties, such as high breakdown electric field, high carrier mobility, and broad optical transparency.1–8 Recently, a negatively charged nitrogen-vacancy (NV) property in diamond has been used for biosensing, magnetometry, and quantum computing.9–12 However, device applications are severely hampered by the extreme hardness and chemical inertness of diamond, which make sophisticated microfabrication processes difficult. Among the challenges associated with diamond fabrication, etching for structural definition, isolation, or any other feature is especially difficult because of the severe demands on the required etch masks. For this reason, most diamond etching studies have mainly focused on enhancing etching chemistry by tuning a mixture of etching gases together with processing pressure and power.13–20 Through these efforts, a robust diamond etching process has been optimized using Inductively Coupled Plasma—Reactive Ion Etcher (ICP-RIE) based on argon, carbon tetrafluoride, or oxygen gases.13–20 However, these ICP etching processes require a substantially thick mask due to physical damage caused by a required excessively high ICP power during the etching process. To minimize physical damage caused by the energized plasma during the etching process, various inorganic etching masks, such as Al, Al2O3, SiO2, or Ti, have been used. For example, Tran et al. used an Al mask and achieved an etch rate of 121–189 nm/min with 35 – 56 selectivity ratios.13 Tran et al. also employed SiO2 or Ti etching masks and achieved etch rates of 123 and 114 nm/min with a selectivity of 8 and 12, respectively.13 Golovanov et al. employed an Al2O3 etching mask and achieved an etch rate of 83 nm/min with an etching selectivity of 2–3.21 

However, these conventional single layer etching masks have several issues. First, ideal masking materials should have low thermal expansion coefficient, high thermal conductivity, and high density (at least higher than the density of diamond) in order to act as an effective mask. However, most of the popular diamond etching mask materials fail to satisfy two or three of these desired material properties. For example, Al2O3 has high density but poor thermal conductivity, and Ti has good thermal conductivity and thermal expansion coefficient but poor density. For these reasons, a conventional diamond etching mask shows poor etching selectivity (typically less than 1:20); thus, a several micrometer thick metal etching mask is required to realize deep diamond etching. Also, this aggressive diamond etching process generates an excessive amount of heat during etching, thus resulting in mechanical failure that is typically observed as pits or cracks on the etching mask. Furthermore, such mechanical failures on the etching mask often substantially degrade the mask integrity, even with a much thicker etching mask; thus, it causes the entire diamond etching process to fail or yield poorly. Second, thermal expansion coefficient values of most masking materials are several times larger than that of diamond; thus, the masking layer experiences a high degree of thermally induced strain, which hinders etches with long times.

In this study, we demonstrate a bilayer metal etching mask that allows us to realize about 1:150 etching selectivity ratios to diamond. In order to realize a tolerant and mechanically durable metal mask with high etching selectively, Cr and Ni were chosen by considering their thermal expansion coefficient, thermal conductivity, and density and then optimized to maximize the etching performance. A finite element analysis was employed to design and understand the physics and working mechanism of the bilayer diamond metal etching masks with different thicknesses. Raman spectroscopy and energy-dispersive x-ray spectroscopy (EDX) on a diamond surface were also performed to investigate the changes in diamond quality before and after the deep diamond etching, and we found no noticeable etch damage or formation of defects. Overall, our bilayer metal etching mask was designed to optimize various aspects of masking materials, such as etching chemistry, heat dissipation, and mechanical strain; thus, it provides a thin but highly tolerable diamond etching mask with the 1:150 etching selectivity that is suitable for deep diamond etching. This bilayer etching mask also offers several technological benefits and advantages in that various deposition methods, such as sputtering and physical vapor deposition of the metals, can be used, and the thinness of the bilayer metal etching mask allows us to realize fine diamond etching or high-aspect ratio diamond etching, which is a critical fabrication process for future power, RF, MEMS, and quantum device applications.

