Patent Grant
7235798
The present invention relates to a focused ion beam (FIB) apparatus.
At present, focused ion beam apparatuses are used in many fields. In particular, a fine focused beam can be applied to a micro area of devices and materials. Furthermore, this apparatus focuses an ion beam from an ion source through a lens and irradiates the beam onto a specimen, and the apparatus is used for milling and observing a specimen in a micro area. For example, it is possible to use the apparatus for observation of a specimen by irradiating an ion beam of a relative low current onto the specimen and detecting secondary particles generated from the specimen. Furthermore, it is possible to mill a specimen by irradiating an ion beam of a relatively high current onto the specimen.
In addition, the focused ion beam is used for structure analysis and failure analysis of microelectrical mechanical systems (MEMS) and semiconductor devices. These elements have been integrated in recent years. In general, therefore, they have a layered structure. For this reason, to inspect them, it is necessary to conduct cross-sectioning up to a specified layer and inspect the cross-section structure. As the MEMS and semiconductor devices become finer, their structures also become gradually complicated. Therefore, the number of cross sections to be inspected increases more and more, and the time which can be used for inspection per cross section becomes shorter and shorter. Since the focused ion beam is applicable to both milling and observation, it is effective to the structure analysis and failure analysis of the MEMS and semiconductor devices. First, the current of the ion beam is increased and milling is conducted from the surface down to a specified layer. Then, the current of the ion beam is decreased and the specified layer is inspected.
For example, a conventional apparatus described in U.S. Pat. No. 5,852,297 has an optical system in which a total optical length from apex of an ion emitter of an ion source to a specimen is in the range of 300 to 450 mm. The distance between the ion emitter and the center of a condenser lens center is 45 mm or less. The distance between the objective lens center and the specimen is 40 mm or less. A FIB formed by this optical system has a maximum current density Jmax of at least 15 A/cm2. As for a fine milling beam in a milling mode, a beam current Ip is Ip≧several tens pA and a beam diameter d is d≦40 nm. As for an observation beam in an observation mode, Ip≧several pA and d≦15 nm.
In the focused ion beam apparatus, the milling position precision substantially depends on the beam diameter d in the milling mode. This results in a problem that, if d is not small enough, the apparatus cannot be suited for the fine structure and features in the structure analysis and failure analysis of the MEMS and semiconductor devices. On the other hand, the milling speed is substantially proportionate to the beam current Ip. If Ip is not large enough, the milling speed becomes slow. For meeting the needs of still higher throughput of the FIB milling in the structure analysis and failure analysis, therefore, it has become a subject to increase the beam current density, i.e., increase the beam current without making the beam diameter of the milling beam large. In the observation mode, on the other hand, the focused ion beam apparatus has a problem that, if the ion beam is not fine enough, it is not possible to observe the structure and features of fine MEMS and semiconductor devices in the structure analysis and failure analysis. Another problem in the observation mode is that, if the current of the ion beam is not large enough, the obtained signal is too small to detect an image with a sufficient signal-to-noise ratio. As semiconductors become finer, it has become a subject to make the observation beam still finer.
The above-described conventional apparatus example does not meet the needs of still higher throughput in FIB milling especially on the high current side. For example, in 30-kV Ga-FIB in the conventional apparatus, a beam having d≈1 μm is associated with Ip≈16 nA. It is now supposed that box milling with a length 20 μm, a width 20 μm and a depth 20 μm is conducted on the surface of a Si specimen by using this beam. The milling yield Y of a Si specimen using 30 kV Ga-FIB depends on the scanning velocity of the FIB, and it is the range of 0.2 to 0.8 μm3/nA·s. Under the condition Y=0.25 μm3/nA·s, the time taken for box milling amounts to approximately 17 minutes. Therefore, it is not possible to meet the needs of higher throughput in the structure analysis and failure analysis of the MEMS and semiconductor devices for conducting a lot of milling of this kind in a short time.
An object of the present invention is to provide an ion beam apparatus that meets the needs of higher throughput and a finer observation beam especially on the high current side.
