JP3301080B2 - Electron beam melting method and apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to an oxide-based nonmetallic inclusion (hereinafter, also referred to as an inclusion) that is present in a sample by melting the sample with an electron beam. The present invention relates to an electron beam melting method and a device for floating on a surface of a dissolved sample and analyzing the surface.

[Conventional technology]

It has been known that oxide-based nonmetallic inclusions in steel samples deteriorate mechanical properties such as torsion value and fatigue of a product.

When a rolling member such as a bearing is repeatedly subjected to rolling contact stress, a microcrack is generated from an oxide-based nonmetallic inclusion as a starting point, and eventually reaches the flaking and the life is reached. Therefore, as proposed by the applicant,
Long-life bearing steel and rolling bearings exist by limiting the average particle size, the number of inclusions, and the proportion of inclusions within a certain range, and by keeping the oxygen concentration below a certain value (Japanese Patent Application Hei 1-264792).

Therefore, by observing and evaluating the size and number of oxide-based nonmetallic inclusions present in the obtained rolling member, the cleanliness of the material is assured, and the life evaluation of the rolling member is performed. The basis is done.

In order to guarantee such cleanliness, it is, of course, necessary to separate only the oxide-based nonmetallic inclusions in the sample from the base material and observe and evaluate the inclusions. Therefore, there has been a conventional method for observing and evaluating inclusions using an electron beam melting method (electron beam method) (iron and steel, No. 75 (1989) No. 10, p. 83-90).

In this conventional example, a small amount of a sample is melted in the form of a button with an electron beam, the inclusions are collected by floating on the upper surface of the metal, and after cooling and solidification, the inclusions are observed with a scanning electron microscope (SEM). The content is to be evaluated.

[Problems to be solved by the invention]

In the conventional example of the electron beam melting method, the input power at the time of electron beam generation is excessive because it is premised that the entire amount of the sample is melted, and when a part of the sample is melted, it is difficult to control the melting amount. There was a problem that inclusions up to an arbitrary depth could not be selectively dissolved and floated.

Such a problem is particularly remarkable in the case of steel for rolling members such as bearings. That is, the flaking phenomenon that causes the rolling life of the bearing is one of the main causes
It is known that the maximum shear stress position is within a few millimeters from the surface due to large inclusions existing at the maximum shear stress position according to the elasticity theory. Therefore, when guaranteeing the cleanliness of the material to guarantee the life, it is necessary to selectively separate and observe and evaluate only the oxide-based nonmetallic inclusions within a few millimeters from the surface. However, in the conventional example of the electron beam melting method, since the sample cannot be partially melted, there is a problem that the inclusion at such a maximum shear stress position cannot be selectively separated and observed.

Further, the dispersion state of the floating inclusions (rafts) after solidification of the sample melted in a button shape by the electron beam is affected by the surface structure at the time of solidification. That is, when a dendrite-like structure is generated on the surface, the floating inclusions are trapped between the dendrite arms, and the dispersion state becomes non-uniform, and the SEM
There is a problem that obstruction of observation / evaluation of inclusions due to the above-mentioned problems.

SUMMARY OF THE INVENTION Accordingly, it is a first object of the present invention to provide an electron beam melting method and an apparatus capable of melting from a surface of a sample to a desired partial depth, and to oxidize the sample by a dendrite-like structure. It is a second object of the present invention to provide an electron beam melting method in which observation and evaluation of physical nonmetallic inclusions are not likely to be hindered.

[Means for solving the problem]

The invention according to claim (1), which achieves the first object of the electron beam melting method, dissolves a sample by an electron beam and floats oxide-based nonmetallic inclusions on the surface of the melted sample. In the above method, the sample is a sample having a surface corresponding to a rolling surface of a rolling member, and electron beam irradiation is performed under the following conditions to obtain a depth from the surface of the sample.
Dissolves within 2mm.

350V + 100 ≦ J ≦ 700V + 200, where J: input power (W), V; sample volume (cm ³ ). The invention according to claim (2), which achieves the second object, dissolves the sample by an electron beam. In an electron beam melting method in which an oxide-based nonmetallic inclusion is levitated on the surface of a dissolved sample, the sample is dissolved after adjusting the carbon concentration of the sample to 0.06 to 0.6% by weight.

The invention according to claim (3), which achieves the first and second objects of the electron beam melting method, dissolves a sample by an electron beam and interposes an oxide-based nonmetallic material on the surface of the melted sample. The electron beam melting method for floating an object is characterized in that electron beam irradiation is performed under the following conditions and a sample having a carbon concentration of 0.06 to 0.6% by weight is melted.