As shown in Fig. 1, a bilayer etching mask consists of two parts. The bottom Cr layer that has low thermal expansion and high thermal conductivity provides good adhesion and efficient heat dissipation to a diamond substrate and minimizes thermally induced strain. Cr is also an ideal choice because it is widely used on diamond for other fabrication processes and is well-understood from the adhesion and deposition standpoint. The top Ni layer has high density and high thermal conductivity that provides high etching resistance and tolerance against thermal and plasma damage as well as efficient heat dissipation during the etching process. In this bilayer configuration, Cr has the small thermal expansion coefficient [4 μm/(m K) at 25 °C] and high thermal conductivity [93.9 W/(m K)], which is comparatively close to diamond among most of the available metals [the thermal expansion coefficient of diamond is 1 μm/(m K)]; thus, this bilayer etching mask can efficiently dissipate heat to the diamond layer without causing any mechanical failure due to a thermal mismatch during the etching process. The top Ni layer shows superior etching tolerance compared to other etching mask materials that are commonly used for a diamond etching mask. For example, the Ni layer shows twice the density (8.908 g/cm3) and four times higher thermal conductivity [90.9 W/(m K)] than Ti [4.506 g/cm3 and 21.9 W/(m K)]. In this experiment, Cr and Ni bilayer etching masks were photolithographically defined, and 5 nm thick Cr and 150 nm thick Ni were deposited using an electron-beam evaporator, followed by a lift-off process. Diamond samples were etched using a Trion Phantom RIE-ICP etcher using 2 sccm of CF4 and 8 sccm O2 gases with 150 W and 900 W of RIE and ICP power, respectively [Figs. 1(c) and 1(d)]. After completion of the etching process, the bilayer metal mask was removed by Cr and Ni etchants (chromium and nickel etchants from Transene Company).

FIG. 1.

Schematic illustration of forming a Cr/Ni bilayer etching mask and following a deep diamond etching process. (a) Cr layer deposition, (b) Ni layer deposition, (c) plasma dry etching, and (d) a metal removal process.

FIG. 1.

Schematic illustration of forming a Cr/Ni bilayer etching mask and following a deep diamond etching process. (a) Cr layer deposition, (b) Ni layer deposition, (c) plasma dry etching, and (d) a metal removal process.

Close modal

When it comes to the design of a bilayer etching mask, it is crucial to consider effective heat dissipation from the etching mask in order to reduce thermally induced strain between the metal mask and the diamond substrate because most failures in the diamond etching process typically occur when a large temperature mismatch between the metal and diamond exists due to high energy plasma bombardment during the etch. Therefore, a finite element analysis (comsol Multiphysics simulator) was used to investigate the optimal thicknesses of the bilayer etch mask. As shown in Fig. 2(a), the simulated structures in the finite element analysis consisted of the Cr/Ni bilayer stack on top of the diamond substrate layer. To understand the impact of the Cr layer thickness on the strain distribution of the entire structure, only the Cr layer thickness in the Cr/Ni bilayer stack was varied from 5 to 50 nm, while all other dimensional parameters, such as the thickness of the Ni layer in the Cr/Ni bilayer stack and the diamond layer, were fixed to 150 nm and 20 μm, respectively. Regarding the initial thermal boundary condition of the simulated structure, the diamond substrate was set to 17 °C, and the top surface of an Ni layer was set to various temperatures ranging from 200 to 400 °C to mimic various elevated surface temperature conditions of the top surface of the Cr/Ni bilayer stack during the etching process. In addition, a denser mesh near the Cr and diamond interface was intentionally created from 200 to 20 nm in width and from 5 to 0.4 nm in height to improve the accuracy of a thermally induced strain of the structure. Finally, the time-dependent solver in comsol was employed in the simulation together with the “Heat Transfer in Solids” and the “Solid Mechanics” modules to simultaneously simulate changes in the strain distribution of the Cr/Ni bilayer stack on the diamond layer under different temperature conditions with respect to the wide range of time frame (1 ms to 10 s). After the simulation, the heat transfer equations [Eqs. (1) and (2)] were then used to calculate the changes of strain as a function of time,

ρCp(Tt+utransT)+q=αT:dSdt+Q,
(1)
q=kT,
(2)

where ρ is the density, T is the temperature, μtrans is the velocity vector of translational motion, q is the local heat flux by conduction, α is the coefficient of thermal expansion, S is the second Piola–Kirchhoff stress tensor, and Q is the additional heat source. Based on the equations above, the strain distribution of a bilayer structure over time under different temperatures was obtained. It should be noted that the Cr layer thickness was only changed in this simulation because the heat dissipation capability and corresponding thermally induced strain depend largely on the phonon mean free path of the Cr layer, which is proportional to the thickness of the Cr layer. To accurately analyze the strain distribution of the bilayer structure, the strain changes at multiple positions [marked as (1)–(3) in Fig. 2(a)] were traced. Figure 2(b) traces the total strain of the entire bilayer etch mask, and Figs. 2(c)2(e) trace the local strain at the Cr layer, the diamond-Cr interface, and the Cr–Ni interface, respectively. From the strain in the entire structure [Fig. 2(b)], it was found that the strain increases as the Ni surface temperature increases, which happens in the actual etching process. Given the processing conditions, however, it was expected that the maximum Ni surface temperature did not exceed 200 oC because the stage temperature of the ICP etcher was constantly cooled at 17 oC during the etching process. It was also found that the Ni surface temperature is the only major variable that decided the strain of the entire structure because the simulated strain levels with different Cr layer thicknesses almost remained in a similar range as shown in Fig. 2(b). However, the Cr layer thickness plays a more critical role in the strain distribution at the Cr/diamond and Ni/Cr interfaces. As shown in Figs. 2(c) and 2(e), the strain changes dramatically (over 0.5%) when the Cr layer thickness increased from 5 to 50 nm. Such differences in strain are attributed to the difference in thermal conductivity between diamond-to-Cr and Cr-to-Ni. In other words, the thinner the Cr layer, the more effective heat dissipation from the Ni layer is; thus, the thinner Cr layer prevents the bilayer metal etching mask from experiencing a smaller degree of thermally induced strain to the diamond substrate as shown in Fig. 2(d). The requirement of very thin Cr layers as part of the metal stack is a key result of this work, and we are unaware of other attempts to use very thin metal interlayers to achieve high-integrity metal etch masks for diamond.