The present inventors have repeated analyses and experiments with respect to the beam diameter d and the ion beam current Ip of the FIB, for the purpose of the structure analysis and failure analysis of the MEMS and semiconductor devices. As a result, the present inventors have found that the following conditions must be satisfied by the FIB. That is, the beam current Ip must be at least 20 nA for the milling mode on the high current side. The beam diameter d in the specimen position must be 0.5×(Ip/20 nA)3/2 [μm] or less. However, the maximum value of d must not exceed 3 μm. For example, the condition expression d≦0.5×(Ip/20 nA)3/2 [μm] yields d≦0.5 μm when Ip=20 nA, and Ip≧31.7 nA when d=1 μm. The condition that the maximum value of d does not exceed 3 μm has been introduced from the fineness of the subject MEMS and semiconductor devices. As for the maximum current density Jmax of the milling beam, approximately 50 nA/cm2 is needed. On the other hand, as for the observation mode, a FIB having Ip≧several pA and d≦15 nm is needed in the same way as the conventional apparatus example. As for the minimum value dmin of d, however, the present inventors have found that a FIB having dmin≦15 nm and a beam current Ip of at least 0.001 nA is required.
To achieve the above-described numerical values, that is, to realize the failure analysis of semiconductors which is the object of the present invention, a focused ion beam apparatus including a condenser lens and an objective lens to accelerate an ion beam emitted from an ion source and focus the ion beam onto a specimen, the condenser lens including at least a first electrode supplied with an extraction voltage and a second electrode supplied with an earth potential is formed according to the present invention so as to cause a distance from the emitter apex of the ion source to the second electrode included in the condenser lens to be in the range of 5 to 14 mm.
According to the above-described configuration, a FIB having a beam diameter d in the specimen position which is 0.5×(Ip/20 nA)3/2 [μm] or less when the beam current Ip is at least 20 nA can be formed for the milling mode on the high current side. In other words, in the FIB, d≦0.5 μm when Ip=20 nA, whereas Ip≧31.7 nA when d=1 μm. Furthermore, as for the maximum current density Jmax of the milling beam, 50 A/cm2 can be achieved. For example, when conducting box milling with a length 20 μm, a width 20 μm and a depth 20 μm on the surface of a Si specimen by using a FIB with d=1 μm and Ip=32 nA, the box milling time can be shortened to approximately 8.5 minutes, which is approximately half of the milling time in the conventional technique. On the other hand, in the observation mode, a FIB with dmin≦6 nm, where dmin is the minimum value of d, and the beam current Ip being at least 0.001 nA can be formed. As a result, it becomes possible to conduct the structure analysis and failure analysis of the MEMS and semiconductor devices with high throughput and high precision.
It is described in JP-A-2002-251976 that the distance from the liquid metal ion source to the condenser lens is set to 10 mm or less. However, its purpose is to prevent extracted ions spread with the distance from the ion source from striking against electrodes of the condenser lens (outside portion of the center hole). If extracted ions strike against the condenser lens, the place is shaved by sputtering, resulting in a shortened apparatus lifetime. Since this known example is different in purpose from the present invention, potential information of the condenser lens electrodes is not mentioned at all.
According to the present invention, the structure analysis and failure analysis of the MEMS and semiconductor devices can be implemented with high throughput.
Other objects, features and advantages of the invention will become apparent from the following description of the embodiments of the invention taken in conjunction with the accompanying drawings.
Hereafter, an embodiment of the present invention will be described with reference to the drawings.
As schematically shown in
Mode changes caused by condenser lens switching will now be described. Two different focusing modes are available by switching the potential V1 at the third electrode 2_3 in the condenser lens; that is (a) high current density beam mode for FIB milling and (b) fine beam mode for image observation using a scanning ion microscope (SIM). In the milling mode, the lens potential V1 is further adjusted every beam diameter (every aperture diameter of the beam limiting aperture 3). In
A conventional technique will now be described to clarify the embodiment of the present invention. The conventional apparatus has a lens configuration similar to that shown in
(1) The maximum current density Jmax of the milling beam is approximately 15 A/cm2.
(2) The beam diameter d of a finishing milling beam having Ip=several tens pA is d≦40 nm.
(3) In the fine beam (observation) mode, a beam with d≦15 nm can be formed at Ip≧several pA.