350V + 100 ≦ J ≦ 700V + 200, where J: input power (W), V: sample volume (cm ³ ) Further, the invention described in claim (4), which achieves the first object of the electron beam melting apparatus, is in a vacuum vessel. In an electron beam melting apparatus comprising: an electron beam generating device that generates an electron beam at; and an irradiation device that irradiates the sample in the vacuum container with the generated electron beam, the sample is placed on a rolling surface of a rolling member. A sample having a corresponding surface, comprising a control device for controlling the power supplied by the electron beam generator at a depth of less than 2 mm from the surface of the sample and within the range of the following equation: It is characterized by.

350 V + 100 ≦ J ≦ 700 V + 200 where J: input power by electron beam (W), V: sample volume (cm ³ ) [Operation] The inventors of the present invention have conducted intensive studies and found that the input power by electron beam (here, input power and Means the irradiation energy per unit time of the electron beam) within a certain range determined by the volume of the sample, and by controlling the power supply time, the penetration depth of the sample is controlled almost in proportion to the injection time. It became clear that we could do that.

FIG. 1 shows the effects of the input power (W) per unit time (h) and the sample volume (Vcm ³ ) on the melting state of the sample by the electron beam melting method.
0V + 100 ≦ J ≦ 700V + 200) is a range in which the dissolution depth can be controlled in proportion to the injection time. Fig. 2 shows the relationship between the power injection time (irradiation time) and the penetration depth for electron beam generation with the electron beam acceleration voltage and beam current set to certain values so that the input power is within this range. was gotten. In FIG. 2, it can be seen that the penetration depth is proportional to the power-on time after the start of sample dissolution within the range of 2 mm. Therefore, it becomes possible to control the penetration depth of the sample by controlling the charging time. The melting of the sample is partial, which is advantageous because only a small amount of energy is required for melting the sample, and the control of the electron beam energy is simple.

When steel for rolling members is applied to the electron beam melting method according to the present invention as a sample, the penetration depth can be freely controlled in proportion to the injection time up to a range of 2 mm by power supply time management. Therefore, only the oxide-based nonmetallic inclusions at the maximum shear stress position can be observed and evaluated separately from the oxide-based nonmetallic inclusions at other positions. Evaluation accuracy can be improved.

If the input power is larger than the upper limit in FIG. 1, the input power becomes excessive, and it becomes difficult to control the penetration depth by controlling the power input time. If the input power is smaller than the lower limit in FIG. 1, the energy density is insufficient, so that planar melting non-uniformity occurs and it becomes difficult to specify the melting portion, and a clear correlation between the power input time and the penetration depth. It is difficult to find out. In any case, the charging time and the penetration depth are not proportional.

By the way, to analyze the starting point of rolling fatigue, conventionally, the main method was to observe the inside of the flaking with SEM, etc., but in many cases, the starting point is specified by the wear of the fractured surface during flaking etc. I couldn't. To determine the starting point, it is necessary to stop the bearing at the stage of cracking before flaking and observe a fresh fracture surface that has not yet been subjected to friction. Therefore, there is a proposal proposed by the applicant of the present application in which a crack generated in a thrust life test piece is detected by an AE (acoustic emission) method, the test machine is immediately stopped, and the starting point is analyzed (Japanese Patent Application No. 1-318011). No., Proceedings of the 33rd Spring Meeting of the Lubrication Society of Japan (1989), pp. 213-216).

Therefore, by combining the electron beam melting method according to the present invention with a nondestructive inspection technique for predicting various types of damage of a bearing at a micro crack stage (for example, a rolling fatigue starting point analysis using the AE method), rolling fatigue is improved. It is possible to provide a new evaluation technique for the analysis of the starting point of exfoliation caused by nonmetallic inclusions in the service life. That is, it is possible to specify a large inclusion that has become a starting point of the microcrack, selectively separate and float the same, and observe and evaluate the inclusion.

Next, an experiment was conducted on the relationship between the solidified structure after electron beam melting and the chemical composition. The occurrence of the dendrite-like structure and the carbon concentration in the sample were closely related to each other. It was found that there was a high carbon concentration.

When the dendrite-like structure is generated, the smoothness of the observation surface is impaired. Therefore, the smooth surface ratio (the projection area of the smooth surface where no dendrite-like structure is generated / the total projection area of the observation surface)
When the relationship between carbon dioxide and carbon concentration was examined, characteristics as shown in FIG. 3 were obtained.