FIG. 2.

(a) Typical simulation structure of the bilayer etching mask on a diamond substrate. Simulated strain values as a function of time and temperature taken (b) from the entire bilayer etching mask, (c) from the Cr layer, (d) at the Cr/diamond interface, and (e) at the Ni/Cr interface, respectively.

FIG. 2.

(a) Typical simulation structure of the bilayer etching mask on a diamond substrate. Simulated strain values as a function of time and temperature taken (b) from the entire bilayer etching mask, (c) from the Cr layer, (d) at the Cr/diamond interface, and (e) at the Ni/Cr interface, respectively.

Close modal

To experimentally verify the simulated bilayer structures, two samples with different bilayer etching masks (a 150 nm Ni layer either with a 5 nm or a 50 nm thick Cr layer) were prepared. After this step, a 20 μm depth deep diamond etch was performed using the ICP etcher. As described previously, the ICP-RIE etcher was used with a high ICP power (900 W) and CF4 and O2 gases for 160 min. It should be noted that multiple identical diamond samples that have the same Cr/Ni etching mask stack were used, and each sample was etched continuously for a specific amount of time, namely, 40, 70, 120, and 160 min to prevent any possibility of different etching rates that are caused by different surface temperatures when the etching process is paused during the etching process. Figure 3(a) shows the thickness of the diamond and the Ni layer removed by the ICP-RIE etch process as a function of time. Both diamond samples that have either a 5 nm- or a 50 nm-Cr layer demonstrated a consistent etching rate of 126 nm/min regardless of Cr layer thickness as derived from Fig. 3(a) (top). Figure 3(a) (bottom) shows the Ni consumption during diamond etching and indicates that the Ni layer was consistently consumed at a rate of ∼0.85 nm/min. With these etch rates from diamond and the Ni layer, the etching selectivity was calculated to ∼150 as shown in Fig. 3(b), which is the highest diamond-etching mask etching selectivity reported to date.

FIG. 3.

(a) (top) Measured diamond etching depth and (bottom) Ni etching depth as a function of etching time and (b) calculated etching selectivity as a function of etching time.

FIG. 3.

(a) (top) Measured diamond etching depth and (bottom) Ni etching depth as a function of etching time and (b) calculated etching selectivity as a function of etching time.

Close modal

Figures 4 and 5 present the details of the etching regions of two samples with different bilayer etching masks (a 150 nm Ni layer with either a 5 or 50 nm thick Cr layer). Figure 4(a) shows the cross section of the sample. Figures 4(b) and 4(c) show microscopic images taken after 40 min (∼5 μm etching) and 160 min etching (20 μm etching). The dark edge from the etched sample indicates the angled sidewall, which is measured to 84° based on angled scanning electron microscope (SEM) images in Figs. 4(d) and 4(e). The masked region had no visual damages and had a clean diamond sidewall. Also, the clean and smooth etched diamond surface confirmed that no redeposition occurred, which is also confirmed by the EDX spectra in Fig. 6(b). In contrast, the control sample that had a 50 nm thick Cr layer with a 150 nm Ni layer [Fig. 5(a)] clearly shows a rough surface on the covered region. The masked region started to form craterlike defects after 40 min of etching as indicated in Fig. 5(b) and became completely dark after 160 min of etching. As shown in Figs. 5(d) and 5(e), the angled SEM images reveal that the dark diamond surface was completely damaged, which indicates a failed etching mask. Based on these time-lapse observations, it can be speculated that the numbers of defects increased due to the increase in thermally induced strain. To compare the performance of the Cr/Ni bilayer etching mask with other metal etching mask, a 150 nm-thick Al etching mask was formed on a diamond substrate, and we repeated the same etching test. Similar to the 50/150 nm Cr/Ni etch mask, the Al etching mask failed before the diamond samples was etched for 60 min.