(4) As for the curve of milling mode Conv. (Mill.), a beam diameter ds caused by spherical aberration becomes predominant in the beam diameter d on a large current side of Ip≧20 nA. The beam diameter d can be approximated by using a proportional equation of the one and one-half power with respect to Ip, i.e., d=K·Ip1.5 where K is a proportionality constant. In the full logarithmic scale graph, this approximation equation becomes a straight line having a gradient equal to 1.5 (as described later). When μm and nA are respectively taken as units of d and Ip, the value of the proportionality constant K in the proportional equation is always greater than 0.5. (In other words, for example, a beam at Ip=20 nA already has d>0.5 μm.) In
The FIB apparatus according to the present invention provides beams of the following two kinds. In the milling mode, the FIB apparatus provides a beam that makes the K value equal to 0.5 or less on the large current side with Ip≧20 nA. In the observation mode, the FIB apparatus provides a beam with dmin≦6 nm and Ip≧0.001 nA, where dmin is the minimum value of d.
An outline of the ion optical system in the FIB apparatus which is the embodiment of the present invention will now be described with reference to
The optical length and potentials are fixed. The beam emission semi-angle on the incidence side of the condenser lens used to form the FIB is denoted by α∘. In general, the beam diameter d is represented by the following equations using a Gaussian image diameter dg, a beam diameter dc due to chromatic aberration, and a beam diameter ds due to spherical aberration.
d2=dg2+dc2+ds2 (1)
dg=MtDs (2)
dc=Hc·α∘·ΔV (3)
ds=(½)Hs·α∘3 (4)
Hc=Mt·Cc/Ve (5)
Hs=Mt·Cs (6)
Here, ΔV is a voltage converted from an energy width ΔE of extracted ions. Ve is an ion extraction voltage (with the emitter potential taken as the origin). Ds is a virtual source size (≈50 nm) of the ion source. Mt is a lens magnification of the ion optical system, and Mt is equal to the product of magnifications M1 and M2 respectively of the condenser lens and the objective lens. Cc and Cs are chromatic and spherical aberration coefficients (defined on the ion source side), respectively. On the other hand, the beam current Ip is represented by the following equation using an angular current (or emission) density dI/dΩ (≈20 μA/sr).
Ip=(dI/dΩ)(πα∘2) (7)
In a region where Ip is large and the beam diameter ds due to spherical aberration is dominant so that the relation d≈ds holds, the following approximation equation using a proportionality constant K can be obtained from the equations (4) and (7).
D=KIp3/2 (8)
A d-Ip curve obtained by making the lens condition and the optical distances such as distances between the condenser and objective lenses constant and making only the aperture diameter of the beam limiting aperture variable is plotted on a full logarithmic scale sheet as shown in
First, a beam that always makes the value of K equal to 0.5 or less on the large current side of Ip≧20 nA in the milling mode is supposed. This corresponds to making ds small in the large current region. Making ds small is equivalent to making Hs small, i.e., making Cs small together with Mt. Here, Cs can be obtained by the following integration conducted along the optical axis from the ion source emitter on the optical axis z to the specimen.
Cs=( 1/64)∫(V/Ve)1/2[4S′2+3S4−5S2S′−SS″]r4dz (9)
Here, S is equal to V′/V. V is a potential on the optical axis. V′ is a first derivative of V with respect to z (=dV/dz), and r is a separation distance of the beam trajectory from the optical axis. The integrand includes a term of the fourth power of r. The present inventors have found that passing a beam having small r through lenses is important in making Cs small.
The condenser lens and the objective lens contribute to the value of Cs. Contribution of the condenser lens to making the current large is great. Therefore, the present inventors have found that it is necessary for the sake of making Cs small that the fourth power r4 of the separation distance r of the beam (where α∘=1 rad) from the optical axis is small in the z position of the outlet electrode 2_2 in the ion extraction & acceleration region (between the electrodes 2_1 and 2_2) of the condenser lens 2. By the way, the electrode 2_2 serves also as an inlet electrode of the unipotential region (between the electrodes 2_2 and 2_4) disposed thereafter. Specifically, as a result of various calculations and experiments, the present inventors have found that making especially Cs small among the subjects of making both Mt and Cs small is important in forming a beam having a beam diameter d of 0.5×(Ip/20 nA)3/2 [μm] or less. For that purpose, the present inventors have found that restriction of r4≦1600 mm4 (r≦6.3 mm) is necessary.