As is apparent from FIG. 3, the ratio of the smooth surface starts to decrease from the vicinity where the carbon concentration increases and exceeds 0.5% by weight, and almost the whole is covered with the dendrite-like structure around 1% by weight. Further, it was recognized that the smooth surface ratio slightly decreased in the region of 0.1% by weight or less.

In order to improve the dispersion state of the floating inclusions extracted by electron beam melting and to facilitate observation and evaluation, it is necessary to set the smooth surface ratio to 90% or more.
When the smooth surface ratio is 90% or more, the dendrite-like structure is reduced, the number of floating inclusions trapped by the structure is reduced, and the dispersion of the inclusions becomes uniform, so that the inclusions can be easily observed and evaluated. In order to make the smooth surface ratio 90% or more, the carbon concentration is set in the range of 0.06 to 0.6% by weight. In particular, the carbon concentration is 0.1 to 0.5% by weight when the smooth surface ratio is 100%.
Is desirable.

The usefulness by avoiding the generation of such dendritic structures is SCr420H, SCM420H, SNCM220H, SNCM420.
This is particularly noticeable in carburizing bearing steels such as H, SNCM 815 (JIS G4052, JIS G4103).

2. The invention according to claim 1, wherein the power supply is performed under constant conditions when generating the electron beam.
By performing electron beam melting on a sample of 06-0.6% by weight, observation and evaluation of inclusions after melting from the sample surface to a desired partial depth can be performed well without dendrite-like structure. it can.

By feeding back the evaluation results for inclusions to the manufacturing department of bearing steel and bearings using the electron beam melting method according to the present invention, as proposed by the applicant in Japanese Patent Application No. 1-264792, for example, The average particle size of oxide inclusions is 3 to 30 μm or less per unit volume (160 mm ² ),
In addition, the composition ratio of the oxide-based inclusions having an average particle diameter of 10 μm or more is less than 2%, or the number of oxide-based inclusions having an average particle diameter of 15 to 30 μm is 10 or less per unit volume (100 mm ³ ). Or an oxide-based inclusion having an average particle diameter of 10 to 30 μm is set to 100 or less per unit volume (100 mm ³ ), or a combination of two or more thereof. By doing so, it is possible to prevent the occurrence of cracks due to oxide-based nonmetallic inclusions and obtain a long-life bearing.

In the case of a rolling bearing, long life can be attained by forming at least one of the bearing ring and the rolling element from a material in which the number and the particle size of the oxidizing inclusions are regulated.

〔Example〕

 Next, examples of the present invention will be described.

FIG. 4 is a structural diagram showing an example of an electron beam melting apparatus according to the present invention.

This electron beam melting apparatus has a vacuum chamber 41 capable of forming a high vacuum, a filament 42 built therein, and an electron gun 44 for forming an electron beam 43 by accelerating and firing electrons generated from the filament, and about a central axis. A water-cooled copper crucible 46 that can rotate and set a plurality of samples 45; a deflection coil 47 that deflects an electron beam 43 emitted from an electron gun toward the sample side; an observation window 48 that can observe the inside of the vacuum vessel 1; A voltage / current controller 49 for controlling the beam acceleration voltage and beam current of the electron gun 44 so that the angle falls within the hatched area in FIG. 1 and a water-cooled crucible drive motor M at a predetermined time interval for each sample. And a motor controller 50 that rotates at a pitch.

In this device, the sample is dissolved by irradiating the sample with an electron beam, and oxide-based nonmetallic inclusions (raft 5
1) Can surface.

Example 1 Ten samples of 0.41 cm ³ in volume cut out from SCr420H (same base material) as a carburized bearing steel were set on a water-cooled copper crucible of the electron beam melting apparatus shown in FIG. The power supply time was changed as shown in Table 1 to dissolve each sample, and the penetration depth at that time was measured.

Here, the input power was adjusted by fixing the beam acceleration voltage to 10 KV and changing the beam current. Vacuum chamber
The pressure in 41 was adjusted to 10 ^-7 to 10 ^-6 Torr. The power supplied to each sample was set in the voltage / current controller 49 in advance.

In the motor controller 50, the power supply time is set in advance for each sample as shown in Table 1, and after the supply time elapses for one sample, the irradiation of the electron beam is interrupted to rotate the one pitch water-cooled copper crucible. The electron beam melting is restarted according to the input power and input time preset for the next sample.