FIG. 4.

(a) Cross-sectional illustration of the sample with a 5 nm thick Cr layer. Microscopic images of the sample (b) before and (c) after 20 μm deep etching. The scale bar is 50 μm. (d) An angled and (e) zoomed-in SEM image of the masked region after removing the bilayer metal etching mask.

FIG. 4.

(a) Cross-sectional illustration of the sample with a 5 nm thick Cr layer. Microscopic images of the sample (b) before and (c) after 20 μm deep etching. The scale bar is 50 μm. (d) An angled and (e) zoomed-in SEM image of the masked region after removing the bilayer metal etching mask.

Close modal
FIG. 5.

(a) Cross-sectional illustration of the sample with a 50 nm thick Cr layer. Microscopic images of the sample (b) before and (c) after 20 μm deep etching. The scale bar is 50 μm. (d) An angled and (e) zoomed-in SEM image of the masked region after removing the bilayer etching mask.

FIG. 5.

(a) Cross-sectional illustration of the sample with a 50 nm thick Cr layer. Microscopic images of the sample (b) before and (c) after 20 μm deep etching. The scale bar is 50 μm. (d) An angled and (e) zoomed-in SEM image of the masked region after removing the bilayer etching mask.

Close modal
FIG. 6.

(a) Raman spectra taken from (top) the masked region and (bottom) the etched region. (b) EDX spectra showing strong carbon peaks without any Cr (0.6 keV) or Ni (0.9 keV) peaks.

FIG. 6.

(a) Raman spectra taken from (top) the masked region and (bottom) the etched region. (b) EDX spectra showing strong carbon peaks without any Cr (0.6 keV) or Ni (0.9 keV) peaks.

Close modal

To further investigate the impact of deep diamond etching with the bilayered metal etching mask, Renishaw Raman spectroscopy was performed on the masked region and the etched region using Raman spectroscopy with a 532 nm laser and a 50× objective lens. Raman spectroscopy was performed after all diamond samples were etched about 10 μm. As shown in Fig. 6(a), both samples show an sp3 Raman mode at 1332 cm−1 from the masked region or the etched region indicating that no noticeable strains are existed.22 We also examined the full-width half maximum (FWHM) values of each sample because it is known that the FWHM value of diamond is an indicator of the existence of defects in diamond.23–25 The FWHM values of each sample remained the same as 3.4–3.6 cm−1, suggesting that no additional defects were formed by the deep diamond etching process with the bilayer metal etching mask. EDX was also performed to check the chemical residue of the Cr/Ni metal mask and their redeposition during etching using a Hitachi S4000 field emission scanning electron microscope (FESEM). Figure 6(b) shows the chemical analysis of the masked region and the etched region from energy dispersive spectra, which indicates the presence of a strong carbon peak, but no other elements, such as Cr (as indicated in a white line at 0.6 keV) and Ni (as indicated in a pink line at 0.9 keV), were introduced during the etching process. The small peak at 0.7 keV on the EDX plot was identified to be fluorine (F), which is different from Cr (0.6 keV) or Ni (0.9 keV).

In conclusion, we demonstrate a process-tolerant and mechanically durable Cr/Ni bilayer metal etching mask that allows us to realize about 150 of etching selectivity to diamond. The crucial element of this metal mask stack is the use of a very thin Cr layer on diamond. A finite element analysis was employed to design and understand the physics and working mechanism of the bilayer diamond metal etching masks with different thicknesses. Raman spectroscopy and EDX on a diamond surface were also performed to investigate the changes in diamond quality before and after the deep diamond etching and found that no noticeable damages or defects were formed. Overall, our bilayer metal etching mask offers a viable way to realize deep diamond etching at far thinning overall metal mask thicknesses than previously reported or in common practice. This bilayer etching mask also offers several technological benefits and advantages, including the flexibility to use various deposition methods, such as sputtering and physical vapor deposition for the metal layers, and the thinness of the overall bilayer metal etching mask allows us to realize fine diamond etching or high-aspect ratio diamond etching, which is a critical fabrication process for many device applications that have previously been difficult or impossible in single crystal diamond.

This work was supported by the National Science Foundation (Grant No. ECCS-1809077) and partially by the internal funding from Fraunhofer-Gesellschaft.