The condition in the observation mode, i.e., dmin≦6 nm and Ip≧0.001 nA will now be described. In the observation mode, the lens voltage V1 is the earth potential, and dg and dc mainly determine d in the equation (1). In particular, dg indicates a minimum limit of d, and the lens magnification Mt (=M1M2) becomes important. Therein, the lens magnification M1 of the condenser lens 2 is important. As a result of various calculations and experiments, the present inventors have found that the restriction M1≦8 is necessary. The reason is as follows. If an image point position of the condenser lens is disposed on the emitter side (i.e., in the case of a virtual image point), increasing M1 decreases M2. Since the increase rate of M1 excels, however, Mt becomes large. As a result, dg also becomes large.
As a result of various calculations and experiments, the present inventors have found that it is necessary in order to satisfy both r4≦1600 mm4 in the milling mode and M1≦8 in the observation mode that the distance Z2 from the ion emitter apex in the ion source 1 to the earth electrode 2_2 having the earth potential included in the condenser lens is 14 mm or less. If Z2 is too short, electric discharge occurs between the emitter apex and the electrode 2_1 or between the electrodes 2_1 and 2_2. Practically, at least 5 mm is necessary. As a result, the range of Z2 has become 5 to 14 mm.
As for the conventional apparatus described in U.S. Pat. No. 5,852,297, it is presented as the condition that the distance Z3 from the ion source 1 to the center of the condenser lens (the position of the third electrode 2_3 which is dominant in the lens action is used as the representative electrode) should be 45 mm or less. However, this condition alone is not sufficient. According to the present invention, the restriction of the range of 5 to 14 mm is necessary with respect to the distance Z2 from the ion source 1 to the earth electrode 2_2 in the condenser lens which is disposed nearest to the ion source.
Hereafter, this range restriction will be described with reference to calculated characteristic curves shown in
In
The d-Ip characteristic curves (Present (Mill.) and Present (Obs.)) shown in
According to the present embodiment, Ip corresponding to, for example, d≈1 μm in the milling mode can be increased to at least three times as compared with the conventional apparatus even under various optical conditions. In milling using the beam with d≈1 μm (with the milling precision of approximately 1 μm), the milling time can be shortened to ⅓ or less. The throughput in the structure analysis and failure analysis of the MEMS and high function fine devices can be improved to at least three times. On the other hand, in the SIM image observation in the observation mode, its highest image resolution is 6 nm or less. The beam characteristics in the observation mode are superior or equivalent to those in the conventional apparatus.
An embodiment in which the present invention apparatus is applied to a specimen separation method (hereafter referred to as micro-sampling method) described in Japanese Patent No. 2774884 will now be described. According to the micro-sampling method, it is possible, in the structure and failure analysis of high function micro-Si devices or the like, to mill a micro-sample for transmission electron microscope (TEM) with a FIB and take out it by using a micro-manipulator installed in the specimen chamber without dividing the Si device substrate.
The present embodiment is an example in which the present invention is applied to micro-sampling milling. Even if the present invention is applied to cross-sectioning conducted by combining ordinary box milling, an equivalent milling time shortening effect has been obtained. This milling time shortening is effective especially in the section structure analysis and failure analysis of the MEMS parts and semiconductor devices. The number of specimens that can be subject to section fabrication per unit time can be increased to twice to four times. As a result, the precision of estimation factors can be improved in the technique of estimating failure factors by conducting statistical processing on a large number of analyses.
It should be further understood by those skilled in the art that although the foregoing description has been made on embodiments of the invention, the invention is not limited thereto and various changes and modifications may be made without departing from the spirit of the invention and the scope of the appended claims.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2004-177261 | Jun 2004 | JP | national |
| Number | Name | Date | Kind |
|---|---|---|---|
| 5852297 | Ishitani et al. | Dec 1998 | A |
| 20060097197 | Sakaguchi | May 2006 | A1 |
| Number | Date | Country |
|---|---|---|
| 05-052721 | Mar 1993 | JP |
| 2774884 | Apr 1998 | JP |
| 2002-251976 | Sep 2002 | JP |
| 02005063865 | Mar 2005 | JP |
| Number | Date | Country | |
|---|---|---|---|
| 20050279952 A1 | Dec 2005 | US |