The penetration depth was determined from the remaining length after melting of a vertical through-hole (φ0.5) previously processed from the sample surface to the inner surface. The dissolution depth is obtained by subtracting the length after dissolution from the initial length of the through hole.

Since the sample volume is 0.41 cm ³ , the input power range (J) in which the sample penetration depth can be controlled by controlling the power input time is 243.5 W ≦ J ≦ 487 W from the relationship shown in FIG.
It is.

Table 1 shows the relationship between the input power (W), the power input time (S), and the penetration depth (mm) for each sample.

In Table 1, in Sample Nos. 4 to 7, since the applied power is within the range shown in FIG. 1, a substantially proportional relationship is recognized between the applied time and the penetration depth. Therefore, it is shown that the penetration depth can be controlled by the power supply time. On the other hand, in the case of sample numbers 1 to 3, since the input power is lower than the above-mentioned condition,
The sample undergoes heterogeneous dissolution, and no clear correlation is observed between the injection time and the penetration depth. On the other hand, from sample number 8
In Fig. 10, since the input power is excessive, the penetration depth per unit time is large, which indicates that it is difficult to control the penetration depth during the entire melting.

[Example 2] Two types of disc-shaped test pieces of bearing steel were prepared, and edited by "Electric Steel Works, Special Steel Handbook (First Edition), Rigaku Kogyo, published on May 25, 1965, pp. 19-21. The microcrack occurrence part was specified using a thrust life tester described in the above section and an AE device connected to the thrust life tester described in the 33rd Spring Meeting of the Lubrication Society of Japan (1989). The identification of the micro crack is performed as follows.

When a stress is repeatedly applied to a test piece by a thrust life tester, microcracks occur starting from oxide-based nonmetallic inclusions, and AE occurs at this time. Therefore, only AEs caused by micro cracks are
Then, the microcrack occurrence site is specified, and the thrust life tester is immediately stopped.

Next, the thrust life test piece was 10 mm square centered on the part where the microcrack occurred on the thrust life test piece.
The sample was cut into a rectangular parallelepiped having a thickness of 6 mm to obtain a sample used for the electron beam melting method.

Electron beam melting was performed using the apparatus shown in FIG. 4 at an input power of 400 W and an input time of 14 sec. The input power was within the hatched portion in FIG. 1, and by this input time, it was possible to selectively dissolve up to 0.2 mm from the surface of the dissolved sample.

FIG. 5 (1) is an optical microscope image (× 400) of the texture state of the race surface before the end of the AE sample and before dissolution of the sample. According to FIG. 5 (1), a microcrack which can be visually recognized as a black line exists near the center of the figure.

FIG. 5 (2) is a photograph showing an SEM image of the inclusion floating on the surface by the electron beam melting method. According to FIG. 5 (2), it is confirmed that the inclusion visible in white is floating on the surface of the solidified structure of the sample.

FIG. 5 (3) is an SEM image of the inclusions that were accidentally detected in the process of investigating the non-metallic inclusions in the metal surface area, and large inclusions can be confirmed in the center of the figure.

Fig. 5 (4) shows an SEM of the large inclusion shown in Fig. 5 (3), which was selectively levitated by electron beam melting.
It is an image. It can be seen that the inclusions confirmed in FIG. 5 (3) float stably while maintaining the size and shape. Therefore, inclusions that must be the starting point of microcracking can be selectively separated (even without performing a life test) to observe and evaluate them. Can be improved.

[Example 3] Six types of bearing steel samples (sample numbers 11 to 16 described below) and a second type of bearing steel having a high carbon concentration (SUJ-2, sample number 17 described below) were prepared using six types of bearing steel samples having different carbon concentrations from the SCr420H. The sample was set in the electron beam melting apparatus shown in FIG. 4 in the same manner as in Example 1 above, and electron beam melting was performed under the following conditions: input power J: J> 700 V + 200 (W), V; sample volume (cm ³ ). Was completely dissolved. At this time, the sample weight was 2 gr, and the sample volume was 0.26 (cm ³ ). The input power for generating the electron beam was 700 W satisfying the above conditions, and the power input time was 5 sec.

The entire sample was melted with an electron beam, and the oxide-based nonmetallic inclusions were floated.
The number of inclusions of μm or more and the projected area ratio of the smooth surface to the observation surface were determined. Table 2 shows the above results.