The authors have no conflicts to disclose.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

1.
S. J.
Cho
 et al,
AIP Adv.
10
,
125226
(
2020
).
2.
H.
Umezawa
,
K.
Ikeda
,
R.
Kumaresan
,
N.
Tatsumi
, and
S.
Shikata
,
IEEE Electron Device Lett.
30
,
960
(
2009
).
3.
T.
Teraji
,
Y.
Garino
,
Y.
Koide
, and
T.
Ito
,
J. Appl. Phys.
105
,
126109
(
2009
).
4.
J.
Isberg
,
J.
Hammersberg
,
E.
Johansson
,
T.
Wikström
,
D. J.
Twitchen
,
A. J.
Whitehead
,
S. E.
Coe
, and
G. A.
Scarsbrook
,
Science
297
,
1670
(
2002
).
5.
H. K.
Mao
and
R. J.
Hemley
,
Nature
351
,
721
(
1991
).
6.
Y.
Zheng
,
E.
Swinnich
, and
J.-H.
Seo
,
Nano Express
1
,
030010
(
2020
).
7.
Y.
Zheng
,
E.
Swinnich
, and
J.-H.
Seo
,
ECS J. Solid State Sci. Technol.
9
,
055007
(
2020
).
8.
M.
Kim
,
J.-H.
Seo
,
U.
Singisetti
, and
Z.
Ma
,
J. Mater. Chem. C
5
,
8338
(
2017
).
9.
R.
Schirhagl
,
K.
Chang
,
M.
Loretz
, and
C. L.
Degen
,
Annu. Rev. Phys. Chem.
65
,
83
(
2014
).
10.
L.
Rondin
,
J. P.
Tetienne
,
T.
Hingant
,
J. F.
Roch
,
P.
Maletinsky
, and
V.
Jacques
,
Rep. Prog. Phys.
77
,
056503
(
2014
).
11.
C.
Santori
,
P. E.
Barclay
,
K. M. C.
Fu
,
R. G.
Beausoleil
,
S.
Spillane
, and
M.
Fisch
,
Nanotechnology
21
,
274008
(
2010
).
12.
N.
Zhao
,
J.-L.
Hu
,
S.-W.
Ho
,
J. T. K.
Wan
, and
R. B.
Liu
,
Nat. Nanotechnol.
6
,
242
(
2011
).
13.
D. T.
Tran
,
C.
Fansler
,
T. A.
Grotjohn
,
D. K.
Reinhard
, and
J.
Asmussen
,
Diam. Relat. Mater.
19
,
778
(
2010
).
14.
D. S.
Hwang
,
T.
Saito
, and
N.
Fujimori
,
Diam. Relat. Mater.
13
,
2207
(
2004
).
15.
G. S.
Sandhu
and
W. K.
Chu
,
Appl. Phys. Lett.
55
,
437
(
1989
).
16.
F. K.
de Theije
,
O.
Roy
,
N. J.
van der Laag
, and
W. J. P.
van Enckevort
,
Diam. Relat. Mater.
9
,
929
(
2000
).
17.
Y.
Ando
,
Y.
Nishibayashi
,
K.
Kobashi
,
T.
Hirao
, and
K.
Oura
,
Diam. Relat. Mater.
11
,
824
(
2002
).
18.
P. K.
Bachmann
,
D.
Leers
, and
D. U.
Wiechert
,
Diam. Relat. Mater.
2
,
683
(
1993
).
19.
M.-L.
Hicks
,
A. C.
Pakpour-Tabrizi
, and
R. B.
Jackman
,
Sci. Rep.
9
,
15619
(
2019
).
20.
W.
Smirnov
,
J. J.
Hees
,
D.
Brink
,
W.
Müller-Sebert
,
A.
Kriele
,
O. A.
Williams
, and
C. E.
Nebel
,
Appl. Phys. Lett.
97
,
073117
(
2010
).
21.
A. V.
Golovanov
,
V. S.
Bormashov
,
N. V.
Luparev
,
S. A.
Tarelkin
,
S. Y.
Troschiev
,
S. G.
Buga
, and
V. D.
Blank
,
Phys. Status Solidi A
215
,
1800273
(
2018
).
22.
J.-H.
Seo
 et al,
J. Appl. Phys.
119
,
205703
(
2016
).
23.
T.
Bauer
,
M.
Schreck
,
H.
Sternschulte
, and
B.
Stritzker
,
Diam. Relat. Mater.
14
,
266
(
2005
).
24.
D. S.
Knight
and
W. B.
White
,
J. Mater. Res.
4
,
385
(
1989
).
25.
D. N.
Jamieson
,
S.
Prawer
,
K. W.
Nugent
, and
S. P.
Dooley
,
Nucl. Instrum. Methods Phys. Res., Sect. B
106
,
641
(
1995
).