In Table 2, the sample Nos. 12 to 15 had no dendrite-like structure and had a smooth surface ratio of 100%, and the inclusions were uniformly dispersed, so that the number of inclusions was easily measured. I was able to measure. Therefore, the number of inclusions in Sample Nos. 12 to 15 can be measured without being affected by the dendrite-like structure, and the measurement result has sufficient reliability.

On the other hand, in sample No. 11, the carbon concentration was low, that is, when the carbon concentration was too low, it was difficult to obtain a high-purity steel, and in sample No. 16, the carbon concentration was high, and a dendritic structure was generated. The measurement of the number of inclusions was incomplete because the surface ratio was extremely low and some of the inclusions spread over the non-smooth surface. Furthermore, in sample No. 17, there was almost no smooth surface, and a dendrite-like structure was generated almost over the front surface, so that the inclusions were so strongly agglomerated that the number of inclusions could not be measured.

Example 4 SCr420H of Sample 12 and SUJ-2 of Sample 17 in Table 2 were formed into a rectangular parallelepiped of 8 × 8 × 5 (mm ³ ), and the input power by an electron beam was 700 W and the input time was 6 seconds. The sample was completely dissolved.

FIGS. 6 (1) and (2) are photographs of SEM images of the solidified tissue surface obtained by dissolving the entire amount of SCr420H and coagulating, showing the dispersed state of floating inclusions (rafts) which can be visually recognized in white in the photographs. Things. FIG. 6 (2) is FIG. 6 (1)
It is an enlarged image of.

As can be seen from FIGS. 6 (1) and (2), it can be confirmed that the sample surface after solidification has no dendrite-like structure and is smooth, and the inclusions are uniformly dispersed.

FIGS. 6 (3) and (4) show SEM of the solidified structure obtained by dissolving the entire amount of SUJ-2 with a high carbon concentration and solidifying it.
It is a photograph of an image, and shows a dispersed state of floating inclusions (rafts) that can be visually recognized in the photograph. FIG. 6 (4) is an enlarged image of FIG. 6 (3).

6 (3), FIG. 6 (1) and FIG.
What can be confirmed from (4) is that a dendrite-like structure develops on the surface of the sample after solidification, the surface is lost in smoothness, and the sample is in a structure state with severe irregularities. Then, it is confirmed that the dispersion state of the inclusions is affected by the unevenness and is non-uniform.

[Example 5] Two types of bearing steel having a high carbon concentration (SUJ-2, C; containing 0.95 to 1.10% by weight) were heated in the air at 900 ° C for 20 hours, decarburized, and the surface of the melting portion was heated. Average carbon concentration at lower 1mm is 0.06
It was adjusted to be about 0.6% by weight. This test piece is cut into a rectangular parallelepiped having a sample volume of 0.24 cm ³ to prepare a sample used for the electron beam melting method. For comparison, SUJ-2 which is not subjected to decarburization treatment is also cut into a similar rectangular parallelepiped to prepare a comparative sample used for the electron beam melting method.

Electron beam melting is performed by using the apparatus shown in FIG. 4 at an input power of 350 W and an input time of 17 sec. To melt the surface of the sample and then cool it.

FIG. 7 (1) is a photograph showing an SEM image of the surface after the decarburized sample has been melted by the electron beam melting method, and the dispersed state of the floating inclusions (rafts) which can be visually recognized in white in the photograph. It shows. FIG. 7 (2) is an enlarged image of FIG. 7 (1).

As can be seen from FIGS. 7 (1) and 7 (2), it can be confirmed that no dendrite-like structure is generated on the surface of the solidified structure, and a wide smooth surface is generated. Then, it can be confirmed that the inclusions are substantially uniformly dispersed.

FIG. 7 (3) is a photograph showing an SEM image of the surface of a sample that has not been decarburized after being melted by an electron beam melting method. This shows the dispersion state of Fig. 7 (4)
Is an enlarged image of FIG. 7 (3).

As can be seen from FIG. 7 (3) and FIG. 7 (4), it was confirmed that a dendrite-like structure was widely generated on the surface of the solidified structure, the surface was lost in smoothness, and the structure was very rough. it can. Then, it can be confirmed that a part of the inclusion is trapped between the dendrite arms.

As described above, even in a sample having a high carbon concentration (carbon concentration is not 0.06 to 0.6% by weight) such as SUJ-2 or the like, the decarbonation treatment is performed in advance to adjust the average carbon concentration to 0.06 to 0.6% by weight. It was confirmed that the generation of a dendrite-like structure can be prevented by performing such an electron beam melting method. As a result, even for a sample having a high carbon concentration, the decarburization treatment is performed in advance, the average carbon concentration is adjusted to 0.06 to 0.6% by weight, and then the electron beam melting method is performed. The operation can be performed satisfactorily without the possibility of receiving an obstacle.

〔The invention's effect〕

As described above, according to the invention described in claim (1), the penetration depth of the sample can be controlled by controlling the power supply time for generating the electron beam. It is possible to provide an electron beam melting method capable of melting the particles.

According to the invention described in claim (2), since the generation of a dendrite-like structure is extremely small, it is easy to observe and evaluate oxide-based nonmetallic inclusions. It is possible to provide an electron beam melting method in which observation and evaluation of inclusions is not likely to be hindered.

Further, according to the invention described in claim (3), after dissolving the sample from the sample surface to a desired partial depth, observation and evaluation of inclusions can be satisfactorily performed in a state where generation of a dendrite-like structure is avoided. .

Further, according to the invention described in claim (4), it is possible to provide an electron beam melting apparatus capable of melting from a surface of a sample to a desired partial depth only by controlling an injection time by an electron beam.

[Brief description of the drawings]

FIG. 1 is a characteristic diagram showing a relationship between a sample volume and input power for generating an electron beam, FIG. 2 is a characteristic diagram showing a relationship between an electron beam irradiation time and a penetration depth of a sample, and FIG. Characteristic diagram showing the relationship between the smooth surface ratio and the carbon concentration in the sample,
FIG. 4 is a structural view of one embodiment of the electron beam melting apparatus according to claim (4), and FIG. 5 is a photograph showing a metal structure of a sample in embodiment 2, and FIG. Optical microscope images, FIG. 3 (3) shows SEM images before dissolution, FIGS. 2 (4) show SEM images after dissolution, and FIGS. 6 (1) to 6 (4) obtained in Example 4. SEM showing the metal structure of the sample after dissolution
7 (1) to 7 (4) are SEM images showing the metal structure of the sample after dissolution obtained in Example 5. In the figure, 41 is a vacuum chamber, 42 is a filament, 43 is an electron beam, 44 is an electron gun, 45 is a sample, 46 is a water-cooled copper crucible, 47 is a deflection coil, 48 is an observation window, 49 is a voltage / current controller, 50 Is a motor controller, and 51 is a raft.

Continuation of the front page (56) References JP-A-64-70134 (JP, A) Edited by the Japan Institute of Metals / Iron and Steel Institute, “Iron and Steel Handbook”, 2nd edition, 2nd printing, Maruzen Co., Ltd., October 1976 20th, (58) Field surveyed (Int. Cl. ⁷ , DB name) C22B 1/00-61/00

Claims (4)

(57) [Claims] In an electron beam melting method in which a sample is melted by an electron beam and an oxide-based nonmetallic inclusion is levitated on the surface of the melted sample, the sample corresponds to a rolling surface of a rolling member. An electron beam melting method comprising: irradiating an electron beam under the following conditions to melt a sample within a depth of 2 mm from the surface of the sample. 350V + 100 ≦ J ≦ 700V + 200, where J: Power input by electron beam (W), V: Sample volume (cm ³ ) 2. In an electron beam melting method in which a sample is melted by an electron beam and an oxide-based nonmetallic inclusion is levitated on the surface of the melted sample, the carbon concentration of the sample is adjusted to 0.06 to 0.6% by weight. An electron beam melting method characterized by dissolving a sample. 3. In an electron beam melting method in which a sample is melted by an electron beam and an oxide-based nonmetallic inclusion is levitated on the surface of the melted sample, electron beam irradiation is performed under the following conditions and the carbon concentration is reduced. An electron beam melting method characterized by dissolving a sample of 0.06 to 0.6% by weight. 350V + 100 ≦ J ≦ 700V + 200, where J: Power input by electron beam (W), V: Sample volume (cm ³ ) 4. An electron beam melting apparatus comprising: an electron beam generator for generating an electron beam in a vacuum container; and an irradiation device for irradiating the sample in the vacuum container with the generated electron beam. A sample having a surface corresponding to the rolling surface of a rolling member, wherein the power supplied by the electron beam generator is controlled at a depth within 2 mm from the surface of the sample and within the range of the following equation. An electron beam melting device comprising a control device. 350V + 100 ≦ J ≦ 700V + 200, where J: Power input by electron beam (W), V: Sample volume (cm ³ )